In this series, beginning in part 1 , we have looked at distributed impedance, transmission lines, characteristic impedance, and S-parameters. We noted that when we launch a signal into a transmission line with a characteristic impedance Z₀ connected to a two-port network under test, all or some of the signal may reflect back to the source, be absorbed in the network under test, or pass through the network under test.

Q: How do we quantify these potential transmissions and reflections?
A: Figure 1 repeats a figure from part 3 with some nomenclature changes. Here, the subscript to the voltage wave v represents the port, while the plus sign as a superscript indicates a signal moving toward the network, and the minus sign as a subscript indicates a signal moving away from the network. Thus, v1+ is a voltage waveform incident on port 1, v₁^– is a voltage waveform moving away from port 1, and so on.

The equations in the figure show the s values for a single incident waveform on the network. For example, s₁₁ equals v₁^– divided by v₁⁺ when v₂⁺ is zero. The following matrix shows the full relationship of S-parameters and the incident and transmitted or reflected voltage waves for a two-port network:

Q: How do we physically make the measurements?
A: I’ve penciled in one possibility, showing a vector signal generator on the left and a vector signal analyzer on the right. Both instruments are positioned to measure s₂₁, which is v₂^– divided by v₁⁺, essentially the forward gain of the network. We could then switch the generator and analyzer around to measure s₁₂, but this approach is tedious and error-prone.

Q: What’s the alternative?
A: We can use a vector network analyzer (VNA) with an S-parameter test set, as shown in Figure 2. The VNA includes the necessary instruments. For a two-port application, a single source (vector signal generator) and two test receivers (vector signal analyzers) often suffice. The S-parameter test set includes switches and directional couplers that route the test signals to and from the network under test. The reference receiver confirms the source’s output magnitude and phase by way of the power splitter in the test set.

Q: What’s the role of the switch and couplers?
A: With the switch in the position shown in the figure, the switch directs the source’s output (representing v₁⁺ in this case) to port 1 of the network under test. Receiver 1 can measure the reflected voltage v₁^– via the directional coupler highlighted in orange, and receiver 2 can measure the transmitted voltage v₂^– from port 2 via the directional coupler highlighted in blue, enabling calculation of s₁₁ and s₂₁.

Reversing the switch applies the source output (v₂⁺ in this case) to port 2 of the network under test. Then, receiver 2 can measure the resulting reflected voltage waveform v₂^– by way of the blue-highlighted coupler, and receiver 1 can measure the transmitted voltage wave v₁^– via the orange-highlighted coupler. These results enable the calculation of s₁₂ and s₂₂, completing our S-parameter derivation for the two-port network.

Q: Could you show us an S-parameter matrix for a specific network under test?
A: As a simple example, let’s assume that our network under test is a three-inch length of 50-W coaxial cable, our measurement system has a 50-W system impedance, and our operating frequency is 1 GHz. Because our measurement system and network under test both have 50-W impedances, there will be no reflections, so s₁₁ and s₂₂ will be zero. We can assume that our short cable is lossless, so the forward and reverse gain magnitudes will be 1. However, the wavelength at 1 GHz is about one foot, so our three-inch cable represents a quarter wavelength, and we can expect a 90° phase shift in either direction. We can therefore expect our S-parameter matrix to look like this:

Q: Can you briefly summarize what we’ve covered in this series?
A: Sure. In part 1, we looked at the limitations of the lumped impedance model and when we need to move to distributed impedance. In part 2, we looked at adapting a DC model to elucidate Heaviside’s telegrapher’s equations. In part 3, we looked at S-parameters, leading up to this conclusion in part 4 on how to measure them.

Related EEWorld Online content

What’s all this VNA calibration stuff?
Making sense of test circuits with Kirchhoff’s laws: part 3
Measuring signals on transmission lines
What are insertion loss and return loss and how can I measure them?
Should I use a spectrum, signal, or vector network analyzer? part 1
Understanding the basics: What is characteristic impedance?

The post S-parameters and distributed impedance: part 4 appeared first on Test & Measurement Tips.

We concluded part 2 of this series with a look at surge impedance, and we began considering what happens when we launch a signal into a transmission line. Figure 1 shows the basic setup, where we have a transmission line of characteristic impedance Z₀ applying a signal to a device under test (DUT) with an input impedance Z_IN.

Q: What are the possibilities when we apply the signal?
A: If the DUT is a single-port device such as an antenna or dummy load, it can absorb the signal or reflect some or all of it back to the source. If the DUT is a multiport device, it can pass all or some of the signal on to downstream devices and reflect the remainder back to the source.

Q: Where do S-parameters come in?
A: Let’s take a closer look at our DUT from the figure. As shown in Figure 2, the DUT is a two-port device with incident voltage waves a₁ and a₂ applied to ports 1 and 2, respectively, while b₁ and b₂ represent waves reflected from or transmitted through ports 1 and 2, respectively.

The S-parameters, then, are simply the ratios of incident and emitted waves, as shown by the equations in Figure 2. They are usually expressed in matrix form, where an n by n matrix represents an n-port network.  For our two-port network, the matrix is

For the two-port case, s₁₁ is also called the input reflection coefficient; s₂₁, the forward transmission coefficient; s₁₂, the reverse transmission coefficient; and s₂₂, the reverse reflection coefficient. S-parameters are complex numbers having both magnitude and phase components, with the magnitude typically expressed in decibels.

Q: What causes the matrix format to be more suitable for the RF/microwave domain compared with low-frequency circuits?
A: Nothing. Recall in a recent series, we used a matrix format so we could use an online solver to solve Kirchhoff loop and node equations. In fact, we can characterize any circuit at any frequency using matrices, with low-frequency circuits lending themselves to Z-parameter analysis, where the letter Z (for impedance) takes the place of each letter S in the above matrix. For a two-port network, Z-parameters take this form:

Consider the simple circuit in Figure 3. For this circuit, z₁₁ equals v₁ divided by i₁ with port 2 open-circuited, or 7.8 kW, and z₂₂ equals v₂ divided by i₂ with port 1 open-circuited, or 6.8 kW. z₁₂ equals v₁ divided by i₂ for i₁ equals 0, and  z₂₁ equals v₂ divided by i₁ for i₂ equals 0, with z₁₂ and z₂₁ both also equaling 6.8 kW.

We can use these values to construct a Z-parameter matrix equation and solve problems such as, what is v₂ if v₁ equals 1 V and i₂ is zero?

We can solve this equation as follows:

So this is a pretty complicated way of solving a simple voltage-divider problem, but it demonstrates that the matrix approach isn’t limited to RF/microwaves. In fact, if you have an S-parameter matrix, you can convert it to a Z-parameter matrix, and vice versa. Twenty-five years ago, I wrote an article^[1] providing more details on the relationship between S-parameters and Z-parameters. In general, we use Z-parameters for low-frequency circuits, because it’s relatively easy to apply voltages to or insert currents into low-frequency circuit nodes and measure the responses. But when signal wavelengths shrink to the dimensions of our circuit, it becomes easier to apply a waveform and measure the resulting reflected and transmitted signals, and we use matched impedances instead of the open circuits we use for constructing Z-parameter descriptions.

Q: How exactly do we measure S-parameters—that is, what equipment do we need?
A: The vector network analyzer is the instrument of choice. We’ll conclude this series next time with a look at the details of making measurements.

Reference

What are S-parameters, anyway? EDN

Related EEWorld Online content

Making sense of test circuits with Kirchhoff’s laws: part 3
Measuring signals on transmission lines
What are insertion loss and return loss, and how can I measure them?
Should I use a spectrum, signal, or vector network analyzer? part 1
The principle of impedance matching
Understanding the basics: What is characteristic impedance?

The post S-parameters and distributed impedance: part 3 appeared first on Test & Measurement Tips.

Custom test instruments usually mean one of two things: settle for a standard box that almost fits your application, or commit to months of FPGA development and find someone with the expertise to do it. Liquid Instruments is betting there’s a third option.

The company’s new GenInst Studio takes a natural language description of the instrument you need and builds it as a deployable, validated instrument on Moku reconfigurable hardware. You describe the function in a chat interface, an agentic AI workflow walks through specification, design, and deployment, and the result runs in hardware with real-time, low-latency performance. No HDL, no FPGA toolchain.

The workflow is auditable at each step, which matters. A black box generating your measurement chain is not something most engineers would trust, so the ability to inspect what the AI actually built is doing real work here.

Target applications are the ones that typically push teams into custom development: hardware-accelerated DSP, custom triggering, closed-loop controllers, and adaptive signal generation. Early users report that projects which previously took months of specialist effort came together in a single session. Worth treating that claim with the usual caution until independent results show up, but the direction is credible given the platform underneath.

Some context: Liquid Instruments closed a $50 million Series C earlier this year, co-led by Keysight Technologies and Australia’s National Reconstruction Fund Corporation. Its Moku platform is already in use at NASA, NIST, and Stanford, so GenInst Studio is landing on installed hardware rather than asking anyone to buy into a new ecosystem.

GenInst Studio is available now for Moku users at liquidinstruments.com/geninst.

The post Agentic AI turns natural language prompts into working test instruments appeared first on Test & Measurement Tips.

Part 1 of this series covered distributed and characteristic impedance as well as the transmission-line concept as a prelude to reviewing S-parameters.

Q: Before we move on, I have a question. Does characteristic impedance apply at DC?
A: Yes. If frequency f equals 0, the equation we presented last time for characteristic impedance becomes

where R’ and G’ are the series resistance and parallel conductance per unit length.

Q: Say we have 1-m of cable with series resistance of 1 W and parallel resistance of 100 W, for an R₀ of 10 W. If we terminate that cable with 10 W and look in the other end, we should see 10 W, but by my calculation we don’t. What am I missing?
A: You may be calculating based on the lumped-element model at the top of Figure 1. We can easily calculate that the equivalent resistance R_EQ is about 10.09 W, higher than we expect. But keep in mind that the series and parallel resistance are distributed equally throughout the cable—not lumped in one location.

Q: How can we handle that?
A: We can use the telegrapher’s equations, developed by Oliver Heaviside in the 19th century, which are two partial differential equations that describe transmission-line voltages and currents as a function of time t (not relevant for the DC case) and distance x. Alternatively, we can apply brute force computing that wasn’t available to Heaviside.

Q: How does that work?
A: Figure 1 at the bottom illustrates the basic approach. We divide our cable into four 25-mm elements. Then, working backwards from the right, we can calculate the equivalent resistance of each element, starting with element 1 terminated by R₀. Then we move on to the other elements, each terminated by the downstream equivalent impedance R_EQ. For our final calculation, we find that the total R_EQ is about 10.02 W.

Q: So the more elements, the higher the accuracy?
A: Right. Figure 2 shows the calculated equivalent values for one, two, four, 10, 50, and 100 elements. I performed a calculation out to 1,000 elements, and the calculated R₀ shrank to about 10.000091 W.

Q: 1,000 elements must require a lot of loop or node equations.
A: No, it’s recursive, so there is just one equation, solved 1,000 times. Figure 3 outlines the process in Excel. The calculated R_EQ for each element becomes the termination resistance R_T for the next element, as shown by the red arrows. After you complete line 8 in the spreadsheet, you can just drag and drop it until you have 1,000 instances.

Q: Another question: What is surge impedance?
A: Surge impedance is numerically equal to characteristic impedance, but it implies a certain condition. If we apply a pulse or sinusoidal signal to either of the top two transmission lines in Figure 4, the source will see the characteristic impedance Z₀, and it will continue to see Z₀, as it applies additional signals.

