1. What is fiber optic transponder?

In optical fiber communications, a transponder is the element that sends and receives the optical signal from a fiber. A transponder is typically characterized by its data rate and the maximum distance the signal can travel.

>> The difference between a fiber optic transponder and transceiver

A transponder and transceiver are both functionally similar devices that convert a full-duplex electrical signal in a full-duplex optical signal. The difference between the two being that transceivers interface electrically with the host system using a serial interface, whereas transponders use a parallel interface to do so.

So transponders provide easier to handle lower-rate parallel signals, but are bulkier and consume more power than transceivers.

>> Major functions of a fiber optic transponder includes:

• Electrical and optical signals conversions
• Serialization and deserialization
• Control and monitoring

2. Applications of fiber optic transponder

Multi-rate, bidirectional fiber transponders convert short-reach 10 Gb/s and 40 Gb/s optical signals to long-reach, single-mode dense wavelength division multiplexing (DWDM) optical interfaces.

The modules can be used to enable DWDM applications such as fiber relief, wavelength services, and Metro optical DWDM access overlay on existing optical infrastructure.

Supporting dense wavelength multiplexing schemes, fiber optic transponders can expand the useable bandwidth of a single optical fiber to over 300 Gb/s.

Transponders also provide a standard line interface for multiple protocols through replaceable 10G small form-factor pluggable (XFP) client-side optics.

The data rates and typical protocols transported include synchronous optical network/synchronous digital hierarchy (SONET/SDH) (OC-192 SR1), Gigabit Ethernet (10GBaseS and 10GBaseL), 10 G Fibre Channel (10 GFC) and SONET G.709 forward error correction (FEC)(10.709 Gb/s).

Fiber optic transponder modules can also support 3R operation (reshape, retime, regenerate) at supported rates.

Often, fiber optic transponders are used to for testing interoperability and compatibility. Typical tests and measurements include jitter performance, receiver sensitivity as a function of bit error rate (BER), and transmission performance based on path penalty.Some fiber optic transponders are also used to perform transmitter eye measurements.

>> Major Applications of fiber optic transponder

300-pin MSA fiber optic transponders can transparently carry a native 10G LAN PHY, SONET/SDH and Fibre Channel payload with a carrier grade DWDM Optical Transport Network (OTN) interface without the need for bandwidth limitation.

Transponders offer G.709 compliant Digital Wrapper, Enhanced Forward Error Correction (FEC) and Electrical Dispersion Compensation (EDC) for advanced optical performance and management functions superior to those found in DWDM Transponder systems.

They support full C or L band tunability and is designed to interoperate with any Open DWDM line system that support 50GHz spaced wavelengths per the ITU-T grid.

• Enables reach extension on SONET, Storage Area Network (SAN), Gigabit Ethernet, and dispersion limited links
• Wavelength services and Metro optical access overlay
• Agile Optical Networks

>> Other Applications

1) Multimode to Single Mode Conversion

Some transponders can convert from multimode to single mode fiber, short reach to long reach lasers, and/or 850/1310nm to 1550nm wavelengths.  Each transponder module is protocol transparent and operates fully independent of the adjacent channels.

2) Redundant Fiber Path

Each transponder module can also include a redundant fiber path option for extra protection.  The redundant fiber option transmits the source signal over two different optical paths to two redundant receivers at the other end.

If the primary path is lost, the backup receiver is switched on. Because this is done electronically rather than mechanically, it is much faster and more reliable.

3) Repeater

As an optical repeater, some fiber optic transponders effectively extend an optical signal to cover the desired distance. With the Clock Recovery option, a degraded signal can be de-jittered and retransmitted to optimize signal quality.

4) Mode Conversion

Mode conversion is one of the quickest and simplest ways of extending multimode optical signals over greater distances on single mode fiber optics.

Note:  Most receivers are capable of receiving both multimode and single mode optical signals.

3. 10Gb/s transponder block diagram

Fiber optic transponders do the simple conversion from low-speed electrical signals to high-speed optical signals

These optical transceivers with built-in MUX/DEMUX come in a compact package with a multiplexing function converting 622Mbps low-speed electrical signals to a 10Gbps ultra-high-speed optical signal.

They can contribute to significantly smaller and cheaper optical interfaces in communications equipment and switches/routers.

4. How to select a transponder?

Selecting fiber optic transponders requires an understanding of jitter measurements and BER measurements.

>> Jitter measurement

There are three types of jitter measurement: jitter generation, jitter tolerance, and jitter transfer. Jitter analyzers are used with fiber optic transponders and test boards.

• Jitter generation data includes current and maximum values for jitter peak – peak, jitter + peak, jitter – peak, and jitter RMS (root mean squared).
• Jitter tolerance and jitter performance are scaled values.

The following figure shows an example setup for jitter measurements.

>> BER measurement

For BER measurements, test boards with fiber optic transponders are used with pulse pattern generators, error detectors, reference lasers, and reference receivers. Case temperature is an important variable.

The following figure shows an example setup for Bit Error Rate Measurement.

>> Transmitter eye measurement

For transmitter eye measurements, fiber optic transponders are used with pulse pattern generators, reference lasers, and high-speed oscilloscopes.

There are significant differences between the filtered eye and the unfiltered eye. Eye measurements vary by distance and, because of the error rate (ER), may require optical modulation amplitude (OMA) instead.

The following figure shows an example setup for transmitter eye measurement.

>> Path Penalty Measurement

For path-penalty testing, test boards with fiber optic transponders are used with pulse pattern generators, reference lasers, reference receivers, and fiber spools. Original chirp and optimal chirp are important parameters to consider.

The following figure shows an example setup for path penalty measurement.

5. 300-pin Transponder MSA

http://www.300pinmsa.org/

>> Why Do We Need Wavelength Converters?

Wavelength conversion can be used in WDM networks to improve efficiency. Consider the network in figure 1 below. It shows a wavelength-routed network containing two WDM crossconnects (S1 and S2) and five access stations (A through E). Three lightpaths have been set up (C to A on wavelength λ1, C to B on λ2, and D to E on λ1).

Figure 1 – All-Optical Wavelength-Routed network

To establish a lightpath, we require that the same wavelength be allocated on all the links in the path. This requirement is known as the wavelength-continuity constrains. This constraint distinguishes the wavelength-routed network from a circuit-switched network which block calls only when there is no capacity along any of the links in the path assigned to the call.

1) Consider the Example in Figure 2(a) Below.

Two lightpaths have been established in the network:

1. Between Node 1 and Node 2 on wavelength λ1
2. Between Node 2 and Node 3 on wavelength λ2

Now suppose a lightpath between Node 1 and Node 3 needs to be set up. Establishing such as lightpath is impossible even though there is a free wavelength on each of the links along the path from Node 1 to Node 3. This is because the available wavelengths on the two links are DIFFERENT. Thus, a wavelength continuity network may suffer from higher blocking as compared to a circuit-switched network.

Figure 2 – Wavelength-Continuity Constraint in a Wavelength-Routed Network

It is easy to eliminate the wavelength-continuity constraint, if we were able to convert the data arriving on one wavelength along a link into another wavelength at an intermediate node and forward it along the next link. Such a technique is referred to as wavelength conversion.

In above figure 2(b), a wavelength converter at Node 2 is employed to convert data from wavelength λ2 to λ1. The new lightpath between Node 1 and Node 3 can now be established by using the wavelength λ2 on the link from Node 1 to Node 2, and then by using the wavelength λ1 to reach Node 3 from Node 2.

Notice that a single lightpath in such a wavelength-convertible network can use a different wavelength along each of the links in its path. Thus, wavelength conversion may improve the efficiency in the network by resolving the wavelength conflicts of the lightpaths.

2) Wavelength Converter Functions

The function of a wavelength converter is to convert data on an input wavelength onto a possibly different output wavelength among the N wavelengths in the system (see figure 3 below).

