This German translation of the original article in English was very kindly provided by Burkhard Hermes in October 2021 to facilitate access to the content for German-speaking readers and we are pleased to publish it here.

Diese deutsche Übersetzung des englischen Originalartikels wurde uns freundlicherweise von Burkhard Hermes im Oktober 2021 zur Verfügung gestellt. Sie soll den Zugang zum Inhalt für deutschsprachige Leser erleichtern, und wir freuen uns, sie hier veröffentlichen zu können.

Es hat viele Überarbeitungen des Moduls gegeben, und es wäre äußerst interessant zu sehen, welche Konstruktionsänderungen vorgenommen wurden.

Hintergrund

Volvo Penta begann um 2006 mit der Einführung der aktuellen D-Serie von Schiffsdieselmotoren. Während alle größeren Modelle Common-Rail-Motoren mit elektronischer Kraftstoffeinspritzung sind, arbeiten die kleineren Motoren der D1- und D2-Serie mit bis zu 4 Zylindern und 75 PS noch mit herkömmlichen, mechanischen Einspritzpumpen. Volvo hat sich dafür entschieden, diese Motoren über ein elektronisches Schnittstellenmodul an seine elektronischen Anzeigen und Steuerungen anzuschließen: die MDI (Mechanical Diesel Interface) Black Box. Zumindest bei einigen Besitzern erlangte die MDI-Box schnell Berühmtheit als das unzuverlässigste Teil eines ansonsten hervorragenden Motors.

Die abgeschraubte Volvo Penta MDI-Black-Box ohne Kabelstränge.

Die MDI-Box wurde im Laufe der Jahre erstaunlich oft überarbeitet, um Fehlermöglichkeiten und anhaltende Zuverlässigkeitsprobleme zu beheben, wobei einige Versionen hohe Ausfallraten aufwiesen. Die nachstehende Tabelle, die aus öffentlich zugänglichen Informationen zusammengestellt wurde, zeigt die aufeinanderfolgenden Modellnummern und das ungefähre Jahr der Veröffentlichung, wenn es ermittelt werden konnte.

2014 wurde die MDI-Black-Box um eine 15-A-Flachsicherung ergänzt, die durch eine abgedichtete Gummiabdeckung geschützt ist.

Funktionsweise

Die MDI-Box hat im Wesentlichen Überwachungsfunktion, d. h. sie liest alle Motorsensoren aus und gibt die Informationen über den CAN-Bus an das Volvo EVC-Anzeigesystem weiter. Sie hat jedoch auch eine kleine Steuerungsfunktion: Sie schaltet die Glühkerzen, den Magnetschalter des Anlassers und das Kraftstoffunterbrecherventil und schaltet die Lichtmaschine etwa eine Sekunde nach dem Anlassen des Motors ein. Bei einemAusfall der MDI-Box, kann der Motor nicht mehr über das EVC gestartet werden. Das kann jedoch mit einem einfachen Überbrückungskabel oder sogar einem Schraubenzieher (Vorsicht!) umgangen werden, um den Motor vorzuwärmen und zu starten. Das Hauptproblem eines Ausfalles des MDI ist die fehlende Überwachung von Kühlmitteltemperatur und Öldruck. Sollte die Sorge, den Motor nicht starten zu können groß genug sein, könnte man einen unabhängigen parallelen Motorstartstromkreis installieren, indem man zwei externe Relais hinzufügt, um die Klemmen “PREHEAT” und “START” auf Knopfdruck mit der Klemme “BATT” zu verbinden. In diesem Fall muss das Vorglührelais einen Strom von etwa 10A pro Glühkerze, also 40A bei einem 4-Zylinder-Motor verkraften können. Ein kurzes Anlegen von Strom an die Klemme “D” der Lichtmaschine nach dem Anlassen des Motors führt dazu, dass die Lichtmaschine normal zu laden beginnt, aber eine Erhöhung der Drehzahl könnte aufgrund des normalerweise im Rotor vorhandenen Restmagnetismus ausreichen, um das Gleiche zu erreichen.

Zuverlässigkeitsfaktoren: Hitze und Vibration

Die MDI-Box ist werkseitig an der Seite des wassergekühlten Auspuffkrümmers montiert. Wenn der Motor eine Zeit lang gelaufen ist, entspricht seine Temperatur in etwa der des Kühlmittels: zu heiß, um es anzufassen. Außerdem ist es fest mit dem Krümmergussteil verbunden und den Vibrationen des Motors voll ausgesetzt.

Ein Volvo Penta D2-40-Motor in seiner Werkskonfiguration mit der schwarzen MDI-Black-Box, die am wassergekühlten Auspuffkrümmer montiert ist. Bitte beachten sie, dass dieses Foto keine zusätzliche Länge des Kabelstranges zur MDI-Box zeigt.

Hohe Hitze und Vibrationen sind zwei bekannte Ursachen für einen vorzeitigen Ausfall der Elektronik. Glücklicherweise nutze ich den Motor nur wenig und lasse ihn nur sehr selten für längere Zeit laufen. Das mag der Grund sein, warum ich in 10 Jahren und 230 Motorstunden nie Probleme mit meiner MDI-Box hatte. Trotzdem hatte ich immer vor, die Box vom Motorblock zu entfernen, um einen Ausfall zu vermeiden.

Verlegung der MDI-Box aus dem Motorblock

Eine Voruntersuchung vor einiger Zeit hatte ergeben, dass der Kabelbaum, der die Sensoren um den Motor herum mit der MDI-Box verbindet, im Allgemeinen lang genug ist, um die Box an der Seitenwand des Motorraums zu montieren. Das geht dank einer gewissen Überlänge des Kabelbaums, die mit Plastikkabelbindern gebündelt ist. Diese zusätzliche Länge deutete fast darauf hin, dass die Verlegung des MDI-Kastens außerhalb des Motors zumindest bei den frühen Motoren konstruktiv ermöglicht und begünstigt worden war. Das kam nicht völlig überraschend, denn es ist nicht das erste Beispiel für unauffällige, nicht beworbene überlegene Technik an einem Volvo Penta-Motor. Ein weiteres Beispiel ist die Fernüberwachung der Spannung der Lichtmaschine. Ich hätte sonst die Verkabelung verlängert, indem ich sie abgeschnitten und gespleißt hätte.

Volvo Penta D2-40-Motor mit MDI-Black-Box, die zum Schutz vor Hitze und Vibrationen vom Motor abgesetzt wurde.

Ich habe eine Sperrholzunterlage mit zwei M6-Bolzen konstruiert und sie mit Epoxy an die Seitenwand des Motorraums geklebt, um die MDI-Box zu stützen. Ich musste dabei einige Einschränkungen berücksichtigen, nämlich die Länge des Multilink-Kabels, das die EVC-Messgeräte und das Kabel zum EVC-Bedienfeld versorgt, sowie das Vorhandensein einer Zugangsklappe direkt an der Seite des Motors. Ich habe die MDI-Box nach unten und hinten verlegt, in die Nähe des hinteren Motorträgers, und ziemlich tief. Eine niedrige Position kann in einem seltenen und unwahrscheinlichen Fall das Risiko erhöhen, dass das MDI naß wird, aber sie ist wahrscheinlich auch kühler, und ich beschloss, dass dies wenstlicher wäre. Ich musste nur das Kabel des Kühlmitteltemperatursensors verlängern und konnte den Kabelbaum ansonsten ohne Probleme neu verlegen.

Zuverlässigkeits-Faktoren: Elektrisch

Viele Ausfälle, insbesondere kurzfristige, haben elektrische Ursachen. Natürlich abgesehen von einigen seltenen Fällen, in denen Motoren mit eindeutig defekten MDI-Modulen ausgeliefert wurden. Einige frühe Versionen fielen durch die Einwirkung der Gegen-EMK (eklektromotorische Kraft) des Kraftstoffunterbrecherventils aus. Dies kann durch das Hinzufügen einer Diode an dem Gerät verhindert werden, um die negative Spannungsspitze kurzzuschließen. Dies ist eine elektrische Änderung, die ich irgendwann durchführen werde (oder ich werde die Verdrahtung zurückverfolgen und sie innerhalb des Moduls hinzufügen, wenn sie nicht bereits vorhanden ist).

Viele gemeldete oder tatsächliche Ausfälle des Moduls wurden in der Tat durch schlechte Verbindungen zur Batterie verursacht: ein hoher Widerstand führt dazu, dass die Spannung abfällt, wenn der Anlassermagnetschalter und der Motor unter Spannung stehen, was leicht dazu führen kann, dass die Elektronik zurückgesetzt wird und Startprobleme auftreten. Außerdem wird jedes Mal am Ende des Anlassens eine negative Spannungsspitze in das elektrische System des Motors induziert. Normalerweise hat sie nur geringe Auswirkungen, da die sehr niedrige Impedanz der Batterie sie absorbiert, aber wenn die Batterie schlecht angeschlossen ist, kann sie zu hohen Rückspannungen an der MDI-Box-Versorgung führen. Dieses Ereignis hat nachweislich viele Module getötet, oft mehrere hintereinander, wobei die Schuld meist auf die Blackbox und Volvo Penta geschoben wird, obwohl es in Wirklichkeit ein Installationsproblem ist. Verwenden Sie Batterien mit Gewindebolzen und hochwertigen Trennschaltern und stellen Sie sicher, dass alle Anschlüsse fest verschraubt sind.

Denken Sie nicht einmal daran, das Modul zu ersetzen, bevor der Weg zur Batterie nicht über jeden Verdacht erhaben ist.

Es lohnt sich auch, daran zu denken, dass die Lichtmaschine in der Werkskonfiguration an den positiven Hauptanschluss des Anlassermagnetschalters angeschlossen ist. Alles, was nicht zu einer soliden, ununterbrochenen Verbindung des Motors mit der Batterie führt, während die Lichtmaschine lädt, führt zu einem Spannungsanstieg im elektrischen System des Motors und birgt das Risiko, das MDI-Modul zu zerstören.

Einige Module gingen auch durch den Einbau von Masseschaltrelais in Leichtmetallbooten verloren, da die Gegen-EMK der Relaisspulen nicht abgeklemmt war.

Möglichkeiten nach einem Ausfall einer MDI-Box

Angesichts des schlechten Rufs der MDI-Blackbox von Volvo Penta habe ich mich immer gefragt, was ich im Falle eines solchen Ausfalls tun würde, da mir sowohl die Zuverlässigkeit als auch die Wartungsfreundlichkeit im Zusammenhang mit Hochseetörns in entlegenen Gebieten wichtig sind. Mir schwebten einige Möglichkeiten vor, die im Folgenden in absteigender Reihenfolge aufgelistet sind, je nachdem, was wünschenswert ist:

1. Reparieren des Moduls, wenn möglich. Da ich aus der Elektronikbranche komme, ist dies immer der erste Gedanke, der mir in den Sinn kommt, aber die Elektronik könnte in Harz eingegossen und völlig unzugänglich sein. Sobald die Ursache des Fehlers identifiziert ist, sollte sie jedoch behoben werden.
2. Das EVC-Instrumentensystem von Volvo Penta wird ein für alle Mal abgeschafft. Dies würde den Einbau von Standard-Automobilinstrumenten für Kühlmitteltemperatur, Öldruck und Motordrehzahl sowie einiger eigenständiger Schalter und Relais für Glühkerzen, Anlasser und elektrischen Stopp erfordern.
3. Entwicklung eines gleichwertigen Ersatzmoduls. Dies würde mehr Arbeit bedeuten, aber ein besseres Open-Source-Modul eines Drittanbieters hätte eindeutig einen Markt. Anstelle von Messgeräten am CANbus könnte ein einfaches LCD-Display alle Informationen anzeigen. Ein solches Modul könnte viel einfacher sein als die MDI-Box von Volvo, die eindeutig Technologie von den Steuergeräten der größeren Elektromotoren der Baureihe übernimmt.
4. Auswechseln des Moduls. Das ist einfach (und ziemlich kostspielig), aber ich habe Berichte von Leuten gehört, die mehr als ein Modul ausgetauscht haben, weil der Austausch allein die Ursache offensichtlich nicht behoben hat. Das wäre kaum zufriedenstellend.

Die zweite Option war in meinen Augen immer sehr wünschenswert, da sie ein proprietäres System durch ein standardisiertes und vollständig wartbares System ersetzen würde, das zu geringen Kosten mehr oder weniger überall auf der Welt eingesetzt werden kann. Der Bereich der Motorinstrumentierung scheint zwischen europäischen und amerikanischen Standards gespalten zu sein. In diesem Fall würde eine solche Lösung auf europäische Messgeräte angewiesen sein, und die größte Herausforderung könnte darin bestehen, die werkseitig eingebauten Sensoren zu identifizieren, um kompatible Messgeräte auszuwählen. Im schlimmsten Fall müssten einige der Sensoren ausgetauscht werden, um die Kompatibilität zu gewährleisten.

Kann die MDI-Box von Volvo Penta repariert werden?

Da ich die MDI-Box vom Motor abgenommen hatte und sie gut zugänglich war, konnte ich nicht widerstehen, zu untersuchen, inwieweit ein Ausfall für meine weitere Planung reparierbar wäre; dies hing davon ab, ob die Elektronik im Inneren gekapselt war oder nicht. Ich begann damit, alle Multicore-Kabel zu entfernen, um die Handhabung zu erleichtern, aber ich ließ die drei dicken, verschraubten Leitungen an ihrem Platz. Das Gehäuse besteht aus Aluminium, aber die Grundplatte, die die Anschlüsse aufnimmt, ist aus schwarzem Kunststoff geformt und wird von vier Torx-T-10-Schrauben gehalten. Zwischen den beiden Teilen befindet sich eine Kompressionsgummidichtung. Die Schrauben ließen sich problemlos entfernen, und die Aluminiumabdeckung ließ sich mühelos abnehmen.

Auf der Unterseite der MDI-Black-Box befinden sich drei Buchsen für abgedichtete Deutsch-Stecker sowie Klemmen für die Batterieversorgung,  die Leitungen zu den Glühkerzen und dem Anlasser-Magnetschalter. Diese Grundplatte wird von vier Torx-T-10-Schrauben gehalten.

Unter der Abdeckung entdeckte ich zu meiner großen Freude zwei gestapelte Leiterplatinen und keinerlei Vergussmasse. Die vollständige Abdichtung des Gehäuses bedeutet auch, dass die Leiterplatten nicht beschichtet sind und bei Bedarf leicht bearbeitet werden können. Die obere Platine enthält zwei 40-A-Relais zum Schalten des Vorheizkreises und der Anlasser-Magnetspule. Sie enthält auch die Stromversorgung für die Elektronik und zusätzliche Schaltkreise mit einem P-Kanal-MOSFET-Transistor IRF4905, der wahrscheinlich mit dem D+-Anschluss der Lichtmaschine zusammenhängt, aber ich habe ihn nicht vollständig untersucht. Die Versorgung der Logikschaltungen scheint von einem NCV4269 5-Volt-Linearregler zu stammen, der Spannungsspitzen von bis zu 60 V und Sperrspannungen bis zu -40 V verarbeiten kann. Von besonderem Interesse sind hier zwei Elektrolytkondensatoren mit 220μF / 63V in der Stromversorgung, denn diese Bauteile altern bekanntermaßen schneller und fallen bei Hitzeeinwirkung frühzeitig aus. Dies würde sie zu den Hauptverdächtigen im Falle eines Blackbox-Fehlers machen, da eine Verschlechterung dieser Kondensatoren zu einer schlechten Filterung des elektrischen Rauschens von der Lichtmaschine führen würde, was letztendlich die Spannungsregelung und den Betrieb der CPU beeinträchtigen könnte. In diesem Fall waren die Kondensatoren in gutem Zustand, ohne Anzeichen von Schwellungen oder Elektrolytaustritt. Was die Komponenten betrifft, so sollte sich alles andere im MDI-Modul im Allgemeinen als recht haltbar und widerstandsfähig erweisen. Der zweite Feind der Elektronik sind Vibrationen, und hier könnten einige der größeren Komponenten, da sie nicht gekapselt sind, im Laufe der Zeit anfällig für Risse in den Lötstellen sein.

Wenn man den Deckel der MDI-Black-Box abnimmt, kommen zwei übereinanderliegende Leiterplatten zum Vorschein. Die obere Platine enthält Relais und Transistoren für die Leistungsschaltung sowie den Spannungsregler mit zwei Elektrolytkondensatoren und einem 5-V-Regler.

Die untere Platine ist die Steuerplatine, die über eine 12-polige steckbare Stiftleiste mit der oberen Platine kommuniziert. Ich habe zwar nicht versucht, die MDI-Box komplett zu zerlegen, aber es scheint, dass das Trennen und Herausziehen der oberen Platine nach dem Abklemmen der drei Starkstromklemmen recht einfach sein sollte. So hätte man Zugang zu den Lötstellen, um die Kondensatoren oder sogar die Relais auszutauschen, falls dies jemals erforderlich sein sollte.

Die Starkstromklemmen auf der linken Seite sind direkt mit der Schaltplatine verbunden. Die Steuerplatine ist darunter verborgen, wobei die CPU und der Quarzoszillator am Rand gut sichtbar sind.

Die untere Platine trägt die CPU, den Quarz sowie weitere Schnittstellenkomponenten für die Sensorsignale sowie die CANbus-Schnittstelle zu den Messgeräten. Die Stifte, die in die versiegelten Deutsch-Steckverbinder der MDI-Box greifen, sind mit ihr verlötet, und es ist davon auszugehen, dass sie sich einfach durch die Kunststoff-Grundplatte ziehen lassen. Dasselbe gilt für die flachen Hilfsanschlüsse. Bei der Steuerplatine ist es weniger wahrscheinlich, dass sie repariert werden muss, aber unmöglich ist es natürlich nicht. Der Verzicht auf die Demontage des Moduls ermöglicht nur einen eingeschränkten Blick auf die Schaltung. Die CPU der Marke Philips / NXP ist ein LPC2119-Mikrocontroller mit 64kB Flash-Speicher, ein recht leistungsfähiger 32-Bit-Prozessor auf Basis eines ARM7-Kerns mit 2 CANbus-Schnittstellen und einem schnellen 10-Bit-Analog/Digital-Wandler. Die Quarzfrequenz war nicht ablesbar.

Die in der MDI-Black-Box verwendete CPU ist ein LPC2119 Mikrocontroller mit 64kB Flash-Speicher.

Fazit

Es hat zwar eine Weile gedauert, bis ich die MDI-Blackbox endlich von der Seite des Motors entfernt habe, aber es sollte eindeutig das erste sein, was beim Einbau dieser Motoren getan wird. Das Vorhandensein von Elektrolytkondensatoren im Modul scheint ein Risiko für Probleme zu sein, auch wenn solche Kondensatoren technisch für eine Lebensdauer von Tausenden von Stunden bei den betrachteten Temperaturen ausgelegt sind. In allen Fällen begünstigt die kombinierte Einwirkung von Vibrationen und thermischen Spannungen das Brechen von Lötstellen im Laufe der Zeit.

Glücklicherweise ermöglicht die Konstruktion der MDI-Box den Zugang zur Elektronik. Der Austausch von Bauteilen oder sogar das Auffrischen des Lötzinns auf den Platinen scheint problemlos möglich zu sein, so dass ein ausgefallenes Modul repariert werden könnte. Sollte dies nicht gelingen, dürfte ein Ersatzmodul eine lange Lebensdauer bieten, sofern es an einem geschützten Ort installiert wird. Die Tatsache, dass die Elektronik nicht gekapselt ist, macht sie sowohl reparierbar als auch anfälliger für Ausfälle aufgrund von Vibrationen.

Im Falle eines Ausfalls ist der Austausch des gesamten EVC-Überwachungssystems von Volvo Penta durch herkömmliche Messgeräte möglicherweise nicht teurer (in Bezug auf die Materialkosten) als der Austausch der MDI-Blackbox, und damit wären alle Zuverlässigkeitsprobleme ein für alle Mal beseitigt. Dies ist ein Weg, den ich ernsthaft in Erwägung ziehen würde, wenn ich nicht in der Lage wäre, ein ausgefallenes Modul zu reparieren.

This article is part of a series dealing with building best-in-class lithium battery systems from bare cells, primarily for marine use, but a lot of this material finds relevance for low-voltage off-grid systems as well.

Lithium iron phosphate (LiFePO₄) battery banks are quite different from lead-acid batteries and this is most apparent when it comes to charging them. Lithium battery banks charge much more easily and overcharge just as easily. They degrade gradually when kept full for extended periods and can develop memory issues when cycled inadequately.

On the other hand, lead-acid batteries resist charging, are tolerant to – and even require – a degree of overcharging and degrade rapidly when not fully charged regularly. This has given rise to a range of technology to meet these needs: it delivers aggressive charging, always errs on the side of overcharging and tries to keep batteries full. Trying to use this lead-acid charging technology to charge lithium cells certainly charges the battery, but it also damages it, so in other words it doesn’t and can’t actually work properly. No amount of searching for the Holy Grail of Settings can offset inadequate charging system design or the use of inadequate equipment.

Disclaimer

A good understanding of DC electrical systems is needed to build and commission a lithium battery installation. This article is aimed at guiding the process, but it is not a simple blind recipe for anyone to follow.

The information provided here is hopefully thorough and extensive. It reflects the knowledge I have accumulated building some of these systems. There is no guarantee that it will not change or grow over time. It is certainly not sufficient or intended to turn a novice into an electrical engineer either. You are welcome to use it to build a system, but at your own risk and responsibility.

Lithium Battery Charging

At a glance, a lithium battery charges just like as a lead-acid one: its voltage rises as it absorbs current until reaching a limit that must be respected. This means that it follows the well-known bulk and absorption pattern of lead-acid batteries. Because lithium batteries accept charge much more readily, they reach a higher state of charge before absorption begins, no forced charging, “boost” and other gimmicks are required or even desirable and the absorption phase is comparatively short. Its duration depends on the absorption voltage, the charge rate, cycling history as well as the age and condition of the cells. At low charge rates, the absorption time can amount to very little or nothing and when charging at high currents, such as when using powerful alternators or chargers, absorption is very significant. It also becomes more significant as the cells age and their internal resistance increases. The consequence of this is that it is impossible to pin down any duration as to how long the absorption stage should last. There is no correct setting.

A typical charging cycle for a lithium battery on a marine vessel can be summed up as follow: the battery is charged with whatever current can be produced with an upper limit on the battery voltage. The exact absorption voltage limit isn’t important, because of the inherent trade-off that exists between absorption time and absorption voltage. As previously illustrated, given sufficient time, any voltage from 3.40V/cell up will eventually fully charge and then overcharge a lithium iron phosphate battery. Lower voltages will fail to charge it extremely quickly and voltage cannot be used as a mean to control the outcome of charging as a result and absorption voltage alone has no bearing whatsoever on the final state of charge reached. Absorption voltages above about 3.55V/cell quickly exacerbate small differences in cell balance and become impractical to operate at. In my experience, 3.50V/cell has been a very good conservative charging voltage for LiFePO₄ battery cells; it is just high enough to perform some automatic cell balance corrections when needed and also sufficient to properly recharge cells that haven’t seen a full charge in a long time, or cells no longer in their prime. Charging must stop when the absorption current falls below the termination current threshold, because this means that the cells ability to absorb current has reduced down to the point where the cells must be considered to be full. Any further charging constitutes overcharging and leads to a point where the cell completely runs out of free lithium and no current can flow any more. Therefore the battery current is a critical piece of information and it must be known and used to control charging.

While lithium battery cells are capable of charging extremely fast and absorb current at rates of 1C and more during the bulk stage, this is not desirable and can progressively cause irreversible damage as developed here. Newer generations of LiFePO₄ cells are rated for regular charging at up to 0.5C and older ones usually 0.3C, so, in other words, even for modern cells a full charge can’t be achieved sustainably in less than about 2.5 hours when absorption is taken into account. This only becomes a concern when the capacity of chargers or alternators is very significant in relation with the size of the bank, but it also means that ridiculously large alternators are not actually as usable as some hope on small vessels. This can prompt using current-limiting chargers or regulators in some cases. Current acceptance capability also reduces with temperature, usually once outside the window of 5°C to 55°C (40°F to 130°F).

Charge Termination Condition

The charge termination parameters are nominated by the cell manufacturer as a pair of values for cell voltage and residual charging current: typically 3.65V and C/30 = 0.033C (C/20 = 0.05C for some of the newer generation cells). This means that if the charging voltage is limited at 3.65V, then charging must end when the charging current has reduced down to the specified level and the cell is then said to be “full”. However, we are not after the fastest possible charge rate, but rather long battery life, and we don’t need or want to charge to 3.65V/cell, so we need to interpolate down the termination condition for lower voltages. A fully charged LiFePO₄ cell at rest has an Open-Circuit Voltage (OCV) of about 3.37V and the termination current at that voltage would therefore be zero. Since the charging cell at a given SoC can essentially be seen as a voltage source in series with a resistive element, we can easily calculate the residual termination current for intermediate absorption voltages up to 3.65V and these are presented in the table below for both older and recent cells:

Example:

Charging an older generation cell up to 3.500V, we need to terminate the charge when the current is at most 0.015C. When there is no guarantee that the charge current will be at least as high as the termination current at the chosen absorption voltage, adaptive termination still works: if we were charging the same cell with solar power on a dull day and the current was only 0.012C, the above table shows that the termination condition should be deemed to be hit when the cell voltage reaches 3.475V.

Terminating the charge at a higher residual current simply equates to stopping the charge short of 100% SOC; this is commonly done as well as it is perceived to be easier on the cells, but the effective difference only amount to a few minutes of charging in most cases and a tiny fraction of capacity only. Lithium batteries just charge too easily.

The green shaded area depicts the charging envelope of a LiFePO4 cell. Below 2.000V, the cell is not rechargeable as its chemistry becomes damaged. Once the voltage rises and the current drops to the point where the termination limit is reached, the cell must be deemed fully charged and charging must stop. If the charging process is allowed to progress into the lower right corner, the cell is being overcharged. The upper and right edges correspond to increasingly aggressive charging regimes, forcing current and/or voltage.

Battery Cycling Management

Once a lithium battery has been charged, not only charging must stop, but it should be allowed to discharge meaningfully before being allowed to be recharged again. Charge controllers designed for the lead-acid chemistry implement algorithms that recharge periodically whether the battery needs it or not, because this strategy delivers far more benefits than drawbacks with cells that sulfate and deteriorate as soon as they are not kept full. Solar charge controller restart charging in bulk every morning and there is typically nothing the user can do about it. It is not a configurable setting. Alternator regulators will restart a new cycle every time they are turned on, even if the battery is full. Mains-powered chargers blindly initiate new charging cycles periodically to make sure that the battery stays full. Unfortunately, this kind of treatment is very detrimental to lithium cells, especially when associated with high-availability energy sources like solar and it becomes disastrous for unused installations with little or no load at all on the battery. A lithium battery that is not in use should be at a low State of Charge (SOC) and able to spend months or more without being charged at all.

Lastly, partial charging of a lithium battery should not be followed by a period of rest, especially if this is going to happen repeatedly at the same point on the charge curve, because this constitutes a memory writing cycle. Incomplete charging cycles are very common in marine house bank applications simply due to energy running out before the bank is full; these are of no concern because discharge normally occurs right away and charging ends randomly. On the other hand, systematic weak charging followed by a holding period, as it easily occurs when charging systems are misconfigured, gradually leads to near-complete loss of usable capacity from cumulative memory effect. Conversely, the battery must be charged properly from time to time to reset the state of the chemistry and erase any traces left by memory writing cycles.

House Banks and the Charge Termination Problem

Let’s consider a simple application like a battery-powered tool: its battery is either being charged, or the tool is being used and the battery is being discharged; the tool is not used while the battery is also connected to the charger. Similarly, we can’t plug an electric vehicle into the mains and drive it around town. This is not true in marine house bank applications: a lot of the time we simultaneously produce and consume energy and the battery acts as a buffer; it can either be charging or discharging at any time. Charging sources often supply current into loads instead of charging the battery. This is not a simple application and it creates a more complex operating context for the charging equipment.

In the case of a battery-powered tool or electric vehicle, the battery is either connected to one charger and not in use, or it is in use and not charging. In this case, the output current of the charger equals the battery current and a good charger can terminate the charge when the battery is full.

In the case of a power tool or EV, there is one charger and the total charging current (i.e. battery current) is equal to the charger output current:

I_Bat = I_Charger

The charger can regulate its output voltage and measure its output current and it has all the information it requires to terminate the charge when the battery is full.

While this diagram represents the typical installation including a lead-acid battery, multiple chargers and loads, it is totally incapable of charging the battery correctly. None of the chargers can determine when the battery is full and charging should stop because the battery current is an unknown quantity.

In the case of a marine house bank, there are (often) several chargers and loads all operating simultaneously; the battery current is the difference between the sum of all the charging currents and the sums of all the load currents:

I_Bat = ∑ I_Charger – ∑ I_Load

We can immediately see that measuring the output current of any charger (or even all the chargers) yields no usable information whatsoever for controlling charging, because loads can rob some of this current. Yet this is what regular lead-acid charge controllers do when they measure their own output current. Many, like alternator regulators, don’t even do that and blindly apply fixed or “cooked up” charge absorption times, because the algorithm is blind and has no idea of the battery current. This kind of gear is completely inadequate in itself for charging lithium batteries, regardless of what the manufacturer claims. There are no correct charge control settings for it. The only practical way of knowing whether the battery is charging or discharging, and whether the termination current threshold has been reached, is directly measuring the battery current I_Bat using a dedicated sensor at the battery itself. Because knowing the battery current is so essential for charging lithium cells without damaging them, there are only two valid system topologies to control charging correctly:

1. Each charge controller must have a dedicated input (like a shunt input) for sensing the battery current and terminate on charging based on a residual current condition; or
2. Each charge controller must be enslaved to a “master” that controls the charging process by measuring the battery voltage and current and tells the controllers when to stop based on a residual current condition. This master is typically the Battery Management System (BMS).

If all the battery chargers are capable of measuring the battery current and perform a correct charge termination on their own, then a capable distributed charging system can be built.

We will note here that a current measurement shunt can be shared with multiple measuring devices without issues because it is an extremely low-impedance voltage source.

Lithium Battery Charge Control

From the above, we can see that obtaining correct charge termination imposes very strong constraints on equipment selection: either the charge controller must be equipped with an external current sensing input and implement residual current termination, or it must be controllable externally by a “master” using at least some kind of “remote enable” signal, and such a master must exist in the system and it must have the required charge control capability.

A battery management system (BMS) measures both the battery voltage and battery current to determine the state of the battery. There are no “chargers” any more. The charging process is supervised by the BMS, which ensures that correct charge termination takes place. The BMS controls voltage regulators and those ensure that the battery voltage doesn’t exceed the required value.

In most situations, it is necessary to use a BMS that offers one or more charge control outputs simply because charging equipment using external current sensing and a correct charge termination algorithm is not available or cannot be sourced. Using a charge control output also has other benefits because the BMS can disable charging ahead of disconnecting the chargers from the battery in case of problem, an essential aspect discussed under the subject of electrical design for lithium battery systems. The ability of arbitrarily turning off charging sources also makes it possible to prevent charging when it is unnecessary or undesirable, like when the installation is not being used, and this prevents the battery from being abused by being held at a high state of charge indefinitely. It also allows the BMS to pause charging while re-balancing the cells if necessary: if cells happen to have drifted too far apart out of balance between two full charges, there comes a point where a balancing circuit can’t handle enough current to keep up when finally recharging. If pausing charging can’t be achieved, the system is left with no other option besides tripping on a cell high voltage condition.