However, for the illustration on the bottom, the termination impedance Z_T does not equal Z₀. Because of the mismatch, some of the incident input waveform will be reflected back to the source, with the reflected current altering the impedance the source sees. But for our 1-m cable, the reflected wave won’t return to the source until after a delay of 6.6 ns. So for more than six and a half nanoseconds—that is, during the initial “surge” of the signal—the source will see Z₀.

Q: How much of the signal reflects back, and what effect does it have on the impedance?
A: The strength of the reflected signal depends on the scattering parameter (S-parameter) s₁₁, also known as the input reflection coefficient. We will take a closer look in the final part of this series.

Related EEWorld Online content

What are the telegrapher’s equations?
Making sense of test circuits with Kirchhoff’s laws: part 3
The difference between VSWR and the S₁₁ reflection coefficient
How are impedance mismatches at connector interfaces handled in mixed-voltage EV architectures?
What is the voltage standing wave ratio (VSWR) in RF systems?
Using the Smith Chart for impedance matching, Part 1

The post S-parameters and distributed impedance: part 2 appeared first on Test & Measurement Tips.

SiC and GaN introduce broadband electromagnetic interference (EMI) challenges across conducted and radiated emissions testing, shielding, grounding, and high-frequency measurement. This article reviews how EMI testing is evolving to address those issues in EV power architectures.

Why wide-bandgap devices change the EMI problem

As shown in Figure 1, SiC and GaN outperform silicon across the key material properties that drive faster switching speeds and broader EMI spectra.

SiC and GaN transistors can switch with rise and fall times under 20 nanoseconds (ns) in EV traction designs. These steep edges push harmonic energy well above 10 MHz, extending EMI concerns into frequency ranges addressed by conducted, radiated, and immunity testing.

For example, an EV inverter operating at a 20 kHz switching frequency can become a significant EMI source from roughly 50 kHz to 50 MHz as large di/dt and dV/dt transients drive energy into higher harmonics.

These fast transitions can also increase common-mode noise relative to IGBT-based designs. Parasitic capacitances between switching nodes and chassis couple common-mode currents into the high-voltage (HV) bus and motor phase cables, turning them into potential radiating structures.

These coupling paths reduce the margin available for meeting formal EMI limits and electromagnetic compatibility (EMC) requirements, even when the standards themselves have not changed as much as the devices. As a result, simulations targeting parasitic coupling paths and filter effectiveness now enter the design flow before formal EMC testing begins.

Conducted emissions: test configurations and filter design

EV conducted emissions testing centers on CISPR 25 methods and OEM-specific requirements. Measurements typically extend from 150 kHz into the tens of megahertz and, depending on the method, up to 108 MHz.

WBG switching edges generate significant noise above 10 MHz, where resonances can subsequently contribute to radiated failures during validation. Pre-compliance workflows increasingly extend frequency sweeps beyond the required range before full antenna-based testing.

As shown in Figure 2, CISPR 25 defines two primary conducted emissions methods: line impedance stabilization network (LISN)-based voltage measurement and current-probe measurement on harnesses.

Both require probes, LISNs, and receivers with sufficient bandwidth to maintain accuracy with WBG transition speeds.

Test bench configuration is just as important as measurement equipment. Laboratory power supplies with internal Y-capacitors to earth can create unintended common-mode paths that distort measured EMI from onboard chargers and dc-dc converters. To avoid those artifacts, many test setups use isolated sources or dedicated line filtering. This configuration discipline is especially important during filter validation.

Filters must provide high-frequency common-mode and differential-mode attenuation while withstanding full HV bus conditions and peak traction currents without saturation. This creates a design constraint for Y-capacitors. Because IEC and OEM specifications limit HV-to-chassis leakage current, filter designers cannot simply increase capacitance to extend high-frequency attenuation.

High-inductance common-mode chokes and multi-stage LC networks tuned with realistic parasitic models help compensate for leakage-current limits on Y-capacitance. In WBG designs, parasitic capacitance, mutual inductance, and PCB layout coupling paths frequently dominate residual conducted EMI. Bench validation with spectrum analyzers and time-domain oscilloscopes is essential for identifying these effects.

Radiated emissions and near-field diagnostics

Radiated emissions testing under CISPR 25 spans 150 kHz to 2.5 GHz at the vehicle and component levels. WBG harmonics and enclosure resonances in the 30 MHz to 300 MHz range are common sources of compliance risk. Because chamber time is expensive and late-cycle remediation can add significant program cost, many pre-compliance workflows now rely on near-field scanning of boards, modules, and harnesses with electric-field and magnetic-field probes before far-field antenna measurements.

Near-field H-field and E-field probes help separate current-driven and voltage-driven emissions sources. H-field probes identify high di/dt loop-current hotspots, while E-field probes identify high dV/dt nodes. This distinction helps determine whether mitigation should focus on reducing loop area, modifying return paths, adding shielding, or reducing node capacitance.

Near-field scanning helps identify cable-related radiation paths. Traction inverter HV cables and motor phase conductors can carry common-mode currents excited by WBG switching transitions, producing substantial radiation even when differential-mode currents are well filtered.

Radiated emissions can also originate from enclosure apertures and poorly bonded seams that behave as slot antennas. Practical design rules limit the longest dimension of any opening to well below one-twentieth of a wavelength at the highest relevant frequency, leaving very small permissible gap dimensions for WBG harmonic spectra. For cable-related emissions, lossy ferrite common-mode chokes placed at cable ends, where antenna impedance is lowest, help reduce radiation in the 10 MHz to 300 MHz band.

Shielding effectiveness and grounding strategies

Many HV harnesses near EMI-sensitive low-voltage electronics use 360-degree shield terminations at connector backshells. Single-ended, one-point shield terminations become ineffective above a few hundred kilohertz, where WBG inverter harmonics are strongest.

Shields must connect as part of the enclosure structure rather than function as high-frequency current return paths. Routing return currents through the shield causes it to behave as a radiating element.

Thin, highly conductive shields provide high-frequency E-field containment when bonded to chassis with low-inductance connections around the full perimeter. EMC test programs increasingly verify shield bonds around the perimeter rather than relying on assumed performance from spot contacts.

Some EV architectures place WBG power stages near sensitive low-voltage electronic control units (ECUs). This proximity requires formal analysis of common-mode coupling paths from the HV bus through device and cable parasitics to chassis and into low-voltage systems. OEMs define bonding points between HV returns, low-voltage grounds, and chassis to control loop areas and transient voltage differences.

The same return-path discipline applies at the PCB level. Solid, continuous ground planes minimize loop inductance and H-field radiation in WBG converter designs. Split planes remain appropriate only where galvanic isolation requirements force a separation. When splits are necessary, a single well-defined connection point helps prevent large ground loops that increase conducted and radiated emissions.

High-frequency measurement and design correlation

Capturing the fast, high-energy switching events of WBG devices requires higher probe bandwidth and common-mode rejection than IGBT-era test equipment typically provides.

Standard probes and current clamps can underestimate high-frequency content, producing inaccurate EMI data that leads to under-designed filters and grounding strategies. As shown in Figure 3, double-pulse testing of WBG devices, combined with high-bandwidth pulse-isolated probes, characterizes voltage overshoot, ringing, and layout-induced common-mode currents that translate into EMI compliance risk.

These measurements are especially useful when correlated with frequency-domain emissions data. Correlation allows engineers to link gate-drive parameters, snubber values, and PCB layout decisions to specific emissions signatures. An integrated debug workflow identifies which design variables affect compliance margins before formal CISPR and ISO test submissions.

Even with frequency-domain correlation, accurately predicting absolute emissions levels remains challenging because small parasitic variations and environmental coupling can dominate results.

To address this limitation, engineers use electromagnetic and circuit co-simulation to evaluate trends and filter options rather than predict exact emissions levels. Emerging machine learning frameworks may further reduce full compliance test iterations by monitoring and predicting EMC behavior from inverter operating data.

Summary

EMI testing for EV power electronics with SiC and GaN devices emphasizes wider frequency coverage, earlier pre-compliance workflows, and tighter time/frequency-domain correlation.

Conducted testing extends beyond formal CISPR ranges to identify WBG resonances before they contribute to radiated failures. Near-field scanning isolates high di/dt loops and high dV/dt nodes before chamber testing. Shielding and grounding strategies prioritize 360-degree terminations, perimeter-verified bonds, and common-mode path analysis.

This evolution reflects the core WBG challenge: broadband, nanosecond-transient EMI requires design-phase characterization rather than end-of-cycle compliance testing.

References

• Optimizing Wide-Bandgap Semiconductor Switches for EMI Compliance, Rohde & Schwarz
• EV EMC design: shielding, grounding, and CISPR 25, PatSnap
• EMC Tips and Tricks for WBG Devices, Wurth Elektronic
• Power Modules: One Shortcut to EMI Compliance, Wolfspeed
• Addressing EMI Conducted Emissions Challenges in EVs with GaN-Based OBCs, TI
• Silicon Carbide and Gallium Nitride Bring New Challenges for Semiconductor Test, Teradyne
• How to Test Wide-Bandgap Semiconductors, Keysight
• Wide-Bandgap Semiconductors: Performance and Benefits of GaN Versus SiC, TI
• Effective PCB Layout Techniques to Minimize EMI from ESD Events, AllPCB
• Mastering EMI Control in PCB Design: Crosstalk Prevention for Better EMI, Altium

Related EE World content

• What Role Do WBG Semiconductors Play in EVs?
• Integrated Test Cell Supports Demand for SiC and GaN Power Semiconductors
• Trends for GaN Adoption in xEV and Clean Power Systems
• How Semiconductor Materials Affect EV Power Electronics?
• How to Overcome the Test and Measurement Challenges with WBG Devices

The post How is EMI testing evolving for wide-bandgap EV architectures? appeared first on Test & Measurement Tips.

In a recent series, we reviewed Kirchhoff’s voltage and current laws and their application to simple test circuits, including the Wheatstone bridge. The techniques we discussed work well for DC and low-frequency signals, for which we can model our circuit using lumped impedance elements. However, when our signal’s wavelength is of the same order of magnitude as the length of the conductor it traverses, those techniques no longer suffice.

Q: Why does zaSthe lumped impedance concept break down?
A: Consider the test circuit in Figure 1, which includes a battery, switch, a resistor divider, an ammeter, and a voltmeter. Let’s say we close the switch for 10 seconds. That gives us time to record the voltage across the device under test (DUT) and the current through it. But as a thought experiment, let’s assume the battery is located one foot from the DUT, and we close the switch for only one picosecond.

Q: It seems that the current won’t have time to reach the DUT in one picosecond.
A: Right. The speed of light is 186,000 mph, or 982 million feet per second. The inverse of this latter figure is about one nanosecond per foot. That’s a useful fact to keep in mind—it takes about one nanosecond for electricity to travel one foot.

Q: How do we analyze what’s happening?
A: First, we note that our 1-ps pulse approximates an impulse, which has infinite frequency content—including frequencies with wavelengths much shorter than one foot. Consequently, our conductor becomes a transmission line, and we need to move from the lumped-impedance model of Figure 1 to a distributed-impedance model.

Q: Transmission line—would that be a coaxial cable?
A: Yes, coaxial cables are often found in RF and microwave applications, but other types also find use. For example, microstrip and stripline versions^[1] can be found on printed-circuit boards. Transmission-line theory was initially developed in support of telegraphy. And of course it’s also applicable to 50/60-Hz power distribution, where the frequencies are low but the distances are very long.