Figure 3 – Functionality of a Wavelength Converter

In this figure and throughout this tutorial,

• λs = the input signal wavelength
• λc = the converted wavelength
• λp = the pump wavelength
• fs = the input frequency
• fc = the converted frequency
• fp = the pump frequency
• CW = the continuous wave (unmodulated) generated as the pump signal

An ideal wavelength converter should possess the following characteristics:

• Transparency to bit rates and signal formats
• Fast setup time of output wavelength
• Conversion to both shorter and longer wavelengths
• Moderate input power levels
• Possibility for same input and output wavelengths (i.e., no conversion)
• Insensitivity to input signal polarization
• Low-chirp output signal with high extinction ratio and large signal-to-noise ratio
• Simple implementation

>> Wavelength Conversion Technologies

Wavelength converters can be classified based on the range of wavelengths that they can handle at their inputs and outputs. A fixed-input, fixed-output device always takes in a fixed-input wavelength and converts it to a fixed-output wavelength. A variable-input, fixed-output device takes in a variety of wavelengths but always converts the input signal to a fixed-output wavelength. A fixed-input, variable-output device does the opposite function. Finally, a variable-input, variable-output device can convert any input wavelength to any output wavelength.

In addition to the range of wavelengths at the input and output, we also need to consider the range of input optical powers that the converter can handle, whether the converter is transparent to the bit rate and modulation format of the input signals, and whether it introduces additional noise or phase jitter to the signal. We will see that the latter two characteristics depend on the type of regeneration used in the converter. For all-optical wavelength converters, polarization-dependent loss should also be kept to a minimum.

There are four fundamental ways of achieving wavelength conversion.

1. Optoelectronic
2. Optical Gating
3. Interferometric
4. Wave Mixing

The latter three approaches are all-optical but not yet mature enough for commercial use. Optoelectronic converters today offer substantially better performance at lower cost than comparable all-optical wavelength converters.

1. Optoelectronic Approach

In optoelectronic wavelength conversion, the optical signal to be converted is first translated into the electronic domain using a photodetector. The electronic bit stream is stored in the buffer (labeled FIFO for First-In-First-Out queue mechanism). The electronic signal is then used to drive the input of a tunable laser (labeled T) tuned to the desired wavelength of the output.

Figure 4 – An Opto-Electronic Wavelength Converter

This is perhaps the simplest, most obvious, and most practical method today to realize wavelength conversion. This is usually a variable-input, fixed-output converter. The receiver does not usually care about the input wavelength, as long as it is in the 1310 or 1550 nm window. The laser is usually a fixed-wavelength laser. A variable output can be obtained by using a tunable laser.

The performance and transparency of the converter depend on the type of regeneration used. Figure 5 below shows the different types of regeneration possible.

Figure 5 – Different Types of Optoelectronic Wavelength Converters

In the simplest case (figure 5(a)), the receiver simply converts the incoming photons to electrons, which get amplified by an analog RF (radio-frequency) amplifier and drive the laser. This is called 1R regeneration. This form of conversion is truly transparent to the modulation format (provided the appropriate receiver is used to receive the signal) and can handle analog data as well. However, noise is added at the converter, and the effects of nonlinearities and dispersion are not reset.

Another alternative (figure 5(b)) is to use regeneration with reshaping but without retiming, also called 2R regeneration. This is applicable only to digital data. The signal is reshaped by sending it through a logic gate, but not retimed. The additional phase jitter introduced because of this process will eventually limit the number of stages that can be cascaded.

The final alternative (figure 5(c)) is to use regeneration with reshaping and retiming (3R). This completely resets the effects of nonlinearities, fiber dispersion, and amplifier noise; moreover, it introduces no additional noise. However, retiming is a bit-rate-specific function, and we lose transparency. If transparency is not very important, this is a very attractive approach. These types of regenerators often include circuitry to perform performance monitoring and process and modify associated management overheads associated with the signal.

2. Optical Gating

Cross-modulation wavelength conversion techniques utilize active semiconductor optical devices such as semiconductor optical amplifiers (SOA) and lasers. These techniques belong to a class known as Optical-Gating wavelength conversion.

Optical gating makes use of an optical device whose characteristics change with the intensity of an input signal. This change can be transferred to another unmodulated probe signal at a different wavelength going through the device. At the output, the probe signal contains the information that is on the input signal.

Like the optoelectronic approach, these devices are variable-input and either fixed-output or variable-output devices, depending on whether the probe signal is fixed or tunable. The transparency offered by this approach is limited—only intensity-modulated signals can be converted.

Cross-Gain Modulation (CGM)

The principle behind using an SOA in the cross-gain modulation (CGM) mode is shown in figure 6 below.

Figure 6 – A Wavelength Converter Using Co-propagation based on Cross Gain Modulation (CGM) in an SOA

The intensity-modulated input signal modulates the gain in the SOA due to gain saturation. A continuous-wave (CW) signal at the desired output wavelength (λc) is modulated by the gain variation so that it carries the same information as the original input signal.

The CW signal can either be launched into the SOA in the same direction as the input signal (co-directional), or launched into the SOA in the opposite direction as the input signal (counter-direction). The CGM scheme gives a wavelength-converted signal that is inverted compared to the input signal.

While the CGM scheme is simple to realize and offers penalty-free conversion at 10 Gbps, it suffers from the drawbacks due to inversion of the converted bit stream and extinction ratio degradation for the converted signal. But, new advanced developments demonstrate that no signal inversion occurs.

This approach works over a wide range of signal and probe wavelengths, as long as they are within the amplifier gain bandwidth, which is about 100 nm. Early SOAs were polarization sensitive, but by careful fabrication, it is possible to make them polarization insensitive. SOAs also add spontaneous emission noise to the signal.

What makes this interesting is that the carrier dynamics within the SOA are very fast, happening on a picosecond time scale. Thus the gain responds in tune with the fluctuations in input power on a bit-by-bit basis. The device can handle bit rates as high as 10 Gb/s. If a low-power probe signal at a different wavelength is sent into the SOA, it will experience a low gain when there is a 1 bit in the input signal and a higher gain when there is a 0 bit. This very same effect produces crosstalk when multiple signals at different wavelengths are amplified by a single SOA and makes the SOA relatively unsuitable for amplifying WDM signals.

The advantage of CGM is that it is conceptually simple. However, there are several drawbacks. The achievable extinction ratio is small (less than 10) since the gain does not really drop to zero when there is an input 1 bit. The input signal power must be high (around 0 dBm) so that the amplifier is saturated enough to produce a good variation in gain. This high-powered signal must be eliminated at the amplifier output by suitable filtering, unless the signal and probe are counterpropagating. Moreover, as the carrier density within the SOA varies, it changes the refractive index as well, which in turn affects the phase of the probe and creates a large amount of pulse distortion.

3. Interferometric Techniques

Cross-Phase Modulation (CPM)

The same phase-change effect that creates pulse distortion in CGM can be used to effect wavelength conversion. As the carrier density in the amplifier varies with the input signal, it produces a change in the refractive index, which in turn modulates the phase of the probe. Hence we use the term cross-phase modulation for this approach.

The operation of a wavelength converter using SOA in cross-phase modulation (CPM) mode is based on the fact that the refractive index of the SOA is dependent on the carrier density in its active region. An incoming signal that depletes the carrier density will modulate the refractive index and thereby results in phase modulation of a CW signal (wavelength λc) coupled into the converter.

This phase modulation can be converted into intensity modulation by using an interferometer such as a Mach-Zehnder interferometer (MZI). Figure 7 below shows one possible configuration of a wavelength converter using cross-phase modulation.

Figure 7 – Wavelength Conversion by Cross-Phase Modulation Using Semiconductor Optical Amplifier Embedded Inside a Mach-Zehnder Interferometer

Both arms of the MZI have exactly the same length, with each arm incorporating an SOA. The signal is sent in at one end (A) and the probe at the other end (B). If no signal is present, then the probe signal comes out unmodulated. The couplers in the MZI are designed with an asymmetric coupling ratio γ 0.5. When the signal is present, it induces a phase change in each amplifier. The phase change induced by each amplifier on the probe is different because different amounts of signal power are present in the two amplifiers. The MZI translates this relative phase difference between its two arms on the probe into an intensity-modulated signal at the output.

This approach has a few interesting properties. The natural state of the MZI (when no input signal is present) can be arranged to produce either destructive or constructive interference on the probe signal. Therefore we can have a choice of whether the data coming out is the same as the input data or is complementary.