In some situations, we don’t need or want to charge the battery, but we would like to take advantage of available free renewable energy, because we know we will be making use of the stored energy later. Powering loads without meaningfully charging or discharging the cells is achievable by lowering the “charging” voltage below 3.37V/cell for LiFePO₄ chemistry, which is the resting voltage of a fully charged cell. Because keeping cells full is detrimental and they should be allowed to discharge to some extent after a charging cycle, a reasonable practical voltage for supplying loads while preserving reserve capacity for short periods of time is 3.325V/cell, which equates to 13.3V and 26.6V for typical 12VDC and 24VDC systems respectively. This however requires an additional degree of charge control as we now either want to charge, hold, or let the battery discharge.

Whenever reserve capacity is not required, then the bank should be kept at a very low state of charge and loads can be powered by supplying current into the installation at a voltage corresponding to that low state of charge, like 3.2V/cell. This situation is encountered when shore power is available continuously and a mains “charger” is used. In this case, charging is the last thing we want and the equipment is best configured to operate as a constant voltage DC power supply. Some chargers can be persuaded to supply a desired constant output voltage by configuring identical absorption and “float” voltages and the better ones actually offer a constant voltage power supply mode. Others are simply intractable and must be thrown out. Another circumstance where capacity is not needed is when the vessel is laid up; a BMS can then maintain a very low state of charge in the bank through charge control.

This sums up the necessary and sometimes desirable ways of managing the charging process for a lithium house bank on board. The other aspect of battery management is deciding whether the battery should be charged or not, and/or how much; such strategies come with a view of preserving and extending battery life. Fully-engineered commercial solutions simply ignore it and charge the battery to full whenever possible. This maximises reserve capacity – which the end-user notices and appreciates, reduces battery life expectancy – which the end-user only discovers too late down the track, and eventually brings in more business, because it means re-purchasing the proprietary battery with integrated BMS. This is why the very high cost of these systems tends to translate more into superior performance while they last than actual value over their lifetime.

A wise system designer will ensure there are ways of keeping the cells at a low state of charge when capacity is not needed and charge the battery wisely. When energy availability is plentiful, there usually is no need to recharge to full, or the battery can be allowed to cycle much more deeply between full charges, which reduces the frequency of cycling and the time spend at a high state of charge. A battery that spends its life only as charged as it needs to be, rather than full, will last considerably longer.

Overcharging, Power Quality and Cell Destruction from Charging

The most common misconception about charging lithium batteries is believing that the State of Charge and by extension overcharging have anything to do with voltage. They don’t. A cell is being overcharged once the lithium ions are becoming depleted. The telltale that the cells are becoming full and charging must stop is reduced current acceptance, which is why using battery current for charge termination is mandatory. When a battery is being overcharged, its ability to absorb current trends towards zero and its apparent resistance to charging becomes increasingly high. The result is that the battery can no longer clamp the voltage down if it spikes. This has catastrophic consequences when the charger output is not well filtered because an overcharged battery becomes increasingly exposed to the peak ripple voltages. Eventually, the battery can no longer absorb any current at all and it is exposed to the full fluctuations in the supply voltage. If these exceed about 4.20V/cell for the LiFePO₄ chemistry, the electrolyte is broken down into gaseous products and pressure starts to build up into the cells.

We sometimes see claims that charging at 3.60 or 3.65V/cell ruined a bank and caused the cells to swell, but a smooth DC voltage at that level is insufficient to decompose the electrolyte. The problem comes from overcharging and poor power quality.

The worst ripple voltage is produced by solar PWM charge controllers, followed by old-style transformer/rectifier battery chargers, which should not be associated with lithium batteries. In the case of a solar PWM charge controller, the solar array is connected and disconnected from the battery at a fixed frequency. The open-circuit voltage of a solar array charging a battery in a 12VDC installation typically reaches up to about 22V (36-cell panel). Once the battery can no longer accept enough current to keep the voltage down, every time the controller sends a pulse to the battery, the cell voltages are gradually driven towards 22 / 4 = 5.5V. If the pulse voltage reaches 4.2V, there is sufficient energy for the electrolyte decomposition reaction to take place and the cells get rapidly destroyed, even if the average battery voltage as measured by a multimeter appears acceptable.

LiFePO4 cells destroyed by overcharging. The pressure in the cell casings was sufficient to push the cells apart.

Some of the power sources we use for charging batteries do not produce clean, filtered DC power, like alternators. As long as the peak ripple voltage can’t reach dangerous levels and correct charge termination always takes place, the situation is acceptable.

System Architecture and Topology

In a typical small-scale marine lead-acid battery system, multiple charge controllers follow independent algorithms and charge the battery in an approximate way while erring on the “safe” side for achieving battery life, which is overcharging it mildly. Overcharging, which causes gassing and recombination, stirs up the electrolyte and promotes voltage equalisation across the cells. This means a distributed charge control architecture which provides at best a roughly acceptable result. When it doesn’t, the battery gets damaged and the blame simply goes to whoever configured the myriad of “charge control settings” available.

In a lithium battery system, overcharging cannot be allowed to happen because it damages the cells and correct and accurate charge termination is needed. This requires an algorithm with a knowledge of both battery voltage and battery current. As common charge controllers are incapable of performing this function, they are incapable of performing charge control and this function must be implemented by the BMS. This leads to a different charge control architecture where the BMS is the only true charge controller and the slave devices only perform a voltage regulation (and sometimes power conversion) function: they really are just voltage regulators in this context because their only function is limiting the output voltage and wait for the BMS to signal the end of the charge.

As a result, the architecture and topology of lithium charging system is very different from what is typically found in lead-acid systems. However, if lead-acid batteries were being charged properly and with all due care, as they are in large stationary installations where the battery cells represent a very large investment that must last, the charge controller(s) would also feature an input for sensing the battery current, a lot of obscure programmable “charge control settings” would become pointless, and this difference would not exist. It is the result of the industry selling easy-to-install garbage equipment into a DIY marine/RV market and most of the difficulties with building good lithium battery solutions on board trace back to having to try and integrate garbage consumer-grade products. This situation has only been improving very, very slowly over many years and is still far from satisfactory.

The problems arise at the interface between the central BMS and the voltage regulators:

1. Ideally, the BMS should be able to transmit the desired voltage setpoint to all the regulators. A lack of standardisation, compatible communication interfaces and product capability still makes this largely impossible. Until recently, only single-brand, proprietary systems could achieve this. Victron Energy published some details about its communication protocols and interfaces, which is very commendable, but there is a lack of commonality across products and, while remote configuration would often be possible, the required information has not been released. Too many devices are not addressable and therefore can’t be networked properly. This can currently be alleviated by plugging them into a gateway device (Cerbo GX), which implements the open CAN SMA protocol from SMA Solar Technology AG to listen to messages from a growing number of BMS units, but in this case the resulting solution is limited to other Victron products and the question of the added drain on the battery from the Cerbo GX would need to be examined closely. The adoption of CAN SMA on solar regulators, alternator regulators, chargers etc seems to be the most promising pathway forward at this point in time (2022) and developments should be watched closely. This would lead to open networked charging systems with interoperability across brands.
2. The next option is accepting a lower degree of control and flexibility and configure the regulation voltages at regulator level; this we can generally do by using programmable devices, but it then leaves the matter of enabling and disabling the regulator. Some regulators feature a digital “enable” input: use it! When they don’t, problems grow because the “charger” must be either disconnected from the battery (when feasible without risking damage) or its power feed must be interrupted in order to disable it. Neither option is very attractive as both require interrupting a high power path using the like of solid-state relays, but sometimes it is possible and there is no other way.
3. Short of being able to transmit the voltage regulation setpoint to the regulators, we would still like to be able to control whether they should charge or hold, i.e. supply current to the loads without charging. As most programmable charge controllers have voltage setpoints for absorption and “float”, we could achieve this quite simply using the “float” voltage setting if there was a way to force them into “float”. Unfortunately, there generally isn’t. They go into float when they feel like it and the victim installer is left to play the Game of Settings to try and approximate a desired outcome without ever getting there reliably.

Besides being externally controllable by a BMS, the charging sources must also be able to cope with a battery disconnect event. A controllable charger will normally be disabled by the BMS ahead of such an event and this normally takes care of most issues. Some voltage regulators however resist just about all attempts at integration in lithium battery systems. Wind generators are notorious for this, as well as for sloppy, horrible voltage regulation and surging. Any disconnection under load usually destroys them and they must see a battery at all times in order to operate, which unfortunately defeats the strategies discussed here. They are among the worst charging devices to integrate safely and properly with lithium battery systems.

The Losing “Game of Settings”

Most of the DIY lithium installations fail to terminate charging correctly because their design and the hardware employed make them incapable of doing so. They try to hold together by relying on a precarious balance between “charging parameters” and energy consumption. A lot of the time, the battery gets overcharged to various degrees, sometimes every day. Any meaningful change in the operating conditions of the installation throws the balance out: in the absence of consumption, the battery gets slammed to 100% SOC; start the engine when the battery is already charged up and it gets abused by the alternator.

Solar charge controllers are notorious to overcharge lithium batteries. First, they initiate a new charging cycle every morning as the light comes up, so in the absence of sufficient overnight consumption, the bank cannot cycle properly. Configuring a “float” voltage low enough as discussed earlier does allow to create some kind of charge termination for sure… after a fixed absorption time, which is wrong most of the time. Many models won’t allow absorption times short enough to be even hopefully realistic, so they overcharge every time. Some units offer optional termination based on the residual (or “tail”) current as an improvement, but as they measure their own output current, they can’t tell whether the current is going to the battery or a load… This only works if consumption is zero, so let’s say that the average background consumption on board at anchor is 2A and charging should terminate when the battery charge current reduces down to 6A. We configure a tail current setting of 8A to try to account for this and terminate the charge and now it works… until the background load is suddenly higher, because the vessel is under way and a whole lot of instruments are on all the time, and from there on the cells get overcharged because the “controller” is fooled by the extra load.

The Game of Settings doesn’t work. These strategies are not solutions, only hopeful attempts at damage minimisation.

A compounding factor is the fact that people who want to install a lithium battery bank also want to reuse the gadgetry they already own. Reprogramming a typical lead-acid charging system differently, when this is possible at all, doesn’t actually lead to any solution, because it is conceptually inadequate for charging a lithium bank: it can’t possible operate the way it should.

Approximating the Solution

One of the reasons why this article has been a very long time in the making is because the only way to build an actually acceptable lithium charging system for a marine vessel was (and still is to a very large extent) by using custom-built electronics. In the past few years, this state of affairs has started to evolve very, very slowly, but it is far from being satisfactory. It leads people towards trying to approximate the solution and it nearly always falls short of the mark.

Considering that the biggest issue is the absence of correct charge termination, one approach can be giving up altogether on absorbing the battery when it is not possible to do it properly:

1. at least one charging source is capable of performing a full charge with correct termination based on residual battery current every time; and
2. this source (which can be the engine alternator for example) is used occasionally; and
3. all the other charging sources are programmable and can be configured to skip absorption, so they just switch to “float” immediately when the absorption voltage gets hit; and
4. there is nearly always a load on the installation, so the bank doesn’t get partly charged and then rested; and
5. the “float” voltage can be configured low enough to ensure the battery will nearly always discharge; and
6. these sources do not restart charging before the bank has been able to discharge meaningfully,

then all the “bad” chargers will perform partial charges only without leaving any memory in the cell because of the immediate subsequent discharge and the one “good” charger will reset the cell chemistry and erase any trace of memory if needed, as well as allow cell balancing to operate during absorption, whenever it is used. Such a system can hold together without tripping on cell high voltage and offer good battery life if its user has also ensured that the depth of cycling is sufficient by matching up charging capacity and battery size to the average consumption. This means coupling relatively small battery banks to good charging capacity. If the average consumption reduces too much, this fragile equilibrium will suffer and human supervision and intervention become essential at such times.

At this point, the hopeful system builder will discover that most charging sources cannot be prevented from restarting to charge whether the battery needs it or not and many cannot be controlled externally either, so in other words they simply constitute lithium battery overchargers. Most DIY systems in operation today use overchargers and the amount of damage they cause varies with the load patterns experienced by the system.

Many “charge controllers” sold with the word “lithium” in the accompanying pamphlet are simply unusable, some to the point of being purely destructive. Some of the most infamous examples are the “lithium” versions of the Genasun GV-5 / GV-10 solar MPPT controllers: these little marvels of engineering and efficiency respectively deliver up to 5A and 10A at a steady 14.2V output… forever! There is no way to adjust anything and no way to turn them off without seriously hacking the circuit boards. These are the best lithium plating controllers on the market. Genasun exited the lithium arena many years ago now, but they keep marketing some of the garbage technology that was destroying their batteries.

Charging lithium batteries requires precise control because no overcharging can be tolerated. If the topology of the charging system is incorrect, the system is not capable of charging a battery correctly and no amount of “programming” will ever change that.

We need good charging equipment and that is:

1. Charge controllers with a battery current measurement input and a residual current termination algorithm; or
2. Programmable voltage regulators interfaced to a BMS using an external enable input and (ideally) another input to switch them to a lower holding voltage when the charge is complete; or
3. Voltage regulators we can control with a BMS by sending a setpoint via a digital bus.

In all cases, we need regulators that don’t engage in rogue, arbitrary recharging for no reason, so we can allow battery banks to discharge and keep them at a low state of charge when this is desirable.

In this article, we have a look at the Volvo Penta MDI electronic black box while relocating it off the side of a Volvo Penta D2-40B engine in order to protect it from the heat and vibrations.

IF YOU OWN A FAILED MDI BOX, PLEASE CONSIDER SENDING US PHOTOS OF THE INSIDE AFTER LIFTING THE LID! There have been many revisions of the module and it would be extremely interesting to see what design changes were made.

Background

Volvo Penta began the release of the current D-Series marine diesel engines around 2006. While all the larger models are common-rail, fuel injected electronic engines, the smaller D1- and D2-series engines, up to 4 cylinders and 75HP, still operate with traditional mechanical injection pumps. Volvo elected to interface these engines to its electronic gauges and controls by using an electronic interface module: the MDI (Mechanical Diesel Interface) black box. For some owners at least, the MDI box rapidly gained fame as the least reliable part of an otherwise excellent engine.

The Volvo Penta MDI black box once separated from the engine and most of the wiring harness.

The MDI box has had a surprisingly long revision history over the years, aimed at addressing failure modes and on-going reliability issues, with some versions recording high failure rates. The table below, compiled from publicly available information, shows the consecutive model numbers and approximate year of release, when it could be determined.

It is interesting to point out that very few modules labelled 23231607 appear to have ever shipped. Volvo was shipping part number 23195776 in boxes marked 23231607 for some time. Some of these early 23195776 units were the object of a replacement campaign initiated in April 2018 due to erratic behaviour. A wider replacement campaign was initiated in August 2023 for modules with P/N 22458451-P, 22594274, 23195776 and 23231607, also due to erratic behaviour, and is open until the end of November 2030. As usual, only specific engines are affected and owners need to check with a service agent.

Until now (2025), the part number was followed by a manufacturing date code starting with “W” followed by 4 digits yyww, where yy = last two digits of the year of manufacture and ww = number of the week of manufacture. “W1128”, as shown in the photo below, indicates a module manufactured in the 28th week of 2011. This practice seems to have been discontinued on model 24743026. From model 24765169 at least, the firmware enforces limits on the activation time of the starter motor as well as the stop solenoid.

By 2011 already, a 15A blade fuse was added to the MDI box, protected by a sealed rubber cover.

Functionality

The role of the MDI box is largely a supervisory one in the sense that it reads all the engine sensors and outputs the information over CANbus for the Volvo EVC gauge/display system, but it does have a minor control function: it switches the engine glow plugs, the starter motor solenoid, the fuel stop solenoid and it energises the alternator about one second after the engine has started. A failure of the MDI box leaves the engine dead, but it can be bypassed very easily to preheat and start using a simple jumper cable or even a screwdriver, with some caution. The key issue is the lack of monitoring over coolant temperature and oil pressure afterwards. In fact, should the concern of being unable to start the engine be acute enough, one could opt to install an independent parallel engine start circuit by adding two external relays to connect the “PREHEAT” and “START” terminals to the “BATT” terminal at the push of a button. In this case, the preheating relay must be able to handle a current of about 10A per glow plug, so 40A on a 4-cylinder engine. Briefly applying power to the “D” terminal post of the alternator after the engine has started will cause it to begin charging normally, but increasing RPMs can be enough to achieve the same, due to the residual magnetism normally present in the rotor.

Reliability Factors: Heat and Vibration

The MDI box is factory-mounted to the side of the water-cooled exhaust manifold. When the engine has been running for some time, its temperature is approximately equal to that of the coolant: too hot to keep a hand on it. Furthermore, it is hard-mounted to the manifold casting and fully exposed to the vibrations of the engine.

Volvo Penta D2-40 engine in its factory configuration with the MDI black box attached to the water-cooled exhaust manifold. Note that this photo shows no extra length in the wiring harness to the MDI box.

High heat and vibrations are two well-known root causes of premature failure for electronics. Heat causes electrolytic capacitors to age and fail. Vibrations lead to solder joints cracking around the wires of trough-hole components and can also cause the quartz crystal of the oscillator to fail. I make little use of the engine and I very rarely run it for any amount of time; this may be why I never experienced any issues with my early model MDI box in 10 years and 230 engine hours. Nevertheless, I always had in mind to relocate it off the engine block to prevent a failure.

Relocating the MDI Box off the Engine Block

A preliminary investigation some time ago had shown that the wiring harness connecting the sensors around the engine to the MDI box was generally long enough to allow remote-mounting the box on the sidewall of the engine compartment, thanks to some extra length in the loom bundled with plastic cable ties. In fact, this extra length almost suggested that relocating the MDI box off the engine had been made possible and favoured by design, at least on early engines. This didn’t come as a complete surprise as it was not the first instance where I discovered understated, unadvertised superior engineering on a Volvo Penta engine: another one is remote voltage sensing for the alternator. I would have otherwise extended the cabling by cutting it and splicing it, or better by constructing an extension cable using male and female 8-pin Deutsch connectors.

Volvo Penta D2-40 engine with MDI black box relocated off the engine to protect it from the heat and vibrations.

I constructed a plywood pad fitted with two M6 studs and epoxy-glued it to the sidewall of the engine compartment to support the MDI box. I also had to accommodate a few other constraints, namely the lengths of the Multilink cable feeding the EVC gauges and the cable to the EVC control panel, as well as the presence of an access panel immediately to the side of the engine. I moved the MDI box down and back to a location close to the rear engine mount, quite low. A low location may be more prone to see water in a rare and improbable event, but it is generally also cooler too and I decided this would be adequate. I only had to lengthen the coolant temperature sensor wire and I was otherwise able to re-route the loom without issues.

Relocating the MDI module off the engine block can be a simple matter of rotating it and refastening it to the side of the engine bay, as on this D2-60 engine, with minimal disturbance to the cabling.

Reliability Factors: Electrical

Many failures, and especially short-term failures, instead have electrical root causes except for some rare instances where engines shipped with clearly defective MDI modules. Some early versions would have failed from exposure to the back EMF of the engine stop solenoid. This can be prevented by adding a diode over the device to short the negative spike out and this is an electrical alteration I will carry out at some point (or I will trace the wiring and add it within the module if not already there).

Many failures of the module, perceived or real, have in fact been caused by poor connections to the battery: a high resistance path causes the voltage to drop whenever the starter solenoid and motor are energised, which can easily cause the electronics to reset and lead to starting problems. A negative voltage spike is also induced into the engine electrical system each time at the end of cranking. It normally has very little effect because the very low impedance of the battery absorbs it, but if the battery is poorly connected, it can lead to high reverse voltages appearing at the MDI box supply. This has demonstrably killed many, many modules, sometimes several in succession, with the blame usually going towards the black box and Volvo Penta, when it is in fact an installation problem. Use batteries with threaded studs, quality battery disconnect switches and ensure that all the connections are clean, free of corrosion and tightly bolted.

Don’t even think about replacing a failed module until the path to the battery is above any suspicion, or you will likely lose the next one as well

It also pays to remember that, in the factory configuration, the alternator charges into the main positive supply terminal at the starter solenoid. Anything less than a solid, uninterrupted connection of the engine to the battery while the alternator is charging will cause the voltage of the engine electrical system to spike up with the risk of destroying the MDI module.

Some modules were also lost due to the addition of ground disconnect relays to the installation on alloy boats because of the unclamped back EMF of the relay coil. 

Pathways Following an MDI Box Failure

In the light of the poor reputation of the Volvo Penta MDI black box, I always wondered what I would actually do if I faced such a failure, remembering that both reliability and maintainability are important to me in the context of ocean cruising in remote places. I envisioned a few pathways, listed below in decreasing order of desirability:

1. Repairing the module, if possible. With an electronic engineering background, this is always the first consideration that comes to mind. The root cause of the failure, once identified, should be addressed however.
2. Doing away with the Volvo Penta EVC instruments system once and for good. This would require installing standard automotive gauges for coolant temperature, oil pressure and engine speed, as well as a few stand-alone switches and relays to deal with the glow plugs, starter and electric stop. An alarm circuit responding to low oil pressure or high coolant temperature would be essential too.
3. Developing an equivalent replacement module. This would represent more work, but a better third-party open-source module would clearly have a market. Rather than using networked gauges on CANbus, a simple LCD display could present all the information. Such a module could be much simpler than the Volvo MDI box, which is clearly inheriting technology from the ECUs of the larger electronic engines in the range.
4. Replacing the module, which is as easy as it is costly, but there are many reports of people having gone through more than one module, because replacement alone obviously failed to address the root cause. This would hardly be satisfactory. The failure rate of the modules has very significantly dropped since late 2019 however.

Option 2 has always been highly desirable in my eyes, because it would replace a proprietary system with one that is standard and fully maintainable at low cost, more or less anywhere in the world. The field of engine instrumentation seems to be split between European and American standards. Here, any such solution would rely on European gauges (such as VDO for example) and the only challenge would be identifying the factory-installed sensors to select compatible gauges.

Option 3 would be a lot more fun and cheaper than purchasing new gauges, as long as the time required for development is no object. This is often the case while cruising however.

Would the Volvo Penta MDI Box Be Repairable?

While I had the MDI box off the engine and well accessible, I couldn’t resist investigating to what extent a failure would be repairable for my own forward planning; this hinged upon whether the electronics inside were encapsulated or not. I started by unplugging all the multi-core cables to facilitate handling, but I left the three bolted heavy wires in place. The housing is made out of aluminium, but the base plate accepting the connectors is moulded black plastic, held in place by four Torx T-10 screws. A compression rubber seal is present between the two parts. Removing the screws presented no difficulty and then the aluminium cover separated effortlessly.

The underside of the Volvo Penta MDI black box reveals three sockets for Deutsch sealed connectors and terminal posts for the battery supply and cables to the glow plugs and the starter motor solenoid. This base plate is held in place by four Torx T-10 screws.

Underneath the cover, I was very pleased to discover two stacked circuit boards and no potting compound whatsoever. The fully sealed nature of the enclosure also means that the circuit boards are not coated and could easily be worked on if necessary (later versions of the module are now using coated PCBs). The top board contains two 40A-rated relays switching the preheat circuit and the starter solenoid (these relays have since been replaced with solid-state switching). It also includes the power supply for the electronics and additional circuitry with a IRF4905 P-channel MOSFET transistor likely related to the alternator D+ connection, but I didn’t formally trace this. The supply for the logic circuits appears to start from a NCV4269 5-volt linear regulator. The key point of interest here is two electrolytic capacitors rated 220μF / 63V in the power supply section, because these components are well-known to age faster and fail early when exposed to heat. This would make them prime suspects in case of black box failure, because a degradation of these capacitors would result in poor filtering of the electrical noise from the alternator and this could ultimately affect voltage regulation and the operation of the CPU. Here, the capacitors appeared in good condition, without signs of swelling or electrolyte leakage. As far as components go, everything else in the design of the MDI module should generally prove quite durable and resilient. The other enemy of electronics is vibrations and, here, in the absence of encapsulation, some of the larger components in particular could be prone to cracking of the solder joints over time. The large black power resistor mounted off the circuit board is a prime candidate for this; it has since been replaced by a series of surface-mounted resistors on newer modules.

Removing the lid of the Volvo Penta MDI black box reveals two stacked circuit boards. The top board contains relays and transistors for power switching, as well as the power supply section with two electrolytic capacitors and a 5V regulator.

The bottom circuit board is the control board, which communicates with the upper board through a 12-pin pluggable header arrangement. While I didn’t attempt a complete tear-down of the MDI box, it seems that separating and extracting the top board should be quite easy after disconnecting the three heavy-current terminal posts. This would give access to the solder points to replace the capacitors or even the relays if it ever proved necessary.

The heavy current terminal posts on the left connect directly to the switching board. The control board is hidden underneath it with the CPU and crystal oscillator well visible near its edges.

The bottom circuit board carries the CPU and crystal as well as more interfacing components to deal with the sensor signals and the CANbus interface to the gauges. The pins engaging into the sealed Deutsch connectors of the MDI box are soldered to it and they just pull through the plastic baseplate, same for the auxiliary flat blade terminals. The fact that both circuit boards are mounted back-to-back means that extracting the stack is enough to gain access to the components side of both boards without having to separate them. Not dismantling the module only affords a limited view of the logic board: the Philips/NXP-branded CPU is a LPC2119 microcontroller with 64kB of flash memory, quite a powerful 32-bit processor built on an ARM7 core with 2 CANbus interfaces and a fast 10-bit analog/digital converter. The crystal frequency is 10MHz.

The CPU used in the Volvo Penta MDI black box is a LPC2119 microcontroller with 64kB of flash memory.

Conclusion

While I took a while before finally relocating the MDI black box off the side of the engine, it should clearly be the first thing done when installing these engines. The presence of electrolytic capacitors in the module looks like a recipe for trouble, even though such capacitors can technically be rated for service lives in the thousands of hours at the temperatures considered. In all cases, the combined exposure to vibrations and thermal stresses promotes the breaking of solder joints over time.

Fortunately, the construction of the MDI box allows access to the electronics. Component replacement or even reflowing the solder over the boards appears perfectly achievable, so a failed module could be repaired. If this was not successful, a replacement module should arguably be able to offer a long service life, provided it is installed in a protected location. The fact that the electronics are not encapsulated makes them both repairable and more vulnerable to failure from exposure to vibrations.

In case of failure, replacing the whole Volvo Penta EVC monitoring system with conventional gauges may not be more costly (in terms of materials) than replacing the MDI black box and doing so would eliminate any reliability issues once and forever.

In July, I wrote about my experience with a homebrew twin-transducer ultrasonic antifouling system over 3 1/2 years on the aluminium sloop Nordkyn. It didn’t keep the hull clean once the antifouling reached past its prime, but the results were nevertheless very valuable as the system eradicated hard growth like barnacles on the hull, even in the absence of any antifouling. Not having to scrape off hard growth significantly extended the life of the antifouling and made hull cleaning much easier.

I since altered the firmware of the ultrasonic driver module to output a higher average power level and also reach higher frequencies in an attempt to disrupt algae growth, while I had virtually no antifouling paint left over large parts of the hull. Since the hull was already dirty, investigating the ability of the modified system to prevent attachment was not possible. The system was instead put to the tougher test of fighting existing algae growth.

An Approximate Baseline

I had dived on 7 April and peeled off a thick carpet of weed and sponge-like growth which had accumulated since the New Year in Oamaru Harbour in the South Island. The harbour basin is small and the water had reached 25°C at times during the summer. Following this, I returned to the Auckland region via Cook Strait and the West Coast of the North Island before rounding the top and coming back down on the East side, covering about 1100NM. I stopped along the way for about a week in Nelson and then again in Whangaroa Harbour. By 26 April, I was sailing off the Bay of Islands and the boat performance had noticeably reduced already. Following this, I covered the short remaining distance more leisurely and arrived near Auckland on 6 May. By early June, sailing was hardly a proposition due to the state of the hull and I was forced to dive again on 29 June.

In short, the table above shows that it took less than 3 weeks to experience a slow-down due to fouling in spite of covering close to 1000NM in this period. After 2 months, sailing had become problematic and I was forced to dive again less than 3 months after last cleaning.

Firmware Alterations

There is no published scientific research that I am aware of related to the effect of ultrasonic energy in marine antifouling applications. Some research was conducted on the effect of ultrasonic energy on microscopic algae suspended in water [1, 2 and 3] and it showed that:

• Exposure to ultrasonic energy is capable of causing cellular damage to algae;
• The effect appears related to cumulative power exposure: it takes longer to achieve the same results at lower power levels;
• Higher frequencies are more effective.

My system had so far operated like the Silicon Chip design which had inspired the project. The Silicon Chip firmware used frequencies between 19kHz and 41kHz and emitted tone bursts of 1000 cycles. 1000 cycles at 20kHz results in a tone duration of 50ms, which reduces down to 25ms at 40kHz. As output power increases with frequency, this strategy curtailed the amount of power delivered in the higher frequency bands. This hardly seems satisfactory in the light of the above. As a result, I rewrote the code to produce frequencies ranging between 19kHz and 65kHz this time and discovered additional resonance peaks above the nominal 40kHz frequency of the ultrasonic transducers. I also implemented a fixed tone duration scheme, so higher frequencies would not no longer be penalised. This clearly increased the average output power of the system and I had to improve the power management features so the consumption of the system could adapt to the energy available.

The Experiment

The hull badly needed to be repainted, but I wanted to take advantage of these unfavourable circumstances to try and improve performance with regard to algae growth. I was annoyed by the rate at which soft growth covered the hull again after cleaning and I wanted to attempt some alterations to the firmware of the ultrasonic system. By June, I had the new ultrasonic driver in operation, but it was still functionally the same as the prototype unit I had operated since 2014. I dived and cleaned the hull on 29 June and the experiment started about two weeks later, after I commissioned the new firmware on 12 July. As a result, it had to contend with about two weeks of “baseline” fouling. By then, the rudder was covered with sufficient light weed to mask its colour and long strands had attached to the trailing edge.

Results

In the space of a few days only, the blue of the old antifouling seemed to become increasingly apparent over the rudder blade, which was promising, because the rudder is separate from the hull where the transducers are installed and the vibration must travel though the rudder bushes. The rudder gradually shed most of its algae growth over the areas which had antifouling left. The upper part, near the waterline, had no paint left and remained dark. The long strands of weed gradually vanished.

The last paint job was nearing 2 years old. The keel and rudder were in better shape as, three-and-a-half years earlier, a friend had brush-painted them a couple more times while I was rolling antifouling over the hull. Some of this thicker coating still remained and any amount of antifouling, even that old, is still preferable to none.

In September, I started making arrangements to haul out. These fell into place very suddenly and I departed for Tauranga, 120NM away, on the 14th. I had been moored for 6 weeks and intended to dive and clean before leaving, but a good 15-knot following wind sprung up and I couldn’t resist trying to use it. While I had thick long weed around the waterline, it didn’t appear to extend underneath the hull. The rudder looked relatively clean, I surmised that the keel would be the same and so I decided to try sailing and stop in a bay along the way to remediate if necessary. Not only this might save me diving, but it would also allow taking photos of the dirty underwater hull once out.

Tacking out of the anchorage under main alone was somewhat tedious, but I then reached the wind and picked up speed immediately to reach 5-6 knots to my surprise. I sailed 50NM on that day, and 70NM the following one in much lighter conditions with an asymmetric spinnaker. I even exceeded 8 knots at some point. The boat was clearly slow by a good 2-3 knots, but it would have been near-unmovable before.