Q: What does distributed impedance look like?
A: Figure 2 shows distributed impedance modeled as an infinite series of lumped impedances, shown at the bottom as blue-shaded rectangles. The image at the top shows the relevant parameters for one of those lumped impedance rectangles, where L’, R’, C’, and G’ represent the per-unit-length series inductance, series resistance, shunt capacitance, and dielectric conductivity, respectively, of an infinitesimally small segment of length dx.

Q: Is distributed impedance the same as characteristic impedance?
A: They are related, and for any frequency f you can calculate the characteristic impedance Z₀ from the per-unit-length resistance, inductance, capacitance, and conductance:

Q: What does this look like in the real world? It seems like we can’t just take a transmission line of arbitrary length and measure Z₀.
A: Right. Z₀ is what a source would see looking into one end of an infinitely long transmission line, as shown in Figure 3 at the top.

Q: What if we don’t have an infinitely long transmission line?
A: We can shorten our transmission to any arbitrary length. If we terminate it with a fixed impedance equal to Z₀ (Figure 3, bottom), the source will continue to see an impedance of Z₀.

Q: What happens if we terminate with some other impedance?
A: Then, things get interesting. We’ll look more closely next time and investigate how to use S-parameters to characterize transmission-line circuits.

Reference

[1] What is the Difference Between Microstrip and Stripline in PCBs? Sierra Circuits

Related EEWorld Online content

Making sense of test circuits with Kirchhoff’s laws: part 1
How to measure PCB trace or power/return plane impedance
How physics relates signal integrity, power integrity, and EMC
What is a Smith Chart and why do I need one? (Part 1)
FAQ on cable impedance: 50 Ω versus 75 Ω
Why do high frequency signals reflect?

The post S-parameters and distributed impedance: part 1 appeared first on Test & Measurement Tips.

Q: What is a key factor in high-performance mmWave systems?
A: The controlling frequency source for the system’s local or other oscillators is critical. While absolute nominal accuracy is important, small static errors in that signal can be accommodated by use of a phase-lock loop (PLL) or other scheme.

More challenging is phase jitter, the minute and unpredictable variations of the oscillator phase (and thus also frequency)  around the nominal value, which degrades performance. It has several causes, with the predominant one usually being thermally induced random motion of atoms; there are other deep physics causes as well.

Q: How is the mmWave oscillator frequency developed?
A: One widely used way is to generate a microwave carrier is to begin with a high-performance crystal-based oscillator at a lower frequency, typically in the tens of megahertz maximum, as it is not possible to use a crystal directly above that frequency range.

Q: Does this solve the jitter problem?
A: No, but it changes it. The compounding problem occurs when the lower-frequency crystal output is up-converted to the desired frequency, and an action that requires multiplying it by a factor of ten or more using a special circuit and filters.

Q: Again, what’s the problem here?
A: Up conversion inherently multiplies the phase noise of the crystal along with its fundamental, so the resultant oscillation in the gigahertz range now has far more phase noise than the fundamental had.

Q: How do OFCs help with this problem?
A: In a counterintuitive turn, the solution is to use an OFC and somehow down-convert its output, bringing it down to the desired GHz-range value. The down-conversion process actually shrinks the OFC phase noise by the down-conversion “multiplication” factor.

Q: Is this easy to do?
A: It depends on what you mean by “easy.” Nothing is easy in the optical, GHz, or related world, and the challenges are in both the concept and the details of its execution.

Q: With the right arrangement, what sort of down-conversion factor is possible?
A: The down-conversion ratio can be as high as six orders of magnitude, thus bringing the optical frequencies down to RF.

Q: How is this achieved?
A: In very simplistic terms, it is analogous to using a mechanical gear train to reduce motor speed. The OFC output rate is “pinned” (held constant) at two spectrum points and locked to an external laser reference output. The locking action results in a dispersion of new frequencies, all the way down to the gigahertz range, as shown in Figure 1:

Q: What hardware is needed to make this happen?
A: This is not an intuitive down-conversion scheme, as seen in the conceptual block diagram of Figure 2:

Q: That really doesn’t seem so complicated, but what’s the reality?
A: Figure 3 shows a deeper dive into the physical implementation, and there is an even more detailed perspective in the full schematic diagram, of course (not shown).

Q: Is this system available on a single or several electro-optical devices, similar to electronic ICs?
A: At present, this is not implemented as a single-chip device or even as a system with just a few discrete optical components; many of the needed precision functions are only available on individual substrates. A complete high-performance system takes a rack-sized chassis fitting in a single-height bay.

However, there has been significant progress on putting multiple functional locks onto individual substrates, so it wouldn’t be surprising to see a monolithic (or nearly so) device within a decade or perhaps just a few years, as shown in Figure 4:

Q: What sort of performance can such a system deliver?
A: There are lots of numbers and perspectives to consider, and testing these systems at these levels of performance to assess their capabilities is as much of a challenge as fabricating them.  It’s the classic metrology dilemma: how do you test a precision device? How do you validate the testing arrangement itself?

One project test result shows that for a 10-GHz carrier, the phase noise is −102 dBc/Hz at 100 Hz offset and decreases to −141 dBc/Hz at 10 kHz offset. Another characterization compares this performance to that of other available techniques (Figure 5).

Conclusion

The blending of, and synergism between, the historically separated classical RF and optical-band functions and components is occurring at a rapid rate due to the aggressive needs of the signal and data communities. There’s been impressive progress in theory, analysis, modeling, real components, and systems. Highly integrated on-chip photonics is getting a lot of attention, and while much of this is being driven by data center needs, there are many “spillover” benefits as well beyond that area.

References

[1] 20 years of developments in optical frequency comb technology and applications, Communications Physics/Nature
[2] Frequency comb, Wikipedia
[3] Frequency Combs, RP Photonics AG
[4] Fundamentals of frequency combs: What they are and how they work, NIST
[5] Optical Frequency Combs, NIST
[6] Optical and microwave metrology, NIST
[7] Ultrastable Photonic Microwave Oscillators: Purity Verified, Microwave Journal
[8] Photonic chip-based low-noise microwave oscillator, Nature/Springer
[9] Compact and ultrastable photonic microwave oscillator, Optics Letters
[10] Photonic Microwave Sources Divide Noise and Shift Paradigms, Photonics Spectra
[11] A chip-integrated comb-based microwave oscillator, via Arvix
[12] Compact and ultrastable photonic microwave oscillator, ResearchGate
[13] Photonic Microwave Sources Divide Noise and Shift Paradigms, Photonics Spectra
[14] Optical Frequency Combs for Low Phase Noise Microwave Generation, URSI
[15] Integrated optical frequency division for microwave and mmWave generation, Nature
[16] Reaching Unprecedented Sensitivity in Cross-Spectrum Phase Noise Characterization, Menlo Systems
[17] Ultrastable Photonic Microwave Oscillators: Purity Verified, Menlo Systems

Related EEWorld Online content

Review: Tekbox TBCG4 harmonic comb generator
Review: Picotest J2150B comb injector
Lasers, optics, electronics and more yield terahertz sources, Part 3 – Infrared lasers and plasma
Understanding electro-optic modulation

The post Frenemies at last? FAQ on optical combs for microwave oscillators: part 2 appeared first on Test & Measurement Tips.

In this article, I’ll show you how I make a quick measurement of cable shielding effectiveness for coaxial cables using the R&S MXO38 oscilloscope. This is important because poorly shielded coax cables, when attached to your equipment under test, can cause radiated emission failures.

The measurement test setup is shown in Figure 1, where I’m injecting strong 100 MHz harmonic combs via an RF current probe clamped around the cable or cables. The MXO3 can display up to four spectral plots simultaneously, as shown in the figure; however, for accuracy, I’ll test each cable separately.

I’m using the Tekbox TBCG3-RN6 harmonic comb generator [2], which creates harmonics out to 6 GHz at 10, 25, 50, and 100 MHz combs. I elected to use 100 MHz combs to keep the data set reasonable.

These harmonic combs are injected into a Tekbox TBCP2-750 RF current probe [3], which is coupled to the cable or cables to be characterized.

Shielding effectiveness measurement

In order to characterize the shielding effectiveness of the coax cable shield, I’ll compare the resulting comb amplitude versus frequency of the shielded cable with the coupling to a single unshielded wire and then take the difference in readings according to the formula shown.

Shielding Effectiveness (SE) in dBµV = Vwire(dBµV) – Vcable(dBµV)

The MXO38 was configured to 50Ω input on Channel 1. The Spectrum Analyzer was turned on, and Start/Stop frequency was set to 1 MHz and 1 GHz, respectively, and Resolution BW set to 100 kHz. I used Max Hold to steady the resulting amplitudes. I used Autoset to achieve a usable display of the time domain.

All cables under test were approximately 3 feet in length and were terminated with a good 50Ω load. I’ll test each cable separately.

Calibration

To establish a baseline measurement, I injected the combs into a single unshielded wire connected via a Pomona BNC-Banana Plug adapter connected to Channel 1 (Figure 2). We’ll assume this is the maximum coupling at each of the 100 MHz combs. These baseline data were recorded.

Next, I connected a length of semi-rigid coax to demonstrate the optimum shielding. To no surprise, I did not detect any combs coupling inside the coax (Figure 3). This confirmed the test setup was performing as expected.

Cable measurements

Next, I measured each cable under test, comparing several different types and general qualities. The amplitudes versus frequency data were recorded for each. Figure 4 shows the measurement of a standard Pomona 3-foot cable. Altogether, six cables were tested.

Results

The results were surprising. I expected the RG316-type cables to be among the best, but it seems all the cables seemed fairly bunched together up to about 600 MHz. After that, they seemed to differentiate more.

The Aaronia and Amazon cables generally had better SE above 600 MHz. The biggest surprise, though, was the HP cable, which was at least 30 years old and had seen a lot of use over the years.

The comb amplitude for the HP cable was either buried in the noise floor or was just visible above the noise floor. I tried a second HP cable with the same results. Even when testing the four cables simultaneously in Figure 1, you cannot observe any harmonic combs (upper left quadrant). I did notice these cables were not as flexible as the others (the Gore cable being the exception).

Dissection of the cables

While I really hated the thought of opening up one end of my excellent HP cable, I really did want to understand why it measured 20 dB higher SE and decided to compare it with the standard Pomona cable. Figure 6 shows the difference in weave and termination.

Both cable shields are terminated in a 360-degree bond to the connector ground shell. The HP appeared to be crimped as well as soldered. The Pomona used a conventional BNC connector where the shield was clamped all around with a threaded nut.

While the Pomona cable performed as well as most others, the best I could tell was that the HP cable had a tighter weave with very few gaps. Gaps are readily observed in the Pomona cable shield.

This is not the worst I’ve seen. I had purchased a generic unbranded coax cable from Amazon many years ago. Figures 7 and 8 show close-ups of the shield termination and weave. While fairly short, the cable shield termination was a soldered pigtail, which is not nearly as good as a 360-degree bond. Looking at a closer view of the weave (Figure 8) shows one of the loosest weaves I’ve ever seen, with large gaps throughout. Note that this cable was “very flexible”.

Summary

Poorly shielded cables used for radiated emissions compliance testing can be quite risky and usually lead to emissions failures. Quick characterization tests, such as the simple shielding test using a harmonic comb generator and RF current probe, can provide a quick check on the shielding quality and connector bonding of proposed test cables.

With its fast acquisition update rate, the MXO38 spectrum feature works extremely well for general bench-top EMC troubleshooting and debug, and it’s become one of my favorite bench-top tools [4 and 5]. The ease of setting up the analyzer is a step ahead of other leading manufacturers.