The advantage of this approach over CGM is that much less signal power is required to achieve a large phase shift compared to a large gain shift. In fact, a low signal power and a high probe power can be used, making this method more attractive than CGM. This method also produces a better extinction ratio because the phase change can be converted into a “digital” amplitude-modulated output signal by the interferometer. So this device provides regeneration with reshaping (2R) of the pulses.

Depending on where the MZI is operated, the probe can be modulated with the same polarity as the input signal, or the opposite polarity. Referring to Figure 7 above, where we plot the power coupled out at the probe wavelength versus the power at the signal wavelength, depending on the slope of the demultiplexer, a signal power increase can either decrease or increase the power coupled out at the probe wavelength.

Like CGM, the bit rate that can be handled is at most 10 Gb/s and is limited by the carrier lifetime. This approach, however, requires very tight control of the bias current of the SOA, as small changes in the bias current produce refractive index changes that significantly affect the phase of signals passing through the device.

We have seen that the CPM interferometric approach provides regeneration with reshaping (2R) of the pulses. As we saw earlier, while 2R cleans up the signal shape, it does not eliminate phase (or equivalently timing) jitter in the signal, which would accumulate with each such 2R stage.

In order to completely clean up the signal, including its temporal characteristics, we need regeneration with reshaping and retiming (3R). Figure 8 shows one proposal for accomplishing this in the optical domain without resorting to electronic conversion.

Figure 8 – All-Optical Regeneration with Reshaping and Retiming (3R) Using a combination of Cross-Gain Modulation and Cross-Phase Modulation in Semiconductor Optical Amplifiers

The approach uses a combination of CGM and CPM. We assume that a local clock is available to sample the incoming data. This clock needs to be recovered from the data. The regenerator consists of three stages.

The first stage samples the signal. It makes use of CGM in an SOA. The incoming signal is probed using two separate signals at different wavelengths. The two probe signals are synchronized and modulated at twice the data rate of the incoming signal. Since the clock is available, the phase of the probe signals is adjusted to sample the input signal in the middle of the bit interval. At the output of the first stage, the two probe signals have reduced power levels when the input signal is present and higher power levels when the input signal is absent.

In the second stage, one of the probe signals is delayed by half a bit period with respect to the other. At the output of this stage, the combined signal has a bit rate that matches the bit rate of the input signal and has been regenerated and retimed.

This signal is then sent through a CPM-based interferometric converter stage, which then regenerates and reshapes the signal to create an output signal that has been regenerated, retimed, and reshaped.

4. Wave Mixing

Wavelength conversion methods using coherent effects are typically based on wave-mixing properties (see figure 9 below).

Figure 9 – A Wavelength Converter Based on Nonlinear Wave-Mixing Effects

Wave-mixing arises from a nonlinear optical response of a medium when more than one wave is present. It results in the generation of another wave whose intensity is proportional to the product of the interacting wave intensities. Wave-mixing preserves both phase and amplitude information, offering strict transparency. It is also the only approach that allows simultaneous conversion of a set of multiple input wavelengths to another set of multiple output wavelengths and could potentially accommodate signal with high-bit rates.

In Figure 9, the value n = 3 corresponds to Four-Wave Mixing (FWM) and n=2 corresponds to Difference Frequency Generation (DFG). These techniques are discussed below.

Four-Wave Mixing (FWM)

Four-Wave Mixing (FWM) is a third-order nonlinearity in silica fibers, which causes three optical waves of frequencies f_i, f_j, and f_k (ki,j) to interact in a multichannel WDM system to generate a fourth wave of frequency given by:

f_ijk = f_i ± f_j ± f_k

FWM is also achievable in other passive waveguides such as semiconductor waveguides and in an active medium such as a semiconductor optical amplifier (SOA). Four-Wave Mixing (FWM) is a promising technique for wavelength conversion in optical networks owing to its ultrafast response and high transparency to bit rate and modulation format.

For the purposes of wavelength conversion, the four-wave mixing power can be enhanced by using an SOA because of the higher intensities within the device. If we have a signal at frequency fs and a probe at frequency fp, then four-wave mixing will produce signals at frequencies 2fp −fs and 2fs −fp, as long as all these frequencies lie within the amplifier bandwidth (Figure 10 below).

Figure 10 – Wavelength Conversion by Four-Wave Mixing in a Semiconductor Optical Amplifier

The main advantage of four-wave mixing is that it is truly transparent because the effect does not depend on the modulation format (since both amplitude and phase are preserved during the mixing process) and the bit rate. The disadvantages are that the other waves must be filtered out at the SOA output, and the conversion efficiency goes down significantly as the wavelength separation between the signal and probe is increased.

Difference Frequency Generation (DFG)

Difference Frequency Generation (DFG) is a consequence of a second-order nonlinear interaction of a medium with two optical waves: a pump wave and a signal wave. DFG is free from satellite signals which appear in Four-Wave Mixing based techniques.

This technique offers a full range of transparency without adding excess noise to the signal. It is also bidirectional and fast, but it suffers from low efficiency and high polarization sensitivity. The main difficulties in implementing this technique lie in the phase-matching of interacting waves and in fabricating a low-loss waveguide for high conversion efficiency.

>> Summary

In this tutorial, we reviewed the various techniques and technologies used in the design of a wavelength converter. The actual choice of the technology to be employed for wavelength conversion in a network depends on the requirements of the particular system. However, it is clear that opto-electronic converters offer only limited digital transparency. Moreover, deploying multiple opto-electronic converters in a WDM intermediate node requires sophisticated packaging to avoid crosstalk among channels. This leads to increased cost per converter, further making this technology less attractive than all-optical converters.

Other disadvantages of opto-electronic converters include complexity and large power consumption. Regarding all-optical wavelength conversion, there are a lot of various schemes developed in which each has own strength and weak in application. Besides all-optical converters mentioned above, another kind of wavelength conversion is based on crystal material, which has also attracted more attention currently. Specially, periodically-poled LiNbO3 waveguide is preferential in this kind of all-optical wavelength conversion scheme.

>> Light Source Launch Condition to a Multimode Fiber

The device that was developed for high-speed multimode transmission at 850 nm was a VCSEL (Vertical Cavity Surface Emitting Laser), which uses a gallium arsenide substrate for speed.  VCSELs can easily operate at 2 Gigabit and produce milliwatts of power with threshold currents so low that they can be driven directly from logic gates without the need for drivers.

Additionally, they can be manufactured very cost effectively when compared to typical single mode semiconductor lasers. VCSELs are the first laser devices that transmit signals in more than one mode, making them ideal for use with multimode fiber.

Each type of device transmits the optical signal in a different manner; LEDs excite all the different fiber modes, VCSELs excite only a portion and the single mode lasers excite only the fundamental mode.

The following figure shows how LEDs, VCSELs and semiconductor lasers transmit in a multimode fiber.

>> DMD (Differential Mode Delay) and Its Limitation on Multimode Fiber Bandwidth

Multimode fiber’s information carrying capacity typically is rated in terms of a bandwidth length product (MHz-km) that can be used to determine how far a system can operate at what bit rate. A simple intuitive model for the bandwidth characteristics considers the fiber to consist of a number of discrete delay lines, each of which corresponds to a particular mode. A conceptual model is shown in Figure 1 below.

In the figure, the low order modes (LOM) correspond to the modes or rays propagating down the center of the fiber; the high order modes (HOM) propagate near the core/clad interface; and the intermediate modes propagate in between. Although somewhat counter intuitive, the intermediate modes carry most of the power.

In an ideal fiber, all of the delays are tuned to be identical. Thus, when a temporally narrow pulse of light is launched into the fiber, its shape is maintained at the output. However, as the example in Figure 1 shows, the over exaggerated delay error causes the output pulse to be broadened. The high order mode power arrives late relative to most of the power (in the intermediate modes), and the low order modes arrive early. Thus, the bandwidth is reduced and the information carrying capacity is limited.

>> Launch Condition and Multimode Fiber Bandwidth

This model then explains why bandwidth is launch condition dependent. The output pulse given in Figure 1 corresponds to an overfilled launch where all of the modes are excited with the maximum amount of power they can carry. This launch is defined by standards; and a typical source that this might correspond to is an LED.