Coincidentally, I ended up hauling out after the exact same 83-day interval that had separated the previous two dive-and-clean jobs. The 29th of June being very close to mid-winter, the same conditions of light and water temperature prevailed both times. Yet, 77 days after diving and cleaning, this time I had still been able to sail quite reasonably. As the growth appeared to have receded following the alterations made to the ultrasonic driver, I am inclined to think that it was being controlled to an approximately constant level by the system in the prevailing conditions of water temperature and sunlight.

Haul-Out

Needless to say, I was extremely curious to discover what the underside of the hull looked like and I had a camera ready when the boat was lifted out. As anticipated, the long thin stringy weed at the waterline didn’t extend further underneath the hull. While the bottom was mostly filthy, the growth was surprisingly thin and short, which explains why I was able to sail. When I had last dived and cleaned after the same period of time, I had peeled off a layer of underwater carpet containing sponge-like growth. There were none this time.

The difficulty in protecting the waterline from slime and algae is a known limitation of ultrasonic antifouling systems. It may be that the ultrasonic energy cannot be transmitted effectively into the algae close to the free-surface. It is interesting to note that barnacles still can’t develop in that area however.

The bottom is only covered by thin short weed interspersed with isolated leaves and tufts. The waterline is showing long stringy algae all around. No hard growth and no sponge-like growth are present.

The keel and rudder had benefitted from a thicker antifouling coating in the past and fared better than the underside of the hull for this reason. The keel bulb in particular had been heavily coated three-and-a-half years earlier. This light-blue paint is an aluminium-compatible ablative antifouling. This suggests that coating the hull more heavily should allow skipping a haul-out by periodically diving and cleaning underwater. The combined feedback I recently received both from shipchandlers and boat owners converged to suggest that this product is also one of the poorest aluminium-compatible formulations on the market at the moment and, this time, I changed to another product.

The keel bulb was quite heavily antifouled three-and-a-half years earlier and some of this old paint is still present. Note the absence of any shells and the difference with the leading edge of the keel and some areas of the foil, which are down to the dark blue 12-year old base coat keyed in the high-build epoxy paint.

The rudder also showed significant residual antifouling covered by sparse weed. Once out of the water, the weed clings to the surface and the blade appeared more fouled than I was expecting, but the growth is thin and sparse. A better paint or a light wipe of the surface in the water would easily improve on this result. Of interest is the fact that there is no slime at all: the surface either has weed attached, or is clean. If slime is, as often claimed, a necessary precursor for heavier fouling to take hold, then starting with a clean hull could have made a considerable difference, but this remains to be verified.

The rudder looks much worse out of the water than it did while still immersed. Yet the weed is staying very thin and sparse, when it had fully covered the blade only 19 days after underwater cleaning with the original ultrasonic firmware. The light blue paint is three-and-a-half year old antifouling. The dark blue paint is 12-year old antifouling keyed in high-build epoxy at construction. The white areas near the trailing edge are bare high-build epoxy paint. Note the complete absence of slime.

Waterblasting was easier and quicker than ever, because the algae growth was so thin this time. As usual, it produced a smooth and clean surface that could be painted without any further preparation. This has been the case since the ultrasonic system was installed.

Waterblasting the hull yields a clean surface that can be painted immediately as there is no hard growth at all. The dark blue paint is a semi-hard aluminium-compatible antifouling that was keyed into the high-build epoxy coating during the construction of the hull 12 years earlier.

Discussion

The experiment didn’t start with a clean hull and, as a result, the modified ultrasonic system wasn’t able to oppose fouling from its inception. The antifouling paint was also past the end of its useful life and large areas had none left. In this regard, the system was put to the test in challenging conditions. On the other hand, the aluminium hull material is probably optimal for such an experiment and the test was conducted in winter when the water is colder and sunlight hours limited. It is very possible that heavier fouling would have resulted in summer.

The alterations to the firmware of the ultrasonic driver produced very impressive results in terms of reducing the thickness of the algae growth. Considerably less algae variety was also present underneath the hull this time. It is interesting to note that, in spite of the presence of the invasive undaria and fanworm algae species in northern New Zealand waters, I never found any on my hull, even when other vessels moored nearby were being severely affected. If these happened to be sensitive to ultrasonic energy, it would be a valuable piece of knowledge. In recent years, the “pest algae scare” has been exploited by lobbyists in the local marine industry to have frequent, indiscriminate and costly haul-outs mandated by councils for boat owners in some areas like Northland and the Bay of Plenty.

The code changes to the ultrasonic driver had three effects:

1. Increasing the maximum frequency used by the system from 41kHz to 65kHz;
2. Increasing the amount of energy allocated to the higher frequencies;
3. Increasing the overall average output power.

In order to differentiate the effect of frequency from the increase in the average delivered power, the test would have had to be staged and a lot more time would have been required. Unfortunately, this didn’t sit very well with my desire to haul out and eliminate the need to dive and clean repeatedly due to the lack of antifouling paint left before the summer. The research mentioned earlier also indicated that the best results should be sought by increasing both the frequency and the energy exposure.

The performance of a properly built and installed ultrasonic system should be a function of:

1. Its peak power output, as peak power determines the “reach” of the system over the hull surface, as well as the maximum amount of damage it can potentially inflict to marine growth on the hull surface;
2. The average power delivered, as it seems that damage from low-power ultrasonic energy results from cumulative exposure over time;
3. The frequencies used, possibly because higher frequencies tend to be intrinsically more energetic, but it may also be that shorter wavelengths are able to excite damaging resonance effects in algal cells.

The table below summarises the consumption of the system with the original firmware as well as the new one, in its two modes of operation. In this test, the new firmware only ever operated in its low-power mode, due to the limited solar energy available at the time. In summer, it will switch to the higher output at least for a part of most days.

The firmware changes nearly doubled the average output power of the system, and its consumption. The figures were measured at the battery voltage available at the time, around 13.3V, except for the high power readings which were taken at 13.75V. It is interesting to observe that the figures obtained with the new firmware now significantly exceed the output of many high-priced commercial offerings. If we add to this that many of these systems are installed on hulls that are unlikely to transmit ultrasonic energy as well as an aluminium one and some of the transducers are poorly constructed and installed, it may just be enough to explain why so many ultrasonic systems have been found to be largely ineffective by boat owners. For the sake of guidance, the figures reported above relate to a total underwater area just exceeding 37m².

Conclusions

The original ultrasonic system allowed me to reach a record two years between haul-outs this time and this limit was largely the result of the hull running out of paint. This time, I applied 50% more of what should hopefully also prove to be a better antifouling paint, with ample drying time between coats as allowing the solvents to fully evaporate is a key critical factor in achieving longevity. I am hoping to reach three years before having to haul out again, provided the new coating doesn’t ablate away too quickly, and four years would represent an exceptional result.

While still providing a degree of foul-release, the silicone coating on the propeller was in rough shape after two years. This can’t be renewed easily without hauling out properly due to the long drying time. The propeller shaft anode could be replaced underwater or by drying up in a suitable location.

The alterations made to the ultrasonic driver firmware appear to have resulted in significant performance improvements with regard to algae growth. Operating the system starting from a newly painted hull should provide a more meaningful answer as to its long-term ability to keep a hull clean, if this is at all possible. Further increasing the output power of the system as well as the operating frequencies would be possible with alterations to the hardware and it would likely further increase effectiveness; however, the need for such developments should first be demonstrated and the power consumption would also quickly become prohibitive for most vessels without access to shore power.

Hard growth was readily eliminated by operation in the lower frequencies at extremely modest power levels, as an average of 0.16W/m² of wetted surface was already enough to be fully effective. This begs the question of knowing whether keeping these lower frequencies is necessary at all. If operation at 40kHz and above disrupted the development of both hard and soft growth, then a reduction in power consumption could conceivably be achieved by optimising the frequency spectrum to only allocate energy where it is also most effective. While I have sketched schematics for a new and different ultrasonic driver, it may now be some time before I investigate the matter further, at least using my own yacht.

References

[1] “Effect of ultrasonic frequency and power on the disruption of algal cells”, K. Yamamoto, P. M. King, X. Wub, T. J. Mason and E. M. Joyce, (2015), Ultrasonics Sonochemistry 24, 165–171
[2] “Effect of ultrasonic frequency and power on algae suspensions”, E. M. Joyce , X. Wu and T. J. Mason (2010), Journal of Environmental Science and Health, Part A, 45:7, 863-866
[3] “Ultrasonic Irradiation for Blue-Green Algae Bloom Control”, T. J. Lee , K. Nakano and M. Matsumara (2001), Environmental Technology, 22:4, 383-390

Starting Point

I had hauled out the sloop Nordkyn for its less-than-annual bottom paint job. It was November, that year had been long, and so was getting the thin stringy weed clinging to the hull. In this condition, the boat is unusable. I was determined to make the effort last and so I generously sprayed the new antifouling in two thick coats. A tinge of irritation showed when, three weeks later, I noticed light green slime already forming on the rudder blade, and six weeks later this had turned into thick heavy slime over the entire hull – in the cold waters of Southern New Zealand. Nothing that the odd dash at 18 knots didn’t remediate when I sailed off in January, but this of course only postponed the disaster. By March, in warmer waters, I was diving underneath the hull and scraping off barnacles in the hundreds.

At this point, a short discussion was up with the paint company, and I politely suggested they had sold me paint from a faulty batch. “Certainly not”, responded their sales manager, an individual who has earned himself here a well-deserved reputation for always blaming the customer, “you simply didn’t apply enough product for it to be effective”. Not enough paint to last three weeks, as if more of a useless product was going to deliver an enhanced result. In fact, I had applied 50% more paint than the previous year. So I objected, only to be told this time that the water was “special” this year, “not our fault”, but primarily around my boat, because others seemed to cope with the special water quite normally. “Aluminium-compatible antifoulings are naturally less effective, there is nothing wrong with it”. As we were about to debate the concept of fitness for purpose in a venue I was going to kindly organise, they shipped me a very reasonable amount of replacement product. Fine.

Still, it took a while to get there, the new paint didn’t apply itself onto the bottom and I had the pleasure of hauling out again. During these lengthy proceedings, the idea of installing an ultrasonic antifouling system for preventing organisms from attaching too readily found a new incentive. These systems are priced like the rest of the marine electronic gadgetry and there seemed to be no clear verdict out about their effectiveness. They always seemed most effective from the manufacturer’s point of view basically. Aluminium hulls however appeared to show consistently positive results. In these circumstances, I looked at what I had to lose in relation with the cost of painting a 43-footer: certainly not much if I built my own system. The project had therefore literally “started in anger”.

Ultrasonic Antifouling: Principle and Limitations

The theory behind ultrasonic antifouling systems is that the vibration induced in the hull plating, and maybe the surrounding water, disrupts cellular growth in organisms trying to attach and develop against the hull. It creates an adverse environment. What these systems don’t do is produce cavitation and a “layer of micro-bubbles” against the hull as some inept claims suggest. That would require the kind of power found in an ultrasonic cleaning tank, 0.5-0.6W/cm², or a mere 180kW for a vessel like mine, and it would probably strip the antifouling off in a matter of minutes! Interesting scientific research supporting the use of ultrasonic energy to keep marine hulls clean was published in 2016 [1].

Ultrasonic transducers before encapsulation. Note the two ceramic discs and the compression bolt.

The vibration is induced into the hull using a piezoelectric transducer, at the heart of which are ceramic elements sandwiched between electrodes. When a voltage is applied to the electrodes, an electric field forms through the ceramic, which either expands or contracts depending on the polarity. An ultrasonic transducer attached to a hull produces a longitudinal wave on the axis of the device, which passes through the thickness of the hull and dissipates into the water, as well as a guided shear wave that travels radially away from the transducer. This shear wave is the one that can propagate throughout the hull and installing the transducers on a section of unsupported plating maximises it.

Propagation of longitudinal and shear ultrasonic waves in a material (Source: Olympus NDT, “Ultrasonic Transducer Technical Notes”).

The material and construction of the hull has a significant impact on the ability of the shear wave to travel away from its source, which is enough to explain the variability experienced in the results. Aluminium, being light and extremely rigid, is notoriously good at transmitting sound; steel is excellent too. In these materials, the propagation velocity of the shear wave exceeds 3000m/s [2]. Solid fibreglass normally gives good results too. Timber, on the other hand, absorbs vibrations. Cored hulls are usually deemed unsuitable, unless maybe the core material is removed to install the transducer against the outer skin, but I have no experience with them.

Ultrasonic transducers are designed for a specific resonant frequency and they don’t operate well at all frequencies, but they can resonate to some extent at frequencies other than their design frequency [3]. Research suggests that organisms also react differently with regard to frequency and the standard approach in this application appears to be sweeping over a relatively broad spectrum of frequencies to maximise the chances of success, both in terms of transducer performance and impact on growth.

Development

A design for a single-transducer ultrasonic antifouling driver, engineered by Leo Simpson and John Clarke, had been published in the September 2010 issue of Silicon Chip Magazine, #264. Shortly after, it was commercialised in the form of a kit by Jaycar in Australia and New Zealand at an attractive price when compared to ready-made commercial systems. I had heard some good first-hand reports about it. At 13 metres, Nordkyn would require two transducers however and, rather than installing two kits, I decided to build my own modified version of the Silicon Chip design that would be able to drive two transducers instead of one. It appeared easy enough and much more interesting.

Driver Board

The piezoelectric transducers are capacitive loads driven at voltages of a few hundred volts through high-frequency step-up power transformers in a very conventional arrangement [4]. The transformer winding, the cable and the capacitive transducer form a resonant circuit and driving two transducers required a whole new second power stage. Some of the components used in 2010 were no longer in production, I wished to improve some aspects and, altogether, this led to building similar, but different electronics. Still, I expected little trouble as the starting point was a working design, but the circuit handles quite high peak power levels and this assumption didn’t prove entirely correct. When the time came, I decided to test a single channel first: it is not every day that one can freely opt out of 50% of a potential disaster. The circuit failed promptly and the fuse protested with an orange flash, but this was nothing a few alterations couldn’t overcome.

The prototype of the dual-channel ultrasonic antifouling driver in its final state, after a few initial alterations and then three-and-a-half years of continuous service.

It has now operated continuously for over 3 ½ years at the time of writing (mid-2018) without any intervention, so I have decided that it must be robust and reliable.

Transducers

One aspect I was very unhappy with when I reviewed the Silicon Chip article was the presence of soft, rubbery potting compound between the ultrasonic transducer and the hull. Of course, some of the vibration will propagate through, but there is nothing like direct contact between hard incompressible materials to transmit sound. A large amount of efficiency would almost certainly be lost this way. The transducers used in this type of application are primarily manufactured for ultrasonic cleaning machines; in this kind of equipment, they are epoxy-bonded to the outside of the tank wall and literally vibrate with the tank. Gluing the transducers to the hull is perfectly possible and would even be desirable. At the time, I elected to install them in a less permanent way, as the whole venture was very much an experiment.

Some original Jaycar systems showed much improved performance when the potting compound was scraped off the face of the transducer and replaced with a piece of fiberglass board epoxy-glued into place.

As the voltage at the transducer terminals can exceed 800V at times, electrical insulation is essential for safety. On a boat, the transducers are also by definition in the bilge and they need to be completely sealed from moisture. I used synthetic plumbing fittings and polyurethane resin to encapsulate the transducers, similarly to what the article described, but I first bonded a piece of hard fibreglass/epoxy board to the face of the transducer to insulate it and made sure it would come into direct contact with the hull.

Two ultrasonic cleaning transducers encapsulated in plastic plumbing fittings and polyurethane resin. Note the hard fiberglass face on the bottom of the unit on the left. It protrudes slightly from the base of the housing and sits flat against the hull.

I located the transducers by first splitting the underwater hull surface into two equal halves and then determining the approximate geometric centre of each area. This gave roughly 15m² of hull plating per transducer with the keel and rudder in addition to that. I made sure the transducers were installed in the middle of a hull panel, as far away as possible from the frames and other stiffeners. This prompted mounting each transducer off-centre: the front one is forward of the mast and to starboard and the aft one is behind the engine and to port.

A threaded flange was epoxy-bonded to the hull plating away from the stiffeners and the transducers is then screwed tightly into place.

Did It Work?

Yes. I built the prototype presented above in the second half of 2014 and commissioned it in early 2015. In New Zealand waters, we get little creatures we refer to as snapping shrimps because of the sharp crackling noises they make as they crawl under the hull. While they initially dislike the taste of fresh antifouling and always wash off with boat speed, it doesn’t take them long to get comfortable once the boat stops and the music starts. When I powered up the ultrasonic system, their noise started fading away and, after a while, the hull was much quieter, other than for the faint clicking sounds produced by the transducers. It suggested that the vibration was indeed propagating throughout the underwater hull and this appeared positive.

Hard Growth

The system has now operated continuously for 3½ years and it essentially eliminated all of the hard growth on the hull: no more barnacles and coral-like formations. Barnacles migrate through the water as tiny larvae, attach to the hull and then start growing a shell. When the water is choppy, I can sometimes see some larvae beneath the waterline; a few days later they are gone. They simply can’t live and develop against the hull plating any more. The only place where I sometimes find a few grown-up barnacles is at the very aft tip of the keel bulb; there appears to be a dead spot there. The effectiveness on the rudder is slightly less, obviously because of the bushes. In particular, the very leading edge seems more vulnerable.

Weed Growth

When it comes to algae fouling, there doesn’t appear to be any silver bullet there: the antifouling still has a role to play. Once it has completely worn away, the hull can get colonised by algae and even sponge-like growth, but this is always easy to peel off. I was never able to distinguish any difference in fouling nearby or away from the transducers; the effect over my hull appears uniform. This was not the case on a 66′ aluminium fishing vessel fitted with two Jaycar kits. In this case, the owner told me that growth was visibly reduced over a large radius around each transducer and their effectiveness then faded away with distance. He said his boat needed four transducers. Because the effect on my hull appears uniform, it is more problematic to evaluate performance with regard to algae growth. In the first 9 months or so after painting, the antifouling normally keeps the hull free of weed anyway. However, if I dive and clean the hull once the antifouling is essentially gone, slime won’t form again for at least a few weeks, but only over areas that were 100% clean. Even when neglected, my hull has never become as fouled as I had seen it before, so the system may somewhat hinder algae growth, but this assessment is somewhat subjective. If left long enough without any care, the bottom does eventually end up filthy and this has been the case with all ultrasonic antifouling “solutions” I have directly heard of.

Ultrasonic Antifouling and Antifouling Paint

All up, the system eradicated the barnacle problem and it is immensely valuable to me for this reason. In the absence of hard growth, the surface is easy to clean, stays smooth and it can be recoated with minimal effort each time. It extends the intervals between haul-outs for me provided I dive and scrape off the soft growth from time to time once the antifouling is a year old or so. I experimented with applying significantly more paint to the keel and rudder (it is a moderately ablative formulation) and I must say that it has allowed cleaning underwater for much longer without running out of paint. This old and thicker antifouling is not as effective as a new coat, but these surfaces still perform noticeably better than the areas of the hull left with nothing. Before I installed the ultrasonic system, barnacles would attach to the paint as soon as it lost some of its effectiveness and then the surface couldn’t be cleaned without effectively removing most of the paint in the process and my conclusion was that applying a thick coating primarily benefitted the paint company. This is no longer true. I would undoubtedly get better results again if I could use a cuprous oxide-based antifouling, which is both stronger and longer lasting than aluminium compatible products.

Risks or Harmful Effects

The awareness about the toxicity of antifouling paints for the marine ecosystems has been increasing over the years and – maybe for this reason – people often ask whether such a system is actually environmentally friendly, and safe for other marine life, for the hull or even people living on board!

Risks to Aquatic Life

Dolphins, which use a sonar-like system for echo-location, must almost certainly be able to hear the transducers at least at times, as they operate in the 20-40kHz spectrum, starting just above our range of perception. They still come, play, swim underneath the hull and stay with the boat for long periods when I am sailing, so they don’t appear to mind at all. My view is that the power level involved is too small and the power density in the water too low to be an issue. The peak power may reach 100W at times, distributed over an area of about 37m² on my vessel, so less than 3W/m²; a small loudspeaker in a portable radio can operate at 350W/m² or more (60mm diameter cone and 1W output power). I normally turn the system off before diving underneath the hull. I once forgot and realised it when I was already in the water. I decided to still get underneath and approach the hull cautiously on the basis of the above considerations. From inside the boat, I can hear the transducers ticking away if I listen intently, but I couldn’t while underwater and I cleaned the hull normally. I have read claims from commercials that their system would not only protect the boat, but also the surrounding area and even nearby vessels… sure. The power density is just far too small in my experience to have any effect beyond the hull itself.

During the development of the hardware, I sometimes handled active transducers without any effect, until one day when I firmly clamped one of them between the palms of my hands. All I could sometimes feel was the odd vibration accompanying the faint ticking noise as the ultrasonic bursts started and stopped. A day or two later, I noticed a dark circular spot in my palm in the exact area that had been in contact with the face of the transducer. The best way I can describe it is like a dark brown grease stain that hadn’t fully washed off. It wouldn’t wash off or even scrape off, because it wasn’t on the skin. I believe the ultrasonic energy broke the capillaries underneath the skin and caused superficial internal bleeding. It took weeks to fade away. Probably not a great idea. The transducers emit short bursts and a lot of them are at frequencies where the energy output is quite low, but they do punch out some power at times when they hit a resonance peak. Such a transducer continuously driven at high power would most likely be able to cause soft tissue injuries.

Risks to the Hull

A friend, retired engineer, got interested in such a system for his alloy yacht, so I gave him the Silicon Chip article to read as background material, but he eventually decided against it. His concern was that the vibration induced in the hull would cause metal fatigue and cracking. I disagree with his conclusion for reasons I will develop shortly, but the thought process he followed is correct. As there would be no more evidence to either support or rule out hull damage like cracking or delamination from ultrasonic energy, if such hull damage was found, its root cause could probably be debated with no end if someone decided to blame the system for it.

The amplitude of the vibration produced by the transducers is measured in nanometres: millionth of a millimetre. It appears insignificant when compared to the vibrations induced by engines and wave impacts and I know of aluminium passenger ferries that have logged in excess of 30 years of continuous commercial service. Their hulls haven’t fractured into little pieces, so I don’t share the concern that sound waves propagating through the plating can cause sufficient stress to induce cracking. Stainless steel ultrasonic cleaning tanks don’t appear to crack either in spite of comparatively massive ultrasonic energy levels. In ultrasonic antifouling applications, the power level is just far too low to damage materials in my opinion and there is no record of hull damage that I am aware of.

Conclusion

I am not aware of any scientific impact studies for hull antifouling applications. A few papers have been published about the effects of ultrasonic energy on algae in suspension in the water and they indicate that prolonged exposure indeed damages them, but the power levels used were high in comparison and the algae was contained in a tank, not drifting. It appears to be a rather clean and low-impact technology.

Recent Developments

At some point after I built the prototype in 2014, Jaycar appeared to have discontinued their ultrasonic antifouling kit. People who had originally bought the Jaycar kit needed to build the transformer and encapsulate the transducer. Later, the kit apparently shipped with those items ready-made, albeit at a higher cost. Some buyers may have experienced difficulties with assembling it. Winding the transformer was not difficult, but it required care and attention, because mistakes almost invariably resulted in destroyed transistors on the circuit board. The construction of the transducer certainly was at the level of many hands-on boat owners and soldering the board wasn’t any different than building any other electronic kit.

In any case, Silicon Chip design revised their 2010 design in May 2017, with a number of small improvements, as well as the much needed ability to drive a second transducer as a separate option. This is to the benefit of Jaycar only, as the details published would no longer be sufficient to construct a functional unit. Both the transformers and the transducers are supplied ready to be installed. I was disappointed to see that the transducer encapsulation had not improved; it has obviously worked sufficiently well as it is. The driver board is not as powerful as the one I built, because they are still getting away with the very feature that had caused my original prototype to blow up near-instantly. A notable difference between the revised Silicon Chip design and my board is that they drive their transducers alternately, not simultaneously. This reduces the peak current draw, but also eliminates interference effects. These effects can be cumulative or cancelling, but the constant frequency shifts can be expected to move the positive interference zones over the hull. As a result, I tend to think that synchronous drive should increase the local peak power levels in the hull, especially in the region half-way between the transducers. The fact that I never observed any dead zones over the hull surface would support this thinking.

As I recently happened to come across a very rare commodity referred to as spare time, I used it to revisit this old project. I updated the design to use mostly surface-mounted components (SMD) this time, ordered professionally-made circuit boards and built a few very nice-looking new units. These have proved to be quite sought after and I am curious to get feedback from a broader user base.

New dual-channel ultrasonic antifouling driver board, this time mostly using SMD components.

As the new board is much more straightforward to reprogram in terms of operation, I have started thinking about trying to improve the performance of the system with regard to fighting algae growth.

References

[1] “An Acoustic Antifouling Study in Sea Environment for Ship Hulls using Ultrasonic Guided Waves”, Habibi, H., Gan, T.-H., Legg, M., Carellan, I., Kappatos, V., Tzitzilonis, V., & Selcuk, C., in International Journal of Engineering Technologies and Management Research 3 (4), 14-30, 2016.
[2] “Ultrasonic Transducer Technical Notes”, Olympus NDT, 2006.
[3] “Power Converters Design and Analysis for High Power Piezoelectric Ultrasonic Transducers”, Davari, P., Ghasemi, N. and Zare, F., in Power Engineering Conference (AUPEC), 2016 Australasian Universities (pp. 1-5). IEEE
[4] “Power amplifier for ultrasonic transducer excitation”, Svilainis, L. and Motiejūnas, G., in Ultragarsas, Nr.1(58), 2006.

This article is part of a series dealing with building best-in-class lithium battery systems from bare cells, primarily for marine use, but a lot of this material finds relevance for low-voltage off-grid systems as well.

This article discusses the protection of lithium battery banks in the context of marine installations. A previous article detailed the design and assembly of such lithium battery banks. Before a lithium battery can be put into service, it must be protected or the first adverse event to occur will damage or destroy it.

Disclaimer

A good understanding of DC electrical systems is needed to build and commission a lithium battery installation. This article is aimed at guiding the process, but it is not a simple blind recipe for anyone to follow.

The information provided here is hopefully thorough and extensive. It reflects the knowledge I have accumulated building some of these systems. There is no guarantee that it will not change or grow over time. It is certainly not sufficient or intended to turn a novice into an electrical engineer either. You are welcome to use it to build a system, but at your own risk and responsibility.

Battery Protection, Battery Management or Both?

The needs of lithium batteries fall into two categories:

1. Requirements to be met to prevent dangerous developments, damage or the rapid destruction of some or all of the cells.
2. Requirements to ensure a long operating life and an absence of issues over time.

The first category represents acute needs which must covered by battery automated protection functions; the second one pertains to battery management. Battery protection must be seen as the basic subset of functions battery management is built upon: there is no management without protection, but at the lowest acceptable limit a lithium battery can be protected without being automatically managed.

A 200Ah DIY lithium battery back for a yacht connected to a custom battery protection module.
From left to right on the connector, the four cell voltage signals, the reference negative, the two power supply cores for the module itself and the cables from the temperature sensors. Using the right equipment considerably simplifies the task of building lithium battery systems.

Battery Protection

An unprotected lithium battery can become a fire risk or an expensive fiasco: take care of the cells and the battery will take care of itself

A lithium battery cannot be responsibly operated without a layer of protection between itself and the electrical installation because:

• Lithium cells are quickly destroyed by voltage excursions.
• Excessive discharge causes severe and irreversible damage.
• Overcharging a lithium battery, or recharging it in a damaged state, can overheat it to the point where it can ignite and burn extremely hot.

Protection is simply a matter of keeping at all times each and every cell within its allowed operating voltage limits and ensuring nothing ever begins to warm up. Give or take a few degrees at high sustained currents, a lithium bank on board should run at ambient temperature, at all times, no exception.

Integrated short-circuit protection is also commonly found on small battery packs, but it is problematic to implement on larger banks, such as found on marine vessels, because the potential fault currents can exceed the interrupt rating of the disconnectors; fusing must be used for this purpose. Also, due to the relative size of the battery and wiring, the risk is not so much towards the battery: something else is likely to melt or get destroyed first.

A protection system must be able to act automatically at any time in case of detected issue, before a serious situation develops. Alarming is pointless for that matter: alarming relies on someone not only being present, but also understanding what is happening and knowing what to do. The human factor becomes embedded into the reliability of the scheme and momentary inattention or absence instantly results in zero protection. It is only a matter of time.

Battery protection is therefore the first and most essential layer of any lithium system, because it implements all key safety functions and protects the asset. It needs to be built and engineered correctly. A number of devices are available on the market to perform this task, with wide variations in quality, reliability and suitability for a marine installation. Interestingly enough, some people have routinely installed junk-grade gadgets on large expensive assets without even blinking an eye, while others simply dispensed themselves from installing anything… As we all know, accidents always happen to others, but the banks I see destroyed seem to be those, a curious paradox.

Cell Voltage Monitoring

The need for monitoring the individual cell voltage is the main point of distinction between a lithium battery bank and a lead-acid one, where total battery voltage only is ever considered. It is made necessary by the fact that voltage excursions quickly become destructive with lithium cells and lithium cells stop accepting current when fully charged, which prevent a battery from self-equalising the way lead-acid ones do.

In order for a LiFePO₄ cell not to fail, its own voltage must be remain at all times with a prescribed range, above 2.0VDC in all cases and below about 3.65VDC. Practically, cell destruction occurs at about 4.2VDC, when the electrolyte is decomposed, but voltages in excess of 3.55VDC don’t make much sense and the cells are sensitive to the maximum instantaneous voltage they are exposed to, not just the average. Many charging sources deliver a voltage superimposed with a ripple and the peaks from the ripple can destroy the cells even if the average value as reported by a multimeter appears acceptable.

Low-Voltage Protection

The obvious purpose of low voltage protection is ensuring that no cell ever gets excessively discharged. The practical low voltage limit of a lithium iron phosphate cell is typically 2.5-2.8VDC. In the case of a 4-cell, 12V-equivalent configuration, one can argue that sufficient low-voltage protection can be obtained by monitoring the total battery voltage:

The worst-case scenario would be reached if three cells were fully charged (3.35V open-circuit) while the fourth one was very low.

Considering that a balanced pack is below 15% state-of-charge at 12.8V, even adopting 12.6V as battery low cut-off voltage would give a lowest possible cell voltage of:

V_{Cell_Low} = 12.6V – 3 x 3.35V = 2.55V

This is still an acceptable value. However, this result does not extrapolate at all to a 24V installation with 8 cells in series.

Low-voltage protection requires the automatic disconnection of the loads from the battery, so the lowest cell voltage doesn’t sink any further and this is absolutely essential. Such an event is referred to as a Low Voltage Disconnect (LVD) or Low Voltage Condition (LVC) disconnect. Reconnection must not take place until the lowest cell has been recharged up into its normal operating range.

Cell-based low-voltage protection is what allows taking full advantage of lithium batteries, because it allows discharging almost fully when needed without any concerns and it maximises the available capacity and performance of the installation.

High-Voltage Protection

High-voltage protection is obviously related to charging and a similar calculation, even based on a modest end-of-charge voltage of 14.0V for the same 4-cell configuration, shows that protection must be based on individual cell voltages:

If we consider three cells out of four at a mid-range voltage of 3.3V, which is not a worst-case scenario, we can calculate the voltage of the fourth cell as follow:

V_{Cell_High} = 14.0V – 3 x 3.3V = 4.1V

A LiFePO₄ cell at 4.1V is far into destruction territory and it is clearly impossible to guarantee that no harmful voltage excursion can take place if only the overall battery voltage is being monitored.