References

[1] Rohde & Schwarz
[2] Tekbox comb generators
[3] Tekbox current probe
[4] Wyatt, R&S MXO3 Oscilloscope for EMC measurements: part 1
[5] Wyatt, R&S MXO3 Oscilloscope for EMC measurements: part 2

The post Measuring cable shielding effectiveness with a R&S Mxo3 oscilloscope appeared first on Test & Measurement Tips.

Between the two regions lies the terahertz band spanning approximately 100 GHz to 10,000 GHz (0.1 THz to 10 THz). This is a region of the spectrum where there are few viable electronic or optical devices for designers, despite their longing to access that wide swath of unused and largely inaccessible bandwidth.

The reality of the RF and optical gap is intense. For the RF world, there are electronic amplifiers, switches, analog/digital and digital/analog converters, and power devices; the optical world has its more esoteric devices, such as all-optical amplifiers, lasers, Bragg gratings, Fabry–Pérot interferometers, Brillouin spectroscopy, Raman spectroscopy, Pockels cells, and more. Perhaps not surprisingly, the interposed terahertz zone has little to call its own.

Despite the existence of these two non-overlapping worlds, there has been considerable research and progress in enabling each to support the other and leverage their capabilities to enhance the other side. One good example is the optical comb, a fascinating structure with unique output that is being used to create high-performance microwave oscillators for RF applications. This FAQ will look at these combs and how they are used for this critical function.

Q: What’s an optical comb?
A: Despite its name, it is not a device for untangling knots in strings of photons or untwisting convolved, propagating optical waves. Instead, it is a somewhat esoteric and complicated laser-based arrangement for generating a continuous stream of precisely spaced optical pulses. The spacing is so near-perfect and consistent that the optical comb is in many ways superior to the classic “atomic” clock, and is even being evaluated for use as a primary time standard in place of the latest iteration of these clocks.

(Note that it is also called a frequency comb, or optical frequency comb (OFC), but those designations are not as crisp as the simpler, attention-getting term “optical comb”.)

Q: What can these optical combs be used for? Or are they just laboratory demonstrations?
A: Due to the precision of the comb output, it is already widely used as a standard for precision measurement of frequency (and thus wavelength) of optical signals and for establishing an optical frequency source.

A basic frequency comb can be used as an optical ruler: If the comb frequencies are known, the frequency comb can be used to measure unknown frequencies by measuring beat notes, which reveal the difference in frequency between the unknown frequency and the comb frequencies, as depicted in Figure 1.

Optical comb history

Q: What is the history of the optical comb?
A: The development of the optical comb was not an “accidental” discovery, nor was it an obvious intuitive development or even a counterintuitive “flash.” Instead, it was the product of deep insight into optical physics and lasers, and the implications of that discipline.

Optical frequency combs began as part of the ongoing quest for optical atomic clocks.  In the late 1970s, Theodor W. Hänsch originated the idea for the optical frequency-comb technique, in which ultrashort pulses of laser light create a set of precisely spaced frequency peaks that resemble the evenly spaced teeth of a hair comb. John Hall extended this research on measuring optical frequencies. Although a procedure known as the optical frequency chain had already been developed to make such measurements, it was so complex that it could be performed in only a few laboratories.

Hänsch and Hall simplified the process concept, although “simplified” is a very relative term here, of course. Their technique offered a practical way of obtaining optical frequency measurements to an accuracy of 15 digits, or one part in one quadrillion. These two were recipients of one-half of the Nobel Prize in Physics in 2005 (the other half of the prize went to Roy J. Glauber for a totally unrelated advance).

Q: What are some specifics of the actual applications of OFCs?
A: It became a long list once OFCs were available as practical, non-lab instruments. They quickly appeared in a multitude of diverse applications covering optical, atomic, molecular, and solid-state systems, including X-ray and attosecond pulse generation, coherent control in field-dependent processes, molecular fingerprinting, trace gas sensing in the oil and gas industry tests of fundamental physics with atomic clocks, calibration of atomic spectrographs, precision time/frequency transfer over fiber and free-space, arbitrary waveform measurements for optical communication, and precision ranging.

Optical clock principles

Q: So, how does an optical clock work?
A: It would be nice to be able to say that there is an easily presented and understood way to explain it, but that is not the case. In addition to a coarse qualitative explanation, there are also equations related to deep physics, including the almost trivial yet critical “comb equation.” There are several ways to generate the comb:

• Using a mode-locked laser to produce a series of optical pulses separated in time by the round-trip time of the laser cavity. The spectrum of this pulse train approximates a series of what are called Dirac delta functions separated by the repetition rate (the inverse of the round-trip time) of the laser.
• Using four-wave mixing, where intense laser lights at three frequencies combine to produce light at a fourth frequency. If the three frequencies are part of a perfectly spaced frequency comb, then the fourth frequency is mathematically required to be part of the same comb as well.
• Two other techniques involve the use of a single laser that is coupled into a microresonator and the electro-optic modulation of a continuous-wave laser.

Each approach has different attributes with respect to complexity, stability, error sources, noise, and other parameters. Regardless of the technique used, the optical frequency comb products pulses that, under properly controlled conditions, are superior to even the latest atomic clocks.

Q: Is that all there is to the “how it works”?
A: Of course not, and the references give more details about this complex and non-intuitive technique from differing perspectives. Ironically, there is an underlying and deceptively simple equation called the comb equation that links the use of two light frequencies in such a clock:

V_N = N × f_r  + f₀

where N is an integer mode number between 100,000 and 1,000,000 that multiplies f_r from the microwave domain to the optical domain, even though the “hardware” is not simple, as seen in Figure 2:

Q: Is there a summary of this function?
A: While an OFC consists of up to a million optical modes spanning hundreds of terahertz in the optical domain, there are only two degrees of freedom: (1) the repetition rate, f_r, and (2) laser offset frequency, f₀, are needed to define the frequency of each individual optical mode, ν_N.

This ability to completely define optical frequencies in terms of microwave frequencies was the original connection between OFCs and precision optical metrology. To summarize, mode-locked lasers (MLLs) can enable near-perfect coherent division of optical frequencies to the microwave domain, and coherent multiplication of microwave frequencies to the optical domain.

Q: How is an OFC constructed?
A: The first OFC was a rack of instrumentation plus a tabletop assembly of fiber optics, lasers, gleaming metal tubes, and more. In recent years, progress has been made in shrinking this down to fewer, more compact components, many of which are based on the many advances in integrated photonics, where the functional elements are built in and on an optical-compatible substrate, similar to how an electronic IC is made.

Now, you can buy a ready-to-go OFC as a single box, such as the one seen in Figure 3. The Thorlabs Compact Optical Frequency Comb is a compact, fully-automated optical frequency comb housed in a standard 19″ 3U rack housing, designed and built for use both inside and outside of the optics lab. It measures optical frequencies anywhere in the 630 – 2000 nm wavelength range, and can measure the frequency of external lasers with 14 digits of precision.

Q: What are the next steps in system and hardware evolution?
A: There is a considerable amount of ongoing work focused on the development of chip-scale systems based on microresonators, along with semiconductor processes and fabricating, but for optical signals rather than electronic ones. The compact size of these systems shows the possibilities of chip-scale and photonically integrated OFC sources. Advanced university-based photonic research programs are working on this with impressive success, as are independent research companies.

Q: All this OFC progress is impressive with respect to metrology, but what does it have to do with microwave oscillators?
A: Part 2 will explain this relationship.

References

[1] 20 years of developments in optical frequency comb technology and applications, Communications Physics/Nature
[2] Frequency comb, Wikipedia
[3] Frequency Combs, RP Photonics AG
[4] Fundamentals of frequency combs: What they are and how they work, NIST
[5] Optical Frequency Combs, NIST
[6] Optical and microwave metrology, NIST
[7] Ultrastable Photonic Microwave Oscillators: Purity Verified, Microwave Journal
[8] Photonic chip-based low-noise microwave oscillator, Nature/Springer
[9] Compact and ultrastable photonic microwave oscillator, Optics Letters
[10] Photonic Microwave Sources Divide Noise and Shift Paradigms, Photonics Spectra
[11] A chip-integrated comb-based microwave oscillator, via Arvix
[12] Compact and ultrastable photonic microwave oscillator, ResearchGate
[13] Photonic Microwave Sources Divide Noise and Shift Paradigms, Photonics Spectra
[14] Optical Frequency Combs for Low Phase Noise Microwave Generation, URSI
[15] Integrated optical frequency division for microwave and mmWave generation, Nature
[16] Reaching Unprecedented Sensitivity in Cross-Spectrum Phase Noise Characterization, Menlo Systems
[17] Ultrastable Photonic Microwave Oscillators: Purity Verified, Menlo Systems

Related EEWorld Online content

Review: Tekbox TBCG4 harmonic comb generator
Review: Picotest J2150B comb injector
Lasers, optics, electronics and more yield terahertz sources, Part 3 – Infrared lasers and plasma
Understanding electro-optic modulation

The post Frenemies at last? FAQ on optical combs for microwave oscillators: part 1 appeared first on Test & Measurement Tips.

In Figure 1, here’s a purchased PC board module based on the Diodes Inc. PAM8403 stereo class-D power amplifier, 2 watts per channel. I was working on a project that used several of these prefabricated ‘8403 boards, and I needed a way to check their outputs with a ‘scope to make sure they were not “clipping” (in analog-speak) or running out of headroom. I planned on building a monitoring device to easily and quickly run tests on multiple devices. But first, I decided to set up one board all by itself so I could figure out how to run my test procedure. To make working with this one board a bit easier, I mounted it on a larger piece of perfboard. I added a few 0.1″ pin connectors to make it easier to connect power, signal inputs, and speaker outputs.

Since the output from each amplifier is a series of pulse-width modulated rectangular waves (see Figure 2 and Figure 3), you can’t simply put a ‘scope probe on an output pin and see if it’s clipping.

Instead of trying to evaluate the amplifier’s performance by scoping its outputs, you need to perform the same action that the speaker does – you need to integrate the applied waveform (i.e., low-pass filter it) and look at that with the scope.

This brings us to the actual test jig that I built. Recall that I would be testing multiple devices; I needed a way to quickly evaluate the ‘8403 modules that were piggy-backed onto my PC board. The class-D amplifier output is a type of differential output. On one amplifier output, there is a stream of rectangular waves of varying duty cycle from approximately zero to 100% (50% at idle). On the other amplifier output, there is a matching synchronized stream varying from approximately 100% to zero.

The load (the speaker) is tied across the two outputs rather than from output to circuit common or ground. This is commonly referred to as a bridge-tied load or BTL. I needed a way to monitor the amplifier outputs in differential mode and then produce the analog equivalent through integration. In Figure 4, here’s what I built.

There are two boards here. One is a PC board, and the other is a perfboard. The perfboard contains a bit of circuitry: a quad op amp and a handful of resistors and capacitors. I mounted both boards on a piece of scrap plywood using standoffs and 4-40 hardware. I also added a copper foil ground plane beneath the perfboard. The use of scrap lumber to build test jigs is a favorite method of mine – it’s inexpensive and can easily accommodate various amounts of circuitry. It’s not perfectly shielded, but that’s not always necessary.

The PC board is an Instrumentation Amplifier Eval board I acquired during my time working for Analog Devices. I populated it with an AD8421 and powered it from ±15 VDC. I equipped the (true) differential input with a connector to match the pins on the ‘8403 amplifier boards I was building. I also added some small ferrite toroids to reduce the high-frequency content. The perf-board is a four-pole active low-pass filter with a corner frequency of 10 kHz and a pass-band gain of unity (1 V/V). Figure 5 shows the set up in use. Figure 6 and Figure 7 show the ‘scope output with a 2 kHz signal input, below and above the level where clipping and distortion start.