If the launch power distribution is then reduced so that only the lower and intermediate modes are excited, the power in the late peak of the pulse disappears, the pulse width decreases, and the bandwidth goes up. The analogy can be extended equally as well where the launch is restricted to just the lowest order modes and the output pulse becomes very narrow. Thus, if all of the modes of a multimode fiber are not tuned perfectly, the bandwidth can change as the launch power distribution changes.

The tuning of these modes is quantified by the differential mode delay (DMD) measurement. The DMD is a process tuning tool which measures the delay of each of the modes. It has never been standardized because it is difficult to implement, costly to run, and not a functional input measure for system modeling.

Bandwidth, on the other hand, is functional and is used in system modeling. The DMD consists of recording the mean delay as a spot of light formed by a single-mode fiber is scanned across the input of the fiber. The typical technique for this measurement is a variation of the time domain bandwidth measurement – a high speed pulse is used.

>> What is Modal Distribution?

Multimode light transmission is by many different paths (see the following figure).

The paths (or modes) nearest to the axis of the core are called low order modes, and the paths with the most deviation are called high order modes.

It is the high order modes that get attenuated (lost), and as they do so, light gradually transfers from the lower to the higher order modes. Interaction between light in the various modes causes some level of random signal noise, which is called "modal noise".

When light is injected into a core, initially the relative proportion of light travelling in low or high order modes will depend on how the light was injected.

• If an abnormal amount of light is in the low order modes, the core is "under-filled", and losses will be abnormally low. This situation is commonly achieved with a laser emitter. This situation also minimizes pulse spreading due to modal dispersion. A number of multimode transmitters used this effect to transmit dater at a higher rate than the "bandwidth rating" of the fiber being used.
• If an abnormal amount of light is in the high order modes, the core is "over-filled" and losses will be abnormally high, and may appear "out of specification". This situation is commonly achieved with an LED emitter. In this situation the, signal pulse dispersion will be worse than expected from the fiber bandwidth specification.
• After a distance of typically 1-2 Km, the relative distribution of modal power stabilizes and is independent of the initial launch condition. This condition is called an equilibrium mode distribution (EMD), and is the condition traditionally used to measure individual multimode components. This is interesting, but usually irrelevant to today’s multimode systems where most link lengths are much shorter than this. The concept is a remnant of a long past era when multimode was used in long distance Telecom applications.

If attenuation tests are made without first establishing a known modal distribution condition, then the attenuation results obtained will be unrepeatable. It should be noted that performance specifications of multimode systems are only valid for a specific mode condition. In other circumstances, different characteristics will be measured.

So to perform meaningful multimode measurements, it is necessary to first define the modal distribution condition.

Recent standards change the emphasis in multimode plant verification. The traditional approach was to try and measure at an equilibrium modal distribution, e.g. an average result. The new approach is use an over-filled source, e.g. worst case result. The new approach also matches the conditions found in practice in today’s short distance applications using a LED transmitter.

Of the two types of multimode fiber in common use today, the new standards appear to concentrate on 50/125 OM3 fibers. In fact a source that is fully CPR compliant with this type, in our experience may not be CPR compliant with the older 62.5/125 types. Presumably the standards folk will resolve this issue in due course. This incidental anomaly is probably a by-product of the fact that CPR issues primarily focus on CPR and it’s effect on bandwidth, rather than the secondary issue of attenuation measurement.

Use of a mandrel wrap or mode stripper can not create something approximating to a modern overfilled launch condition (a common misconception). This has to be achieved with a suitable compliant source.

>> How Does Fiber Modal Distribution Look Like?

Modal distribution in a multimode fiber depends on your source, fiber, and the intermediate "components" such as connectors, couplers and switches, all of which affect the modal distribution of fibers they connect. Typical modal distributions for various fiber optic components are shown here.

In the laboratory, a lensed optical system can be used to fully fill the fiber modes and a "mode filter", usually a mandrel wrap which stresses the fiber and increases loss for the higher order modes, used to simulate EMD conditions. A "mode scrambler", made by fusion splicing a step index fiber in the graded index fiber near the source can also be used to fill all modes equally. If one has a proper optical system, one can control the launch conditions to very specific levels as desired for the measurements being performed.

A fully filled fiber means that all modes carry equal power, as shown by the line across the top of the graph. A long length of fiber loses light in the higher order modes faster, leading to the gently sloping "EMD" curve. Mode filtering strips off the higher order modes, but provides only a crude approximation of EMD. The microlensed LED , often thought to overfill the modes, actually couples most of its power in lower order modes. The E-LED (edge-emitting LED) couples even more strongly in the lower order modes. Connectors are mode mixers, since misalignment losses cause some power in lower order modes to be coupled up to higher order modes.

>> How To Measure Modal Fill in Multimode Fibers

Modal fill can be measured by either a near field scan or a far field scan. The technique is similar to measuring numerical aperture (NA) by looking at the light exiting the fiber. If light fills more modes the scan (intensity vs. position across the fiber end, either near or far field) will be wider, as shown by the red and green modes and profiles below. The presentation of the data is where changes have occurred over the years. Mode power distribution has been used for years but has been replaced by encircled flux for standards. CPR has also been used as a simple metric but has serious problems and is becoming obsolete.

1. Coupled Power Ratio

This is an old outdated requirement that has been replaced with Modal Power Distribution, which in itself is being replaced with a new requirement called Encircled Flux.

The measurement of Coupled Power Ratio (CPR) is specified by international standards as a means of quantifying the amount of modal filling in multimode fibers. A CPR measurement is made by comparing the light output from a patch cord with the light coupled into a single mode fiber that is butt coupled to the end of the patch cord. The following diagram shows the principle of the CPR measurement:

CPR = P1 – P2 dBm

CPR was divided into classes. The rated category values in dB for both 850nm and 1300nm into a 62.5/125 multimode fiber, are as follows:

2. Mode Power Distribution

Mode power distribution (MPD) has been used for a long time as a metric for modal distribution. It is a result of a far field scan of the output of a fiber with some mathematical manipulation to show the power in the modes, all normalized, as shown in the graph below. The dark lines are the limits set for test sources. Needless to say, it’s not a very easy function to visualize, leading to searches for other ways to measure and define modal distribution.

3. Encircled Flux (EF)

Encircled Flux is a method of characterizing the launch conditions of a multimode light source such as a light emitting diode (LED) or laser. EF is the percentage of power within a given fiber core radius when light is launched by a transmitter into a multimode fiber and is determined from the near-field measurement of the light coming from the end of a reference-grade test cord attached to the test instrument.

Encircled Flux compliance reduces loss measurement variation to a goal of +/- 10%. Reducing variability by up to 75% compared to the preceding standard, it is the most recent standard that increases multimode testing accuracy and repeatability.

While the lab is the ideal environment to meet EF launch conditions, there are now external solutions available for field testing purposes. Such accessories, called “launch controllers”, are specially constructed test-grade reference cords fitted with modal conditioners. These launch controllers work by restricting the number of mode groups launched from the test cord to within EF specifications, ensuring that the resulting measurements are precise and repeatable according to the standards.

Encircled flux (EF), defines the integral of power output of the fiber over the radius of the fiber. When you look look at the graph below, consider that the vertical axis is the total amount of optical power from the source coupled into a fiber core inside the radius shown in the horizontal axis. EF was defined during the development of 10 GB Ethernet as a way to define the light output from an ideal VCSEL source which concentrates more of its power in the center of the fiber than a LED.

The EF definition was used for bandwidth simulation only at that point. Of course a real VCSEL may be (is likely to be) different, but this model allowed calculating the bandwidth of this ideal VCSEL in various types of fibers of various lengths to determine their capability of supporting 10 GBE. It was later decided that EF would be a better way to define mode fill for loss testing.

This method of measuring mode fill should be more precise than other methods like CPR. EF should be easier to measure using imaging devices that can be calibrated.

>> The Gigabit Ethernet Optical Link Model

The IEEE 802.3z committee’s Gigabit Ethernet Link Model spreadsheet is available for public use, to enter assumptions for a given link configuration.