There is no alternative to cell-level monitoring and automated protection with lithium batteries

Cell level protection is required here for ensuring charging safety and, should any cell drift towards an unacceptable voltage, charging must stop. This can involve automatically disabling the charger(s) and, if this doesn’t appear effective, the battery must be automatically disconnected from all charging sources: this typically means disconnecting the charge bus, an event referred to as a HVC disconnect. This should always be treated as a serious issue: a HVC event indicates that something is not in order in the installation and the charge bus should not reconnect unless the battery is significantly depleted.

Since cell-level voltage measurement is mandatory for charging safely, it can just as well be used for over-discharge protection too. Cell-level protection is the minimum standard for operating a lithium bank.

Cell Over-Temperature Monitoring

Monitoring the temperature of the cells is highly desirable for the simple reason that temperature is always involved when the situation starts going seriously wrong with lithium battery cells and it provides a second view of what is happening in a battery pack that is completely independent from the voltage information.

In most instances, voltage deviations will highlight problems first, especially in simple series topologies such as 4S or 8S. When parallel blocks of cells are used, voltage monitoring applies to each block only (because the parallel cells all share the same voltage by definition) and temperature monitoring can provide a more granular view of what is happening. Temperature sensors can be inserted between the cell casings and therefore any one sensor can monitor a pair of cells. Full temperature monitoring can therefore be achieved with a number of sensors equal to half of the number of cells.

LiFePO₄ cells should ideally be operated between about 10°C and 25°C, but safety concerns only develop at very high temperatures. The absolute maximum temperature a bank should be allowed to reach depends a lot on the environmental conditions it is exposed to. A LFP bank should ideally not be exposed to temperatures exceeding 30°C to prevent premature aging; in this case, a temperature as low as 40°C could be used for alarming. A high temperature should eventually lead to a complete system shutdown and disconnection of the bank. The consistency of the temperature readings within a bank is a lot more valuable piece of information. A marine house bank should always operate within a few degrees from ambient temperature and discrepancies between cells are highly suspicious. The recommended maximum allowed temperature deviation within a lithium battery bank is 5°C from highest to lowest [1]. Any value exceeding this should be investigated, regardless of what the actual absolute temperature is.

Do Not Ever Rely on a Battery Monitor for Protection

Battery monitors with configured high and low state of charge limits are not suitable devices for protecting lithium batteries. The state of charge displayed is merely a completely unreliable estimate obtained by accounting for charge and discharge currents over time. If the battery is cycled for a period without ever being fully recharged, the monitor has no opportunity to reset its capacity count (provided it was configured to do so in the first place).

Protection must be implemented based on the voltage of each individual cell and the capacity estimate made by the instrument is irrelevant. Cell voltages are either within range, or they are not, and they are all that matters.

Do Not Confuse Protection and Control

A protection system with automated disconnection is just that: a last line of defence that should never be activated. Using the disconnection device(s) to terminate charging (a suggestion often formulated by DIY implementers) is out of the question. It breaches the system design boundaries, where the battery protection layer’s role is to mitigate any failure in the charge control system. More specifically, one role of the BMS is defending against a failure in maximum charging voltage regulation.

Several owners of lithium banks I know came to grief with the battery protection layer after making alterations to their charging infrastructure: as it should be. In such circumstances, people sometimes complain about the protection when they should in fact be looking at the consequences of what they have been doing.

Battery Disconnection

Protecting the battery as an adverse event is developing eventually entails disconnecting it. There are three types of devices that can be employed to this effect and, each time, they need to be able to cope with all the current that can be expected in the installation. Battery disconnection is a feature in all battery management systems and most of them can only handle a specific type of disconnector, so discussing the pros and cons of the disconnection schemes is relevant. One may rule out a BMS because it cannot control the type of disconnector best suited for the application.

Regardless of their type, high-current contactors can be designed for the current to flow in a specific direction. This can be somewhat disconcerting, as a simple metal contact obviously conducts independently of the direction of the current, but this is ignoring the effect of the magnetism produced by very high currents. This magnetism can interfere with the one from the energising coil and affect the operation or reliability of the contactor. If the datasheet specifies the direction of the current, follow it.

Mechanical Contactors, or Relays

Tyco Electronics EV200 contactor.

In a mechanical contactor, a coil holds a contact closed when energised. Such devices are also termed “monostable” as they can only be in one state when not energised. Their drawback is the continuous drain arising from the coil current. There are wide variations in coil power between contactors and this will determine the standby power drain of the system, so it is very important to pay attention to the coil current data. Some contactors have built-in measures to reduce the coil current once the contact is closed (holding requires less power than closing), or the addition of an external capacitor and resistor can deliver a similar result, see below. In this case, the capacitor must be large enough to cause the relay to close while it charges and then the resistor must be low enough to still reliably holding the contact closed while reducing the current.

When power is switched on into a coil, the voltage increases relatively gently, but upon turn off, the abrupt collapse of the magnetic field causes a spike (also known as back EMF). The higher the coil power, the stronger the spike. This spike can cause the equipment controlling the relay (i.e. the BMS) to fail, so it should be neutralised using a free-wheeling diode. Some relays have this diode built-in and the coil has a polarity as a result, some haven’t and it should be added externally.

Monostable contactor with external RC network to reduce the holding power consumption. When the control switch is closed, the inrush charging the capacitor C closes the relay contact. The current then reduces to the value determined by the coil resistance and the external resistance R. Without the additional components, the coil keeps drawing full current continuously.

The expression “fail-safe” is sometimes used in relation with (and to promote) simple contactors, because a power failure results in disconnection. That is all very well, but there are far worse and more insidious failures modes than a simple power loss, as discussed further in the section about reliability, and this type of configuration doesn’t automatically fare very well in a Failure Mode and Effects Analysis (FMEA). The outcome of the most common control failure (rather than power failure) usually is that the relay can no longer open because power to the coil can no longer be interrupted. Since the normal state of the contactor is closed when the system is operating normally, such a failure only becomes apparent when a need arises to disconnect, i.e. in an emergency.

Solid-State Relays

Solid state relays emulate the operation of a mechanical relay by using MOSFET transistors instead of a coil and a contact. The advantages are immediate: the control current is very low (technically zero after the device has switched, but the drive circuitry usually adds a small consumption) and there is no arcing or contact damage, as there are no contacts.

However, MOSFET transistors can easily be destroyed by voltage or current spikes and, when they fail, they nearly always do so by short-circuiting completely. Another technical consideration is that such a switch normally has a polarity and it only operates “one way”. When polarised in reverse, the device conducts like a diode and a high current will result in overheating and the complete destruction of the unit, unless it was specifically designed to also block reverse current. This is far from always the case as it requires using back-to-back MOSFETs, i.e. twice as many transistors for twice the losses, heat and cost.

Solid-state relays have no contacts and offer minimal power consumption, but they are usually polarised – like this Victron unit – and can only interrupt the current in one direction.

In a typical small craft marine installation, the disconnectors must be able to handle several hundreds of amps and they can be exposed to voltage transients from starter motor or windlass motor switching. Should the unit fail as a result, it will almost invariably fail short and leave the battery permanently connected. This is not a good prospect in a safety system. Disconnecting the charge bus must also prevent the battery from draining itself into a damaged charger for example, so the disconnector must fully isolate the circuits. Building a suitable solid-state switch is possible, but far from cheap. Solid-state switching is best suited for relatively small battery packs operating in well controlled conditions.

Latching Relays

A typical latching relay has two coils instead of one and a mechanical contact that can stay either opened or closed. Power is only required in the form of a short pulse to one of the coils to change the state of the relay, so there is no on-going consumption associated with the device. This, and the fact that the contact itself can withstand tremendous abuse compared to a solid-state switch, tends to make it the preferred solution in marine systems that are in service 24/7. The coils used in latching relays are not usually rated to remain energised for long periods and they will quickly burn out if left powered. Some larger latching relays sometimes use a small motor to open and close their contact.

The drawback of latching relays is that they require a different control logic with two commands instead of one and so they are not interchangeable with standard holding relays.

Tyco Electronics BDS-A latching relays with zero standby consumption and a 260A continuous current capacity. The peak current is 1500A.

A control failure means that the device stays in its last state, either opened or closed, and this is generally not a pleasing prospect either. However, since the control circuitry is only operating briefly for switching and spends its near-whole life inactive, ensuring and monitoring its integrity can be achieved much more successfully.

Battery Management

Battery protection doesn’t guarantee that the bank will offer a long service life; it only ensures that no dangerous development can take place and that the battery can’t be quickly destroyed. Battery protection systems operate by enforcing operating limits on cell voltages and temperatures to prevent accidents, but they are unable to prevent lithium plating damage due to reduced-voltage overcharging, or loss of capacity resulting from keeping the cells at a high state of charge for example.

Battery management includes functions beyond protection. The purpose of a battery management system is ensuring that the battery is operated as close as possible to the optimum. This means:

• Deciding whether charging is necessary
• Deciding whether charging should be prevented
• Deciding whether discharging is acceptable
• Ensuring the battery doesn’t remain idle at a high state of charge for long periods of time
• Ensuring that the cells remain balanced over time
• In some applications, provide thermal management of the battery to keep it within an acceptable temperature range

Most battery management functions rely on knowing the state of charge and therefore battery management is not possible unless battery current information is provided to the BMS. Basic battery protection is quite simple and straightforward, battery management is not; it can involve sophisticated decision-making algorithms.

Battery Management Systems

The term Battery Management System (BMS) is used indifferently for devices that provide protection functions only or protection and management functions. Regardless, the most essential function here is a safety one, as stated earlier. Management, if present and well implemented, can ensure longer cell life and consistent performance over time regardless of the regime of utilisation of the bank.

Critical Features

The following must be achieved for effective battery protection on board:

• Individual cell voltages must be measured. Problems with lithium batteries always occur at cell level.
• Overheating must be detectable. When things go very wrong with lithium cells, heat is nearly always involved. Heating is what can eventually lead to thermal runaway and a battery fire.
• The system must be able to automatically act to prevent cell damage. This means eliminating the cause of the problem by disconnection. Alarming is not good enough. Human-powered systems, involving alarms, watching, monitoring and taking action when something goes wrong will always fail in short order: the first time no one is around or paying attention.
• The protection system must be reliable and resilient, it must be protected from voltage transients and not be able to fail unnoticed. Having a level of redundancy and a self-checking ability is highly desirable in a marine BMS, because we are protecting large assets worth much more than the battery.
• The BMS must not induce uneven drain on the cells, or it will throw the bank out of balance and create problems in very short order. This was an issue with some early junk BMS contraptions and it caused a few people to start clamouring that “BMS were harmful”. Using with decent equipment to begin with goes a long way.
• The intrinsic consumption of the BMS solution must be as low as practically possible, or it may drain the battery to destruction in the absence of power for charging. Boats can end up spending winters with snow covering solar panels at times, with little or no energy input and the installation must tolerate these conditions. This generally weighs against solutions that require holding contactors closed. Battery banks drained flat by the BMS also caused some people to claim that any electronics added to the battery were “harmful”. If the cells get low to a point where it is obvious that recharging is not taking place, a BMS must be able to shutdown the installation as well as itself to preserve the battery.

A system encompassing the features listed above will prevent an accident with the battery and ensure the cells cannot get damaged. It could however cause damage to other equipment on board if it gets triggered, so a few additional features are also desirable and correct electrical design is essential.

Essential Features for Marine Installations

Marine applications, especially on ocean-going vessels, tend to place higher requirements on the systems than what would be found in other applications where power availability is more of a convenience.

Availability and Resilience

Batteries on marine vessels are relatively important and energy storage is often seen as an area where high availability is desirable. For this reason, the BMS should be designed to control a dual DC bus system: charge disconnection should be distinct and separate from load disconnection, otherwise a charge regulation issue can cause the vessel to lose all power unexpectedly. Conversely, an over-discharge condition would effectively prevent recharging, which would be senseless and defeat recovery.

Any BMS not engineered to control split buses can essentially be ruled out in serious marine applications.

Support for Advanced Charger Disconnection

The disconnection of the charge bus must be able to be performed without resulting in damage such as the destruction of alternators or wind generators. This require the management system to be able to provide “advanced notice” of a charge disconnect.

Advanced Warning

A protection action should not take place unexpectedly, especially if the problem develops gradually, such a low battery. There must be an output to indicate an issue, whether it stems from cell voltage or temperature so the user has time to react appropriately.

A Word about “Non-Essential” Features…

If a BMS is any good, gimmicks such as displaying cell voltages should be entirely irrelevant: the end-user wants reliable, trouble-free energy storage, not a subscription to the Lithium Channel on Battery TV, always wondering what is going to come up next.

Protection Voltage Limits

Which Limits?

The battery chemistry sets the absolute outer limits of what voltage a cell can tolerate. The manufacturer sets more conservative absolute limits to discourage abuse of the product. The system designer may opt for more conservative cell voltage limits again.

So, which one should the BMS use? It depends… but it is interesting.

Junk-grade BMS often use upper cell voltage limits well outside manufacturer recommendations, like 3.7-3.8V for high-voltage cut-off. It allows the product to be used “successfully” in electrical systems that shouldn’t be charging lithium batteries. Some implementers sometimes expect an alarm just outside their control limits and then the BMS not only protects the battery, but also flags any anomalies in the charging system. In this approach, it is important that the BMS doesn’t try to enforce charging voltage limits that are in fact too low to sustain good cell health on the long run.

Debating at long length what the “magic” voltage numbers should be is rather pointless

If the expectation is that the BMS doesn’t only protect against serious events, but also ensures long battery life and sustained performance, then it needs an algorithm much more sophisticated than enforcing simple cell voltage limits and needs to play a role into charge control.

Cell Upper Voltage Limit

My experience so far indicates that the cells must be able to be charged up to 3.50V routinely. Systems without cell balancing experience more cell-to-cell variations at that voltage and an alarm limit below 3.55V would be asking for trouble. Disconnection may happen at 3.60V, with 3.65V absolute maximum as it is the manufacturer upper limit.

Cell Lower Voltage Limit

The low end is much simpler to deal with. The remaining capacity drops off very quickly below 3.00V, manufacturers usually indicate 2.50V as end of discharge voltage and the chemistry becomes unstable below 2.00V. Other than that, there is no reason to leave much unused capacity at the low end. The chemistry is most stable and the cells age the slowest when heavily discharged.

I alarm for any cell at 3.00V and disconnect at 2.80V. It almost allows using the full capacity of the cell when needed.

Battery Management Functions

Some BMS units offer limited battery management functions. The most common (and often only) one is automated cell balancing. A lot of myths and misunderstandings exist around automatic cell balancing, so we will discuss the subject in some detail.

Cell Balancing

The concept and the importance of balancing cells were treated at considerable length in the context of assembling a lithium battery bank. Over time, cells that were initially balanced drift apart and a time eventually comes when something needs to be done about it. If the cells are of good quality, they were never abused in service and the manufacturing process was consistent, it can take many years before rebalancing becomes necessary. With a bit less luck at the time of purchase, cell balance adjustments can be necessary every year or so. If cell balancing becomes frequently necessary, there is a major problem: cells are shorting internally and failing. A cell with high self-discharge must be replaced.

Properly implemented automatic cell balancing has its place in most lithium installations, because it prevents charging problems with high cell voltage events from developing on the long run. If the cells could be hand-picked from lab test data to all be virtually identical, then a pack could possibly be constructed that would remain balanced until the end of its life. In common real-world applications, this is not possible.

Part of the answer to the question “should it balance the cells or not” may depend on who will own and operate the installation. People who build their own systems and actually understand what they are doing may decide to check on cell balance now and then and perform any adjustments manually. For everyone else, the general answer is that balancing is an integral part of battery management. Only installations with unduly low charging voltages get away with sloppy cell balance and they do so at the expense of other problems developing, such as large capacity reductions from memory effects after a few years in service.

There are passive and active cell balancing circuits: passive circuits “burn” excess energy from the higher charged cells and active circuits transfer it to the lower cells, but not without losses.

Active Balancing

In theory at least, active balancing is superior in applications with restricted recharging opportunities or where gaining access to the maximum possible capacity is essential. Even so, starting with a matched set of cells of equal capacity beats transferring energy. Active balancing is also much more complex and costly to achieve.

Active balancing systems further split up into inductive and capacitive energy transfer systems, the latter being much less common. Some active balancing circuits continually transfer energy between cells throughout charging and discharging with the aim of maintaining all cells at the same voltage at all times. While such a strategy results in operating top balanced cells at the top and bottom balanced cell at the bottom and the available capacity is theoretically maximised, efficiency is nowhere near 100%. Energy can also be transferred uselessly during the cycle because of differences in cell internal resistance causing discrepancies in voltage. It is a crude strategy that requires the ability of quickly transferring a lot of charge between cells in order to succeed. Mastervolt MLI batteries operate this way.

High-capacity inductive charge pumps on top the cells inside a Mastervolt MLI lithium battery pack.

The photo shows a 30A blade fuse and the transformer performing the coupling between cells at different voltages on the top circuit board. In front of the transformer (yellow and red), we have 4 MOSFET transistors (typically in a push-pull arrangement) flanked by two electrolytic “tank” capacitors. The transistors switch the current going to the transformer and the capacitors provide the peak of the current. Switching high currents into inductive loads like transformers at high-frequencies is hard on electronics. The most likely failure mode is one of the MOSFETs failing short from stresses, at which point a second transistor might then get destroyed by current overload before the fuse blows.

Note the control board installed vertically on the front and the ribbon cable to control the balancers. More electrolytic capacitors (blue and cylindrical) there.

Passive Balancing

Passive balancing can be used with top or bottom-balanced packs and its aim is maintaining cell balance either at the top or bottom by shunting some current over any cells registering an excessive state of charge compared to the others. As the battery banks of interest to us here are basically always top-balanced, we will consider shunt-balancing at the top.

A 4-cell battery with cell shunt balancers. The control signal operates each transistor like a switch.

Shunting involves connecting a resistor across the cell terminals: if the cell is charging, some of the charging current will flow through the resistor instead of through the cell and the cell will be charging at a reduced rate. If the cell is not charging, then the resistor simply contributes to producing a discharge current for that cell only. The amount of current flowing through the shunt is the ratio between the cell voltage V_cell and the resistance R of the shunt circuit. Either way, a resistive balancer operates by “wasting” some energy. The amount of power the balancer needs to dissipate as heat is P = V_cell ²/ R.

Shunt balancer during charging. The cell current I_cell is reduced by the amount of the shunt current I_shunt and becomes smaller than the battery charging current.

Shunt balancer in discharge. The cell current I_cell is increased by the amount of the shunt current I_shunt and becomes greater than the battery discharge current.

Smart Shunt Balancing

Ideally, a shunt balancer should waste exactly the amount of charge required to bring any high cells down to the level of the others and therefore restore balance. In this case, shunting can happen during discharge and at normal operating voltages. Practically, determining how much energy this represents cell-by-cell is a non-trivial task that first requires knowing the state of charge of each cell at the end of a full charging cycle. Then it becomes possible to selectively deplete any cell requiring it by the precise amount needed to achieve balance. A computer-controlled centralised BMS with a current sensor is required to do this. This type of balancing only needs to occur occasionally and only needs to perform very small corrections. The shunting currents can be very small because once the magnitude of the correction has been calculated, the adjustments themselves can be done over hours or days while the bank is operating normally. Factory-engineered EV battery packs typically operate this way.

Example:

As we were finishing to charge a pack, we determined that we need to perform a balance adjustment of -1Ah on a specific cell. The cell voltage is now V_cell = 3.35V and the shunt has a resistance of R = 10Ω. The shunt current is equal to I_shunt = 3.35V / 10Ω = 0.335A and the amount of shunting time required to deplete 1Ah is t = 1Ah / 0.335A = 3 hours.

However, if the cell voltage drops during this period, the shunting time will be extended. The BMS needs to continuously track and accumulate the amount of energy shunted and every cell effectively has its own capacity monitor embedded in the BMS.

The shunt would dissipate P = 3.35V x 3.35V / 10Ω = 1.12W in heat, which is a very modest value.

Voltage-Based Shunting

Voltage-based shunting is a crude shortcut sometimes used with cheap equipment and is commonly found on BMS systems built around cell boards. Shunting begins whenever the cell voltage exceeds a threshold. This strategy can work if the shunting voltage is only ever reached at low currents and the cells are near-fully charged each time. It can work for charging DIY EVs overnight at C/10 using CC/CV chargers. Charging at high currents with alternators on boats results in the cell voltages rising too early, because of the parasitic effect of the internal resistance of the cells. At this point, the shunts engage and operate not only based on the state of charge of each cell, but mainly on its R x I voltage drop. Cells with higher internal resistance get shunted and receive less current for no valid reason. This typically results in upsetting balance. If charging continues until the current tapers down to a very low value, the shunts may find enough time to correct for the upset they caused earlier. On boats, we often shut the engine down before this has happened and start discharging again and the shunting cell boards just mess up the bank.

Voltage-based shunt balancers commonly create significant heat, because there is very little time to fix up cell balance at the top before charging terminates and they need to handle a lot of current compared to a calculated shunting strategy. In practice, shunting cell boards cause so much havoc that their threshold voltage is usually set high enough to make them virtually useless. Their main pitfall is that they have no idea of what the battery current is doing and they shunt while the current is still high.

For doing the same work as above, a shunt balancing cell board with a capacity of I_shunt = 2A starting to operate at V_cell = 3.6V would produce P = 3.6V x 2A = 7.2W of heat for t = 1Ah / 2A = 30 minutes. The cell voltage would also need to be held at 3.6V for that duration by the charging system, which means overcharging.

In other words, a 2A-rated shunt isn’t enough to do the job, because it takes too long.

A voltage-driven shunt balancer trying to balance a high cell (3) out needs to carry all of the charging current to prevent the high cell from getting further stressed by overcharging. Balancing needs to continue until cells 1, 2 and 4 have caught up with cell 3.

Voltage-based shunts are voltage limiters with a limited current capacity, i.e. their rating. In order for them to work properly, the charging source also needs to cooperate: if the charger supplies more current than the shunts can carry, then some current must go through the cell and its voltage will keep increasing. Also, in order to balance the pack, all the other cells eventually need to be taken as high as the overcharged cell, which takes a long time and is undesirable. Voltage-driven shunting is a generally a poor strategy unless it can operate at lower voltages with a charging regime controlled to make it work. It is never the case when it is just blindly thrown into a system.

Shunt Balancer Reliability

A passive (resistive) balancer can also fail, of course, in one of two ways: open-circuit, and then it simply can’t operate any more, or fail short and then it can’t be stopped. The first failure mode is less likely and could go unnoticed for a long time. The second failure mode is the harmful one as it would cause the affected cell to get discharged uncontrollably to destruction. There is much hype around about “cell balancers destroying banks” and very little in the way of supporting cases once all the cases of brand new faulty cell boards are discounted. A passive shunt balancer operates at cell voltage only and switches a resistive load, i.e. no spikes. This means virtually no stress on the switching transistor and it makes it immensely reliable compared to active balancing circuits.

Other Battery Management Functions

Reporting the state of charge of the battery is a BMS function sometimes present. A BMS with a chemistry-specific SOC algorithm will massively outperform any ordinary battery monitor in data accuracy and reliability.

Disabling charging in excessively low temperatures or battery temperature control are management functions sometimes available on upper end units. A BMS should ideally direct charging as it is the only piece of equipment in the system actually capable of determining what the battery needs.

BMS Topology Considerations

BMS solutions come in two main flavours: either distributed or centralised. The difference would not matter much if the functionality was identical, but distributed BMS systems have slipped towards the cheap and nasty end of the spectrum in recent years.

Distributed BMS Architectures

Distributed BMS systems typically have the advantage of being scalable over a large and variable numbers of cells in series. These systems use one “cell board” per parallel cell group and usually some kind of head unit to implement the control functions. In order to retrieve the information pertaining to each cell, a communication bus linking all the cell boards is necessary. This increases the complexity and cost of the cell boards and the communication loop can be seen as somewhat vulnerable: this is not automatically true and cell boards themselves are commonly vulnerable due to their lack of physical protection, especially in marine systems. Regardless, this was “addressed” by some designers focused on the low end of the market by using a simple current loop over a single wire, starting from the master board, passing through every cell boards, and returning to the master. It is no longer possible to transport individual cell information, so the loop just carries information about one or two conditions typically, like over-voltage/under-voltage. The master can see an alarm condition, but can’t identify the cell it is originating from. It makes battery management impossible, but protection without any diagnostics can certainly be obtained this way.

As most of these “one wire loop” systems are first and foremost designed for lowest engineering and production cost to the vendor, many other aspects tend to leave much to be desired. Close attention must also be paid to the protection limits implemented, because the clear intent is often allowing drop-in lithium battery replacements on automotive installations where battery-damaging (but not dangerous) charging voltages must not trip the “BMS” and cause grief to the installer.

Some early distributed BMS solutions implemented digital communications across all the cells and were scalable up to fairly high voltage pack configurations, typical of EV systems. This can be quite good, but it is more complex to design than a centralised BMS. In recent years, distributed solutions have drifted towards the bottom end of the scale.

Centralised BMS Architecture

On a centralised BMS, voltage sensing wires run from each cell to the module, as well as wiring for the cell temperature sensors. This wiring needs to be arranged neatly, but it is easy to protect. On marine installations with 4 or 8 (groups of) cells, it is always straightforward. If battery management is the objective, then the BMS must also have access to battery current information through a suitable sensor. Depending on what the objective of the management algorithm is, the BMS may need access to the installation load current as well as the battery current, or currents through the charging and discharging paths separately.

A centralised BMS has access to every piece of battery information in one place and at any time: this is necessary for implementing many management strategies, including and especially smart cell balancing. It becomes possible to track capacity cell-by-cell, track cell internal resistance, predict time-to-empty, issue charging parameters, manage charging sources and loads and much more.

Centralised BMS modules tend to include more engineering effort, especially on the firmware side. A vendor interested in marketing BMS for maximum profits immediately looks in the direction of a distributed system with cell boards and an analogue loop, because it can easily adapt to any number of cells; a centralised unit can only offer a predetermined number of inputs for cell voltage measurement and will only fit installations within these limits.

BMS Reliability and Failure Modes

From a reliability point of view, a BMS is just an electronic circuit with external connections. It can fail in two ways: internally, or due to an external event reaching it through its connections.

The reliability of a battery management system (BMS) can be challenged by design and construction factors (internal) as well as external events such as voltage spikes or short-circuits outside.

Internal BMS Failure

An internal, or intrinsic failure, is a failure “for no apparent reason”. Ageing of the components, component defect, manufacturing defect can all lead to an internal failure. Quality of the parts and construction have a role to play, but also design: components like electrolytic capacitors have a finite life, which gets shorter at high temperatures. After 10-15 years, most electrolytic capacitors aren’t looking too good in general. In a device like a BMS, they can be easy to avoid altogether and ceramic capacitors can be used instead, with a near unlimited life expectancy. They may just cost a little more. The maximum voltage rating of the parts used also matters a great deal: 16V-rated components exposed to battery voltage in an automotive-like system have almost no margin for surviving voltage transients.

Electronic devices are most likely to fail “for no reason” when they are either very new or very old

Component and manufacturing defects tend to come out at the start, while ageing effects appear at the far end. The ageing limit can be pushed so far back by design and component selection that it can move past obsolescence and become irrelevant: the Voyager space probes are still sending back information after 40 years.

The probability of failure of electronic equipment is highest immediately following manufacturing due to the risk of defect and at the end of the product life due to ageing.

The failure probability curve of electronic hardware has a characteristic U-shape. Quality control and factory testing at the start are intended to take out most of the bad samples and prevent them from reaching the market. From the user’s point of view, the failure risk keeps decaying immediately after purchase and reaches a low level that remains somewhat constant over most of the product’s life. This level is very much a function of the quality of the product.

The prospective life of well-managed marine LiFePO₄ banks already appears to be in excess of 10 years (as of 2017), based on data from early installations, and this figure keeps increasing all the time. As a BMS should logically outlast the cells it is protecting, this is placing increasing demands on the long-term reliability of the units. Even though we don’t yet see this into the BMS market, it is starting to make more and more sense to design for 20 years of operation now.

BMS Failure from External Event

On one side, a BMS measures small signals and, on the other, it actuates remote devices. These connections and the associated wiring can pick up transient voltages, RF energy or be exposed to unfortunate events caused by the installer or user. How resilient the BMS is to such events at its terminals depends on how much protection was built into its design.

Devices that include protections at their terminals are sometimes referred to as being “hardened”, like industrial-grade control equipment. Hardening electronic circuitry means adding components that serve no functional purpose and may never actually do anything over the life of the device. They exist as an insurance policy, and should an abnormal event take place one day, they can prevent damage to the device. An industrially-hardened circuit can include a lot of such additional components for protection, they take board space and they add to cost without contributing anything to functionality. For this reason, they are the first ones to be found “missing”.

A particularly vulnerable part of BMS modules is the outputs driving the battery disconnectors. Relay coils are inductive in nature and produce a current surge on energisation and a reverse voltage spike upon de-energisation. These are hard on the MOSFET switching transistors handling the current. When such a MOSFET fails, 99% of the time it turns itself into a short-circuit. It is very bad news for a lot of BMS designs:

The BMS is holding a normally-open battery disconnector closed. It looks fail-safe: lose the power and the contactor releases, disconnecting the battery. One day, the voltage transients caused by switching the relay coil finally manage to kill the transistor, which can’t be turned off any more. If the contactor is supposed to be closed, the installation has power and the failure is both hidden and undetectable, but the system has no protection any more. This is the worst kind of failure mode.

If the BMS uses latching relays instead, then a transistor failure becomes detectable, because the coils should never be energised under normal circumstances. However, should the BMS suddenly “stop” functioning, it may leave the battery connected indefinitely (in both configurations, by the way). This can be addressed too, but the intent here is not to delve deeply into electronic engineering. This type of thinking and analysis is part of failure mode analysis. It needs to be performed by system designers both at component and installation levels.

The designer of a BMS can go to great lengths to mitigate or eliminate these risks and design up to a standard, or instead ignore them and design down to a cost, and the product may have no more and no less functionality for the user and yet be vastly different in suitability. Quality does matter in relation with the application and marine applications include situations where high reliability matters, and others where people or large assets can be at risk.

Last, but not least, a BMS module is just one component in the system and its role is ensuring that no single failure can result in a hazardous situation. The designer of the battery bank and associated electrical system should ensure that his installation does not depend on the BMS to operate safely, but instead relies on the BMS to protect itself if something goes wrong. This way, two cascaded failures need to take place before risks develop. Meanwhile, a single failure must be both visible and detectable, or it will go unnoticed until a second one follows and causes some disaster big or small, but usually costly!

General Guidelines for BMS Installation

Protecting the bank is an essential part of installing it on board. The details of the installation of a BMS are specific to the solution retained, but it must always involve establishing dedicated connections for sensing the cell voltages and should also involve adding temperature sensors to the cells.

This 2P4S 400Ah lithium battery bank during installation on a sailing catamaran is occupying a fraction of the space previously allocated to the deep-cycle lead-acid batteries. The owner must still clamp and secure the cells. The cell voltage sensing and temperature sensors wiring is going to a Gen 1 Nordkyn BMS module controlling remote latching disconnector relays in the electrical panel several metres away.

The best location for the BMS module is generally close to the battery, so the length of the voltage sensing wiring is minimised. Keeping this wiring short makes it less susceptible to pick up electromagnetic disturbances. Keeping the cell-sensing wiring into the battery compartment also eliminates considerations around fusing that wiring, a questionable idea as it negatively affects reliability.