This setup made it easy to evaluate my devices. I could ensure they were all built correctly with no defective components.

I built another piece of audio equipment to make it easy to test microphones that needed phantom power. A phantom power supply provides (usually) +48 VDC to microphones that have built-in preamplifiers. The term phantom and the use of 48 VDC derive from the early days of telephony and tube preamplifiers. A low-power tube amplifier’s plate circuitry could be operated from 48 V effectively. Phantom circuitry allows a second circuit (the power, in this case) to be overlaid on the primary circuit (the audio).

With a balanced audio line – the kind that typically uses XLR connectors on the cable – there are two wires carrying the audio inside the grounded shield. The signal on one wire is 180° out of phase with the signal on the other. Stated another way, the audio is sent differentially through the cable. The 48 VDC is coupled onto both these wires through two almost identical resistors (1% tolerance) sized at a few thousand ohms. This means the DC is sent via common mode (with respect to ground) on the balanced audio line. The resistors don’t appreciably load the audio from the microphone, which typically has a source impedance of perhaps 50Ω to 150Ω. The circuit I built to provide the phantom power is shown in the schematic in Figure 8. Power comes from a 48 VDC wall-wart plugged into J1. I added a network of one more resistor and three capacitors to minimize the power supply ripple.

The apparatus is shown in Figure 9, again, built on a scrap of plywood with some copper foil added as a low impedance ground plane. The left cable goes to the female XLR connector (J2); the right cable goes to the male XLR connector (J3). The power cable comes in at the top and plugs into J1.

Another useful jig to have for audio work is a microphone matching/isolating transformer, again built on plywood and using parts I had available to control costs. See Figure 10. The microphone plugs into the cable that comes in from the top of the picture (via J1 as seen in Figure 12). There is a lo-Z to hi-Z transformer mounted upright on the board, followed by a ¼” audio connector. This connector is a two-circuit connector, sometimes referred to as a stereo audio jack or a T-R-S connector. T-R-S refers to tip-ring-sleeve, the three electrically isolated sections of the jack and corresponding plug. As before, the terminology derives from telephony technology.

This jig is set up as balanced in, balanced out. Occasionally, I used this jig simply as an adapter to convert an XLR connector to a ¼”audio connector with no need for an impedance transformation. To completely bypass the transformer, I needed a 4PDT switch. I had an abundance of DPDT slide switches, so I used two and ganged them together as seen in more detail in Figure 11.

Since I sometimes plugged the hi-Z side directly into an amplifier that was configured as single-ended, I added a switch (SW2) to ground the ring circuit. Refer to Figure 12 to see how switches SW1 and SW2 function.

As a variation of the previous transformer jig, and to address my concerns regarding the pickup of stray electrical fields (RF and audio frequency), I built the test jig seen in Figure 13. Electrically, much like the version in Figures 10 to 12, but without the switches, I built this to minimize extraneous pickup. It is configured as Lo-Z balanced in and Hi-Z unbalanced out. To provide the needed shielding and to minimize cost, I wrapped sections of cardboard (from the back of a pad of paper) with adhesive-backed copper tape and soldered these panels together. Not surprisingly, it’s fabricated on a scrap of plywood.

In Figure 14, note the added RCA pin jack. I installed that to provide an easy ‘scope test point to monitor the signal going to my amplifier while still maintaining my shielding.

Speaking of amplifiers, if you are testing high-power units and don’t want to blast your eardrums into oblivion, you need to operate the amp into a dummy load. Typical loads are made of 4Ω, 8Ω, or 16Ω resistors. I built a simple one that I can use as either a stereo 8Ω load or, with suitable jumpers, a mono 4Ω or 16Ω load. I happened to have four 15Ω, 1%, 50W power resistors. With two in parallel, I had a 7.5 Ω, 100W resistor – close, but no cigar. I also had a couple of 0.5Ω, 10%, 5W resistors on hand. With one of those in series with each 7.5Ω, I had my dual 8Ω dummy load resistors. An additional advantage of the 7.5Ω+0.5Ω arrangement: It provides a convenient voltage divider point that can drive headphones without damage – it’s functioning as a 1:16 voltage divider.

Referring to Figures 15 and 16, the input is applied via the left two screw terminals on the barrier block. The amplifier’s output can be monitored via a ‘scope at the far-left RCA jack or via headphones at the second from the left RCA jack. The second or right-hand channel is a mirror image of the first with respect to the connections.

While the power resistors have a rating of 100W, realistically, that presumes an infinitely large heat sink. Since I don’t have one of those available in my lab, I used a substantially smaller than infinity aluminum plate. Heat sink size and thermal conductivity between the resistors and the heat sink; and the heat sink and the ambient air surrounding the heat sink; and the temperature of the ambient air affect how hot the resistors will get versus the power being dissipated in them. So that smaller size heat sink means the resistors’ power rating must be derated from 100W, but the rating is still big enough for most of my testing, especially with short duty cycles of applied power. The schematic of the dummy load is shown in Figure 17.

Also on the topic of amplifiers, sometimes you need to amplify low-level signals (e.g., from a guitar) enough to be useful (i.e., be able to hear them). I built a small headphone amplifier using an LM386. It actually can drive a speaker to a quite usable level. One minor detail that makes it slightly problematic: The typical LM386 circuit has a fairly low input impedance of 50kΩ. To work around this, I added a simple JFET source follower circuit to buffer the input of the LM386. The JFET input is bootstrapped to make the input impedance super high. I built this all up on a perfboard and made provisions for gain switching between 20 V/V and 200 V/V. I also provided a direct output to drive a speaker and a resistively padded output to drive stereo headphones. The completed amplifier is shown in Figures 18, 19, and 20.

The schematic is shown in Figure 21.

Please post suggestions of test jigs and apparatus you’ve constructed for your own specific applications.

The post Test jigs and bench tools to make your life easier: part 2 appeared first on Test & Measurement Tips.

Q: In the webinar, you walked through a mixed-signal example where a known I²C packet wasn’t producing the expected output. How do you approach something like that systematically?
A: The first thing I try to do is stop thinking of it as one big mystery and start treating it like a black box with inputs and outputs. You have something going on, in this case a known I²C packet from a debugger, and you’re not seeing the expected behavior at the output: an LED that should be turning on. So instead of guessing, you interrogate the circuit one step at a time. We traced back from the output toward the input. The I²C bus looked suspicious first, so we dug into that. Then we noticed the microcontroller was resetting sporadically; its reset pin was occasionally dropping. That pointed us away from the bus and toward the MCU’s power supply. Sure enough, the supply was dipping below the minimum required voltage, causing brownouts. Traced that back one more step and found a failed capacitor on the input. That’s the process: go back one node at a time until you hit the root cause.

Q: What tools made that chain of investigation practical?
A: A combination of the digital bus decode and the power integrity techniques from Part 1, really. The serial decode let us see what was happening on the I²C bus and rule it out. Then we used the same AC coupling and measurement approach to characterize the supply rail and confirm it was dropping out. What advanced tools give you in these situations is speed. On a simple system, yeah, maybe two quick probes and you’re done. But on a more complex one, being able to layer triggers, decode protocols, and log measurements simultaneously is what gets you to the answer in minutes instead of hours.

Q: Once an engineer has identified a fault and captured the data, how should they think about sharing it?
A: The biggest mistake I see is treating all audiences the same. Who you’re sharing with completely changes what you should show them. Other engineers? Give them the full waveforms. They’ll expect it, they know what they’re looking at, and that detail is exactly what they need to give useful feedback. A technical manager who’s somewhat familiar with the system probably just needs the key numbers: peak-to-peak, RMS, pass/fail, not a waveform they have to sit down and interpret. And a technician on a production floor, or a non-technical stakeholder, just needs a clear pass or fail. Each format has its place; you just have to pick the right one for the person in front of you.

Q: What about communicating urgency to someone non-technical, where a subtle anomaly is actually a serious problem?
A: That’s a real challenge. You can put up a beautiful waveform showing a tiny blip on a power supply, and a non-technical manager will think, ‘That doesn’t look that bad.’ So you have to translate it. Boil it down to the consequence, not the measurement. Say there is a fault, and I can prove it, but the point is, this will cause failures in the field. And if that still doesn’t land, talk about dollars. Recalls. Schedule slips. When you frame a 10mV anomaly as ‘this will cost us three months and a recall,’ people pay attention. A technical manager who understands the data can help carry that message, and sometimes it’s better to go into that conversation as a team.

Q: You talked about a real-world example with a pipe manufacturer where oscilloscope data fed directly into a production control loop. Can you describe how that worked?
A: Sure. They’re manufacturing pipe continuously: sheet metal comes in, gets rolled, welded, and cut to length. They needed a way to detect faults in real time and maintain traceability on every section that shipped. The approach was to monitor the current draw from both the induction coil and the electric welder as the pipe moved through. If the material is heating and welding consistently, those current profiles are stable. A deviation, say a thin spot in the sheet metal burning through or a section that’s not welding cleanly, shows up as a change in that current signature. Using the scope’s API, we can flag the fault, mark the section, and log the data automatically. Every good section gets serialized with the proof to back it up. In a medical or government application, that traceability isn’t a nice-to-have. It’s a requirement.

Q: I’m not a strong programmer. Is there a middle ground between the standard scope interface and a full SDK or API?
A: Absolutely. In PicoScope, we call it measurement logging. You set up whatever automatic measurements you want: peak-to-peak, RMS, phase, frequency, whatever applies to your signal, and the scope continuously writes those to a CSV in the background. So instead of trying to store and analyze eight hours of raw waveform data, which would bring Excel to its knees, you’ve got a manageable file of metadata about your signal. Pull that into a spreadsheet, build a histogram, and you can immediately see whether your signal is behaving consistently over time or whether something is slowly drifting. Taking Measurements one step further, we have Actions that can then DO something useful in reaction to specific criteria, such as a failed limit on a measurement. It’s a very accessible middle ground, and no programming is required.

Q: Why would an engineer choose an off-the-shelf oscilloscope over building their own data acquisition hardware?
A: It really comes down to where you want your engineers’ time to go. Yes, a capable team could design a DAQ system from scratch: ADCs, analog front end, firmware, calibration standards, the whole thing. But that’s a side quest, not the main quest. The scope manufacturer has already solved that problem, maintains calibration, provides an API or SDK, and stands behind the hardware. That engineer’s specialty might be medical sensors, robotics, or RF. Diverting them to build a DAQ is a waste of their talent, and it might cost you months of schedule. Buying the right off-the-shelf tool and integrating it can get you operational in days or weeks instead.

The post From root cause to report: mixed-signal debugging, data sharing, and automation appeared first on Test & Measurement Tips.

Spectrum Control has introduced the Weinschel 8401-16E 16×16 Butler Matrix, a passive beamforming network for MIMO systems and wireless test applications across 2.4 GHz to 7.125 GHz. The 16-input, 16-output architecture generates fixed orthogonal beams without active phase shifters, supporting repeatable beamforming, multi-channel validation and simulated OTA testing for Wi-Fi 6E, Wi-Fi 7 and other high-channel-count RF systems. Designed for test and measurement, radar and EW/SIGINT environments, the unit helps engineers evaluate throughput, beam selection, MU-MIMO behavior and interference under controlled spatial conditions.