The Gigabit Ethernet model is based on the concept of power budgeting or Power_out =  Power_in – Power Attenuation Losses – Link Power Margin.

The GbE link model was created by using worst case models and extensive experimental testing, resulting in an empirical estimation of the characteristic of a fiber optic link. The multimode fiber modal bandwidth is estimated by this empirical spreadsheet. The units of measure for a multimode fiber bandwidth is expressed in MHz * km to designate the bandwidth-distance product. See Figure 2 and 3, which provide a graphical representation of the bandwidth– distance product calculated from the IEEE 802.3z GbE link model spreadsheet.

Figure 2 shows that 160 MHz*km modal bandwidth at 850 nm for 62.5 micron multimode fiber is limited to a maximum link length of 220 meters. This is due to the limiting factor of the ISI power penalty on the bandwidth-distance product.

Figure 3 shows you that 500 MHz*km modal bandwidth at 1300 nm for 62.5 micron multimode fiber is limited to a link length of 550 meters, which has been set for power budgeting. There is still approximately 1.5 dB power margin left, which if this were utilized the maximum ISI limited link length would become approximately 690 meters.

In addition, there are other power loss and noise factors that effect the transmission bandwidth– distance product.

>> Gigabit Ethernet Standard Link Length Specification

Faced with an installed base of multimode fiber, whose bandwidth was unknown for such restricted launches, the Gigabit Ethernet standards group chose to control the launch distribution produced by the source. After evaluating various launch specifications and control mechanisms, it was agreed that sources must meet a coupled power ratio (CPR) requirement and that 1300 nm operation would require an offset launch patch cord.

CPR measurement is described in OFSTP 147 and consists of taking the ratio of the power coupled into SMF compared to MMF. The requirement ensures that a source launches a power distribution which is fairly overfilling. The offset patch cord achieves a significantly over filling power distribution by launching from a single-mode fiber into a multimode fiber through a controlled offset created by a special connector. The source must be designed to couple efficiently into the SMF launch fiber.

By controlling the source launch conditions as well as adding a receiver bandwidth requirement and reallocating the jitter budget, agreement was achieved on the link lengths in the current standard and reproduced in the following table.

>> Modal Distribution’s Effects on Multimode Fiber Loss Measurements

If you measure the attenuation of a long fiber in EMD (or any fiber with EMD simulated launch conditions) and compare it to a normal fiber with "overfill launch conditions " (that is the source fills all the modes equally), you will find the difference is about 1 dB/km, and this figure is the "transient loss". Thus, the EMD fiber measurement gives an attenuation that is 1 dB/km less than the overfill conditions.

Fiber manufacturers use the EMD type of measurement for fiber because it is more reproducible and is representative of the losses to be expected in long lengths of fiber. But with connectors, the EMD measurement can give overly optimistic results, since it effectively represents a situation where one launches from a smaller diameter fiber of lower NA than the receive fiber, an ideal situation for low connector loss.

The difference in connector loss caused by modal launch conditions can be dramatic. Using the same pair of non-PC biconic or SMA connectors, it is possible to measure 0.6 to 0.9 dB with a fully filled launch and 0.3 to 0.4 dB with a EMD simulated launch. PC connectors (ST, SC or LC) will have smaller but measurable differences, ~0.1-0.2 dB. Which is a valid number to use for a connector pair’s loss ?

That depends on the application. If you are connecting two fibers near a LED source, the higher value may be more representative, since the launch cable is so short. But if you are connecting to a cable one km away, the lower value may be more valid.

>> First Generation 100GbE

>> Second Generation 100GbE

>> Front Panel Interface Density Trends

>> 28 Gbps Common Electrical Interfaces (CEI)

• Optical Internetworking Forum (OIF) is doing fundamental work on 28 Gbps electrical signaling which will make newer interfaces and pluggable media modules possible
• Lower power, Very Short Reach (VSR) 4" interfaces are being defined for several new applications
  (a) 1 lane for 32 Gbps Fibre Channel at 28.05 Gbps
  (b) 4 lanes for 100 GbE at 25.78125 Gbps
  (c) 16 lanes for 400 GbE at 25.78125 Gbps
• CEI-28G-VSR is approaching technical stability, and is expected to be finished soon

>> 25 Gbps and 28 Gbps Common Electrical Interfaces (CEI)

1. Backplane: CEI-25G-LR – 30"

2. Chip to Chip: CEI-28G-SR – 12"

3. Chip to Module: CEI-28G-VSR – 4" (used by 2nd generation 100GbE media modules)

>> 10Gbps Module Review – 3 Generations of 10 GbE Over 7 Years

>> 100 Gbps Module Evolution

Two generations of 100 GbE expected to take 5 years

>> 100 Gbps CFP Module Evolution

>> 100 Gbps Module Evolution (Graphical View of Module Form Factors)

>> 100 GbE Module Technologies Comparison

>> Recent 100 GbE Developments

1. 2nd Generation projects based on 4×25 Gbps electrical signaling have started

2. New IEEE Copper Study Group was approved in November, 2010

• 100GBase-KR4: 4×25 Gbps over backplane
• 100GBase-CR4: 4×25 Gbps over copper cable

3. 10×10 MSA is growing and working on several projects

• Up to 25 members including AMS-IX, Facebook and Google
• Finishing 10X10-10km and 10×10-40km standards, expected to be approved
• Investigating muxing 8 bands of 40 km links to carry 8 x 100 Gbps over a single fiber pair

4. IEEE is expected to start work to define new interfaces that are expected to be available in 2013+

• 100GBase-SR4: 4×25 Gbps over OM3 MMF
• 100GBase-FR4 : 4×25 Gbps over SMF for 500 m – 2km
• CAUI-4: electrical signaling to the CFP2
• CPPI-4: electrical signaling to the 25 Gbps QSFP and CFP4
• 25 Gbps QSFP and CFP2/4 will be competing for the highest front panel density

>> 100 GbE Line Card Architectures

>  1st generation 10 Gbps and 25 Gbps signaling

>  2nd generation 25 Gbps signaling

>> 100 GbE Technology Overview

>> Beyond 100 GbE

>> Next Higher Speed Ethernet (250 GbE, 300 GbE, 400 GbE, or TbE?)

• Using 10×25 Gbps signaling the next speed could be 250 GbE – However, the industry wants a larger jump
• 12×25 Gbps signaling matches the number of fibers in a high density MMF cable for 300 GbE – Unpopular too
• The likely candidate for the next speed is 400 GbE using 16×25 Gbps signaling
  16×25 Gbps wavelengths can be easily muxed/demuxed onto one SMF
  MMF solutions would need 32 fibers in a high density cable MPO/MTP assembly
  Evolution to 10×40 Gbps signaling
• TbE is simply impractical in the near future
  40×25 Gbps lanes in and 40×25 Gbps lanes out would make a gigantic media module
  40 Gbps serial lanes aren’t expected to be economical until after 2016, and will take considerable work as electrical losses grow exponentially with super high frequency signaling

>> 400 GbE Module

The 400 GbE module could be 16 channels wide and would be larger than the current 100 GbE CFP

>> Summary

• The 1st generation of 100 GbE uses 10×10 Gbps electrical lanes and large CFP media modules
• The 2nd generation of 100 GbE will use 4×25 Gbps electrical lanes and smaller CFP2/CFP4/25 Gbps QSFP modules
• Industry is working on 2nd generation 100 GbE for the next few years
• 400 GbE work may start in 2013+ and could finish by 2016+
• TbE is currently technically and economically unfeasible until 40 Gbps electrical lanes are defined after 2013 with a possible standard following many years later

Fiber strands and cables are manufactured with a standard color coding. This allows for easy, effective management and identification of strands.

An example; a loose buffer tube cable with 144 strands would have 12 tubes colored as indicated in the image below. Within each buffer tube would be 12 fiber strands using the same color scheme. Therefore, strand number 61 would be in the white buffer tube, blue fiber.

>> The Need for Bend-Insensitive Fiber

1. Single Mode Bend Insensitive Fiber Applications

Fiber-to-the-Home (FTTH) is becoming the main stream broadband technology  of choice.