The DC disconnectors associated with the BMS can be relocated to meet wiring constraints. It is sometimes simpler to run a feeder cable from the battery to the distribution panel and split the system into a charge bus and a load bus over disconnectors at the panel, rather than inside the battery compartment. In some circumstances, an argument can be built for installing the BMS and disconnectors literally into the battery pack with all cell terminals inaccessible when there are any risks of third parties interfering with the battery. This type of construction makes bypassing the battery protection layer impossible without first resorting to tools and some dismantling.

Last Words

Over the last two years or so, I received a number requests to write this article. I resisted and deferred because I have been using my own BMS modules, but the focus here is on the technology.

This article now logically complements the construction of lithium battery banks and all the changes required in the battery compartment for transitioning from a lead-acid battery to a protected and operable lithium iron phosphate battery have been covered. Over a year ago already now, I was writing about the all-important aspect of redesigning the electrical system to operate with a lithium battery and a BMS and this now just leaves the interesting subject of charge control and charging the cells on board. Lithium batteries require full system integration to offer the best performance and longest life, something even high-cost commercial solutions have commonly overlooked. After a while, some of these vendors have had to face the longer-term consequences and cost associated with their engineering shortcuts that sacrificed battery life and pulled out of the market.

References:

[1] DNV GL Guideline for Large Maritime Battery Systems, DNV GL – 10/03/2014 No. 2013-1632, Rev. V1.0

This article is part of a series dealing with building best-in-class lithium battery systems from bare cells, primarily for marine use, but a lot of this material finds relevance for low-voltage off-grid systems as well.

Integrating a lithium battery bank on board a vessel introduces a few additional constraints and challenges that don’t exist with lead-acid batteries. Let’s consider two key statements:

A key difference between a lead-acid and a lithium battery is that the former can be damaged safely

While this may come across as provocative, it is nevertheless very true. Overcharging or flattening of a lead-acid battery is detrimental to its life. That’s about it. A lithium battery quickly gets totally destroyed and becomes a fire risk in the same circumstances.

Another main difference between a lead-acid and a lithium electrical system is that, in the second instance, the battery may become entirely disconnected from the installation, which can result in considerable damage

Protecting a lithium battery from damage may ultimately require isolating it from the system following a dangerous adverse event. A charge regulation failure or a complete discharge, for example, are such events. Unfortunately, there tend to be charging sources in marine DC electrical systems that are typically not designed to operate or cope without a battery in circuit in most instances: disconnecting the battery has a strong potential for causing malfunctions and sometimes considerable and very expensive collateral damage.

The battery is the base load in the charging system and is required to prevent the voltage from spiking up, sometimes considerably; many charge regulators cannot function or regulate properly without it.

In this article, we will discuss some avenues and options to design systems taking care of these aspects.

Disclaimer

A good understanding of DC electrical systems is needed to build and commission a lithium battery installation. This article is aimed at guiding the process, but it is not a simple blind recipe for anyone to follow.

The information provided here is hopefully thorough and extensive. It reflects the knowledge I have accumulated building some of these systems. There is no guarantee that it will not change or grow over time. It is certainly not sufficient or intended to turn a novice into an electrical engineer either. You are welcome to use it to build a system, but at your own risk and responsibility.

Basic Electrical System Design for Lithium

Due to the above considerations, the electrical system on board needs to conform with a model that allows battery disconnection without creating additional problems. In nearly all instances, alterations need to be made to the existing installation before a lithium battery bank can be considered. This assessment should take place before anything else.

There are absolutely no issues with electrical consumers on board; the voltage out of a lithium battery bank not only is within the range of what is experienced with lead-acid systems, but also exhibits less variation. A typical lead-acid system operates between 11.5V and 14.4V (less for gel cells). While the practical voltage range of a lithium system extends from 12.0V to 14.2V at the very most, the bulk of the cycling takes place between 13.0V and 13.4V only.

The challenge resides with charging sources and the risk of seeing them being disconnected, including under load, or even worse, directly feeding into the boat’s electrical system without a battery present.

Dual DC Bus Systems

Dual DC bus systems represent the optimal solution in reliability, resilience and functionality with lithium batteries:

• Power on board is not lost if an issue is detected with a cell reading excessive voltage. This can happen if a charger regulates poorly, cell imbalance is developing, or there is a system setup issue.
• A low-voltage disconnect doesn’t compromise recharging and the system has a chance to recover by itself.

This makes the dual DC bus topology very desirable on board marine vessels, but it also comes with higher engineering requirements.

The conversion of an existing installation to use a lithium battery bank with a dual bus system first entails segregating charging sources from electrical loads. Skipping this step is not really possible unless another (lead-acid) battery remains in circuit after the lithium bank is disconnected.

Twin battery disconnectors are at the heart of all dual DC bus lithium systems. Those are top-quality Tyco Electronics latching relays that offer zero standby consumption and a 260A continuous current capacity. The battery bank connects on the middle post, while the load and charge DC buses tie on the sides.

Creating a separate charge bus and load bus normally requires some rework of the heavy current cabling. Choosing a judicious location for the disconnector relays goes a long way towards minimising the impact of the changes. Electrical distribution is normally either carried out close to the battery compartment, or a feeder cable runs from the batteries to a distribution panel where the main positive and negative busbars are located.

Occasionally, marine electrical systems conform to another topology known as a Rat’s Nest. Those need to be pulled out before any further considerations

In essence, the positive busbar must be duplicated to separate charging sources from loads; the negative busbar normally stays as it is. The battery disconnectors are inserted close to this point to tie the bank into the system and any feeder cables normally remain unaffected.

The split DC bus configuration is the gold standard in terms of reliability and functionality for lithium battery installations. It is the preferred pathway for engineering elaborate lithium-only systems and for critical applications as it allows for specific and optimal responses to both excessive charge and discharge situations. Achieving this result requires capable equipment and good system design.

Controlling a dual DC bus system requires a BMS offering suitable outputs: this is not commonly found on solutions intended for electric vehicle (EV) conversions, which tend to rely on a single “disconnect all” contactor.

Attempting to build a dual bus system with an inadequate BMS all too often results in installations where both buses can (and therefore will, sooner or later) end up connected with no battery to charge; at this point, an unregulated charging voltage usually gets fed straight through into the boat’s electrical system, leading to a memorably expensive wholesale fry up. The ultimate in terms of the depth of thoughts afforded by the incident is when it happens at sea.

Key Challenges with Dual DC Bus Lithium Systems

It is fair to say that, today, a majority of DIY dual DC bus lithium systems contain critical design flaws their owners are often unaware of, or have decided to ignore because they could not solve them properly. This is often related to the use of some junk-grade or unsuitable BMS solution, carefully selected for no other reason that others have used it, coupled with a lack of design analysis.

A system is not good because it works, it is only good if it can’t malfunction or fail under any unusual circumstances

Dual DC bus systems come with two challenges associated with the potential disconnection under load of the charge bus or the load bus. A charge bus disconnect event is typically associated with a high-voltage event, while the load bus normally drops out due to an under-voltage situation at the battery.

Issues Associated with a Charge Bus Disconnect and Possible Solutions

In case of a high-voltage event causing a charge bus disconnection, charging sources can end up:

1. Disconnected under load, which can destroy some charging devices by causing their output voltage to spike; and
2. Subsequently linked together with no battery to charge, which can also result in damage due to excessive voltages for some devices. Many charge controllers require the presence of a large capacitive load (the battery) to operate correctly.

These two situations need to be analysed carefully and mitigated if required.

Typical examples:

1. A simple PWM solar charge controller switches the panels on and off rapidly to keep the battery voltage at a setpoint. The voltage varies very little because the battery absorbs the current while the panels are turned on. If the battery is removed, the open-circuit voltage of the panels is directly transferred to the output and injected into the charge bus: this means about 22V at times with the standard 36-cell panels used in 12V nominal installations.
   While this doesn’t really matter in itself and the controller can always take it, if other charging devices are also connected to the charge bus, they suddenly get exposed to that voltage that may prove excessive.
2. Many simple wind generators can be disconnected under load without getting damaged (as long as they don’t reach excessive speeds afterwards), but a very significant voltage spike can result, high enough to damage other electronic charge controllers that would happen to share the charge bus.
   High voltages also keep being produced at the output afterwards if the unit spins up. This is generally completely unacceptable.
3. Some modern wind generators can’t be disconnected at all under load, or their charge controller will be destroyed by the resulting voltage surge.
4. Some, but not all, MPPT charge controllers can fail from an output voltage spike if disconnected under (heavy) load. Good quality units use buck stages implementing cycle-by-cycle limiting and can in fact regulate their output even under no load.
5. Alternators nearly always fail with considerable damage to the rectifiers and regulator if disconnected under load. Interrupting the current causes a collapse of the magnetic field in the stator, which induces an intense surge, sometimes in excess of 100V.

The best and the simplest avenue, by far, would be using charging equipment that can be disconnected under load without issues and won’t output wildly unregulated voltages if there is no battery to charge. Unfortunately, this is not always practical, like in the case of alternators, or economics can favour trying to keep pre-existing gear: this is not always feasible, for a number of reasons, and can considerably increase the cost of a system conversion from lead-acid to lithium-ion.

Typical solutions to address these problems fall into three categories.

Disabling the Device in Advance

This involves turning off the charging device before it gets disconnected:

• Alternators can be disabled by interrupting the field circuit with a relay.
• Shore power chargers can be disconnected on the mains side.
• Wind generators often need to be diverted into a dump load or a short-circuit, which stops them.
• If concerns exist with solar systems, disconnecting the panels before the charge controller is an effective measure and normally always safe to do.
• Many externally-regulated wind generators are best disconnected (and short-circuited) before the charge controller as well.

In all cases, powering a relay or other disconnection device to disable a charging source is completely unacceptable. These systems must be fail-safe and not charge by default in the absence of control signal, so disabled charging sources can’t restart producing power after the battery has been disconnected and an additional layer of protection is created. This requires – for example – using relays with normally open (NO) contacts or bistable latching relays, so even a loss of control power can’t lead to a reconnection.

The best is often using fail-safe solid-state switching devices to minimise the current consumption while held on and maximise reliability.

In order to implement an advanced disconnection scheme, the BMS must support it and provide an adequate signal to act upon at least a fraction of a second before the DC charge bus gets isolated.

This can take the form of a “OK to charge” control signal and/or some kind of dedicated “charger enable” output, which would both get turned off long before a high-voltage (HV) protection event occurs.

Here again, junk-grade BMS products typically never offer such functionality and are therefore completely unsuitable to build such systems.

Individual Disconnection

If damage to other charge controllers is the main concern, disconnecting a device on its own is effective. This equates to giving it its own charge bus and disconnector. This can work very well for some unregulated wind generators, which are notorious for producing voltage surges and very high open-circuit voltages. Units featuring external charge controllers (in contrast with those equipped with built-in regulators) can be disabled by intervening upstream of the controller.

The drawback is the cost of an additional disconnector.

Absorbing/Deflecting the Surge

Another very effective option is ensuring that the current has somewhere to go following a disconnection: the output of a charge controller can be split over an isolator (diodes) and shared between the lead-acid starting battery and the lithium battery charge bus.

In this case, the presence of the lead-acid starting battery becomes essential to the safe operation of the system.

Not all charge controllers accept being wired this way however, because it effectively “hides” the battery voltage until charging begins. Some controllers draw on the battery to power themselves and operate in standby before starting to charge. Many wind generators fall into this category and simply refuse to operate when cabled this way.

More relevant information can be found further below under charge splitting, because the strategy can be, partially or wholly, applied to the charge bus of a dual bus system.

Issues Associated with a Load Bus Disconnect and Possible Solutions

Disconnecting the load bus presents no hazards at all as long as all loads connected are resistive and/or capacitive in nature. Loads falling outside this definition are inductive and therefore include electromagnetic devices like coils, motors and solenoids. The disconnection of a powered inductive load results in a reverse (i.e. negative) voltage spike (also known as back-EMF) produced by the collapsing magnetic field. The amount of energy released is proportional to the square of the intensity of the field, so the primary offenders are high-current devices like winches, windlasses or starter motors.

When the load bus is disconnected from the battery to stop further discharge, any energy surge released in the load circuit will potentiall reach all connected equipment on board from lights to electronics and these will be exposed to a brief, but possibly intense reverse-voltage pulse. A lot of marine electrical equipment is protected against reverse polarity connection and, up to a point, voltage surges, but the back-EMF from a large DC motor tripped under heavy load still has the potential to take out a lot of equipment on board.

Back-EMF Suppression

Suppression involves shorting out the spike at the source and it is very commonly implemented for small coils by the addition of a free-wheeling diode. A free-wheeling diode is wired to conduct from the negative towards the positive and therefore does nothing (blocks any current) in normal operation, but it clamps out the negative voltage spike to a value below 1V typically.

Suppression is best implemented as close as possible to the source by adding a diode across the terminals of the offending winding, but a limit exists to the amount of energy a diode can take in a pulse without getting destroyed. This energy is equal to W = 0.5 x L x I ², where L is the inductance of the motor or coil and I is the current at the time of the disconnection and, as large motors are significantly inductive and the energy increases with the square of the current, this approach is only really practical for small loads like relay coils or a fridge compressor DC motor due to the cost of very large diodes.

Disabling the Device in Advance

Disabling the device while the battery is still in circuit is here again a sensible and highly effective solution. It is best implemented as a low-voltage disconnect of the control circuit (i.e. control solenoid etc), which is low-power and low-current and therefore doesn’t require any high-capacity equipment.

This preventative action also has the advantage of potentially avoiding a low-voltage disconnect under high load with a general loss of power on board. While new, fresh, lithium cells have very low internal resistance and the voltage doesn’t sag much even under heavy loads, it increases over time and older installations become more susceptible to experiencing low-voltage disconnects under heavy loads when the cells are at a low state of charge.

Care must be taken to ensure that that this early action will always precede the disconnection of the load bus and the best and most reliable way to achieve this is getting the BMS itself to supply this signal. This eliminate potential conflicts between the reaction time of an independent low-voltage disconnect device and the BMS dropping the load bus in the event of a sudden and significant voltage drop.

Voltage Sensing

As long as a power source only charges the lithium bank, the reference voltage can normally be obtained from the bank.

The alternative is getting it from the DC charge busbar, which is the same, but upstream of the feed line and disconnector. The benefit is that it keeps reflecting the charger output voltage after a disconnection and can prevent over-voltage on the charge bus; the drawback is that it ignores the losses in the feeder cable and disconnector relay.

Many charging devices fall back on regulating their own output in the absence of a signal at the voltage sensing input, but this usually needs to be tested on vase-by-case basis if the installation is going to rely on it for proper operation.

These two strategies can be mixed and matched as required by charging devices, but the analysis needs to be carried out.

If a charge splitting strategy is used, then the corresponding guidelines apply to the chargers featuring a split output.

Simplistic Alternatives to the Dual DC Bus Topology

Building and commissioning a dual DC bus system can be demanding. It requires a good understanding of the behaviour and capabilities of the equipment used on board and some kind of “what-if” analysis must be carried out to ensure that simple unusual events are not going to result in serious malfunctions.

For these reasons, there appears to be no shortage of dangerous and irresponsible advice to be found under the KISS moniker when it comes to building lithium battery banks and installations. Let’s just say that, provided the cells have first been balanced, it always “works” – until something suddenly goes very wrong. Badly engineered lithium battery systems are still causing enormous amounts of electrical damage on board vessels, which typically doesn’t get reported back. I do hear about those however, quite regularly.

System design doesn’t lend itself to browsing around and averaging; it needs to be consistent and robust

Here, we will try and explore a couple of actually valid avenues to “simplify” the construction of a lithium system without creating additional risks.

The simplest way of resolving the issue of the disappearance of the battery in the electrical system following a safety disconnect event is… ensuring that a battery remains afterwards.

Two examples of simplistic, but safe and functional, topologies are provided below. In each case, we deflect and negate the problems instead of eliminating them at the source. While these schemes can easily be implemented successfully, they remain workarounds with some drawbacks and limitations.

There is no simplification down to the point of just dropping some lithium battery cells in a battery box

Regardless of the system design retained, all the charging voltages still need to be adjusted in order to stay clear of over-voltage problems at cell level and due care still needs to be taken not to overcharge the lithium cells.

The new battery also needs to be protected just the same, because of its different electrochemical nature.

Alternative 1 – Lead-Lithium Hybrid Bank

The simplest way of resolving all the challenges mentioned at the beginning of this article is running the lithium bank in parallel with some standard lead-acid capacity. If any issue arises with cell voltages or temperatures, the lithium bank can be disconnected and the installation will revert to a simple lead-acid system. In some instances, this lead-acid capacity could get damaged or destroyed if the event that resulted in the disconnection of the lithium cells was severe, like an alternator regulation failure.

The simplest safe lithium installation: leaving a sealed lead-acid battery in parallel with the lithium bank at all times allows disconnecting the lithium capacity in case of problem without any issues. The additional SLA doesn’t contribute to any meaningful capacity; its function is ensuring charging sources always see a battery in circuit.

The practical result of such an arrangement is that the lithium battery ends up doing virtually all the work, because it is first to discharge due to its higher operating voltage. The charging voltages are no longer high enough to provide effective charging for the lead-acid cells, but as those are being trickle-charged above 13V all the time, they can be expected to remain essentially full and it hardly matters.

The lead-acid battery needs to be able to absorb whatever “unwanted” current may come its way if the lithium bank gets disconnected due to a high voltage event for example. In some instances, a single sealed lead-acid (SLA) battery can be sufficient. SLAs are the best choice for this application as they don’t consume water and are very inexpensive; gel cells should be avoided as they are costly and a lot more intolerant to overcharging and AGMs would be a complete waste of money in this role.

The drawbacks are:

• Some charge gets lost trickling continuously into the SLA, more so in a lead-acid battery in poor condition.
• It doesn’t fully eliminate the lead and associated weight.
• Removal of the SLA from the system, at some point in the future, would create an unexpected liability.

Some advantages are to be found as well:

• Disconnection of the lithium bank can be managed with a single contactor; there is no need to implement a split bus. This can allow using some small BMS solutions incapable of managing a dual DC bus.
• The lithium bank is literally added to the installation in place, normally without cabling alterations required, but not without voltage and regulation adjustments.

With this in mind, it certainly is the simplest fully functional design one can build, as long as protection and automatic disconnection are still very properly implemented for the lithium bank.

Should the lithium bank ever become heavily discharged, the additional lead-acid capacity can start contributing, but this would also leave it at a reduced state of charge for a time afterwards and cause it to start sulphating. This is not automatically much of a concern, because it may not happen (this depends on the BMS low-voltage disconnect threshold) and it doesn’t actually result in much harm if it does. The SLA needs to remain in a reasonable condition however, in order to be able to absorb any transients if the lithium bank gets dropped off due to excessive voltage and not continuously discharge the lithium cells at an excessive rate.

Voltage Sensing

NEVER, EVER, SENSE THE CHARGING VOLTAGE DIRECTLY AT THE LITHIUM BANK TERMINALS IN THIS CONFIGURATION

The sensing voltage required for charge control must be sourced upstream of the lithium battery disconnector, or in other words from the SLA battery, so it remains valid even after a disconnection of the lithium capacity. This is very important, otherwise uncontrolled, unlimited charging of the lead-acid battery will occur after the lithium capacity gets isolated.

Alternative 2 – Split Charging

Considering that, in most instances, good system design practices lead to keeping a separate SLA battery for starting the engine, one can be tempted to derive similar benefits from it, instead of carrying one or more additional SLAs as required by the Lead-Lithium Hybrid topology.

Charge isolators are extremely useful devices for building lithium battery systems and can be found in a variety of configurations, 1 or 2 inputs connected to 2 or 3 outputs. They are extremely rugged and robust. The best ones all seem to be manufactured in the USA: Sure Power Industries, Hehr and Cole Hersee are all excellent sources for quality units. Inferior products generate considerably more heat.
If efficiency is a key concern, isolators using MOSFET transistors instead of diodes are available, albeit at significantly higher cost.

Using a charge isolator (also known as blocking or splitting diodes) can provide at least a partial solution, depending on the nature of the charging devices present. It is a good option with alternators and any chargers that don’t need a voltage originating from the battery to begin operating.

Since most of the electrical issues with the integration of lithium batteries in traditional marine systems arise with battery disconnection, splitting and sharing a common charge bus with the engine starting SLA battery is a very simple and effective way of addressing the matter.
Unfortunately, some battery charging devices refuse to operate behind an isolator; this prevents adopting this configuration as a universal solution, but it is nevertheless valuable.

Alternators and unregulated/crudely-regulated wind/tow generators are usually happy to function this way behind a diode. Internally-regulated generators commonly refuse to start unless they can “see” the battery voltage, because they require a small amount of power to first “release the brake”.

If this configuration can be achieved, then again the lithium bank can simply be dropped using a single disconnector without any ceremony, should some adverse event occur. One side-benefit is that the charging systems feed both into the lithium bank and the start battery, even though the voltage isn’t ideally quite high enough for the latter. This can be remediated by the addition of a small dedicated charger for the lead-acid battery, either solar or through step-up DC/DC conversion from the lithium bank.

Note that the charge bus still feeds into the positive bus after the lithium bank has been disconnected. The voltage from the charge bus is limited by regulation and the presence of the lead-acid battery, but the power quality may not be adequate with possible brown-outs. Also disconnecting the feeder line to the distribution panel in a battery protection event is one way of remediating this.

In such a configuration, it is very important that the lead-acid battery always remains present in the charging path. A battery switch to isolate the engine circuit is fine and desirable, but the charge isolator(s) should remain directly connected to that battery at all times to provide a pathway to dissipate any surge, as well as a nominal base load for the charge regulators.

Voltage Sensing with Charge Isolators

Any serious charge controller comes with a battery voltage sensing input. When the charger output is split to charge multiple banks, this becomes even more important as any losses over the charge isolator must be compensated for and a quandary always arises as to where to source the charging reference voltage.

Accurate battery voltage control is only going to be achieved for the battery being sensed, because there are voltage losses proportional to the current in charging systems. With a lithium bank in the system, sensing should reflect the voltage of the lithium bank and this will result in best performance for charging it; this is usually the desired outcome.

NEVER, EVER, SENSE THE CHARGING VOLTAGE DIRECTLY AT THE LITHIUM BANK TERMINALS IN THIS CONFIGURATION

Voltage sensing for a lithium battery in a split-charging topology must be performed at the output of the charge isolator, upstream of the battery disconnector, so disconnection of the battery doesn’t dissociate the sensed voltage from the charging voltage altogether: this would otherwise lead to uncontrolled, unlimited overcharging of the remaining lead-acid batteries in the system.

Voltage sensing can sometimes be performed at the input terminal of the charging isolator instead, for some equipment such as alternators typically. In this case, the charging voltage adjustment must be made for the lowest voltage drop that can be experienced over the isolator. This is normally about 0.3-0.4V for Schottky diode type units and essentially zero if a MOSFET-based isolator is used instead.
The difference in system performance is subtle and yields a less aggressive charging characteristics with lithium cells in particular.

General Electrical Installation

Fusing & Feeder Cables

A heavy-duty fuse should normally be found very close to the bank to protect the feeder cables. This fuse should be sized so it will never blow unless an intense short-circuit occurs, or it may create the potential for at least accidentally destroying the alternator, and often much more.

ANL fuses are cost-effective, easy to source and can offer interrupt ratings up to 6kA at 32V, but some are only good for 2kA.

The nominal current capacity of a fuse reflects the current it can conduct indefinitely without blowing. Currents above this value will cause the fuse to heat and eventually blow; the time it takes for this to happen is related to the ratio of the over-current and can range from minutes or more to milliseconds.

The interrupt rating of a fuse is considerably higher than its current capacity and defines how much current the fuse can successfully interrupt by blowing; values beyond this figure may result in continued arcing over the fuse after it has blown. The interrupt rating is very voltage dependent, for obvious reasons, and increases significantly at lower voltages.

Unless the feeder cable leaving the battery compartment is of an exceptional size and the battery bank is very large, a common low-voltage ANL fuse with an interrupt rating of 6kA at 32VDC is normally adequate. There is too much resistance in the cells, connections and cables to sustain the hypothetical currents (and associated apocalyptic predictions) that would supposedly arise from a short-circuit.

For a 13.3-volt source to supply in excess of 6000A, the total circuit resistance would need to be below 2.2 milliohms. Small lithium battery systems of interest for pleasure crafts normally fall short of such capability simply due to the size of the cabling used and number of bolted connections involved.

In the case of larger installations, a proper prospective fault current calculation should be carried out and the fusing should be selected to match the required interrupt rating.

Class T fuses offer much higher interrupt ratings (20kA) than the common ANL fuses and can become necessary to protect the feeder cables in large lithium battery bank installations.

The feeder cables should be sized according to the maximum acceptable voltage drop they can induce under normal operation. Quite often, alternator charging currents and inverter loads represent the maximums the installation can be expected to see.

Using unreasonably heavy cables or seeking negligible voltage drops at peak current also increases the maximum prospective short-circuit current the installation can produce and results in a higher level of risk. The cables need to be able to hold until the fuse blows and, until then, their resistance is precisely a good part of what limits the fault current: it pays to keep this in mind and take advantage of it.

Common Negative

In the case of a system with more than one battery bank – a very common configuration due to the presence of at least a starting battery – it is usually wise and sensible to tie all the negatives together, because it simplifies the integration of any device connected to more than one bank.

If charge splitting is to be used one way or another, then a common negative to these battery banks is mandatory.

Battery Sensing

Battery Voltage

If not already present, a dedicated battery voltage sensing cable with its own small fuse at the battery end should be run from the source of the sensing voltage, which often is not at the battery itself, to wherever the charging equipment is/will be located. All voltage sensing can then be consolidated onto a dedicated terminal block rather than having multiple wires all running back to the same location for an identical purpose.

A great deal of damage and destruction can result from sourcing the charging reference voltage inadequately in an installation with a lithium bank

Where the voltage sensing cable should be connected in the system depends on the topology of the installation and the subject was discussed on case-by-case basis earlier.

Battery Current

Many systems also include a current measurement shunt associated with a random number generator battery monitor or amp meter. The shunt is almost always found on the negative side, because it is technologically simpler and cheaper to measure the current there. Run a twisted pair cable from the shunt block directly to the measuring instrument.

Other than for the negative voltage sensing core and any BMS wiring, there should be nothing else than the lithium bank connected to the battery side of the shunt. This includes the negative of other batteries, such as a starting battery: failure to observe this will result in the current of the other batteries to also be measured, when it shouldn’t.

Temperature Sensors

Any battery temperature sensors associated with charge controllers and pre-existing lead-acid cells must be disconnected from all charge controllers and removed altogether. Some controllers may signal a fault as a result, but normally keep operating assuming a default constant battery temperature: this is exactly what we want. Occasionally, an ill-tempered controller may refuse to operate without its temperature sensor. Most temperature sensors are 2-wire negative temperature coefficient (NTC) thermistors (resistors whose value is temperature-dependent). Measure it at ambient temperature with a multimeter and replace it with an approximately equivalent fixed resistor (the nearest standard value will do) at the controller terminals.

This aspect is in fact part of the integration of lithium batteries with other equipment, but as the task of removing the sensors takes place within the battery compartment, it seemed logical to include it here.

Temperature sensors have their place in a lithium battery bank, but they are part of the battery protection circuitry and completely unrelated to the charging voltage. Lithium batteries in marine installations should always operate within a degree or two from ambient temperature, without exhibiting meaningful differences between cells.

Battery Switches

Simple heavy-current battery switches are a much better choice than combining switches with lithium batteries, as paralleling of batteries is usually most undesirable.

On dual DC bus systems, it is highly unadvisable to leave or install a battery master switch in the feed line between the batteries and the bus disconnectors. The correct way of achieving battery isolation is by opening both the charge and load bus disconnectors, which is a function that is normally provided by the BMS; failing to observe this point would again result in removing the battery while leaving both buses linked together as described earlier.

The only acceptable function for a manual battery isolator switch is turning the power off to the vessel, i.e. disconnecting the load bus.

If complete manual battery disconnection is desired, then either two single-pole battery switches or a 2-pole switch must be used to isolate both positive buses. Some analysis must be carried out to determine whether leaving the charging sources tied together at the “floating” charge bus with nothing to charge could result in equipment damage or not.
While the BMS may be able to provide “advanced notice” of a charge disconnect and turn the chargers off, a manual disconnect typically won’t.

Paralleling Switches

Paralleling batteries is a concept that evolved from trying to crank diesel engines with proverbially flat lead-acid batteries. One good engine starting battery is all it takes to do the job. Unless the engine is truly large, a single battery is normally ample, and more is just dead-weight.

If either the lithium or the lead-acid battery is heavily discharged, closing a parallel switch can initially result in an intense discharge current, with a risk going towards the cabling and the lead-acid battery due to the formation of explosive gases.

Systems including isolated banks of each type normally also include provisions for charging the lead-acid capacity properly (i.e. at higher voltages, using a temperature-compensated voltage and float-charging) and this makes the paralleling switch a very dubious proposition, because it exposes the lithium cells to a completely inadequate charging system. The fact that you “won’t leave the paralleling switch on” only means that it will happen anyway, sooner or later, because it can.

On a dual DC bus system, there is also the question of where to connect the switch: the tie-in can typically both consume and supply energy and it can only be cabled to either the charge or the load bus, leaving the system vulnerable to discharge through the charge bus, or overcharge through the load bus afterwards.

I personally prefer having the option of using jumper cables if ever warranted, rather than creating an unnecessary and permanent liability by having a paralleling switch in a dual DC bus installation.

Simple systems that don’t feature a dual DC bus can actually be designed with a paralleling switch, but it must join past the lithium bank disconnector relay, not on the battery side. This ensures that the BMS can break the parallel link if trouble is coming from there. Regardless, it is still a bad idea.

Voltage-Sensitive Relays (VSR)

Voltage Sensitive Relays (or VSRs) are always poor solutions in marine electrical systems and, at best, next to useless with lithium batteries. The one depicted above, with a cut-in voltage of 13.7V and a cut-out threshold of 12.8V, would essentially remain closed until deep discharge has occurred.

Voltage-sensitive relays are another plague of modern marine electrical systems. They gained ground after people experienced issues with diode-based charge isolators due to the voltage drop they induce and because VSRs are seemingly easier to deal with and understand.

Each battery bank has it own state of charge and needs in terms of charging profile. Paralleling banks together is never a great idea, even when the batteries are of the same type and require the same voltages.

Some VSRs sense the voltage on one side only, others on both; some offer adjustable thresholds and others not. Unless the unit is fully adjustable and includes both low and high voltage disconnection points, it is normally completely useless (and equally harmful) around lithium batteries.

Forwarding a charging voltage from a lithium bank to a lead-acid battery won’t result in a good charge characteristics. Doing the opposite requires observing both a connection and a disconnection voltage threshold, because lead-acid battery charging reaches excessive voltages. The resulting charge characteristics for the lithium battery is typically not good either, because no absorption time can be provided. It keeps getting worse: should one of the banks become heavily discharged, closing of the VSR can easily result in a sustained discharge current way beyond its current capacity, leading to some catastrophic failure.

On dual DC bus systems, VSRs normally bring all the same issues as paralleling switches: there is no correct place to wire them in and they have no place there.

Regardless of brand or type, VSRs never seem to lead to any good solutions in systems with both lithium and lead-acid cells. Fortunately, there seems to be an endless queue of ill-inspired people keen to buy them and this makes them very easy to get rid of.

The best answer to charging auxiliary engine starting SLA batteries is using a battery isolator, if an alternator is present, and DC/DC chargers from the lithium bank (or an auxiliary solar panel) to ensure full charge can be reached. The installation can then simply be configured to charge the lithium bank optimally.