The post 16×16 Butler Matrix supports MIMO beamforming tests appeared first on Test & Measurement Tips.

Engineers are problem solvers. It’s what we do. When working on a project at your bench, you probably have the usual collection of hand tools, test equipment, and a soldering station or two. Sometimes you need more than these tools. Sometimes, you have to solve a practical problem before solving the problem at hand. That’s where adding and building jigs, holders, compartments, and other things can help.

Some common add-ons are the alligator clip “helping hands” (Figure 1). The clips let you hold PCBs, wires, connectors, or components, which certainly helps when soldering. A magnifying glass enhances the experience, especially when working with small parts.

Figure 2 shows a PCB vice, another device that can hold a PCB during soldering. Several such tools are on the market. See EEWorld’s review of the PCBite kit from Sensepeak.

You may also need special tools, stands, or jigs to assist in construction and testing. One of my projects involves adding on-board electronics to electric guitars. I do enough of this work that I don’t want to prop up the guitar body on whatever cardboard or plastic project boxes I have in my lab and start soldering. Instead, I built a couple of simple jigs. One allows me to hold the guitar upside down with strings facing the bench but not touching anything (so the strings aren’t muted). See Figure 3 and Figure 4. This makes it easy to tinker with the circuit boards I’m installing on the back of the body. A second jig allows me to clamp the guitar body sideways (without marring the body) when that’s needed. See Figure 5.

Another handy device is a compartmentalized component and hardware holder: an egg carton. There’s one visible in Figures 3 through 5. These help keep mounting hardware from vanishing under the various bench equipment or, worse yet, into the vacuum cleaner.

Preventing ESD damage

Let’s now consider assembling a PC board populated with surface-mount components that might be damaged by electrostatic discharge (ESD). Let’s assume you know that ESD should be mitigated with a conductive mat, upon which you place your PCB. You could buy a conductive mat, but an inexpensive mat can be equipped with a piece of black conductive IC foam, shown in Figure 6. Note the clip lead that attaches and grounds the mat. I added a small piece of brass shim stock to make it easier to clip onto the mat without chewing through it.

On my bench, I added the grounding connectors (banana jacks), also visible in Figure 6. The plate is connected to the same earth ground connection that my circuit breaker panel uses. It’s a combination of a ground stake and a connection to the incoming copper water supply line. I’m assuming there will never be any high magnitude ground fault currents flowing while I’m working at my bench, so my ground connection should really be at a (true) earth potential.

I also realized I would scrape my knees on the tips of the banana connectors on the underside of my ground block, so I added a piece of ¼-in. rubber tubing (slit the entire length and attached with hot melt adhesive), shown in Figure 7.

Besides these ESD features, you will likely want a magnifier to help place tiny surface-mount components. A magnifying glass can provide some help, but I prefer a video microscope. These are available in various magnification levels and price points. I’m using an inexpensive device from Amazon with a mag range of 40 X to 1000 X. I mounted it on a used bench light boom I got from Goodwill. See Figure 8. For the video monitor, I’m using a laptop computer that originally ran Windows 10. Since that’s no longer supported, I wiped the hard drive and converted the laptop to a Chromebook.

Sometimes, you need a camera to document your work or to see things on a large screen. You surely need something to hold the camera, such as a boom or tripod. Figures 9 and 10 show the details of how I attached a camera to the end of the boom.

Hold components in place while soldering

For around $30, I had a useful setup that mostly made soldering tiny components with my hot air soldering station pretty easy. Occasionally, when attaching parts using a soldering iron and 0.020″ solder, I found I would sometimes bump the component out of position. For those occasions, I built a tool to hold the component while soldering the first few pins. See Figure 11. I built this using a tuna fish can (without the tuna), some pea gravel, epoxy glue (acting as a sort of potting compound), a piece of threaded stock plus nuts and washers, a piece of 14 AWG copper wire, and an old ballpoint pen to act as a spring-loaded pin. I changed the ink cartridge and spring locations around to get the desired spring action. This applies just enough pressure to keep things in place.

For soldering with hot air, solder paste is the preferred method. I was using some paste that had much higher viscosity than I would have liked, but I didn’t want to waste it, so I built a small heater box to keep it warm, which lowered its viscosity. See Figure 12. It’s simply a spare electrical junction box, a [grounded] power cord, a 15 W candelabra light bulb, and a scrap of stainless-steel screen. Once again, a low-cost solution.

Returning to my engineering work on guitars and guitar electronics, I needed a way to evaluate different styles of pickups and methods of inducing sympathetic vibration into the strings. Referring to Figure 13, I built a jig with some used strings, a piece of MDF (medium-density fiberboard), a piece of pine trim board, miscellaneous bits of hardware, small blocks of (scrap) oak (acting as bridges), and a piezo (PZ) disc salvaged from a PZ beeper.

In use, I placed my electromagnetic drivers beneath the strings at different points between the left and right bridges and noted their effectiveness as drivers.

When testing line-powered equipment, you may find that the equipment suffers from internal short circuits. This problem commonly arises from power transformers with shorted turns within windings, shorted rectifier diodes, shorted filter capacitors, or other supply rail shorts in the equipment under test (EUT). We need a good way to detect and analyze these problems without repeatedly blowing the EUT’s built-in fuses or tripping its circuit breakers.

The clever way is by using a Variac to power the EUT and inserting a medium to high wattage incandescent light bulb in series with the power being supplied from the Variac. For line-powered equipment, we don’t want to use a suicide cord and a bunch of Radio Shack clip leads. A more elegant solution is shown in Figure 14.

With the appropriate wattage bulb, normal current draw causes little heating of the bulb’s filament. Based on the positive temperature coefficient of tungsten, the barely warm filament’s voltage drop is very low, and the EUT receives close to the full line voltage. You can crank up the line voltage slowly, and if excessive current flows, you’ll know because the light bulb (but not the fuses) will glow brightly.

The schematic for this device (Figure 15), like many of my preceding test jigs, is simple. I recommend having several different bulb wattages on hand, from 40 W to 250 W. You should procure several now before they become obsolete.

Lastly, consider those times when you’re trying to spray contact cleaner into a volume control whose opening is obscured due to its mounting position. Trying to bend the small-gauge tubing that comes with the spray can rarely work — it will probably kink. I worked around this by fabricating an elbow from some scraps of larger-gauge plastic tubing. I was able to bend it by carefully heating it with my hot-air soldering station. The larger gauge made it far less likely to pinch shut at the bend. See Figure 16.

In part 2, we’ll take a close look at some of my custom-built test jigs to simplify some of the audio testing that I do.

What test jigs or time-saving tools have you devised? Tell us in the comments.

The post Test jigs and bench tools make your life easier: part 1 appeared first on Test & Measurement Tips.

IMS 2026 took place the week of June 7-12 at the Menino Convention Center in Boston’s Seaport District. Over 400 large and small exhibitors displayed their components and equipment in the exhibition hall. EEWorld was there. Through videos and photos, here’s a roundup of the exhibits:

Aaronia

Aaronia’s large booth featured spectrum analyzers, antennas, and software. The booth featured large screens showing real-time spectral analysis and, in my opinion, stole the show for the most color. The photo shows Spectran Ultra baseboard, which produces 16 phase-coherent outputs, each at 40 Gsamples/sec with up to 20 GHz bandwidth per channel. It connects to networks and other equipment through eight Ethernet and eight USB ports.

Analog Devices

The photo shows a demonstration of the AD9084 Apollo MxFE Quad, 16-Bit, 28 Gsample/sec RF DAC and Quad, 12-Bit, 20 Gsample/sec RF ADC. In this application, the oscilloscope shows the boot sequence ending once the RF output starts.

Anritsu

Anritsu showed the latest in its Tensor series of vector-network analyzers, announced on June 9 at IMS. In the video, you’ll see the latest hardware improvements to this four-port VNA: four signal sources for four ports. Watch the video to the end for a preview of the instrument’s special new features.

Bird

Known for its handheld testers for RF installation and maintenance, Bird displayed the SiteHawk SK-6000 Cable and Antenna Analyzer, which operates at frequencies up to 6 GHz. The company also offers models for frequencies to 4.5 GHz and 9 GHz. The instruments use frequency-domain reflectometry to detect faults such as breaks in cables.

Copper Mountain

The company known for its USB-connected vector-network analyzers (VNAs) came to Boston with several instruments, including the BFx-02 Frequency Extension Base for S-Parameters Measurement. Using this instrument and frequency extenders, engineers can make S-parameter measurements on components, systems, and materials at frequencies beyond those usually accessible for traditional VNAs.

Emerson NI

The company formerly known as National Instruments produces modular RF instruments such as its PXI Vector Signal Generators, Analyzers, and Transceivers. In this exhibit, the company used its vector-signal transceiver in combination with a software-defined radio to develop an RF test bench.

Keysight

Long known for its RF test equipment, Keysight Technologies demonstrated its 32 GHz XA5 Signal Analyzer, among other equipment. The swept-spectrum analyzer features 2 GHz analysis bandwidth. As the video shows, the instrument’s measurement speed provides a significant improvement over previous swept-spectrum analyzers.

Mathworks

Working with Analog Devices and Leonardo, Mathworks demonstrated an RF digital twin. The model simulates a radar and satellite RF system using components from Analog Devices. With the model, engineers can alter parameters and see how different settings affect system performance.

Microchip

Havily invested in wireless and wired network system timing, Microchip exhibited oscillators for GNSS timing systems.

Pico Technology

Known for its USB-connected oscilloscopes and network analyzers, Pico technology exhibited several instruments including the PicoScope 500 Series featuring what the company calls it’s FlexRes signal resolution, which produces from 8 bits to 16 bites with the tradeoff being sample rate (62.5 Msamples/s at 16 bits). The model shown features four analog inputs with a connector for adding logic-level signal capture.

Pickering Interfaces

Specializing in PXI modular switching instruments, Pickering Interfaces came to IMS with (l-r) a PXI/PXIe Triple SP6T, 18 GHz, 50 Ω, SMA, Failsafe Microwave Multiplexer, a Quad 4-Channel, 4GHz, 50Ω, SMB MEMS RF Multiplexer, and a Single SP6T, 67 GHz, 50 Ω, SMA-1.85, multiplexer. This model uses a cable to connect a remote multiplexer, which lets you install a single-slot module in the chassis.

Qorvo

Fixed wireless access (FWA) has emerged as one of 5G’s successes and one of the few use cases for mmWave signals. Qorvo makes parts that you can use to up convert and down convert carrier frequencies. The demonstration uses the AWMF-0224, a 24 GHz to 30 GHz Dual-channel IF transceiver that performs the frequency conversion. The demonstration also uses the AWMF-0221 Dual Polarization Quad 4×2 Beamformer. The combination can achieve data rates of 3 Gbits/s at distances up to 3 km.

Samtec

The company known for connectors and cables demonstrated its Bulls Eye BE130A, a 130-GHz test-point system. The video provides insight into the system, which can accommodate up to 32 channels.

Rigol

While RF engineers likely gravitated to Rigol’s DNA6264 four-port VNA, what caught my eye was the MHO984 oscilloscope. This 800-MHz, 12-bit model on the MHO900 series features four channels. Lower-cost models start at 350 MHz. What the photo doesn’t show is the oscilloscope’s footprint, or lack thereof. It’s just 77 mm (about 3 in.) deep.

Rohde & Schwarz

IMS is the company’s biggest event, and the venerable company was out in force. As always, the company exhibited RF test equipment that included spectrum analyzers, vector-network analyzers, signal generators, and oscilloscopes. The video shows a demonstration using a signal generator and spectrum analyzer to characterize a 3-GHz RF amplifier.