In FTTH, fiber is “home-run” from exchanges all the way to the subscriber premises up to the Termination Point (TP) on the wall of the subscriber’s home.

Sharp bends are unavoidable in last mile cable installation in FTTH deployment. Fiber Patch cord connecting TP point to ONT (Optical Network Terminal) also requires ruggedized bend-insensitive capability.

2. Multimode Bend Insensitive Fiber Applications

In enterprise network, multimode fiber is becoming more popular in the horizontal cabling in the Fiber-to-the-Zone (FTTZ) architecture. Bandwidth is shifting from 1Gbps to 10Gbps, therefore shrinking the power loss budge.

>> Bend Insensitive Fiber to The Rescue!

1. Fundamentals of Macrobending in Multimode Fiber

Multimode fiber has many modes of light travelling through the core. As each of these modes moves closer to the edge of the core, it is more likely to escape, especially if the fiber is bent. In a traditional multimode fiber, as the bend radius is decreased, the amount of light that leaks out of the core increases.

2. How Does Bend Insensitive Fiber Work?

Bend-insensitive fiber cables are designed for improved bend performance in reduced-radius applications. The fiber cables employ a moderately higher NA than standard single mode telecommunication fiber cables, and offer improved bend performance for applications in the 1310- and 1550-nm range.

Optical fiber manufacturers used a refractive index “trench”, which means a ring of lower refractive index material, to basically reflect the lost light back into the core of the fiber. This “trench” configuration is shown below.

This specially engineered optical trench can be used to trap the energy in the many modes which propagate within the fiber core. Keeping the light in the core, even in the most challenging bending scenarios, significantly reduces the bend-induced attenuation.

In regular graded index multimode fiber, there are many modes (about 400 of them) being transmitted down the fiber. The inner modes are "strongly guided" which means they have little sensitivity to bending stresses. But the outer modes are "weakly guided" which means they can be stripped out of the core when the fiber is bent.

Bend-insensitive fiber adds a layer of glass around the core of the fiber which has a lower index of refraction that literally "reflects" the weakly guided modes back into the core when stress normally causes them to be coupled into the cladding. Some early single mode fibers (depressed-cladding fibers) used a similar technology to contain the light in the core of the fiber but this design has a much stronger effect.

The trench, or moat as some people call it, surrounds the core in both single mode and multimode fibers to reflect lost light back into the core. The trench is just an annular ring of lower index glass surrounding the core with very carefully designed geometry to maximize the effect. See the red ring around the core on this fiber drawing.

When you look at the end of a bend-insensitive fiber in a microscope with angled lighting, you can sometimes actually see the trench as a gray ring around the core.

Bend-insensitive fiber (or BI fiber as it is now called) has obvious advantages. In patch panels, it should not suffer from bending losses where the cables are tightly bent around the racks. In buildings, it allows fiber to be run inside molding around the ceiling or floor and around doors or windows without inducing high losses. It’s also insurance against problems caused by careless installation.

BI fibers are available in 50/125 MM (OM3 and OM4) and single mode versions.  Considering the advantages of BI fiber and the small incremental cost to manufacture it, some manufacturers have decided to make all their 50/125 MM fiber bend-insensitive fiber. But there are no real advantages for BI single mode fiber in long distance applications, only in premises installations like apartment buildings or for patch cords, so manufacturers offer both regular and BI single mode fiber.

>> Comparison of Standard OM3/OM4 Fiber vs. Corning’s ClearCurve OM3/OM4 Fiber

• Up to 10x better bend performance than standard 50um fiber
• High bandwidth OM3 and OM4 capability
• Improved optical performance
• Fully standards compliant; compatible with installed base
• May be spliced/connectorized with commercially available equipment

>> Bend Insensitive Fiber Standards

1.  Bend Insensitive Fiber Standard for Single mode Fiber

ITU-T G.657 covers two categories of single mode bend insensitive fibers as Category A and Category B as listed below.

G.657 Class A fiber conforms to the widely regarded G.652.D standard, while G.657 Class B fiber does not. Class A fibers will serve the majority of applications requiring a G.652.D fiber with improved bend performance. Class B fibers can support unique or special tight bend applications, but with  the added  challenges of a non-standard fiber (splicing, reliability, higher  cost, etc.).

1) Category A:

Category A is fully compliant with the G.652 single mode fibers . They emphasize on backward compatibility with ITU-T G.652.D Loss specified at 1550nm and can also be used in other parts of the network.

Class A fibers are suitable to be used in the O, E, S, C and L-band (i.e., throughout the 1260 to 1625 nm range). Fibers and requirements in this category are a subset of G.652.D fibers and have the same transmission and interconnection properties. The main improvements are improved bending loss and tighter dimensional specifications, both for improved connectivity.

Features:

• Fibers with improved bend loss and required compliance to G.652
• Typically these fibers have a slightly lower mode-field diameter, but are still within G.652 specification range
• Intended for use anywhere in the network

Class B fibers are suitable for transmission at 1310, 1550 and 1625 nm for restricted distances that are associated with in-building transport of signals. These fibers have different splicing and connection properties than G.652 fibers, but are capable at very low values of bend radius.

Features:

• Fibers with greatly improved bend loss, not necessarily G.652 compliant
• Several technologies such as photonic crystal fibers and hole-assisted fiber, which have very low bending loss can be used because G.652 compliance is not required
• No water peak attenuation specification
• No chromatic or polarization mode dispersion recommendations
• Intended for short reach applications where dispersion and PMD are not an issue

Here is the complete fiber attributes specification from ITU-T G.657 for category B fibers.

2. Bend Insensitive Fiber Standard for Multimode Fiber

Currently there is no standard which defines tighter bend radius for multimode fibers. Here is the regular bend radius specification from IEC 60793-2-10 and ITU-T G.651.1.

Note:

Bend insensitive multimode fibers, available from Corning, OFS, Draka, OFS, and other fiber manufacturers, achieved by keeping most modes in the core of the fiber. But this may disturb the mode distribution which is vital in the high performance multimode fibers such as OM3 and OM4.

>> Compatibility with Conventional Fibers

One question that often arises is are these fibers compatible with regular fibers. Can they be spliced or connected to other conventional (non-BI) fibers without problems? How does the inclusion of higher order modes affect bandwidth? That answer seems to be yes for all SM fibers; it seems you can mix and match regular and BI fibers with no problems.

For MM fibers, it may depend on the manufacturer but for the most part, it appears that BI MM fiber can be made to be compatible to other non BI fibers by modifying the core design slightly.

For BI MM fibers, the fact that they support more higher order modes means that they have a larger effective NA and core size than conventional MM fibers of similar design. Modifying the core index profile slightly to reduce the higher order modes makes them compatible to non BI fibers without otherwise materially affecting the performance of the fiber.

>> Testing of Bend Insensitive Fibers

Testing BI MM fibers or using them for reference cables for testing is another matter. Most multimode fiber testing standards call for modal conditioning, often using a mandrel wrap mode filter. The mandrel wrap specified in most standards doesn’t affect BI fibers the same way as conventional fibers, so either special sources or very small mandrels are required. The encircled flux standard for modal fill which is being adopted by many new standards may address some of these concerns. Another problem arises when testing BI MM fibers with an OTDR. The BI fiber has a larger scattering coefficient than regular fibers and can create large directional loss differences when testing splices or connectors with an OTDR, causing gainers in MM fiber OTDR tests, a major confusion.

1. What is Cladding Mode?

As shown in the above figure, certain light rays in the incident radiation are not captured by the core of the fiber, but pass through the core-cladding interface into the cladding region. Because of the finite radius of curvature of the outer cladding surface, some of this light at this boundary will be reflected back into the cladding, where it can be trapped and propagated. This light forms the cladding modes of the fiber, and appreciable coupling can occur with the higher-order modes of the core, resulting in increased loss of the core power.

2. What is Mode Stripping?

Cladding modes are suppressed by placing a high-loss material outside of the cladding surface that will absorb the light as it strikes the interface, by increasing the scattering at the cladding interface to extract the cladding modes, or by surrounding a portion of the fiber with a material whose index of refraction matches that of the cladding, causing the cladding light to transmit into the index-matching material (see the figure below). This latter technique is called mode stripping.