Engine Starting Batteries

Internal combustion engines can be cranked with LiFePO₄ batteries, very successfully at that, and even when the battery is low on charge, within reason: a lithium bank down to 3.0V/cell can struggle to crank a diesel. There are however a number of good reasons for not doing it when the vessel is large enough to sustain a dual bank installation:

• Redundancy and the ability to still start the motor with a discharged house bank are lost.
• Unless the lithium bank is huge and a current of some 100A means little, engine cranking still causes the voltage to sag at the battery and creates transients in the system.
• Lithium batteries are harder on engine glow plugs, because they supply a higher voltage under load.

Unless low weight is everything, using a lithium battery as a separate starting battery is possible, but usually not sensible:

• A SLA purely used as a starting battery is very easy to keep at full charge and commonly lasts 8 years or more on a marine vessel. A very small solar panel can be dedicated to floating that battery at the appropriate voltage if needed.
• The comparatively very high cost (and added complexity) of a lithium battery in this application cannot be justified.
• A lithium starting battery should be kept at about 50% SOC in order to age well; it introduces a new lithium charge control regime in the system.
• As highlighted earlier, there are often technical benefits to be found in still having a SLA in the system and dedicating one to cranking the engine is a good use for it.

Next Steps

Once the new battery bank has been balanced, assembled, protected and installed in an electrically correct configuration as described above, it needs to be integrated with existing charging equipment.

Due to the large variety of gear found on the market, with hardly any of it ever intended or properly designed to charge lithium batteries, chargers require a lot of attention in order to function without tripping the high voltage protection limit or overcharging the bank over time.

The subject is extensive enough to be treated separately.

This article is part of a series dealing with building best-in-class lithium battery systems from bare cells, primarily for marine use, but a lot of this material finds relevance for low-voltage off-grid systems as well.

Here, we detail the hands-on process of building a lithium battery bank from individual single prismatic cells. There is more to it than just arranging and connecting the cells, because those can only be assembled into a battery after they share a common state of charge. They also need to be protected before anything can be done with the battery, which is the object of another article.

Before that, preliminary decisions also need to be made: how much capacity to install and what cells to source? What is the most suitable interconnection scheme to adopt?

A 200Ah DIY lithium battery back for a yacht, balanced and instrumented for cell voltages and temperature. A standard 12-pin plug connector provides the interface to the battery protection module.
Cell clamping arrangements can be very simple and effective.

Buying cells and assembling the bank is not the beginning. Learning about lithium cells and understanding their properties and their risks is, before committing to building anything.

As it is an extensive topic in itself, the integration of a lithium battery on board is also dealt with separately.

Disclaimer

A good understanding of DC electrical systems is needed to build and commission a lithium battery installation. This article is aimed at guiding the process, but it is not a simple blind recipe for anyone to follow.

The information provided here is hopefully thorough and extensive. It reflects the knowledge I have accumulated building some of these systems. There is no guarantee that it will not change or grow over time. It is certainly not sufficient or intended to turn a novice into an electrical engineer either. You are welcome to use it to build a system, but at your own risk and responsibility.

How Much Capacity?

Generally speaking, a LiFePO₄ bank will offer about twice the usable capacity of equivalent deep-cycle lead-acid cells in good condition, and much more when such lead-acid cells have deteriorated. This can provide a rough guideline when considering the purchase of lithium cells. In practice, it only suggests the maximum capacity that should be considered as a starting point: no more than 50% of the lead-acid capacity.

In the traditional lead-acid way of thinking, more capacity meant smaller cycles and longer life and a justification was found there: the situation is almost the exact opposite with Li-ion batteries

Many lithium banks installed on yachts nowadays are in fact not only much larger than they need to be, but also much larger than they should be.

The oversize bank approach can in fact deliver less value: there is nothing suggesting that a bank twice as large will last twice as long: it will more than likely just result in twice as many old buggered cells at the same point down the track if not earlier. The first consequence of installing an oversize battery bank, especially when sustained charging is involved as with solar panels, is that the bank remains at a higher state of charge much longer, if not most of the time. This is very detrimental to its ageing for reasons that were developed earlier. Lithium cells like cycling because it means they don’t spend any amount of time near full; alternatively, they can sit happily half-discharged, or even lower, for years.

Invest in energy efficiency or charging capacity, not in unnecessary storage

The bank needs to be large enough to provide the capacity needed between recharges, but beyond that, all what comes out needs to go back in and the size of the battery makes no difference there. Money is best invested in energy efficiency on board and charging capacity than storage.

The question therefore revolves around the cycle duration that must be accommodated. A yacht spending all its time in the tropics with considerable solar supply available on a daily basis doesn’t technically need to store much more than its overnight consumption, strictly speaking. The ability to accommodate a 2-day or 3-day cycle may be valuable however, but this calls for adapting the management of the battery to suit. Consumption can also be reduced in adverse conditions, extending cycle duration and this is a sensible way of looking at the matter, compared to calculating everything on maxima and worst-cases.

In practice, lithium banks of about 200Ah are easily capable of supporting yachts with an electric refrigeration system and auxiliary loads in the mid-latitudes and it is very difficult to present a valid case for installing more than 300-400Ah on a sensibly outfitted pleasure craft. Some, however, are fitted out and operated as if they were permanently tied to the power grid.

Some of the installations I built and commissioned included a provision for expansion by adding an extra set of cells later if needed, in order to alleviate the owner’s concerns. None of them were expanded afterwards

While a lithium battery bank can easily be expanded by adding more cells later if needed, unneeded capacity cannot be returned for a refund. Best long-term value is achieved when both the installed capacity and the management of the installation are correct and adequate.

Sourcing Cells

Manufacturers

Those are all common cells on the market today: the CALB SE-series in blue and CALB CA-series in grey (now identical other than for the casing). Sinopoly cells are black and Winston cells are yellow.

There are many manufacturers of LiFePO₄ prismatic cells, mostly located in China, but the only well-known ones are those imported and available in the Western countries. Some smaller players like Hipower and Thundersky have disappeared. Some of the oldest names in the game today are Sinopoly, CALB (China Aviation Lithium Battery) and Winston, the latter having had a troubled history in recent years. Short of having a significant amount of time and access to a lab, it is very difficult to differentiate these products from a quality point of view.

Sinopoly and CALB operate their own research and development labs. CALB in particular has also established a very strong reputation for product quality control with each cell being measured and labelled with its actual capacity before being shipped. Yet, issues with CALB cells are not unknown to occur. Winston has been making reliable and long lasting cells for a very long time. In spite of being virtually unknown, Lishen also makes very good cells, which were selected by a large customer in Switzerland following lab tests, ahead of the better known brands.

While it is often possible to source unusually cheap cells with obscure brand names, such bargains might not represent long-term value. The ageing behaviour of the cells is extremely dependent on the quality of its manufacture and trade secrets associated with electrolyte composition and, in this regard, even the best known brands are not all equal.

Cell Sizes

Single 3.2V prismatic LiFePO₄ cells can nowadays be obtained in huge capacity, as high as 10000Ah. Commonly available cells range between 40Ah and maybe 1000Ah. It should be pointed out that the larger sizes are intended for stationary applications where no accelerations, vibrations or shocks are ever experienced.

A sales manager at Sinopoly I was talking to was adamant about using 100Ah or 200Ah cells only for assembling marine battery banks, with 100Ah being preferred and 200Ah acceptable. Large cells simply don’t have the structural strength-to-weight ratio required to be taken to sea on board small crafts and would exhibit shortened life due to internal mechanical damage arising from on-going vessel motion. It is common sense: as a cell becomes larger, its internal weight increases much faster than the rigidity and surface area of the casing and the casing is all what holds the plates together in a prismatic cell.

Failures have been reported on vessels equipped with 700Ah cells following ocean passages: some cells were suddenly found to be losing charge inexplicably, rendering the battery bank completely unmanageable and the matter ended in a complete write-off. All big-brand commercial marine lithium battery packs on the market today are built from cells no larger than 200Ah.

While there certainly are examples of marine DIY systems that were built with large cells in series without issues, closer inspection usually also reveals a houseboat usage or infrequent good weather, sheltered waters sailing. In other words, the data point is null and void if the intent is sailing and designing upon the assumption that the boat won’t be going anywhere would be questionable.

Physical cell dimensions, space availability on board and interconnection topology are the other factors that influence the final choice of cell model. 200Ah cells are usually taller and require more “headroom”.

Condition Check

As much as possible, when sourcing cells from a local agent, I try to physically go there and check the cells as they come out of the crate. I normally decline buying cells that are no longer factory-packaged and may have been tampered with.

A set of brand new Sinopoly cells just out of the factory crate. All are reading within less than 1mV and their state of charge is just over 40%.

• I ask for cells from the same production batch, with consecutive serial numbers. Those should hopefully exhibit more consistent characteristics than randomly chosen cells.
• All cell voltages must read below 3.300V. Pay attention to multimeter calibration there, there is a vast difference in terms of state of charge between 3.31V (over 75% SOC) and 3.29V (less than 45% SOC). This is to ensure that I am not getting cells that have been sitting around at a high SOC. This is not normally a problem with factory-shipped cells, but more caution applies with cells on the retail market.
• All cell voltages must be very closely matched. I like to see differences of 1mV or less, but sometimes accepted up to 2-3mV. At the SOC cells ship at, there is no justification for deviations in voltage, which could indicate a defective cell or prior tampering.
• Obviously, no cell must show unusual signs of physical use or prior connection. All cells are connected, charged and discharged at the factory following manufacturing, so there is no reason for any of them to appear any different, unless the cell is in fact second-hand.

If I place an order and cannot physically check the cells myself prior to purchase, I explicitly state all these conditions in writing with my order, so they become contractually binding if the order is filled. It can go a long way with eliminating the temptation to slip a “perfectly good” second-hand cell in a batch to get rid of it, knowing that returning it would be a major hassle for the buyer.

Warranty Conditions

Prior to purchase, I also get a warranty statement from the supplier. While warranty is usually limited to one year, this should cover any problems arising from major manufacturing issues.

Warranties on lithium battery cells are tricky, because the cells can easily be damaged through misuse and suppliers know that only too well. Chances of making a successful claim for a ruined bank or on an installation where cell-level protection didn’t exist would be near-zero (and rightly so), but it would be very difficult for a supplier to push back in the case of a single-cell failure on a properly engineered system.

In some countries, warranty clauses offered by vendors in general deliberately conflict with applicable consumer protection laws, so a one-year warranty doesn’t automatically mean that all bets are off after 12 months.

Pre-Balanced Packs

Some resellers sometimes offer “ready-to-go” balanced cell packs. While this could appear simpler than having to carry out cell balancing, the large amount of uncertainty existing around how the cells were balanced and treated prompts for extreme caution. These packs may have been exposed to excessive voltage and then left fully charged for extended periods of time, which makes them rather undesirable to own: the cells are already damaged.

The considerations about cell balancing further below contain all useful information required to validate the process used by the vendor, should one ever be tempted to go this way.

Cell Links

Consider sourcing cell links and stainless steel bolts in the same time as the cells. Cell manufacturers nearly always offer those. Use solid copper links in marine installations. Braided straps, such as earthing straps, even tinned, are not a good idea. They have a lesser cross-section than a solid conductor and will not age as well in the marine environment. They are bound to corrode and heat up severely one day.

Copper cell interconnection link. Those are readily available from battery manufacturers and resellers.

Alternatively, source 40 x 6mm (1 ½ x ¼’’) aluminium flat bar, cut it and drill it to suit. Sand the contact areas bright to remove the thin oxide skin. If using DIY links, consider insulating the sections between cell terminals using heat shrink tubing; it will greatly reduce the risks of causing an accidental short while working around the cells afterwards. Such home-made long links are particularly effective when dealing with blocks of cells in parallel and work out much cheaper than copper single links.

In all cases, the bolts must be long enough to thread deep into the cell terminals and be fitted with locking washers.

Transportation Considerations

Shipping crate for lithium cells

Shipping of lithium-ion batteries currently falls under very restrictive rules as they are classified as Dangerous Goods UN3480 Class 9. This determination can significantly increase freight costs and makes air freight essentially impossible today (2016).

Sea freight costs, on the other hand, are usually calculated on a minimum quantity of 1 cubic metre or 1 metric tonne (break bulk or LCL) and obtaining a good honest quotation for a small one-off shipment can be more than problematic at the best of times. Unless the order is large enough to approach 1 cubic metre, it is typically uneconomical to consider international sea freight, from China typically, because of the multitude of fixed processing fees and charges associated with landing and clearing the cargo.

The most practical pathway for sourcing small numbers of cells is often going through a company already importing such batteries for a purpose or another, such as electric vehicle conversions.

Battery Bank Topology

Once the system voltage and intended capacity have been established and a source/manufacturer has been identified for the cells, the topology of the bank can be determined according to cell size.

Electrical Interconnection

The principle is always the same: a 12V nominal system requires four identical blocks of 3.2V nominal cells, and a 24V installation requires eight. Each one of these blocks must offer the capacity sought after. Cells in the 100Ah to 200Ah range are relatively small building blocks and assembling larger banks requires creating parallel configurations.

Cell terminals and link plates must be sanded clean and bright prior to assembly: high resistance connections immediately result in hot spots at high current with the heat flowing straight into the cells. Connections should always be very tight for the same reasons.

In its simplest expression, a 12-volt lithium bank is built out of 4 cells connected in series; this is also the safest configuration. If more capacity is required, two main options are available in terms of architecture and interconnection schemes.

Parallel First, Then Series

The most common and simplest scheme is creating parallel blocks of cells of the required capacity, and then linking them in series to reach the voltage sought.

A 200Ah LiFePO₄ bank can be assembled using four 200Ah cells connected in series, or four groups in series of two 100Ah cells in parallel. The first topology would be referred to as 4S (four in series, figure A below) and the second as 2P4S (two parallel, four times in series, figure B below).

Cell interconnection schemes for 12V systems. “A” represents a 4S configuration, “B” a 2P4S arrangement and “C” is the same, but fused.

The main advantage of these configurations is that they minimise the complexity of the protection required. It is also very easy to physically interconnect cells this way. The drawback of configuration B is that, should one cell fail by shorting internally, the others connected in parallel will discharge into it, potentially aggravating the situation.

This introduces a low, but additional, risk into the system that doesn’t exist with a pure series interconnection scheme as in figure A. Connecting cells in parallel to achieve large capacities is very commonly done however, even at industrial scale in stationary installations.

Fusing Individual Cells, or Block of Cells

A variant on the parallel blocks scheme of figure B is fusing some or all of the parallel cell links (figure C above). The challenge resides in sizing the fuses as small as possible, while still large enough to carry the normally expected currents without undue voltage drop and risk of blowing. We will note that in the case of configuration C, the fuses should never see much more than half of the bank total current. Fusing can’t prevent good cells from discharging into a faulty cell, it can only prevent them from heavily discharging into it, so the outcome and effectiveness of such schemes is uncertain.

On board marine vessels where loads such as inverter and windlass commonly draw in excess of 100A, the fusing requirements can be placed so high that they undermine the value of such arrangements for small banks.

In larger parallel interconnection schemes, such as this 3P4S configuration, individual fusing of the cells becomes increasingly important and provides a degree of fault tolerance.

The more cells connected in parallel, the higher the amount of energy available for heating a failed cell uncontrollably, but the smaller the individual cell currents. This can make individual cell fusing schemes more effective for larger installations. In the 3P4S configuration in figure D, each cell only contributes to one third of the total current and failure of a cell fuse doesn’t immediately compromise electrical supply to the vessel.

Large vessels can use banks comprising 8 to 10 cells in parallel in each block and then individual fusing can become very effective.

Parallel Banks

Deviating from parallel group topologies leads to building completely separate banks then connected in parallel. This requires a complete duplication of the protection/management system, but can be justified.

An active protection scheme with two independently managed 4S banks provides both redundancy and the highest degree of protection, but at the cost of duplicating the management system.

The approximate 200Ah physical cell size limit determines the capacity of each individually protected pack if no parallel discharge risk is the goal. This is the way commercial marine lithium offerings are usually constructed, as it minimises associated liabilities. A cell failure causes disconnection of the associated pack and the only energy involved is the one contained within the failing cell.

Summary

A simple 4S configuration (diagram A) offers both simplicity and maximum safety for a 12-volt nominal system. It allows building systems up to 200Ah.

The majority of the DIY lithium battery banks built to date have used the parallel+series configuration (figure B), occasionally with partial fusing as shown in diagram C. These topologies are not uncommon in large stationary installations either. At the time writing, I am not aware of any incidents arising from isolated cell failure within a bank on a marine DIY system. This doesn’t mean it couldn’t possibly happen.

Prospective owners of very large lithium battery banks should seriously consider using individual cell fusing, as shown in figure D, or going to multiple parallel banks as depicted in diagram E. The large number of cells increases the chances of seeing an isolated failure and the small size of the cell in relation with the bank suggests greater potential effects.

Production automotive battery packs are commonly made of very large numbers of small cells and typically fused as per figure D, and also broken up in separately managed and protected blocks. Those are usually connected in series afterwards to obtain high DC voltages, which is out of scope here.

Configuration E is arguably the best when it comes to minimising risks and maximising reliability while achieving a larger target capacity. It is more costly due to the duplication of the battery protection equipment and high-current disconnectors, but in the context of building a large lithium battery bank, the cost of protection should be seen as small.

Using larger cells in order to remain with a simple 4S configuration while achieving higher capacity would probably constitute a very dubious choice. Cells suffering internally from physical stresses and damage are much more likely to fail and short out than smaller, more robust cells simply connected in parallel.

Mechanical Installation

It was once thought prismatic cells could be operated in more or less any position as they do not really contain free liquid. Nowadays manufacturers are a lot more prescriptive with installation position. In most instances, the only acceptable position is upright, vent cap and terminals on top (Sinopoly, Winston). Sometimes it may be acceptable to mount them on edge, with the terminals on the side (CALB). This may vary not only between manufacturers, but also between cell models, so seeking specific guidance is a sensible step if an odd installation position is being considered.

When questioned, Sinopoly indicated that installing the cell in any other position than upright would cause some of the plates to run dry after a while, damaging it. Installing them flat on their side is out of the question in all cases.

The cells must be installed securely in such a way that no movement is possible in relation with each other, or it will stress the terminals and link plates. Prismatic cells should also be clamped together between compression plates as the application of a modest amount of pressure helps with preventing electrode delamination, even more so in the presence of shocks and vibrations as found on marine vessels. It also helps with preventing the internals of the cells from shifting in case of violent shock, which can lead to internal cell short-circuits. Clamping is a common warranty condition from manufacturers. Strapping the cells together is simply not good enough for that matter.

The bank must also be installed in such a way that it can’t shift and nothing can come and short-circuit the cell terminals. This can involve fitting a cover over the cells.

Location

Ambient Temperature Considerations

Lithium batteries age at an accelerated rate and degrade very quickly at high temperatures. For this reason, installing a bank in an engine compartment is completely out of the question. Ambient temperatures in the battery compartment should not exceed 30°C.

Conversely, exceedingly low temperatures can lead to temporarily reduced performance and capacity on discharge and cell degradation during charging. Marine house batteries are not normally operated at very high currents, but charging below 0°C is an issue that can arise for some vessels in some areas and needs to be prevented, especially at high currents.

Volume within the accommodation space and below the waterline is often the most suitable in terms of ambient temperature conditions for housing a lithium battery bank. Vessels operating in polar waters or facing harsh winters may require special dispositions ranging from heating the battery compartment to disabling charging.

Shocks and Accelerations

Prismatic cells are made of thin plates stacked together within a semi-rigid plastic housing. The stack itself hardly has any structural strength other than in compression. The edges of the plates are weak and can be prone to damage if the cells are exposed to violent impacts. Installing prismatic lithium cells into the bow section of a marine vessel is out of the question, no matter how tempting it may be to power a windlass. Wound cylindrical cells would be far for robust for that matter, but the battery assembly containing a large number of such small cells may not be. Shaking it loose over time could turn the battery into a fire risk.

A lithium battery bank should be installed aft of midships typically, in the most comfortable part of the vessel and the cells must be firmly clamped as discussed earlier. In the case of offshore vessels, the prospect of falling off a wave in heavy weather cannot be entirely excluded, hence the importance of selecting cells of modest size and weight for building marine battery banks.

Measuring Cell Voltages

Before moving ahead with building a lithium battery bank and balancing cells, make sure you have access to a good quality, calibrated digital multimeter: cheap, junk-grade instruments are little else than voltage-inspired random number generators. It should read at least within 10mV of the true voltage in the 3 to 4 volts range with perfect repeatability and regardless of changes in ambient temperature.

Many of the “marinised” multimeters I have come across over the years were out by 0.1V or worse. If you happen to own one of those, complete with the proverbial bent probes or broken leads, do yourself a favour and place it carefully in a rubbish bin if you can’t give it away. While most multimeters can be adjusted internally, the cheap and nasty ones resist calibration attempts by drifting all over the place afterwards. The internal voltage reference they measure against is worthless and the measurement circuits are not temperature-compensated.

An instrument with a range of 4000 counts, rather than the more common 2000 counts found on low-end units, also means that it is capable of displaying differences down to a single millivolt between 3 and 4 volts.

Always measure cell voltages directly from the cell metallic terminals themselves, rather than the cell links or bolts. The readings are much more reliable. And keep your instrument in a sealed plastic freezer box, with the leads neatly folded and a spare battery: this also makes for more reliable readings on the long run!

Safe Handling

New cells come out of their crates fitted with insulator caps over their terminals to prevent accidental short-circuits. The extraordinary discharge current capability of lithium battery cells has been discussed already. Accident risks are very high while repeatedly connecting, disconnecting and re-arranging cells for balancing and building a battery bank. Keep the insulator caps on the cells terminals for as long as nothing is connected and insulate the tools used for making the connections. Most of the cells in the sizes suitable for building marine banks use M8 bolts and require a 13mm spanner or socket drive. Wrap this tool with insulating tape, or better, heat-shrink tubing, if it is going to be used in a sustained way, only leaving exposed metal at the working end. On the same token, cover the top of the cells you are not working on.

Shorting cells while working on their connections with a spanner could result in intense burns and the offending tool might weld itself to the terminals before failing “fuse-style”, sending molten metal flying around. Watching people working on cell connections is one thing that always makes me nervous.

Cell Balancing

Before a bank is physically assembled into place, the cells must be balanced. This step is absolutely critical, because if a cell becomes fully charged ahead of the others, its voltage and resistance increase very rapidly, the charging current collapses and the other cells can’t be charged any further.

When cells are manufactured, their actual capacity always deviates more or less from their intended nominal capacity; next the cells are cycled, tested and then left partly charged by the factory before shipping. There is every chance that cells even belonging to the same production batch won’t all share the exact same capacity and will not land in a state that would allow simply connecting them in series to obtain a balanced battery bank.

Trying to operate an unbalanced battery bank, such as when assembling new cells without precautions, results in problems both with charging and discharging. Here, cell 4 gets over-charged while the others are not yet full and cell 2 reaches an over-discharge condition while the others still have some capacity left.

At best, on a well-designed system, cell imbalance causes a reduction in available capacity and potentially some kind of alarming or even disconnect; on an unprotected, unmanaged system, it leads to cell destruction and can result in dangerous developments.

A bank can be top-balanced or bottom-balanced, but never both, because the cells never share the exact same capacity. The choice depends on the application and type of service.

Bottom Balancing

Bottom balancing is normally very undesirable for marine house banks as they hardly ever, if ever at all, get fully discharged and it creates most unwelcome difficulties with charging. Charging and managing bottom-balanced banks will not be developed here for these reasons.

For the sake of completeness and understanding only, some information is provided here about bottom balancing.

All cells are first discharged to a common low level prior to being assembled together. Charging results in one cell reaching full charge before the others, but deep discharge is no issue on the other hand.

The goal of bottom balancing is ensuring that all cells get to their low charge limit evenly together. Bottom balancing makes most sense in applications where deep discharge happens routinely, like in the case of electric vehicles that are driven almost to the point of running out of energy. For this reason, bottom balancing was introduced (and rather successfully at that) by people building DIY electric cars, such as Jack Rickard at EVTV; until then, not only they lost a lot of cells, but some also managed to incinerate a few vehicles.

In order to bottom-balance a set of cells, each cell must be discharged down to a voltage that is at or below what the low voltage cut-off setting will be. Typically, this would mean a value of about 2.5V. The best and quickest way to achieve this is wiring all the cells in parallel and discharging them through an automatic low voltage disconnect device. Power resistors or light bulbs are all usable loads for discharging.

One should remember that if the cells are accidentally over-discharged in this process, they will be destroyed. Over-discharge means reaching below 2.0V for LiFePO₄ chemistry.

Once all cells are down to the same low stabilised voltage, the bank can be assembled and charged.

A bank that has been bottom-balanced will invariably go out of balance at the end of the charge. This is unavoidable. The voltage of the smallest cell will peak up ahead of the others and throttle the charging current. If, at this point, charging is not immediately discontinued, this cell will quickly get damaged through over-charging.

Top Balancing

The goal of top balancing is ensuring all cells get full together at the end of the charge instead. Top balancing is almost the rule for all applications where very deep discharge essentially never happens, and this precisely includes marine house banks. Top balancing makes the task of recharging the bank more straightforward, because the total battery voltage is distributed quite evenly across the cells near the top end.

At the bottom end, one cell will invariably drop out first and if the bank is discharged beyond this point and the voltage of the weakest cell falls below 2.0V, it will be destroyed by over-discharge.

All cells are fully charged before being assembled together. A deep discharge can cause the smallest cell to “hit the bottom”, but charging is normally no issue and all cells can easily be stopped short of over-charging.

Top balancing is by far the most common process used for building a lithium battery bank, because cell imbalance issues at the low end normally never become apparent, on the basis that cycling that deep doesn’t normally happen; at this point, the bank hardly has any stored energy left and cutting it out becomes a simple and logical response.

In order to top balance the cells, they need to be charged in parallel until well into the upper “knee” region of the voltage curve, where small differences in state of charge become very visible in terms of cell voltage.

How the cells actually get charged is irrelevant as long as they are kept within their voltage limits throughout. Unlike often stated, there is no point pushing the cells to voltages far exceeding 3.6V to balance them. It is just a good way of starting with electrochemical damage and achieve absolutely nothing else.

We are going to present two options for top-balancing a set of cells.

Method 1: Charging and Balancing Cells Using a Regulated Power Supply Unit

There are a few options available for first charging and balancing the cells. Using a regulated bench top power supply unit (PSU) is the commonly promoted approach and also the least practical and accessible for a one-off job on board – which is often the context in place when building a DIY system on an ocean cruising yacht. This process is very slow, inefficient and requires a regulated power supply unit and mains power for several days.

In some cases these constraints don’t apply or this method can be combined with the second method to “finish off” the cells, so the process is explained below, but you should prefer the second method described.

Never use a crude battery charger: its output is unregulated and, even if it is able to hold without overloading and tripping, it cannot limit the voltage as the cells charge up. The guaranteed outcome will be a totally destroyed set of cells at best, or a fire. Don’t imagine for a second that you will be able to “see it coming” and prevent it. The voltage seems to remain constant forever and then rapidly rises without any warning.

You need an adjustable, regulated power supply unit to follow this process.

Parallel charging and top-balancing cells using a regulated power supply unit (PSU).
Voltage regulation is essential to ensure the target voltage cannot be exceeded.

First of all, power the PSU before connecting anything to it and never interrupt the mains for as long as there are batteries connected to it. Some PSUs are not well protected against reverse current flow and not intended for use with large capacitive loads!

If possible at all, use a PSU that is explicitly suitable to charge a battery; in doubt, use great caution as a mishap can easily damage it. If smoke escapes from it, you will never get it back in.

1. With the output disconnected, set the voltage regulation limit at 3.40V and preset the current limit (if any) to a value that won’t overload the PSU. Refer to the manual as required. In doubt, always start with a low current limit and don’t exceed 80% of the rated output.
2. With all the cells wired in parallel, connect the PSU, bulk charge and absorb until no current flows any more. The voltage will stay around 3.3V for a very long time before starting to rise. Charging this way can take several days. This will near-fully charge the cells without stressing them unduly, but don’t hold them at that voltage indefinitely. Keep checking up on them at least a couple of times each day. Briefly disconnect the cells and recheck the voltage limit setting on the PSU: better safe than sorry. Avoid charging the cells individually, or in batches; the whole process would take just as long, but would also result in some fully charged cells lying around for several days.
3. Once the voltage has reached the PSU output regulation limit and there is no apparent charging current any more, disconnect the cells from the PSU and increase the output voltage regulation limit to 3.60V.
4. Then, while standing by only, reconnect the cells and allow the voltage to rise up to 3.60V and stabilise; this doesn’t normally takes long, provided the cells were fully absorbed at the lower voltage. The current restarts high and then quickly tapers down; stop when it reaches a small value, such as 2-3% of the cell capacity. Whether you target 3.60V, 3.65V or even 3.70V is of no consequence or interest if you are actively monitoring the process, because these values are often reached seconds apart only.

That’s all. The balancing algorithm of the BMS will take things from there if necessary and there is no need to insist beyond this point. A balancing BMS is a necessity with a lithium battery as cells tend to slowly diverge over time; sometimes they don’t appear to, but this situation is a very rare exception.

As with all unattended charging of lithium batteries, some very careful thoughts must be given to the potential consequences of a failure somewhere

Using a regulated PSU, a failure of the unit – no matter how unlikely – cannot be entirely excluded and there is no other line of defense in place. Hopefully, it would just trip, but if it didn’t, it could lead to a battery fire. Alternatively, the battery could then discharge into the PSU and possibly burn it out. Someone could also come past and interfere with the equipment during charging with an adverse outcome.

Method 2: Charging and Balancing Cells on Board

After going about charging and balancing cells in a few different ways, I devised this method. It has since become the solution of choice for one-off lithium battery projects, because it is much more efficient and doesn’t require equipment that is not already available on board.

The idea is rapidly bulk-charging the cells using the boat’s engine and alternator and then addressing the balancing part separately.

I also consider it as potentially safer, because it is short enough to be fully supervised.

While the process usually takes a couple of hours only, it requires unfailing vigilance. This is only feasible because the time frame is short. If you have access to a regulated PSU, proceed up to step 6 and then consider finishing using the first method.