Sage Instruments

Known for field installation and maintenance measurement equipment, Sage Instruments exhibited its WSA-408 Wireless Signal Analyzer. This handheld instrument includes an 8-kHz-to-8-GHz spectrum analyzer, an interference analyzer, and a 5G signal analyzer.

Siglent

Siglent exhibited its line of equipment, including spectrum analyzers, oscilloscopes, and RF signal generators.  

SignalHound

The RF test equipment company’s USB-connected equipment includes VNAs, signal generators, phase-coherent receivers, and phase-noise testers. The photo shows its VNA400, a two-port, 40 GHz instrument connected to a bandpass filter.

Poster session: DC-DC converter switches at 200 MHz

The exhibit hall featured more than just companies showing off their products.

Tabor Electronics

This maker of portable, benchtop, and modular RF test equipment demonstrated equipment designed for software-defined radio (SDR) testing. On the left in the photo is the LSX2091D, a 20 GHz microwave signal generator.

Uni-Trend

Known for basic test instrumentation, including RF equipment, Uni-T exhibited one of its oscilloscopes, as well as (top-to-bottom) a 60 MHz function generator, a 5-1/2 digit multimeter, and a power supply.

IMS 2027 will take place from May 23rd to 28th in San Antonio, TX. It returns to Boston in 2031.

EEWorld Online related articles

IMS 2024 roundup: test equipment
Spectrum measurements shed light on sign malfunctions
Antennas to bits: Modeling real-world behavior in RF and wireless systems
Tryout: Uni-T UPO1202 200 MHz oscilloscope
Tryout: Uni-T UTG962E Function/Arbitrary Waveform Generator

The post IMS 2026: RF engineers come to Boston appeared first on Test & Measurement Tips.

B&K Precision has introduced the DML Series DC electronic load system, a modular platform with a 4U mainframe that supports up to four channels and 800 W. The system uses swappable single- or dual-channel load modules with front-panel inputs providing up to 80 V and 80 A, or up to 160 A in parallel operation, and supports constant current, constant voltage, constant resistance and constant power modes for testing DC power supplies, batteries, fuel cells and photovoltaic arrays. The front panel allows independent control of each module and setup of parameters including voltage, current, slew rate and pulse width, with settings saved to internal memory for recall. Remote control options include LAN, USB, RS232 and GPIB with SCPI support, and the system also offers fast transient operation up to 25 kHz in CC mode, sweep modes and built-in protection features.

The post DC electronic load system supports four channels appeared first on Test & Measurement Tips.

The instrument captures heat flow, evolved gas, and electrochemical data (voltage monitoring plus charge/discharge) simultaneously, in a single run on a single intact cell. That’s a meaningful change from running separate tests on separate samples and trying to correlate results afterward. The temperature range spans -80°C to 600°C, which the company positions as best-in-class for this category, wide enough to cover both low-temperature performance work and high-temperature thermal runaway events in one sweep.

TA Instruments is positioning this as filling the gap between materials-level thermal analysis and full-cell DSC testing. Compared to accelerating rate calorimetry, which needs large-format cells, dedicated facilities, and long test cycles, this approach lets researchers work with intact coin cells earlier in development. For labs running early-stage characterization, that’s a faster, lower-overhead path to the same kind of insight.

It can also be coupled with mass spec, FTIR, or GC-MS for more detailed gas analysis when needed.

Available for order now, with shipments starting August 2026.

The post Coin cell calorimeter skips teardown, spans -80 °C to 600°C appeared first on Test & Measurement Tips.

Tektronix® has introduced wideband shunts for IsoVu isolated current probes and the THDP0400 high voltage differential probe for power electronics validation in EVs, industrial systems, AI data centers and power conversion applications. The shunts support bandwidth up to 250 MHz for low-current transient measurements and include built-in communication, temperature compensation and fuse protection, while the THDP0400 provides 400 MHz bandwidth and ±2000 V measurement capability for floating high-voltage measurements, wide-bandgap semiconductor development and double-pulse testing. Together, the two accessories help engineers capture fast transients and low-current behavior with less setup complexity by working within a single oscilloscope-based measurement workflow.

The post Probe measures floating signals to ±2000 V appeared first on Test & Measurement Tips.

Ranatec AB has introduced standard 8-channel and 12-channel attenuator box configurations for RF and microwave test systems. The off-the-shelf units are intended for telecom production, electronics manufacturing and automated test environments where users need faster deployment and channel counts matched to specific setups without custom development. The added configurations give test engineers more flexibility when scaling RF measurement systems while helping reduce sourcing time and system integration effort. They are aimed at applications that require repeatable signal conditioning in production test, verification and troubleshooting workflows.

The post RF attenuator boxes target automated test systems appeared first on Test & Measurement Tips.

Pico Technology has introduced the PicoScope 5000E Series USB-C oscilloscopes for low-amplitude analog, digital and mixed-signal measurements. The series provides 16-bit resolution, bandwidths up to 200 MHz, sample rates up to 2.5 GS/s and capture memory up to 1 GS, while Plus models add switchable 8-bit operation with bandwidth up to 500 MHz, sample rates up to 5 GS/s and memory up to 2 GS for faster edge and serial-bus analysis. The scopes are designed for applications such as power integrity, sensor measurement, analog front ends, medical electronics and mixed-signal debug, with a noise floor below 22 µV RMS and better than -73 dB THD for low-level signal analysis. Additional features include 4 analog channels with optional 16 digital channels on MSO models, more than 40 serial protocol decoders, advanced triggering, a 200 MS/s 14-bit AWG, FFT analysis and PicoSDK® support for C, C#, C++, Python, MATLAB and LabVIEW.

The post 16-bit scope captures low-level analog signals appeared first on Test & Measurement Tips.

In this series on strain gauges, we’ve looked at quarter- and half-bridge configurations. In this final part, we will look at a full-bridge implementation with four active elements.

Q: Where do these active elements get placed on our test specimen?
A: Figure 1 shows one possibility. With stress applied in the direction of the arrow, the red elements on top, each of length l in the unstressed state, will expand by Dl, and the blue ones of the same initial unstrained length will contract by Dl.

Q: How do we connect these elements?
A: Figure 2 shows these elements connected in the Wheatstone-bridge circuit, where VEX is the excitation voltage, and V_O is the output voltage proportional to strain e.

Q: How do we derive strain from VO?
A: We basically have two voltage dividers with the voltages at the positive and negative terminals of our voltmeter as follows:

Then, V_O is V₊–V_–:

Given a gauge factor (GF), we want to find V_O as a function of strain. From part 3, we know we can substitute R(GF)e  for DR:

For GF = 2, the Figure 3 traces show V_O as a function of e  for full-, half-, and quarter-bridge circuits along with the corresponding equations. Note that for the quarter-bridge case, the equation represents a linear approximation (black trace) of the actual nonlinear response (dashed black trace) that we calculated in part 2. Note that as we double the number of active elements, V_O doubles, thereby increasing our measurement sensitivity.

Figure 4. Lead resistance RL can drift with ambient temperature (a), but an additional sense lead can provide compensation (b). (Image: Rick Nelson)

Q: Couldn’t we also increase sensitivity by increasing the excitation voltage?
A: Yes, increasing the excitation voltage would increase the sensitivity and signal-to-noise ratio. However, this approach has a downside: it increases power dissipation in each strain-gauge element, with power increasing with the square of the excitation voltage. As power increases, the gauge becomes susceptible to self-heating, which can cause the gauge to expand and contract relative to the test specimen [1], thereby introducing a thermal error that is difficult to compensate for. Consequently, the excitation voltage should remain as low as possible while maintaining an adequate signal-to-noise ratio.

Q: What about lead-length resistance while measuring strain on large structures such as wide-body airframes?
A: As illustrated in Figure 4 in a quarter-bridge configuration, our strain gauge R_X attaches to our data-acquisition system through two leads, Lead 1 (red) and Lead 2 (blue), of resistance R_L each, so our meter will respond as if R_X=R_X+2R_L. If we know R_L, we can compensate, but R_L will vary with ambient temperature, making it difficult to know its value exactly.

The gold-standard approach for compensating for lead resistance is to use a four-wire Kelvin measurement, but that incurs additional cost. With the Wheatstone bridge, however, we only need to add one additional sense lead and make a minor wiring change, shown in orange in Figure 4b. Note that with this approach, we’ve moved the Lead 1 resistance from the bottom right of the bridge to the top right, while the Lead 2 resistance remains in the bottom right. Consequently, the bridge will cancel out any resistance changes due to temperature, as described in an earlier article. Of course, the sense lead has some resistance, but it’s negligible compared to the voltmeter’s impedance.

Reference

[1] Strain Gauges: Selecting an Excitation Voltage, Siemens

Related EEWorld Online content

Defining and measuring strain: part 1
The basics of Kelvin connections
Making sense of test circuits with Kirchhoff’s laws: part 4
What’s the difference between 2-, 3-, & 4-wire RDT sensing?
Stress & Strain, Part 1: fundamental principles
How to avoid errors in low-voltage measurements

The post Defining and measuring strain: part 4 appeared first on Test & Measurement Tips.

In this series, we’ve been looking at the strain gauge, a type of sensor that can measure how a test specimen deforms as a function of applied stress. In part 2 of this series and in an earlier article on the Wheatstone bridge, we looked at how a single active strain-gauge element coupled with a passive element in a half-bridge configuration can measure tensile and compressive axial strain while providing temperature compensation. In this article, we’ll look at the use of two active strain-gauge elements.

Q: Could we first review the basic Wheatstone bridge topology?
A: Yes, and in fact, this is a good time to update one of our labels. Figure 1 shows the basic Wheatstone-bridge circuit. So far, our unknown resistance of interest has been R_X on the lower right. With multiple active strain-gauge elements, however, we can have multiple unknown resistances in our bridge, so I’ll rename R_X as R₄ to emphasize that it’s not the only potential unknown. Note also that V_EX is the excitation voltage, and V_O is the output voltage, with V_O varying with strain e.

Q: So how do we make use of multiple strain-gauge elements?
A: We concluded part 2 of this series with a figure similar to Figure 2a, with strain gauges mounted on the top and bottom of a test specimen, and with the long sense conductors aligned in parallel with the direction of axial strain. Both will increase in resistance when in tension and decrease in compression. This, in and of itself, can be useful, but first let’s look at a slightly different configuration. In Figure 2b, our test specimen becomes a cantilever subjected to a bending stress, as shown by the red arrow. Here, the top strain-gauge element of length l increases in length by Dl, and the bottom element decreases in length by the same amount.

Q: How do we connect this to our bridge?
A: Figure 3 shows one possibility. Let’s say the fixed resistors and the unstrained resistance values of our strain gauges are all equal to R. We can see immediately that the voltage at the negative terminal of our V_O meter is V_EX/2, and the stressed resistances of the R₂ and R₄ strain gauges are R–DR and R+DR, respectively. Note that this configuration retains the temperature-compensation characteristic of our previous implementation with one active and one dummy element.

We can then calculate V_O:

From part 1, we know that

And therefore:

We can now substitute R(GF)e  for DR in our equation for V_O:

The Figure 4 trace in blue plots V_O vs e  for the Figure 3 circuit. The red trace shows the relationship for our circuit from part 2 with the single active element. Note that for a given e, the voltage for the two-active-element circuit approximately doubles, boosting measurement sensitivity.