3. What is Leaky Mode?

Another type of mode is the leaky mode, which is a nonpropagating mode with significant power shared between the core and the cladding. They are predicted by theory and occur near the cutoff conditions for propagating modes. They are attenuated in long fibers but can carry significant power in short fibers. These modes also play a role at connectors and splices, occurring in fibers where the connector or splice can cause conversion of energy from a propagating mode to a leaky mode.

>> Introduction on the first Generation 40/100 Gigabit Ethernet CFP Modules

With the completion of 40/100 Gigabit Ethernet (GbE) optical interface standards (IEEE 802.3ba-2010) and pluggable optical transceiver module specifications (CFP-MSA Rev 1.4), and with the production shipment of first-generation 40GbE/100GbE CFP products underway, optical module vendors are focusing on developing technologies and proving design-ins for their next-generation 40/100GbE pluggable optical transceivers.

The key objectives include significant reductions in module power dissipation and size, which are critical to increasing system port density and reducing overall optical port cost for system vendors and their customers.

The 100GbE CFP module provides the highest faceplate density (in terms of Gbps per faceplate aperature-module pitch area) for MSA-specified pluggable optical modules to date. However, from a systems point of view, the CFP port density most likely will be limited by thermal constraints on power dissipation, which may typically be greater than 25 W for first-generation 100GbE CFP modules.

In this tutorial, we will discuss some of the key challenges facing optical module vendors considering these design objectives and outlines some of the more promising technical approaches to tackle and overcome these challenges.

>> First Generation 40/100GbE Pluggable Optical Transceivers

Let’s first review the technologies and designs of choice in the first generation of 40GbE/100GbE pluggable optical transceivers. The 40GbE and 100GbE optical interfaces specified in IEEE 802.3ba-2010 are summarized in table 1 below.

From a market application perspective, the 40GbE and 100GbE LR4 singlemode fiber optical interfaces are high priority and present the greatest technological challenges when it comes to significant reductions in transceiver power dissipation and form factor size.

1. CFP 40GBase-LR4 Design

The first-generation 40GBase-LR4 optical transceiver is based on a 4x10G architecture that comprises the following discrete components:

• 10G CWDM 1310nm uncooled directly modulated distributed feedback (DM-DFB) transmit optical subassemblies (TOSAs)
• 10G PIN photodiode (PD) with integrated transimpedance amplifier (TIA)
• Receive optical subassemblies (ROSAs)
• Four-channel optical multiplexing/de-multiplexing filters
• Quad dual-channel clock/data recovery (CDR) IC. The quad CDR IC provides the XLAUI 4x10G electrical interface defined in the IEEE Std 802.3ba-2010 specification

These components are packaged into the pluggable CFP module; the module’s mechanical, electrical, and management interface specifications are given in the recently completed CFP MSA rev. 1.4, as shown in figure 1 below.

The CFP module-host system management interface is based on the IEEE Std MDIO/MDC interface and includes several new features, such as programmable controls and alarms, module state transitions, and error rate calculations. The first-generation design leverages existing 10G optoelectronic device technology and uses innovations in packaging to realize high-performance, low-cost modules for high-volume production.

The CFP 40GBase-LR4 module power dissipation is typically in the range of 6 W, which fits well within the CFP module’s 32-W power maximum. Thus, there is considerable interest in reducing the 40G-LR4 module form factor in next-generation designs for increases 40GbE port density. This will be addressed later in this tutorial.

2. CFP 100GBase-LR4/ER4 Design

The first-generation 100GBase-LR4/ER4 optical transceiver architecture is similar to that of the 40GBase-LR4, but with the speed of the active optoelectronic components increased to 28Gbps for realizing a 4x28G optical interface. Additionally, the CAUI electrical interface defined in IEEE 802.3ba-2010 is widened from 4x10G lanes to 10x10G lanes. A 10:4/4:10 “gearbox” serializer/deserializer IC is used to implement the electrical interface between the 10-lane host data path and the four-lane optical data path.

The optical interface defined in IEEE 802.3ba-2010 uses a four-wavelength LAN-WDM 800-GHz wavelength grid in the 1310-nm band and optical multiplexing/de-multiplexing on single mode fiber. The transmitter optical specifications for LR4 and ER4 are based on cooled electro-absorption modulation with integrated DFB (EA-DFB) laser technology, but were written to allow eventual implementation with directly modulated DFB lasers for smaller size, lower power consumption, and lower cost TOSAs.

The receiver optical specifications for LR4 and ER4 are based upon PIN-PD detector technology with integrated TIA. The receiver specification also includes optical amplification, such as from a semiconductor optical amplifier, to compensate for optical fiber attenuation loss in the ER4 40-km application.

These components are packaged into the CFP pluggable module (previously shown in figure 1) with non-coaxial, 28-Gbps electrical connections between the discrete component TOSAs, ROSAs, and gearbox IC, as shown in figure 2 below.

The first-generation 100Gbase-LR4 module power dissipation is typically in the range of 24 W, which poses significant thermal management challenges for system designers, particularly as they seed to increase 100GbE optical port density. Thus, there is strong motivation to significantly reduce the 100GBase-LR4 optical transceiver module power dissipation in the next-generation design.

>> Next-Generation 40GbE/100GbE Optical Module Design Targets

For the next-generation 40GbE/100GbE optical transceiver modules, system designers want significant reductions in power dissipation and form factor size. These objectives are particularly critical as system houses work to scale their core switching and routing input/output capacities and reduce constraints on port densities due to thermal management limits.

For 40Gbase-LR4, the priority target is module form factor reduction. In terms of faceplate density, the current CFP form factor for 40GbE ports is 2.5X less efficient compared to the 100GbE CFP.

One approach the CFP design enables is to double or possible triple the number of 40GbE ports within a single CFP module. While this approach increases port density, it suffers from reduced port provisioning modularity.

A more feasible approach is to make use of the existing QSFP+ form factor (SFF-8436) and the non-retimed XLPPI electrical interface specified in IEEE 802.3ba-2010. This approach increases faceplate density by more than 60 percent over the CFP while retaining port modularity. To make this switch, however, optical module vendors need to not only reduce the physical size of their optical components, but they must reduce component power dissipation by over 50 percent so as to fit into the 3.5-W maximum power envelope od the QSFP+.

For 100GBase-LR4/ER4, the priority target is power dissipation reduction. System makers are looking for power consumption reduction on the order of 50 percent or more. This will ease system thermal management and enable 100GbE port count scaling in the short term. For the longer term, however, system designers seek 100GbE transceiver roadmaps with significant reductions in both power dissipation and form factor size.

>> Next-Generation 40GbE/100GbE Technologies

Several promising technological advances in progress could be used by optical module designers to achieve their next-generation 40GbE/100GbE module design targets. These include:

• Laser array/planar lightwave circuitry (PLC) hybrid integrated TOSA
• PD/TIA array/PLC hybrid integrated ROSA
• Low-power BiCMOS ICs and CMOS gearbox IC
• Narrow 4x28G-VSR electrical interface, electrical connector, and 28G CDR IC
• CFP2 electro-mechanical module development

Hybrid integration of DFB discrete or array lasers with optical multiplexing PLCs has been investigated intensively across the industry. Some of the key challenges to using this technology in TOSA development include laser/PLC device alignment and optical coupling loss minimization. Use of a laser array is desired, as it minimizes the number of process steps in active alignment with the PLC device. However, a DFB laser array is particularly challenging to realize with sufficient gain across all channels for a wide temperature range. Nevertheless, four-channel devices appear to be feasible for realizing an optical hybrid integrated TOSA.

Similarly, hybrid integration of PIN-PD and TIA arrays with optical de-multiplexing PLC has also been investigated. Early process using this type of ROSA was made with 10GBase-LX4 module designs, so it appears feasible to scale ROSA rates up to 4x10G and 4x28G. Challenges still remain in PD/PLC device passive alignment, control of attenuation and polarization mode dispersion losses, and temperature stability for realizing an optical hybrid integrated ROSA.