1. Assemble the lithium cells in the final topology the bank will be using and bolt the on-board battery cables to it, as if performing a direct replacement. In some instances, this requires shifting the old lead-acid cells out of the way first. It is important that a heavy-duty connection is made between the lithium battery and the alternator.
2. Start the engine normally, run at idle for a couple of minutes and then rev it up. This will immediately result in a high alternator output. Check that the B+ (output) post of the alternator doesn’t heat up; this would indicate a bad or dirty connection and easily cause the alternator to fail. Also check all the cell connections for any temperature rise. All electrical connections should remain cold. Next, be mindful of alternator temperature. It is advisable to keep the charging current no higher than 80% of the alternator rating. Keep the engine compartment open if necessary and reduce engine revs if required. Twin-engine vessels like catamarans can (and usually should) charge with both engines. Keep a voltmeter connected directly to the bank.
3. Make a cup of tea and watch the voltage. New cells normally ship at 40-50% SOC, so a simple initial calculation can provide an idea of charging time. It is normally a matter of 1-2 hours. After remaining stagnant around 13.40V for a long time, the voltage will eventually start to rise. Periodically measure the individual cell voltages to ensure they don’t diverge abnormally and all remain below 3.60V. If this becomes tedious or distraction sets in, shut the engine down, disconnect the bank and carry on later. Should any cell reach 3.60V prematurely or, conversely, clearly lag behind the others, it is an indication that the cells weren’t in a consistent state of charge at all when sourced. This should be seen as a warning flag about a potential quality issue, like a significant difference in self-discharge rate or internal resistance.
4. The voltage will eventually reach the alternator regulation limit, normally 14.20-14.40V. If this was set higher (through the use of an external regulator typically), don’t allow it to exceed 14.40V. At this point, the individual cell voltages should still appear very even, because the cells were charged at a fairly high rate and are not full yet; only the bulk charge has completed.
5. From this point onwards, differences in cell voltages are going to start appearing. Only individual cell voltages matter. Identify the highest cell(s) and gradually reduce engine revs, so none exceed 3.60V. Keep reducing revs until down to idle, then shut the engine down. On a twin engine vessel, cut back and shut down one engine first. After about 30 minutes, most of the absorption phase is complete and the unbalanced bank cannot be charged any further without experiencing excessive cell voltage issues.
6. Disconnect and break up the bank, and now connect all the cells in parallel. If a cell is reading more than 0.1V higher or lower than the “pack”, parallel it with a small jumper cable at first to prevent any large current inrush, and connect it with the heavy link plate once the difference has subsided.
7. Once all the cells are wired in parallel, they need to be properly topped up and balanced.
   1. If significant solar capacity is available, take the solar feed from the panels (before any charge controller!) and connect it directly to the lithium bank. Solar panels are current sources and don’t care about their output voltage. They will contribute about the same current at any voltage.
   2. Alternatively, bridge from the old lead-acid batteries (or a basic battery charger) using a few metres of electrical wire (not cable!). The wire acts as a resistor, dropping the voltage and limiting the current. Depending on the length available, 2.5mm2 (12AWG) or 4.0mm2 (10AWG) are normally suitable choices. If there are 6V batteries available, bridge from 6V, otherwise bridge from 12V. The wire will heat up as a result of the voltage drop. If it gets too hot, stop and use a longer or smaller wire. Use caution and common sense.
   3. Bring all the cells up to 3.60-3.65V, disconnect and keep recharging this way until they all hold above 3.50V for at least 10 minutes. They are then full and balanced. Don’t leave the circuit closed and unattended under any circumstances; it would very quickly destroy all the cells.

This method is many times faster than parallel charging throughout, but more labour-intensive and requires continued attention. It is only feasible because maintaining complete focus for a period of 2 to 3 hours is not unreasonable. If you are negligent or over-confident and leave the process unattended for any amount of time, you will likely damage or even lose the cells completely.

Additional Considerations

A few additional notes regarding cell balancing:

1. Refrain from pushing cell voltages above 3.60-3.65V. You can trade a little more time for less voltage and achieve the same without stressing the cells. Balancing cells is not a hazardous process involving excessive voltages, infrared temperature guns and a fire extinguisher in standby: it is just a one-off, parallel full charge within normal voltage limits.
2. Use heavy link plates or substantial cabling in relation with the charging current to connect the cells in parallel, and preferably feed “in the middle”. The objective is keeping all the cells at the exact same voltage while they are charging. If the cell interconnections are dropping voltage, the cells away from the feed point will see a reduced charging voltage, at least until near the end of the charge.
3. Don’t waste time leaving cells sitting around connected in parallel. They don’t balance or equalise unless they are being charged in parallel at the end. I have specifically tested that. There is not enough voltage difference to keep driving current between cells until they balance out “over time”. It simply doesn’t work!

Discharging Afterwards

Once the cells have been charged and balanced as described above, they should be assembled into their final topology immediately and the bank must be discharged at least down to the equivalent of 3.325V/cell: that is 13.3V for a 12V nominal system, or 26.6V for a 24V nominal system. Don’t allow the cells to sit around at 100% SOC after balancing! If the balanced pack is going to remain out of service for some time, it must be further discharged until it doesn’t read more than 13.15V (or 26.3V) after balancing.

There are dealers who offer – for a premium – “pre-balanced” lithium battery packs that were fully charged in parallel, interconnected and then stored on a shelf… buying those often equates to paying more money for deteriorated cells, because the packs are typically never discharged again once charged and balanced. They may also have been exposed to excessive voltages for long periods while being charged. This why I only ever source factory-packaged cells.

Next Steps

The tasks described in this article lead to building a top-balanced battery bank of a given capacity. Before anything can be done with it, the cells must be protected from voltage excursions outside their safe operating range and automated measures must be put into place to take action if any abnormal conditions, whether it is voltage or heat, are detected.

The electrical system on board must also be altered in order to separate charging sources from loads, so both can be isolated independently at any time to protect the battery if necessary.

Last but not least, charging sources must be made compatible for charging lithium batteries in terms of voltage and operation. This is not always possible and some devices may need to be replaced.

This article is part of a series dealing with building best-in-class lithium battery systems from bare cells, primarily for marine use, but a lot of this material finds relevance for low-voltage off-grid systems as well.

Batteries are about voltage, current and capacity first and foremost. This article discusses the performance characteristics of lithium iron phosphate cells in service and the key concepts associated with them. It is very important in the context of setting up lithium battery systems, but also useful when living with and operating one. The chemistry and internal construction of the cells is detailed in a separate article of a more fundamental nature.

Disclaimer

A good understanding of DC electrical systems is needed to build and commission a lithium battery installation. This article is aimed at guiding the process, but it is not a simple blind recipe for anyone to follow.

The information provided here is hopefully thorough and extensive. It reflects the knowledge I have accumulated building some of these systems. There is no guarantee that it will not change or grow over time. It is certainly not sufficient or intended to turn a novice into an electrical engineer either. You are welcome to use it to build a system, but at your own risk and responsibility.

What is a Battery?

A battery stores electricity and the question may appear trivial, but it is not. An ideal battery would supply any current at a voltage purely dependent on its state of charge. Real batteries don’t. Real batteries see their voltage drop under load and suddenly step up while being charged. The reason for this phenomenon is that they have an internal resistance. The higher the current flow, the higher the voltage lost to this internal resistance. The electrical symbol for a single battery cell looks like this:

Ideal battery representation. Here, the voltage would purely be a function of the state of charge of the battery, at any current, which is obviously incorrect.

In order to represent the variation in voltage caused by changes in current and understand the behaviour of batteries, we need to add internal resistance to this ideal battery:

A simple model for the non-ideal battery. The ideal battery is in series with an internal resistance element that causes the voltage to change with the current.
At rest, the output voltage reflects the state of charge of the battery.

If no current is flowing, the internal resistance has no effect on the output voltage; this is why it is important to measure cell voltages at rest if the objective is obtaining an idea of the state of charge. Otherwise, the effect of electrical resistance is skewing the voltage proportionally to the current according to the relation:

ΔV = R x I

Upon discharge, we can now observe the following effect, which does model the reality of a battery discharging at a steady rate:

Under discharge, the voltage at the terminals is lower than the true voltage of the cell because its internal resistance is introducing a loss equal to R x I in the direction of the current.

As a consequence, the voltage measured at the terminals of the battery no longer reflect its state of charge. This is why the state of charge of any battery can only be deduced from a stabilised voltage measurement taken at rest: it is called the stabilised open-circuit voltage (OCV). A similar situation arises when charging:

While charging, the voltage at the terminals is higher than the true charging voltage of the cell because its internal resistance is introducing a loss equal to R x I in the direction of the current.

Now, the internal resistance of the battery is making the charging voltage at the terminals look higher than it actually is in terms of actual state of charge of the battery. Here is a real-world illustration of this behaviour:

We were in the process of building a brand new lithium iron phosphate battery bank on a sailing catamaran, charging 400Ah of cells for the first time with both engines running. The charging current had been a solid 180A for almost an hour. The cell voltages, which had initially jumped up around 3.40V, were gradually rising. When they reached 3.60V, we shut one engine down in order not to exceed this value, reducing the current by half, down to 90A.

The cell voltages instantly dropped down to 3.45V.

We therefore lost 0.15V in cell voltage by reducing the current by 90A. We can use these figures to calculate the internal resistance of the cells using the relation presented earlier, ΔV = R x I:

In this case, we have ΔV = 0.15V and I = 90A. As a result, we can write R = ΔV / I = 0.15 / 90 = 1.66mΩ

1.66 milliohms is a very small resistance figure typical of lithium battery cells, but it is nevertheless enough to significantly skew the voltage reading at high amperage. At a current of 10A, its contribution becomes only ΔV = R x I = 0.00166 x 10 = 0.0166V = 16.6mV, but still enough to be measured. We will refer to this again when discussing alternator voltage for charging, low voltage cut-off limits and cell balancing boards amongst other topics.

In a bank, all cells don’t share the exact same internal resistance, so their voltage doesn’t automatically read the same when there is current flowing, even when their state of charge is identical. It becomes increasingly true as cells age.

Before moving on, let’s point out that the battery model we used above featuring the cell internal resistance is correct as long as the current is steady and the voltage at the terminals has had a few seconds to stabilise. A more complex electrical model would need to be used if the transitions when the current varies were of interest, because of capacitance effects.

Battery Currents

Current measurements related to batteries in general are expressed in relation with their capacity rather than in absolute terms: a 100Ah battery operated at 100A is said to be charging or discharging at 1C: one time its capacity rating. A 10A current would only amount to 0.1C; a full charge at a rate of C/5 would represent a 5-hour (approximately) charge, etc.

Charge and Discharge Ratings

Prismatic LiFePO₄ battery cells were once rated for charge at up to 1-2C and discharge to 3C, and this seemed to imply they could theoretically be charged in 30 minutes and discharged in 20 minutes. We since realised that most lithium-based chemistries didn’t last long when subjected to this kind of treatment. Excessive charging currents in particular are damaging and even more so at lower temperatures.

The maximum recommended routine charge and discharge rate became about 0.3C for long-term, sustained operation. Some of the newer-generation aluminium-cased LiFePO₄ cells are rated for 0.5C and this results in minimum long-term sustainable charging times of about 2.5 hours when absorption is factored in. This can lead to having to limit charging currents and can discourage the use of ridiculously oversized alternators or chargers.

Beware of Short-Circuits

The short-circuit current capacity of LiFePO₄ cells can easily exceed 20-30C, which is far more than needed to cause catastrophic heat damage. The greatest of precautions must be taken when working around cell connections as dropping a non-insulated tool onto any battery bank can result in molten metal flying around, a fire, disastrous burns or any combination of the three.

The practical difference between working near the common deep-cycle lead-acid batteries on board and working around lithium cells is that there are a lot more exposed connections in much closer proximity and even small tools or metallic objects can be long enough to cause a short-circuit. Furthermore, in the event of short-circuit, even relatively small lithium cells are capable of delivering extremely intense and sustained currents.

Incidentally, manufacturer tests have repeatedly shown that a healthy LiFePO₄ cell can be bluntly short-circuited to complete destruction without reaching ignition temperature: this is due to the fact that its internal resistance is very low. The same may not hold for a previously damaged cell with an elevated internal resistance and the outcome could then be extremely different.

A short-circuit test on a fully charged Sinopoly cell. The current is exceeding 1800A as the cell is venting profusely.

This image was extracted from a video released by Sinopoly Battery Ltd, China, where other common battery failure modes were investigated, such as when a crew shoots into the battery with an automatic pistol.

Typical Cell Operating Limits

Manufacturers ratings for LiFePO₄ battery cells have become more conservative in recent years as more experience was gained with the practical operation of these cells. Nowadays, the typical operating specifications for LiFePO₄ prismatic cells [1] look as follow:

Back in 2007, Thundersky, a manufacturing company later absorbed by Sinopoly Battery Ltd, was advertising its prismatic cells for charge up to 4.25V using a current of 3C and their maximum rated discharge current was 10C. Those who followed these guidelines quickly came to a great deal of grief, first and foremost with charging, destroying cells left, right and centre while charging up to the 4.25V “target” at low current.

Today’s charging specifications may still appear as being on the high side, but they must be understood in the context of a constant current/constant voltage (CC/CV) charge regime with charge termination and charging to maximum capacity as the aim. The recommended upper SOC limit is 90% however, not 100%, and charging to 100% SOC in this context means absorbing the cells at 3.65V until the residual current is C/30. Anything short of this will not – by definition – achieve 100% SOC.

All maximum ratings must be understood as absolute limits, not standard operating values, which is why the simplistic reasoning suggesting that 4 cells in series can be charged at 4 x 3.65V = 14.6V couldn’t be more wrong. Just as wrong as the suggestion that any old lead-acid charging system is fine for operation with lithium cells “because the voltage range is compatible”. The voltage range can be quite close, but the charging process required is very different because it needs to provide for charge termination.

The specifics of charging lithium cells on board will be the subject of a separate article due to the extent of the subject, but the essential charging characteristics of LiFePO₄ cells are discussed further below.

Battery Capacity

Peukert’s Law and Lithium Batteries

The capacity of a battery is not a constant figure: it depends on the charge and discharge current. The phenomenon was documented whilst working with lead-acid batteries as Peukert’s Law in 1897. In simple terms, Peukert’s Law states that the available capacity shrinks as current increases.

The answer to the question of whether Peukert’s relation can really be applied to lithium chemistries is essentially negative [2], but the capacity of Li-ion batteries does also vary with discharge current and Peukert’s Law is all we have at present. Peukert’s Law was only ever formulated to be valid at constant temperature and we do know that Peukert’s effect in LiFePO₄ batteries becomes increasingly noticeable as temperature drops below 15°C.

Peukert’s relation is characterised by a supposedly constant exponent k and, in the case of LiFePO₄ batteries in house bank applications, experimental data at modest temperatures has suggested a value of k=1.04. An exponent of k=1.00 would indicate no dependency between storage capacity and current, i.e. an ideal battery, and lead-acid batteries often score around k=1.25, with the figure getting worse as they age.

Configuring Battery Monitors

This value of k=1.04 can make for a useful starting point when configuring battery monitors, but temperature variations (which are never accounted for) can easily throw the calculation out, especially when large swings from winter to summer are involved. In the tropics, with batteries at 25°C or over, a value of k=1.02 for the exponent may be more appropriate.

Trying to configure battery monitors designed for lead-acid batteries – where they already perform suspiciously at the best of times – to operate with lithium cells is fraught with uncertainty: the supposedly constant exponent k has been shown to be anything but constant with lithium-ion chemistry [3]. Provided the temperature doesn’t change significantly and the currents in operation are reasonably consistent, a set of parameters can be derived to obtain seemingly sensible readings.

Rated Capacity and Actual Usage

Lithium cells are usually capacity-rated at much higher currents than lead-acid batteries and the battery is deemed discharged when it can no longer supply the discharge current. Capacity rating for discharge at 0.5C (2-hour discharge) or 0.3C are common for prismatic lithium cells, while lead-acid cells are normally rated at C/20 (20-hour discharge). The practical consequence of this is that lithium batteries commonly appear to exceed their capacity ratings at the average currents normally run on board a yacht.

Peukert’s Law can be formulated as: C2 = C1 x [ C1 / (I2 x T1) ] ^(k-1), where:

C1 is the battery capacity when discharged in T1 hours at a current I1, and

C2 is the calculated capacity when discharged at a current I2. k is the Peukert exponent discussed earlier.

What can we expect from a 100Ah lithium battery rated at 0.5C = 50A when used as a house bank and discharged at C/20 = 5A instead?

We have C1 = 100Ah, I1 = 50A, T1 = 2 hours, I2 = 5A and we will use k = 1.04:

C2 = 100 x [ 100 / (5 x 2) ] ^{(1.04 – 1)} = 100 x 10 ^0.04 = 109.6Ah

The same 10% gain stands for a 200Ah battery discharged at 10A, etc.

These differences can become quite significant in larger banks, as a 400Ah battery discharged at 10A only would now exhibit a theoretical capacity of 476.6Ah. Such calculations are fraught with uncertainty however due to the temperature dependency for the value of k, but matching results have been demonstrated experimentally. At very low temperatures, some of the battery capacity simply becomes inaccessible altogether.

Low Temperature Effects

Capacity Reduction

Capacity is also quite sensitive to temperature effects. Lithium cells offer more capacity and higher performance at higher temperatures, including at excessive temperatures causing accelerated ageing. At freezing temperatures, the available capacity upon discharge shrinks quite significantly [4], but is recovered once the cell warms up again.

Available capacity as a function of temperature for a low voltage discharge cut-off threshold of 2.5V/cell (Plot courtesy of Tsinghua University)

This phenomenon highlights the fact that lithium ions become more and more difficult to extract from the graphite matrix of the anode as temperature drops and only relatively superficial charge carriers are available at low temperatures; the balance of the capacity effectively becomes “locked-in” out of reach. This loss of available capacity also translates into a lower discharge voltage, with the low voltage cut off point being reached earlier.

Constant current voltage discharge curves at different temperatures. The low voltage cut-off threshold leaves significant capacity locked into the battery at freezing temperatures (Plot courtesy of Tsinghua University)

Accepting a lower low voltage cut-off threshold would be a way of regaining access to some of this locked-in capacity in sub-freezing conditions, but the matter is only of real interest for automotive applications.

For all practical purposes on marine vessels, battery temperatures below freezing should be uncommon unless the water also freezes around the hull. Capacity reduction is then limited to about 10% only in the worst case, which should be negligible. While discharge at low temperature yields both reduced power and capacity, it is harmless to the cell. The same cannot be said of low temperature charging.

Cold Temperature Charging

Cold temperatures are known to be detrimental to the cells if they are exposed to charging. Cycling performance tests at varying temperatures showed the apparent existence of a threshold below which capacity fade with cycling suddenly accelerated. This threshold appeared to be above the temperature of 0°C often suggested as limit for recharging, but the data available was limited and the exact details of cell manufacture are likely to influence this value.

The intercalation of lithium ions into the graphite matrix of the anode becomes more difficult as well at low temperatures and lithium ions ejected out of the cathode and unable to soak into the anode instead plate its surface and edges; this lithium is then irreversibly lost. This suggests that fast charging in particular becomes increasingly harmful to the cells as temperature drops.

Impact of State of Charge and Temperature on Capacity and Cell Life

The cell voltage naturally increases with their state of charge and the higher voltage makes the electrodes chemically more reactive and this appears to encourage undesirable side reactions with the electrolyte that harm the cell health. The practical consequence of this is that keeping lithium cells at a high state of charge for long periods reduces their working life by causing capacity loss and an increase in internal resistance. Higher temperatures are well-known to reduce cell life.

The combined effects of state of charge and temperature on cell health during a storage period of 9 to 10 months were studied in detail by Keil et al. [8] and illustrated below.

LiFePO₄ cell capacity and internal resistance changes over a 9-10 months storage period as a function of state of charge and temperature (plots from Keil et al. [8]).

The first plot clearly illustrates that cell capacity is best preserved by storing the cells at the lowest state of charge and temperature, with virtually no degradation occurring at 0% SoC and 25ºC. If we compare the plot with the graph of the cell open-circuit voltage presented in the next section below, we can see that capacity fade then increases in direct relation with the cell open-circuit voltage and this translates into a step change above 70% SoC and a flat between 40% and 70% SoC. This indicates that LiFePO₄ battery banks should be stored at a very low voltage, like 3.0V/cell and in cool conditions, with temperatures up to 25ºC being acceptable, but lower being even better.

The second plot depicts the effect of SoC and temperature on the DC internal resistance of the cells and shows that this is unrelated to the state of charge and dependent on temperature only. Not only banks stored in warm conditions lose capacity at an accelerated rate, but they also develop increased voltage sag under load and this reduces their ability to supply high power levels.

Voltage and State of Charge Characteristics

A LiFePO₄ cell has a rated nominal voltage of 3.2V. In practice, 3.2V is only reached when heavily discharged (or under significant load) and the normal operating voltage is about 3.3V. This implies that a 12V nominal lead-acid battery made up from six cells in series for a total of about 12.7V in operation can be substituted with four LiFePO₄ cells instead, for a resulting voltage of about 13.2V.

On-board power from a lithium bank shows an improved and much more constant system voltage; most of the equipment runs noticeably better, from pumps to SSB transceivers. Lights don’t dip either when a load is turned on, because its low internal resistance translates into much less voltage sag.

The state of charge (SOC) of a lead-acid battery can normally be deduced from its voltage, but only as long as the battery has been at rest long enough for the reading to stabilise. Lead-acid batteries have significant internal resistance, especially when no longer in their prime and drawing current from them immediately skews the reading to the downside.

Lithium batteries are similar, other than for their much lower internal resistance and a more complex relation between state of charge and voltage, which exhibits a prolonged flat when the cells are in the 40% to 65% SOC range. Outside of this region, voltage readings do provide very useful indications of the state of charge.

Single LiFePO4 cell stabilised open circuit voltage as a function of the state of charge (Data courtesy of Tsinghua University)

The cell voltage differs depending whether the cell was being charged or discharged before the voltage was allowed to stabilise. In nearly all instances on board yachts, small loads quickly bring the voltage back in line with the discharge curve.

If this higher resting voltage following charging appears to dissipate very quickly, it is a tell-tale sign that the cells have been abused and suffered electrochemical damage.

LiFePO4 4-cell battery stabilised open circuit voltage as a function of the state of charge (Derived from data courtesy of Tsinghua University)

At rest, or for low charge and discharge currents, the above plots are extremely useful for estimating the state of charge, even just by glancing at the voltmeter:

The owners of installations cycling moderately who can refrain from making an automatic beeline to the nearest marine electrical retail store can be pleasantly surprised to discover that the addition of a random number generator battery monitor to the system can be completely superfluous with lithium, as long as a simple voltmeter and a little knowledge are available.

Current and Power Efficiency

Lithium batteries in general are near 100% current efficient: this means that charging 1Ah yields a typical discharge of 0.997Ah at a similar current. This is hugely higher than what lead-acid chemistry can achieve and often results in gains of 30-50% in charging efficiency when a lead-acid house bank is replaced by LiFePO₄ cells on a yacht.

The net effect with solar arrays is as if the size of the array had suddenly become significantly larger and a change to a LiFePO₄ bank can be a more sensible answer to energy issues than adding more panels or running an engine.

Power efficiency, on the other hand, sits around 95%, but varies with current: expend 100Wh charging and you will retrieve about 95Wh on discharge. The difference stems from the fact that the charging voltage needs to be a little higher than what is available afterwards during discharge.

In marine use, current efficiency is what matters, because finding a little more voltage is never an issue.

Charging Characteristics

With regard to charging, lithium cells are both far simpler to charge and totally different than lead-acid cells. As a consequence, they should also be managed differently. Another important aspect is that recharging a fresh, new cell can be very different and much easier than recharging a cell which has just seen a large number of partial charge and discharge cycles, due to memory effects which are discussed further below.

The most commonly documented charging regime for lithium cells is constant current, constant voltage (CC-CV). It is also one that is essentially never achieved with marine installations: on-board systems deliver variable current, limited voltage mixed with partial charge/discharge cycles.

As a result, the only parameters that actually matter are the maximum voltage the battery is allowed to reach during charging and the way the charge is terminated, because those determine the outcome of the charging process.

What is Charging Voltage?

The charging voltage is basically the voltage at the battery terminals during charging. The battery user essentially has no control over this voltage for most of the charging process: the battery absorbs all the current provided and the voltage rises at its own pace, as the state-of-charge increases.

The voltage can only be controlled – by reducing the charging current – once it would start to exceed a limit.

I remember once reading a senseless post about an alternator. The author was complaining that the regulator was “useless” because “it limited the voltage instead of charging at the desired setpoint”.

What this person didn’t understand is that the voltage reaches a value that depends on the state of charge of the battery and, with the alternator at full output already, there is nothing more the regulator can do until the voltage naturally rises enough to warrant limiting it.

The parameter the user has control over is the end-of-charge voltage. The end-of-charge voltage is simply the voltage limit used by the charging system before the charge is terminated. Because of the higher internal resistance of lead-acid batteries, the charging voltages rises both earlier and a lot more rapidly than what is observed with lithium cells.

Lithium cells commonly charge at 3.4V or less for very long periods of time while soaking up full current and, when the voltage finally begins to increase, the battery is already significantly charged.

The Relation Between End-of-Charge Voltage and State of Charge

The relation between the end of charge voltage and the state of charge eventually achieved by a LFP cell can be explored by charging battery cells using a range of maximum voltage limits until the current has reduced down to a very small value each time before discharging them again to assess capacity.

Such an experiment was conducted by Powerstream [5] in 2014 with four different brands of LiFePO₄ cells of the same size, which were charged until the current had reduced down to about 0.013C. This is quite a low charge cut-off current and it must have resulted in extended absorption times.

I used their published experimental data to plot a more interesting graph showing the state of charge reached against the absorption voltage limit.

The graph above illustrates that while 3.3V is insufficient to recharge a cell, 3.4V is enough to obtain near 100% capacity already and limiting voltage cannot realistically prevent overcharging without also compromising charging.It also highlights that using high absorption voltages essentially fails to achieve anything as far as capacity is concerned, but charging times would certainly be reduced if also shown.

LFP cells simply don’t really charge at voltages up to 3.3V and then fully charge already at 3.4V and upwards. The transition is so abrupt that claiming to control the charging process by adjusting the voltage is purely and simply bound to fail.

Charging at reduced voltages, down to 3.4V/cell, only increases the absorption time and therefore the overall charging time, but achieves strictly nothing in terms of preventing the battery from getting fully charged and then overcharged. It only takes longer for this to happen. Furthermore, low-voltage charging opens the door to severe longer term performance issues which arise from memory effects in the cells.

Memory Effects

Memory effects in LiFePO₄ cells were discovered and studied by Sasaki et al. [6] and the results published in Nature Materials in 2013. The authors illustrated that, under specific circumstances, the prior cycling history of a cell alters the voltage curve during charging by causing the voltage to increase faster and earlier than expected.

Memory effect in LFP cell following different incomplete charge and discharge cycles. Note that the voltage is referred to the potential of a lithium electrode (plots from Sasaki et al. [6]).

For a memory effect to appear, an incomplete charge cycle followed by a rest period and a discharge must have taken place earlier (memory-writing cycle). A partial charge followed by an immediate discharge is not sufficient to record a memory of the incomplete cycle [7]; this is important because the practical consequence is that a charge-and-hold strategy is particularly harmful when full charge was not achieved. It is not uncommon for DIY lithium battery systems to implement deficient charging strategies which in fact result in this scenario taking place and it is detrimental to the long-term performance of the battery bank.

When a memory-writing cycle has been completed, an abnormal increase in voltage can be observed afterwards as the charging process approaches the point where charging had stopped earlier; this creates a bump in the charging curve. Partial charging of all common types of lithium cells (with the notable exception of lithium titanate oxide Li₄Ti₅O₁₂) leaves the cell with divided lithium-rich and lithium-poor phases which persist during and after discharge. In order to erase the cell memory of the previous interrupted cycle(s), a full charge must be performed (memory-releasing cycle) and this requires overcoming the bump caused by past partial cycles.

The memory effect was found to strengthen with the number of incomplete charge cycles performed before the erase cycle. It was also strengthened when a partial charge was followed by a shallow discharge, rather than a deep discharge.

These latter aspects have proved to be of key significance when considering the longer term performance of LiFePO₄ batteries in house bank applications, because incomplete charge cycles are common when relying on renewable energy sources and shallow discharge cycles are also frequently experienced. These have the potential to render battery banks near unusable after as little as 2-3 years in regular service in the absence of memory-releasing cycles. Ineffective memory-releasing cycles are very common in DIY installations where the charging process is not properly controlled and/or configured incorrectly by fear of overcharging or due to widespread mythologies.

An absence of memory-release cycles caused by ineffective charging allows the voltage bump caused by the memory effect to grow over time. If the absorption voltage and/or the absorption time are insufficient to overcome it, the charging process gradually terminates earlier and earlier. This has a compounding effect as memory-writing begins to occur at lower and lower values of SOC over time and the available capacity of the battery can disappear almost completely without any loss of lithium or chemical degradation as such. Recovering battery banks in this state can be challenging and require many memory-release charging cycles using high absorption voltages, followed by deep discharge. For these reasons, LiFePO₄ batteries should be charged properly whenever the opportunity arises, so the effects of unavoidable previous partial cycles can be wiped out while it is still relatively easy to do so. This calls for a robust absorption voltage and a charging strategy providing adequate charge absorption. Anything else falling short of this will eventually result in significant performance and capacity issues.

While we showed earlier that voltages as low as 3.4V/cell were able to fully charge and even overcharge a LFP cell, this must now also be considered in the context of memory effects altering the charging curve of the cells. My experience so far has been that any termination voltage below at least 3.5V/cell should be considered as inadequate if the installation experiences incomplete charge cycles. Any charging system that is unable to provide an adequate absorption down to at least C/20 or less when required should also be considered as unfit for purpose, because it will fail to deliver charge cycles capable of erasing the cell memory.

Overcharging

Overcharging means applying a charging voltage to an already fully charged battery. As we just highlighted the fact that – given enough time – lithium batteries always fully charge at 3.4V/cell or above, any voltage from 3.4V up can most definitely overcharge and damage a lithium battery.

How quickly this happens certainly depends on how high this voltage is, but – unlike what is observed with lead-acid chemistry – there is no such thing as a safe charging voltage that can be maintained continuously with lithium cells. All charge cycles must end when or before the battery becomes full.

A lead-acid battery benefits from what is known as a shuttle reaction, which does (within reason) allow excess energy to be absorbed and dissipated. This mechanism is not present in lithium batteries and it makes them very intolerant to overcharging.

A lithium battery that is being held at an elevated voltage with zero current flowing in has been overcharged and is getting damaged. This situation commonly happens with many marine charge controllers, including and especially some supposedly designated for lithium banks.

The “lithium” versions of the Genasun GV-5 and GV-10 MPPT solar charge controllers are prime example of this as they maintain 14.2V on the battery indefinitely (based on units inspected in 2015)

Charge Termination

Since absorption voltage can’t practically be used to limit charging, it becomes a matter of determining when to stop. Charge termination ideally needs to occur before the battery is completely full, because most of the stress on the battery happens when it runs out of lithium to transfer, or when it can’t transfer lithium ions fast enough, such as when the charge rate is very high and the voltage is allowed to rise excessively.

The tell-tale sign of a fully charged (or overcharged) battery is that it is no longer able of absorbing any significant current, or even any current at all

Voltage-Based Termination

If charging at very low currents, such as 0.05C, where internal resistance doesn’t meaningfully skew the voltage reading, termination can be implemented based on a voltage threshold on the basis that the current is then known to be low. A small solar system charging a sizable bank can fall in this category. In this case, charging must stop when the target voltage is reached and not resume until the voltage has dropped to a level indicating that the battery can and needs to be recharged again.

At higher currents, this strategy would err on the safe side by leaving an undercharged battery, but it is unsatisfactory, because charge absorption is still essential with lithium cells in order to erase the memory from previous partial cycles and make a good use of the capacity installed.

Time-Based Termination

Schemes involving a timed absorption period perform an approximate charge termination only. If the battery requires bulk charging and the duration of the absorption period has been determined wisely, a good charge cycle may result. If the battery is already full when charging begins, it will invariably suffer throughout the undesirable absorption phase; using a lower absorption voltage limits the stress placed on the cells, but fails to properly address the issue, increases the overall charging time and opens the door to long-term capacity problems resulting from memory effects.

Nearly all so-called “smart” alternator controllers typically implement a time-based absorption strategy to provide a charge termination that is anything but smart… any charge termination is still a lot better than none however.