Q: Why say “approximately doubles”?
A: The equation above for V_O for two active elements is clearly linear, and the equation for the single-element version is nearly linear over the ranges of strain found in many applications, especially ones with metallic test specimens. However, if we look at a wider range of strain, as shown in Figure 5, we see that the single-active-element response is actually nonlinear—the dashed trace is a straight line connecting the end points of the red trace. Because of the nonlinearity of one of our traces, it wouldn’t be accurate to say the blue trace voltage is exactly double the red trace voltage for any value of strain.

Q: What else should we know about strain gauges?
A: We will conclude this series with a look at full-bridge implementations as well as considerations regarding excitation voltage levels.

Related EEWorld Online content

Sensors expert talks data acquisition, IoT, wearables, and AI
Stress & Strain, Part 2: Implications for electronics
Wheatstone bridge, Part 2: Additional considerations
Defining and measuring strain: part 1
Making sense of test circuits with Kirchhoff’s laws: part 1
Quantifying and measuring non-electrical phenomena: Heat

The post Defining and measuring strain: part 3 appeared first on Test & Measurement Tips.

Welcome to the April edition of Engineering Deep Dive — a curated selection of the most engaging technical threads from the Electro-Tech-Online community’s Electronic Projects Design/Ideas/Reviews category. Questions are selected based on view counts, reply depth, and educational value. Each entry below has been expanded with context, key concepts, and suggestions for further exploration.

Questions at a glance

Q1  Why Does My Window Comparator Output Stay ON at 0V Input Despite Correct Threshold Voltages?
Q2 Does Cable Velocity Factor Affect Pulse Propagation Delay or Only Phase Shift?
Q3  How Does Pulse Ignition and Flame Rectification Work in Gas Water Heater Ignition Systems?
Q4  How Do You Identify Whether a PCB Failure Is Caused by Design Errors or Manufacturing Defects?
Q5  Why Is Reflection Cancellation Still Considered Resonance in Time-Domain Analysis?
Q6  How Can You Measure and Isolate PCB Trace S-Parameters Without RF Connectors?
Q7  What Is the Most Convenient Way to Implement Real-Time Audio FFT Analysis?

Q1  Why does my window comparator output stay on at 0V input despite correct threshold voltages?

A window comparator circuit based on the LM339 is behaving unexpectedly — the output LED stays ON even when the input voltage is 0V, despite threshold voltages appearing correct. This thread walks through threshold calculation, LM339 open-collector output behavior, and systematic troubleshooting of component failures that cause latched outputs.

Key technical topics covered

• Window comparator design with LM339 (open-collector output stage)
• Upper and lower threshold voltage calculation
• Diagnosing latched or stuck outputs caused by faulty components
• Pull-up resistor selection and LED drive circuit

Why It Matters | Window comparators appear in battery monitors, motor-speed controllers, temperature alarms, and ADC over-range detectors. Misunderstanding open-collector outputs is one of the most common LM339 pitfalls for beginners.

Topic tags

Circuit Design | LM339 | Comparator | Troubleshooting

Supporting data: Circuit schematic image

Community thread: Thread link

Q2  Does Cable velocity factor affect pulse propagation delay or only phase shift?

When feeding a digital pulse into a coaxial or transmission-line cable, does the cable’s velocity factor (VF) introduce a propagation delay, or does it only shift the phase of a sinusoidal signal? This thread distinguishes group velocity (relevant to pulse delay) from phase velocity (relevant to sinusoidal phase shift) and explains why both are numerically identical in non-dispersive media.

Key technical topics covered

• Velocity factor and its physical origin (permittivity of the dielectric)
• Group velocity vs. phase velocity — when they differ
• Propagation delay calculation: t_d = length / (VF × c)
• Practical impact on digital timing in long cable runs

Why It Matters | Signal integrity engineers designing high-speed serial links, RF engineers building phased arrays, and hobbyists working with long cable runs all need to understand how VF affects their signals.

Topic tags

Signal Integrity | Transmission Line | RF | Digital Timing

Supporting data: Oscilloscope waveform image

Community thread: Thread link

Q3  How Does Pulse Ignition and Flame Rectification Work in Gas Water Heater Ignition Systems?

This thread goes beyond simple ignition spark generation to explore how a flame rectification sensor confirms combustion. The discussion covers the ionization current produced by a gas flame, how it is used as a half-wave rectifier in the safety circuit, and the risks of DIY modifications to gas appliance electronics.

Key technical topics covered

• High-voltage spark generation via a pulse ignition module
• Flame rectification: DC bias through an ionized gas column
• Safety interlocks and why repeated ignition failures must not be bypassed
• Troubleshooting the sensor electrode (fouling, misalignment, cracked ceramic)

Why It Matters | Gas appliance faults can be dangerous. Understanding the intended safety logic helps technicians and advanced hobbyists diagnose faults responsibly, without disabling protective interlocks.

Topic tags

Power Electronics | Safety Systems | Sensors | Gas Ignition

Supporting data: Module photograph

Community thread: Thread link

Q4  How do you identify whether a PCB failure is caused by design errors or manufacturing defects?

When a PCB batch fails, the root cause might be in the Gerber/drill files or in the fabrication process itself. This thread provides a structured methodology: cross-referencing design files with fab specifications, identifying tell-tale defect signatures (trace opens, plating voids, layer misregistration), and communicating findings to the PCB house.

Key technical topics covered

• DFM (Design for Manufacturability) review checklist before ordering
• Common fab defects: plating voids, trace opens, drill inaccuracies, solder-mask misalignment
• Layer misregistration detection via cross-section or X-ray inspection
• How to document and report defects to get boards replaced or credited

Why It Matters | PCB fabrication failures are costly in both money and schedule. A systematic approach reduces finger-pointing between design and manufacturing teams and speeds up root-cause resolution.

Topic tags

PCB Design | DFM | Manufacturing | Quality Assurance

Supporting data: PCB microscopy image

Community thread: Thread Link

Q5  Why Is Reflection Cancellation Still Considered Resonance in Time-Domain Analysis?

Resonance is traditionally taught in the frequency domain as a sharp peak at a natural frequency. This thread unpacks the conceptual bridge to the time domain: how delayed reflections, constructive/destructive interference, and oscillating energy exchange between inductance and capacitance all manifest as what we still call ‘resonance’, regardless of domain.

Key technical topics covered

• Time-domain view of resonance: energy oscillating between L and C
• Reflections on transmission lines and how they create standing waves
• Fourier duality: why time-domain oscillation maps to a frequency-domain peak
• Practical examples: stub resonance, via resonance in PCBs

Why It Matters | Signal integrity engineers and RF designers often switch between domains. Understanding why the same physical phenomenon appears as both a time-domain ringing and a frequency-domain peak prevents analysis errors.

Topic tags

Signal Integrity | RF Theory | Frequency Domain | Time Domain

Supporting data: Simulation waveform image

Community thread: Thread Link

Q6  How can you measure and isolate PCB trace S-parameters without RF connectors?

Characterizing a PCB interconnect with a VNA is straightforward when SMA connectors are available — but what if there are none? This thread covers probe-based measurement, the de-embedding process to remove fixture and pad parasitics, and the 2x-Thru method for extracting a single trace’s S-parameters from a back-to-back structure.

Key technical topics covered

• Probe landing and transition de-embedding concepts
• 2x-Thru and Short-Open-Load-Through (SOLT) calibration strategies
• Reducing fixture discontinuities with careful pad geometry
• Software tools: IDEM, OpenDEKit, or VNA manufacturer utilities

Why It Matters | As PCB speeds push into multi-GHz territory, accurate S-parameter extraction without connectors is essential for channel simulation, equalizer design, and compliance testing.

Topic tags

RF Measurement | S-Parameters | PCB | Signal Integrity

Supporting data: VNA measurement image

Community thread: Thread Link

Q7  What is the most convenient way to implement real-time audio FFT analysis?

Real-time FFT analysis of audio turns a time-domain waveform into a live frequency spectrum. This thread compares hardware (dedicated FFT modules, FPGA), microcontroller (ARM CMSIS-DSP, ESP32 FFT), and PC-based (Python, MATLAB) approaches, weighing latency, cost, and complexity.

Key technical topics covered

• FFT fundamentals: window functions, bin resolution, sample rate requirements
• Microcontroller options: ARM CMSIS-DSP library, ESP32 FFT example
• Dedicated modules: MSGEQ7 7-band analyzer IC, OpenMusicLabs FHT
• PC/software approaches: Python (numpy.fft), MATLAB, Audacity spectrum view

Why It Matters | Audio FFT is used in music visualizers, hearing aid design, acoustic testing, and voice-command pre-processing. Choosing the right platform depends on the required resolution, update rate, and available hardware.

Topic tags: Audio DSP | FFT | Embedded Systems | Signal Processing

Supporting data: N/A — community discussion

Community thread: Thread Link

Join the conversation

If any of these questions sparked an idea or you have hands-on experience with a related problem, jump into the thread — the community benefits most when engineers at all levels contribute. You can also start your own question in the Electronic Projects Design/Ideas/Reviews category on Electro-Tech-Online.

Browse all categories: electro-tech-online.com

The post Engineering deep dive-monthly forum highlights April edition appeared first on Test & Measurement Tips.

The Arbitrary Waveform Generators from Spectrum Instrumentation now include a new Sequence Restart Mode for the 65xx and 66xx series, allowing the full sequence of looped and linked waveforms to restart automatically on a trigger event with fixed trigger-to-output timing. The mode is intended for automated test environments where repeatable sequence control can help reduce test time and improve measurement efficiency. The feature is available at no additional charge through the latest driver installation and is supported on Windows and Linux, with programming examples for Python, MATLAB, C++, and LabVIEW as well as a high-level Python API.

The post New AWG mode restarts sequences on trigger appeared first on Test & Measurement Tips.

The update expands NI Nigel AI with prompt-based code generation in the LabVIEW+ Suite and extends AI-assisted capabilities across FlexLogger, InstrumentStudio, TestStand and SystemLink. The tools are designed to support engineers throughout the test lifecycle, offering context-aware suggestions for development, debugging, validation and system reuse.

According to Emerson, the AI features are built specifically for test engineering environments, where traceability, repeatability and system visibility are required. Engineers retain control over generated code and workflows while using AI to reduce manual effort.

The NI platform combines modular instrumentation hardware with open software and a shared data framework, enabling teams to manage diverse signal types, scale test systems and reuse data across projects and locations.

In internal use, Emerson reports reductions in some test development and troubleshooting tasks from hours or days to minutes. Availability of the new capabilities is expected later in 2026.

The post Emerson adds AI-assisted code generation and workflow support to NI test platform appeared first on Test & Measurement Tips.

Emerson announced updates to its NI test software portfolio, adding AI-assisted features designed to improve test development efficiency and system integration. The NI Nigel AI technology will expand to include prompt-based code generation in LabVIEW+ and broader support across tools such as FlexLogger, InstrumentStudio, TestStand, and SystemLink.

The updates are intended to help engineers develop, validate, and deploy tests more quickly while maintaining visibility into system behavior. Nigel AI is designed for test environments, providing context-aware suggestions across the workflow, from development and code reuse to validation and deployment.

The NI platform combines modular hardware, open software, and a shared data framework to support complex test requirements. Engineers can configure systems to handle a range of signals, manage large data sets, and integrate evolving computing technologies over time.

According to internal testing, AI-assisted workflows reduced some development and troubleshooting tasks from hours or days to minutes. The platform is used in industries including aerospace, semiconductor, and transportation, where reliability, traceability, and performance are critical.

The post Emerson introduces AI capabilities for test automation at NI Connect 2026 appeared first on Test & Measurement Tips.