Significant reduction of 40GbE/100GbE optical transceiver power dissipation will come from improvements in the component ICs. For next-generation 40GBase-LR4, use of the non-retimed XLPPI electrical interface enables elimination of the quad CDR device, which results in more than 30 percent power consumption savings. Further process and design improvements in laser drivers and TIAs will assist with an overall module power consumption reduction of over 50 percent.

For next-generation 100GBase-LR4, EA-DFB driver IC process improvement and CMOS gearbox ICs will be a major factor in module power consumption reduction. With these factors, plus improvements in TIA and module DC-DC power conversion, 50 percent overall CFP module power dissipation looks feasible in the near term.

For the longer term, it is desirable to narrow the module electrical interface to four parallel lanes operating each at 28 Gbps. This would enable replacement of the gearbox IC with a quad 28G CDR IC, thus reducing electrical interface IC complexity and power consumption. Work is underway in the Optical Internetworking Forum (OIF) to specify a host chip to optical module electrical interface, called CEI 28G-VSR. Electrical connector suppliers, physical layer IC suppliers, host system vendors, and module vendors are working together in the OIF to confirm application requirements and specify a 28G channel module and electrical interface characteristics.

Even with all of these developments, it still appears that 100GBase-LR4 power dissipation will be on the order of 10 W, which is still too high to fit into the existing QSFP+ form factor power envelope. To reduce module form factor size, consideration of a next-generation CFP module, “CFP2”, is underway that would be compactly sized for sub-10-W power dissipation and support a narrow 4x28G electrical interface.

With the above-noted technology advances, the next-generation 40GBase-LR4 QSFP+ conceptually could be realized as shown in figure 3 below.

A future 100Gbase-LR4 CFP2 would look similar architecturally, with operation at 4x28G and inclusion of a quad 28G CDR electrical interface. The CFP2 module dimension specifications are an open point of study at this time, but past design experience suggests the CFP2 may look mechanically similar to the existing X2 form factor.

>> Some More Thought

First-generation 40GbE/100GbE CFP optical transceivers are now completing customer qualification and shipping in production. Key design targets for next-generation optical transceivers are: significant reduction of module power dissipation and form factor size.

Critical technologies for tackling these design targets include 4x10G and 4x28G hybrid integrated TOSAs/ROSAs and process improvements in 28G gearbox and CDR ICs. There also may be consideration of uncooled CWDM 28G laser technology for realizing 100GbE optical transceivers in a QSFP+ like form factor for short singlemode fiber (<2km) applications.

>> What is 3M’s 6851-E Hot Melt Polishing Machine for?

3M’s 6851-E Duplex Polishing machine is intended to polish two (2x) 3M Hot Melt connectors at a time for LAN applications. This duplex polishing machine virtually eliminates the need to change the holder if you are switching from one connector type to another.

6851-E duplex hot melt connector polishing machine can be used with 3M’s Hot Melt ST, SC, or FC connectors.

>> General Instruction

The 3M™ Duplex Polishing Machine 6851-E with the universal dual holder is a compact, portable machine designed to polish 3M™ Hot Melt Connectors for LAN applications. The universal dual holder allows for polishing of two ST, SC or FC PC connectors at once (see Figure 1).

The polishing machine package includes the following items:

• 3M 6851-E Polishing machine
• Universal 120/230V AC to 12V DC converter
• 120V plug changeover
• 3M™ Polishing Film 6192 B
• Instruction Manual

>> Installation

1. Plug power converter 12V DC output jack to input connector on the polishing machine. Put power converter into your outlet. The red LED will switch on (see Figure 2).

Note: For 120V use, remove the attached 220V plug and replace it with the 120V version (see Figure 3). The 120 V plug will ‘click’ into place when seated.

2. Put the cable holder straight and move the weight in the direction of the cable holder (see Figure 4).

3. One or two drops of isopropyl alcohol on the rubber disk will help secure the polishing film to the rubber polishing disk. Place a fresh piece of 3M™ Polishing Film 6192 B, shiny side down, onto the rubber polishing disk (see Figure 5).

Note: Carefully follow health and safety information on container label or Material Safety Data Sheet for isopropyl alcohol.

4. Assemble fiber optic connector onto fiber per standard procedure. Score and remove glass fiber extending out of hot melt adhesive. Lift the polishing arm and install connectors into the universal dual holder. Fix both connectors by moving the lever approximately 15° clockwise (see Figure 6).

Note: Both connectors must be placed into the universal dual holder so that the ferrules protrude evenly. Bottom ferrule/collars to ensure this. Do not mix connector types.

Note: The dual holder cannot polish single connectors one-at-a-time. Apex offset drift may result in poor optical contact of fibers.

Gently lower the universal dual holder onto the abrasive film. Squirt a few drops of isopropyl alcohol between the slots of the holder to start the polishing process smoothly.

5. Set timer for 30 seconds and press the red I/O start button (see Figure 7).

6. When machine stops, lift connector holder off the abrasive film and place a new fresh piece of 6192 B polishing film onto the rubber polishing disk. Lower the connector holder onto the abrasive film and gently move the weight toward the connector holder (see Figure 8). Start I/O button again. When machine stops, remove weight first. Remove connectors from connector holder. Clean ferrules with isopropyl alcohol, inspect, and place protective caps on the connectors.

Fiber optic systems have always have the problem of ruggedness. Although bare fibers are protected very well within a fiber cable, when you have to mate two fiber ends together with fiber connectors, the exposed fiber endfaces are easily contaminated and susceptible to mechanical damage too.

In order to solve this problem, fiber connector manufacturers have designed dozen of different types of fiber connectors. Generally, there are two main types:

1. Physical contact connectors (PC)
2. Expanded Beam Connectors

So let’s take a look at both connector types and compare the differences.

1. What is Physical Contact (PC) Fiber Connector?

Optical beam exits the fiber core in a cone-shaped manner, like shown in the following picture.

So in order to capture all the light emitted, the receiving fiber must be in close proximity with the emitting fiber. Just like its name, physical contact connectors have a physical contact (PC) between two fibers. Each fiber is bonded inside a ferrule, and the two ferrules are mated within a ceramic sleeve to assure radial alignment between them for minimum signal loss.

Physical contact connectors use spring force to make sure the two fibers are in physical contact. This is shown in the following picture.

Physical contact connectors include LC, ST, FC, SC, and MT connectors. They provide excellent optical performance but their applications are mostly for indoor, controlled environment.

LC Connector

SC Connector

2. What is Expanded Beam Connector?

Expanded beam connectors have a optical lens that can expand and collimate the light beam emitted from the launch fiber. This is shown in the following picture.

As shown in the above picture, the lens on the emitting fiber side expands and collimates the light beam, and the lens on the receiving fiber side focuses the light beam into the receiving fiber.

Typically the lenses are a 3mm ball lens. In each connector, the fiber ferrule is mounted in a precise position relative to the lens inside a rugged housing. The inside structure of a typical expanded beam connector is shown in the two pictures below..

Expanded beam connectors are usually designed genderless, which means each connector has a pin and a hole, and when they mate, one connector’s pin mates with the other’s hole. In this case, they don’t need a mating sleeve in between any more, and rather connect to each other directly. This is shown in the following picture.

For a 62.5/125um multimode fiber, a 3mm lens can expand the light beam’s cross section area by a factor of 200 times. For a 9/125um single mode fiber, a 3mm lens can expand the cross section area by 900 times.

3. Advantages of Expanded Beam Connectors

Expanded beam connectors have no mechanical physical contact between two mated connectors. Thus they don’t need a fragile mating sleeve in between as in physical contact (PC) connectors. This makes EB connectors especially good fit for rugged environment applications such as industrial, aviation, medical, and military applications.

The advantages of expanded beam connectors include:

• No contact between optical elements of the mated connectors

• Less sensitivity to soil and contamination particles

• Connector misalignment and vibration have little effect on the signal loss

• Consistent performance during repeated matings

The key aspect of the expanded beam solution is the consistent, repeatable insertion loss performance over mating cycles and environmental conditions. While the initial insertion loss may be fractionally higher than the physical contact solutions, over time this parameter remains constant versus the increase in IR seen with the physical contact.

In addition, the expanded beam connectors can easily be cleaned in the field with slightly more than a bottle of water and a shirt sleeve as shown below.