Absorption times with lithium iron phosphate batteries are typically in the 30-40 minutes range in most situations when charging with high-current sources, and much less if the battery is being charged at low current. If a time-based termination is going to be implemented, then the absorption time should be determined experimentally by monitoring the current taper. If the battery is suffering from memory effects due to previous repeated partial charge cycles, then the required absorption time can increase very significantly and a time-based termination will interrupt the charge before the cell memory has been cleared.

Optimal Charge Termination

In all instances where significant charging currents are present or where the battery has seen a large number of interrupted and partial charge cycles, correct termination can only be obtained by monitoring both current and voltage to make an informed decision.

The voltage must be up at the absorption setpoint while the current is down at the charge termination limit; this indicates that the ability of the battery to absorb further charge is near its end. The final state of charge achieved depends on the combination of maximum voltage and minimum current, but changing the termination current is the only reliable way of altering the state of charge obtained and the voltage must always be sufficient to ensure memory effects from previous partial cycles can be overcome.

Charging equipment intended for lead-acid batteries is hardly ever able to perform a proper charge termination, because overcharging lead-acid cells (with the exception of gel-cells) is acceptable to some extent, there are no real safety considerations arising and batteries are relatively inexpensive. The functionality required is not present and the addition of the word “lithium” in the product brochure typically does exactly nothing to remedy to this situation. While battery voltage is always available, battery current is either not measured or the information is not exploited by the equipment. For this reason, the only place for realistically determining charge termination in a lithium battery system is at the BMS and the BMS should supervise the charging process.

References:

[1] CALB CA180FI and Sinopoly LFP200AHA cell datasheets.

[2] D. Doerffel, S.A. Sharkh, A critical review of using the Peukert equation for determining the remaining capacity of lead–acid and lithium-ion batteries, Journal of Power Sources, 155 (2006) 395–400

[3] N. Omar, P. Van den Bossche, T. Coosemans and J. Van Mierlo, Peukert Revisited—Critical Appraisal and Need for Modification for Lithium-Ion Batteries, Energies 2013, 6, 5625-5641; doi:10.3390/en6115625

[4] L. Lu, LiFePO₄ battery performance testing and analysis for BMS, Department of Automotive Engineering, Tsinghua University (2011)

[5] http://www.powerstream.com/lithium-phosphate-charge-voltage.htm

[6] T. Sasaki, Y. Ukyo and P. Novak, “Memory effect in a lithium-ion battery”, Nature Materials, Vol. 12, June 2013; doi:10.1038/nmat3623

[7] H. Kondo, T. Sasaki, P. Barai and V. Srinivasan, “Comprehensive Study of the Polarization Behavior of LiFePO₄ Electrodes Based on a Many-Particle Model”, J. Electrochem. Soc. 2018 165(10): A2047-A2057; doi:10.1149/2.0181810jes

[8] P. Keil, S. F. Schuster, J. Wilhelm, J. Travi, A. Hauser, R. Karl and A. Jossen, “Calendar Aging of Lithium-Ion Batteries, I. Impact of the Graphite Anode on Capacity Fade”, Journal of The Electrochemical Society, 2016 163 (9) A1872-A1880

This article is part of a series dealing with building best-in-class lithium battery systems from bare cells, primarily for marine use, but a lot of this material finds relevance for low-voltage off-grid systems as well.

Hundreds, if not more, of research papers have been published about lithium batteries, as well as numerous books. Still, a lot of this material is not very applicable to lithium batteries on board marine vessels and can be misleading: the type of application and usage considered are too different, the chemistry is not the same… the reasons are numerous. Some of this research is very relevant however. I lost track long ago of the number of publications I read or studied about lithium batteries in the process of designing and building such systems. One of the best books available on the subject at present is Design and Analysis of Large Lithium-Ion Battery Systems [1]. This article is an attempt at synthetizing the most essential fundamental information about lithium battery cells for those wanting to use them to build marine house banks.

Lithium batteries absolutely hate being treated like lead-acid ones

This material is essential to read and understand if you are seriously thinking about changing over to a lithium bank: lithium batteries are very, very different from traditional lead-acid ones. It is not just a matter of swapping.

Lithium-Ion Batteries

Firstly, to clarify the terminology, lithium-ion refers to a family of battery chemistries, not a specific type of battery. There are numerous types of lithium-ion batteries, each with its strong points and weaknesses, as shown in the following spider chart [6]. Their common feature is just that Li⁺ ions act as charge carriers.

Strength and weaknesses of common lithium-ion battery chemistries: LCO – lithium cobalt oxide (1991), LMO – lithium manganese oxide (1996), NMC – lithium nickel manganese oxide (2008), LFP – lithium iron phosphate (1993), NCA – lithium nickel cobalt aluminium oxide (1999), LTO – lithium titanate oxide (2008).

Lithium Iron Phosphate (LiFePO₄, sometimes also referred to as LFP) and Lithium Titanate Oxide (LTO) are by far the most robust types of lithium batteries developed so far, but they both feature relatively low energy densities. The superior performance and potential lifespan of LTO is problematic to justify due its high cost and this makes the LFP chemistry the most logical choice for marine house bank applications. The other types of lithium-ion batteries are mostly completely unsuitable and unadvisable for use on board a yacht due their low thermal stability and they would not represent a sensible, safe choice on board at all. As a result, the information presented on this site specifically applies to lithium iron phosphate cells and battery banks.

What is a LiFePO₄ Battery Cell?

Before discussing the “do and don’t” aspects of LiFePO₄ batteries, developing a basic grasp of their construction and chemistry is very useful. Somehow, we tend not to overlook so readily what we understand. In the case of lithium batteries, cutting corners is not an option.

Cell Internal Structure

Similarly to a lead-acid cell, a LiFePO₄ battery cell is formed of positive plates (cathode), negative plates (anode), porous insulating separators preventing them from shorting out, and a conductive liquid (electrolyte) surrounding them. The differences reside in the materials used and the fact that a lead-acid battery operates through chemical reactions transforming its components, whereas a lithium battery just relocates lithium ions during charge and discharge, leaving everything else in the battery largely unaltered.

By definition, the anode of a device is the side where current flows in; the cathode is where it flows out. For this reason, the positive terminal of a battery is the cathode. This can feel counter-intuitive at first, because the anode of a diode is on the positive side.

The anode of a LiFePO₄ cell is made of a highly conductive copper sheet coated with porous graphite. The cathode is made of aluminium coated with lithium iron phosphate material, a kind of ceramic. The specific capacity of lithium iron phosphate material is about 140mAh per gram, so a 100Ah cell needs a little over 0.7kg of active material in its cathode.

Internal structure of a lithium iron phosphate battery cell in a partly-charged state.

In order to create a battery, a conductive electrolyte must be provided to permit the transfer of charges between the anode and the cathode. The electrolyte is formed of a lithium salt (lithium hexafluorophosphate, LiPF₆ typically) dissolved into an organic solvent: some combination of ethylene carbonate, dimethyl carbonate, propylene carbonate with various other additives. This solvent is very flammable and fully absorbed into the porous plates and separator. Unlike in a flooded lead-acid cell, there isn’t free liquid in a LiFePO₄ cell.

A battery in the state described above is fully discharged, with all of the lithium present into the cathode and the electrolyte. This corresponds to a newly manufactured cell. The manufacturer then performs formation cycles to obtain a cell that can be subsequently used and will last. The formation cycles control the initial development of the Solid/Electrolyte Interface (SEI) layers at the surface of the plates. The SEI layer originates from chemical reactions between the electrolyte and the electrodes; it stabilises the cell chemically, but also keeps growing gradually over time, until it becomes detrimental to its operation. An important notion is that the SEI layer grows a lot more rapidly over time at elevated temperatures. Protecting lithium cells from heat is essential in terms of achieving a long life. The objective should be trying to keep them around some 15-25°C (59-77°F) as much as possible.

Prismatic versus Cylindrical Cells

Cells can be built by stacking parallel plates, as illustrated above, or from single long strips rolled onto themselves into a cylinder or flattened cylinder. The former are referred to as prismatic (box-shaped) cells, the latter are cylindrical or wound cells. Their chemistry is exactly the same, the main difference resides in their construction and ability to dissipate internally generated heat and to some extent cost. Prismatic cells are sometimes encapsulated into soft synthetic pouches instead of a hard casing, but pouch cells are a lot more fragile mechanically and more problematic to work with.

Prismatic cells come in much larger sizes than small cylindrical ones. Batteries are sometimes internally assembled from a large number of small cylindrical cells.

Wound cells, and small cylindrical cells in particular, are cheaper to manufacture than the larger prismatic ones for a given capacity. They also have a higher volumetric energy density, but their round cross-section prevents from packing them together without gaps and this advantage doesn’t extend to the assembled battery. The gaps between the cells can present an advantage for cooling when thermal management is necessary due to very high currents, but, in marine applications, the currents are modest and the battery cells never seem to get more than a few degrees above ambient temperature. Mechanically, cylindrical cells are very robust and very resilient to mechanical damage from shocks and vibrations, which is good in electric vehicles.

Prismatic cells seem to age better and offer more cycles in operation than small cylindrical cells, but this may be specific to high-current usage profiles. Marine lithium battery systems are normally built from prismatic cells, for practical reasons first and foremost: they are much easier to interconnect than an extremely large number of small cylindrical cells.

Charge and Discharge Mechanisms

When the battery is charged, the lithium leaves the cathode, migrates through the electrolyte and into the anode to form LiC₆ as the lithium inserts itself within the graphite matrix; in the same time, the LiFePO₄ cathode progressively turns into iron phosphate FePO₄ only. Conversely, discharge depletes the graphite anode and re-inserts the lithium into the iron phosphate crystalline structure of the cathode.

Migration of the lithium ions in a lithium iron phosphate cell throughout charge and discharge.

The insertion of lithium ions into either material is called intercalation. Both the graphite and the iron phosphate materials are very stable structurally, with or without the presence of lithium in them, which is why these batteries can be so durable. They don’t structurally suffer from the lithium intercalation and de-intercalation process, and by extension they are largely immune to cycling.

Never allow lithium batteries to sit around fully charged

When the lithium salt is initially dissolved into the electrolyte, it naturally breaks up into lithium ions Li⁺ and hexafluorophosphate ions PF₆^–. Near the end of the charge, the cathode runs out of lithium, the amount of free lithium ions in the electrolyte reduces rapidly and the cell voltage begins to rise very fast. The electrodes also becomes a lot more reactive chemically with regard to the electrolyte: this latter phenomenon is harmful to the cell and the longer this condition persists, the more internal degradation it causes. A high ambient temperature further accelerates it. Electrolyte composition and additives are the main points of differentiation between manufacturers; additives have a huge bearing not only on cell performance, but also on eventual life expectancy, because they alter the way capacity fades over time by controlling the growth rate of the SEI layer.

Current Capacity

During charging, when lithium ions are extracted from the iron phosphate cathode and inserted into the porous graphite structure of the anode, these ions are initially inserted near the surface and then gradually migrate deeper into the graphite. They need some time to be able to do so and this creates a limit to the rate at which a lithium cell can absorb current. If this limit is exceeded, lithium plating occurs and it is irreversible, so this lithium is subsequently lost to the cell and this translates into a  permanent loss of capacity. For this reason, a limit exists on the maximum allowable charge rate and this limit is also temperature-dependent, because lithium ion insertion is slower and more difficult at lower temperatures.

In discharge, ion mobility is also impeded by low temperatures, but discharging is safe at any temperature because it only happens as fast it can and the cell voltage drops if the current is too high. At temperatures where the performance of the cell is not affected, discharge rates can be limited by thermal effects; this can be an issue in electric vehicle applications during high accelerations, but not normally for marine house banks where discharge currents are mostly modest in relation with the size of the battery.

Lithium Plating

Lithium plating is a phenomenon that deserves its own paragraph because it originates primarily from improper management of the cells. It is very harmful and results in irreversible capacity loss. Lithium plating refers to the formation of solid lithium metal within the cells. Lithium metal forms when the intercalation mechanism of the lithium ions fails to take place normally and solid lithium can be deposited on the surface of the anode or around its edges typically. There are a small number of well-known situations that promote lithium plating:

• Excessive charge rates. If the charge rate exceeds the cathode’s rate of absorption for lithium ions, lithium metal is instead deposited onto the surface of the electrode. Most manufacturers recommend operating the cells at or below 0.3C in spite of much higher charge rates being readily achievable. The most recent generations of cells are rated for sustained operation at up to 0.5C.
• Cold temperature charging. Temperatures close to or below freezing dramatically reduce the absorption rate of lithium ions into the graphite cathode, because lithium insertion becomes more difficult. In other words, charge rates that are acceptable at normal ambient temperatures become excessive in cold conditions.
• Trickle charging. Trickle charging causes all of the lithium present in the cells to be transferred to the positive electrode, whether it can be normally inserted into the carbon cathode or not. Any lithium that cannot be absorbed normally ends up plating the cathode.

From the above, it should be obvious that an operating regime combining fast charges followed by a trickle charging is particularly harmful. A fast charge tends to saturate the anode surface with lithium ions and then any additional charging results in lithium plating. This scenario can quite easily be met on marine installations where a focus was placed on fast recharging through the use of an engine and alternators (or high-capacity DC chargers) and renewable energy systems then fail to properly implement a charge termination and hold the cell voltages up.

Safety Considerations with Lithium-Ion Cells and LiFePO₄ Chemistry

Safety is an important topic in the context of lithium batteries, for the simple reason that they are capable of igniting violently and burning extremely hot and can be problematic to extinguish.  

A lithium iron phosphate battery fire totally destroyed this DIY electrical vehicle, melting glass and aluminium. (Photo Greg Fordyce, February 2009)

There are many, many types of lithium-ion chemistries on the market, with more coming undoubtedly. So far, only one is truly eligible for installation on boats, and even more so for building DIY systems, because of its much higher ignition temperature: lithium iron phosphate (LiFePO₄).

The devil you know versus the devil you don’t

LiFePO₄ cells are incredibly stable and robust chemically compared to the other variants and they are also manufactured in sizes that are practical for assembling house banks. This primarily translates into a battery cell that is exceedingly difficult (but not impossible) to ignite. Considering the explosive gases produced by lead-acid cells when charging and the risks associated with them, LiFePO₄ cells should probably be seen as significantly safer in comparison, but they are lacking the long track record of lead-acid chemistry.

Batteries, regardless of their chemistry, hold significant amounts of stored energy, not only electrical but also chemical, and can cause considerable damage if things go very wrong

A marine engineer I know once went down into the engine room of a fishing boat that wouldn’t start. Flat battery was deemed to be the culprit. He and the owner brought along a spare fully-charged battery and heavy-duty jumper leads.

The starting bank was made of two regular 12V lead-acid batteries. A strong spark was observed when they connected their new battery in parallel. They saw it as a confirmation that the bank was really flat. Before they could even try and crank the engine, their battery exploded, showering them in debris and acid within the confined space of the engine room. They barely made their way out, blinded by the acid burning their eyes.

The engine starting system was in fact 24V, the two batteries were in series and there wasn’t much wrong with them either. They discharged heavily into the single 12V battery, causing an intense release of highly explosive oxygen and hydrogen until something triggered ignition.

In order to reach thermal runaway and a battery fire with LiFePO₄ chemistry, the oxygen held within the iron phosphate material needs to be released. Unlike with other lithium compositions, this is extremely difficult to achieve and requires very high temperatures to be reached. There have been many instances of inadequately installed LiFePO₄ cells that overheated and caused great concerns about a possible fire, but none that actually ignited and burned yet on marine vessels – as far as I am aware of, at the time of writing.

The fact that the technology is actually remarkably forgiving should not be an incentive for taking more risks or abusing it, but this point has already been lost by a few and a serious incident with a marine installation may well be around the corner, just like some DIY electric vehicles caught fire in the past.

The photograph above should be seen as a sobering reminder of what can happen, even with lithium iron phosphate chemistry. In this instance, the charging regime was well into over-voltage territory and the system lacked some of the most basic protection features. Ignition took place during recharging and the car burned with such intensity that glass and aluminium melted and pooled on the ground underneath the vehicle. The only notable fuel present in this purely electric car was the lithium iron phosphate batteries… Needless to say, there wasn’t enough left to be too specific about the definite root cause of the fire and such an event on board a yacht would write it off. LiFePO₄ cells can provide extraordinary performance and service on board marine vessels, but unlike some of their proponents like to claim, abusing them always causes irreversible damage and their safety can become severely compromised afterwards.

Cell Failures

How LiFePO₄ Cells Fail Catastrophically

This Winston cell built up pressure until the casing exploded.

Catastrophic failures of LiFePO₄ cells are near-invariably associated with recharging and result from over-charging and excessive voltages. These failures are normally limited to the destruction of the cells. If the cell voltage rises to 4.3V or above, the electrolyte is chemically decomposed into gaseous products that pressurise the sealed casing of the cell. If the pressure build-up is sufficient, it can cause the swollen cell to vent these gases. In some instances, cell casings ruptured violently.

The gases, once released, are highly flammable, but they don’t spontaneously ignite; most of the time, they just dissipate. Should they get ignited, they would cause some kind of gas explosion external to the cells by combining with atmospheric oxygen. Some robust form of entertainment would certainly be provided, but a battery fire would still be most unlikely. One should however note that, should the LiPF₆ salt present in the vented electrolyte find moisture to react with, hydrogen fluoride (HF) can form. This gas is fluid enough to pass through the skin, extremely corrosive and a known carcinogen.

When things go very wrong with lithium batteries, heat is nearly always involved

When enough current is involved to severely overheat the cell, the electrolyte can boil off and also vent out of the cell. In each case, the root cause can normally be traced back to a charge regulation failure with an absence of overcharge protection and an absence of temperature sensing.

A friend who had simply dropped 200Ah of LiFePO₄ batteries in his cruising yacht a few months earlier, following advice from the battery dealer who posed as an “expert”, faced an alternator fault. Twenty minutes after leaving the marina under power, standing alone in the cockpit, he noticed a most abnormal smell coming from down-below.

The cabin was filled with mist and the air was almost unbreathable. He killed the engine immediately. Lifting the lid of the battery compartment, an intense heat was coming off the destroyed bank. Two out of the four cells had swollen completely out of shape. The heat was such that he feared a fire would break out any minute and he notified the local Coastguard of the situation.

Several nearby vessels, including a 450-tonne passenger ferry, came to stand by until a police patrol vessel took the yacht in tow.

The cells took several hours to cool down to the point where they were no longer too hot to be touched.

In order to discover the root cause of the incident, the engine was later restarted at idle. The voltmeter read 17.5V at idle: the alternator had failed and was no longer regulating. Had the problem gone unnoticed a little longer, there is very little doubt that the bank would have eventually caught fire with the full output of the alternator getting dumped into it continuously.

If the elements in the cell are heated above some 200°C (for LiFePO₄ cells only, other chemistries have a lower value), the oxygen bound within the iron phosphate material of the cathode gets released. This is extremely dangerous, because this free oxygen then recombines with other elements inside the cell (like carbon) acting as “fuel”; this suddenly causes much more heat to be released and thermal runaway quickly follows. At this point, combustion has become the dominant source of heat, the cell bursts into flames and an extremely hot fire quickly propagates to neighbouring cells. While this scenario can be difficult to reach on a yacht, some installations on multihulls can feature solar arrays in excess of 1kW, other vessels are equipped with large alternators or shore power chargers, and such systems certainly have the capability of overheating a bank to ignition point, should something seriously malfunction without an independent protection mechanism to act as a line of defence.

Spontaneous Internal Failure

The other failure mechanism is by internal fault developing within the cell; in other words, the cell shorts internally. This seems to be the rarest of occurrences to say the least (I am not aware of a single instance that resulted in significant consequences with prismatic LiFePO₄ cells) and it should be noted that the same can just as well happen to a lead-acid cell and those have a tendency to explode then.

There is little actual data available to indicate what happens next, but it is worth noting that driving a nail through a cell for example and thus causing it to short-circuit internally is enough to cause it to heat up and vent profusely, but not sufficient to cause ignition.

Most other lithium chemistries spontaneously ignite in the same circumstances because, as described just earlier, the bonded oxygen gets released by thermal decomposition, but at much lower temperature. This is the main reason for not considering any other lithium chemistry than LiFePO₄ for on-board electricity storage.

Over-Discharge Failure

There is one well-known pathway promoting internal cell failure: excessive discharge. If a LiFePO₄ cell is discharged below 2.0V, at some point its polarity suddenly reverses and the anode copper substrate starts dissolving into the electrolyte. Upon recharging, this highly conductive copper gets precipitated out of the electrolyte and deposited onto the cathode surface where it forms dendrites, or crystals with very sharp features. These copper dendrites are at risk of piercing the thin insulating separator between anode and cathode and short-circuit the cell [2].

The first consequence of dendrite formation is increased self-discharge through internal micro short-circuits; those also induce cell heating during recharging. A cell that has faced such treatment can potentially fail at any time afterwards and there is no assurance any more that it might fail “nicely”. The root cause goes back to an absence of automatic over-discharge protection.

When several cells are connected in series to form a higher voltage string, if a single cell in the string gets fully discharged and cannot contribute any current any more, it then only acts as a resistor while the remaining cells keep powering the circuit, forcing current through it backwards, from anode to cathode. This type of event is disastrous for the cell and has been the #1 mechanism that resulted in fires and many more near-misses on electric vehicles being recharged afterwards. The root cause is a lack of protection at cell level once again.

Monitoring overall battery voltage is not good enough. Electric vehicles are much higher voltage systems, where dozens of cells can be connected in series; this makes it much more difficult to ensure no cell ever get into over-discharge territory, because a flat cell doesn’t show as much in terms of overall pack voltage and individually monitoring that many cells has been more difficult technically, especially in DIY systems. Specialised integrated circuits have now become available to facilitate such a task, but their integration into a system is a non-trivial task at DIY level.

In conclusion, don’t ever overcharge lithium cells and never, ever take them into over-discharge territory. A cell which has suffered a complete voltage collapse or polarity inversion is only good to be discarded immediately, even though it could appear that it can be somewhat recharged. Don’t try, don’t risk it.

The same applies to swollen, overcharged cells. Some have been irresponsible enough to crush them back flat in a press and return them to service, calling this “recovery process”, taking an unlimited risk for the sake of saving a few hundred dollars after screwing up in the first place.

Don’t operate damaged or “recovered” lithium cells on board either, ever.

Failure Following Mechanical Damage

A very interesting report [4] into the safety of lithium batteries in general was published in 2011 by the Fire Protection Research Foundation. Besides covering a lot of the information already presented here, it provides interesting and uncommon insights:

Mechanical damage (crush or penetration) that occurs at electrode edges is significantly more likely to cause cell thermal runaway than damage perpendicular to electrode surfaces.

This was demonstrated in specific tests in which the narrow edges of the cells were mechanically challenged, rather than their main faces. The explanation is simple and logical: damage at the edges of the plates tends to lead to folding-over of the electrodes, with significant risks of heavy short-circuit; compression or punctures perpendicular to the plate separators results in much more benign damage in comparison.

I am only aware of one instance where marine lithium batteries may have failed following mechanical damage on a yacht: the crash of the Volvo 70 racer Team Vestas Wind on offshore coral reefs in the Indian Ocean in 2014. The investigation report only contains two very succinct mentions of the Mastervolt lithium battery packs present on board, one to indicate that the batteries started venting following the accident and were transported to a nearby beach, and the second to mention that they were later found to have burned themselves out.

The document blames the failure on immersion in seawater, but a simple calculation based on the conductivity of sea water shows that this would only result in discharge currents of a few amps. Water ingress within the cells is also unlikely due to their sealed nature, the lack of water pressure and the fact that the relief valve is designed to open on internal over-pressure. A tentative explanation is that the shock from the sudden impact at speed caused the elements within the cell casings to shift, bruising the edges of the plates and creating a short-circuit. The cells themselves were of standard prismatic construction, 180Ah/3.2V LiFePO₄. If the violent impact indeed caused them to short-circuit internally, then it illustrates the foolishness of using large format LFP cells for building marine house banks as some have already done, sometimes resorting to 400Ah or even 700Ah individual units.

Thermal Runaway and Ignition of Lithium Chemistry Battery Cells

As already developed earlier, thermal runaway is a chemical reaction that follows dissociation of the oxygen contained in the cathode material. This reaction requires heat: overcharging in itself is enough to destroy cells, but not sufficient to set a lithium battery on fire. The amount of current and duration involved must to be sufficient to raise the internal temperature before thermal runaway can occur.
This is why high-powered charging sources such as alternators and – worse – shore power chargers can represent significant potential hazards when no reliable, effective and independent automated trip mechanism is implemented.

An interesting study [5] of thermal runaway of three different lithium cell chemistries was published in 2014. In this work, a number of small cylindrical 18650 lithium/graphite-based cells were first fully charged and then gradually heated in a specially designed chamber:

“At a critical temperature, a chain of exothermic reactions can be triggered. The reactions lead to a further temperature increase, which in turn accelerates the reaction kinetics. This catastrophic self-accelerated degradation of the Li-ion battery is called thermal runaway.

During thermal runaway, temperatures as high as 900°C can be reached, and the battery can release a significant amount of burnable and (if inhaled in high concentrations) toxic gas.”

The battery cells tested featured three different types of cathode materials:

1. LCO/NMC, a blend of lithium cobalt dioxide and lithium nickel manganese cobalt dioxide: LiCoO₂ / Li(Ni_0.50Mn_0.25Co_0.25)O₂. This chemistry is an attempt at taming the violent nature of lithium cobalt dioxide alone and the cells were rated at 2.6Ah at an average voltage of 3.8V, with a mass of 44.3 grams.
2. NMC, lithium nickel manganese cobalt dioxide: Li(Ni_0.45Mn_0.45Co_0.10)O₂. The cells had a capacity of 1.5Ah, an average voltage of 3.8V and a mass of 43.0 grams.
3. LFP, lithium iron phosphate: LiFePO₄. The cell capacity was 1.1Ah at an average voltage of 3.3V and its mass was 38.8 grams.

The gives LFP the lowest power density with 1.1Ah x 3.3V = 3.63Wh. LCO/NMC and NMC respectively deliver 9.88Wh and 5.7Wh, or 272% and 157% of the value of LFP for the same geometric volume. The calculation can be continued to include the mass of the cells, in which case LFP comes out a little better off due to its lighter weight, but still remains in last position.
This alone explains why some applications stray from the safest, most robust option represented by lithium iron phosphate. The now famous failure of a lithium cobalt dioxide backup battery on board a Boeing 787 Dreamliner comes to mind, together with questions as to why we should accept such elevated risks for the sake of saving a few kilograms on an aircraft.
In the case of marine batteries, it simply makes no sense, and even more so for DIY systems where design and installation errors can occur.

The results of the heating tests are edifying:

Thermal runaway and temperature readings for three different lithium battery chemistries.The red curves relate to lithium iron phosphate cells under fast (1) and slow (2) heating rates. Ignition always occurs around 200°C.

Lithium iron phosphate cells required the highest temperature to ignite, burned more gradually and reached much lower temperatures than the other types.

The sharp increase in heating rate indicates the onset of thermal runaway. With an ignition temperature of about 200°C, lithium iron phosphate batteries (in red) are more difficult to set on fire and then don’t heat up as violently than other types. While the Lithium-Cobalt cell reached a rate of 400°C per second while bursting into a fire ball, its LFP equivalent only recorded a rate 8°C per second.

Another graph of interest shows gas discharge rates:

Venting behaviour of various lithium battery chemistries. Unlike the other samples tested, which gased suddenly and violently, lithium iron phosphate cells (red curve) begin to release gases very gradually until reaching a steady rate.

The graph clearly highlights the lack of initial explosion in the case of the LFP cell, but also shows that combustion is still on-going after 100 seconds. A look back at the first plot in this series clearly shows that lithium cells burn long and hot once ignited and all the ingredients to spread the fire to nearby materials are present.

While LiFePO₄ batteries are remarkably safe and stable, they are not an option for reckless and/or negligent installation

Once again, this data was collected for 18650-size cells, which are very small; anecdotal experience such as the vehicle fire mentioned earlier shows that a bank of prismatic LFP cells on fire is capable of producing much higher temperatures than recorded in the study.

This information is sobering and some may decide to stay with lead-acid batteries and their drawbacks as a result: it is fine. There is irresponsible advice to be (easily) found, suggesting that building a lithium battery is just a matter of balancing a set of cells and throwing it into the boat after tweaking a few charging voltages. Those are the consequences that potentially come with such practices.

Life Expectancy and Terminal Failure of LiFePO₄ Cells

At the end of their life, lithium battery cells eventually fail by running out of capacity. Chemical damage and plating of the electrodes inside the cell eventually consumes some of the lithium originally available for energy storage, consumes electrolyte and prevents the remaining lithium from migrating between the anode and cathode; the battery can no longer be charged or discharged enough to be usable and it must be discarded.

Elevated temperatures, 40°C and over typically, greatly accelerate these adverse chemical reactions and shorten the life of the cells. In the case of installations on board marine vessels, strict care must be taken not to install them in hot environments such as engine compartments for this reason.

While small cylindrical cells mostly used in portable devices usually fail relatively rapidly due to repeated heating during recharging and electric vehicle batteries in general can suffer from heating caused by intense discharge currents caused by accelerations, prismatic cells in more conservative applications such marine house banks have been exhibiting an increasingly phenomenal track record. The oldest ones in service may be around 7 or 8 years old now (2015) and no one has yet come up with a clear figure in terms of life expectancy. It is becoming increasingly clear that the 10-year mark will be reached and exceeded by some marine lithium house banks.

Accelerated cycle testing is of no relevance in the context of marine house banks as it stresses the cells far beyond normal operation in this type of service and fails to model calendar ageing, in other words the passing of time. In a house bank application on a yacht, end-of-life will occur when there is either insufficient capacity left to cycle, or the discharge current capability is no longer sufficient to support high loads, like inverters.

To my knowledge, no properly managed lithium house bank has yet reached a natural end of life. All those that were discarded or replaced failed due to external and completely avoidable events.

References:

[1] Design and Analysis of Large Lithium-Ion Battery Systems, Shriram Santhanagopalan, Kandler Smith, Jeremy Neubauer, Gi-Heon Kim, Ahmad Pesaran, and Matthew Keyser, 2014, ISBN: 978-1-60807-713-7  

[2] Failure Investigation of LiFePO4 Cells in Over-Discharge Conditions, Hao He, Yadong Liu, Qi Liu, Zhefei Li, Fan Xu, Clif Dun, Yang Ren, Mei-Xian Wang and Jian Xiea, Journal of The Electrochemical Society, 160 (6) A793-A804 (2013)

[3] Failure Investigation of LiFePO4 Cells under Overcharge Conditions, Fan Xua, Hao Hea , Clif Dunb, YaDong Liua, Mei-xian Wanga, Qi Liua, Yang Renc and Jian Xiea, ECS Transactions, 41 (39) 1-12 (2012)

[4] Lithium-Ion Batteries Hazard and Use Assessment, Exponent Failure Analysis Associates, Inc., published by the Fire Protection Research Foundation (2011)

[5] Thermal-runaway experiments on consumer Li-ion batteries with metal-oxide and olivine-type cathodes, Andrey W. Golubkov, David Fuchs, Julian Wagner, Helmar Wiltsche, Christoph Stangl, Gisela Fauler, Gernot Voitic, Alexander Thalera and Viktor Hackere, Royal Society of Chemistry, RSC Adv., 2014, 4, 3633

[6] State of the Art of Lithium-Ion Battery SOC Estimation for Electrical Vehicles, R. Zhang, B. Xia, B. Li, L. Cao, Y. Lai, W. Zheng, H. Wang and W. Wang, Energies 2018, 11, 1820; doi:10.3390/en11071820