How to choose the right dwdm test instrument
Optical spectrum analyzers and multiwavelength meters have different strengths that must be evaluated when choosing a dwdm test system.
Lynn Mason, exfo Electro-Optical Engineering Inc.
Dense wavelength-division multiplexing (dwdm) technology has been the hottest topic in the fiber-optics industry for the past few months. Small wonder, since dwdm systems will offer unprecedented bandwidth that will inevitably increase information transmission capacity. However, whether dwdm is installed on new or existing networks, this new technology presents test and measurement challenges that must be overcome.
The basic idea of this powerful technology is quite simple. As opposed to traditional systems, in which a single fiber carries a single-wavelength signal, the optical fiber in a dwdm system carries several signals at the same time. Each signal must operate at its own wavelength. Therefore, dwdm systems multiply the information-carrying capacity of a fiber by the number of wavelengths it carries.
The operating window of the optical amplifiers used in dwdm networks is quite narrow, typically from 1525 to 1570 nm. Being restricted to such a narrow spectral region leads to the use of very closely spaced carriers. To get the maximum capacity out of a network, designers want to exploit as many carriers as possible. As a result, the spacing between carriers must be as small as possible. The International Telecommunication Union (itu) was expected to release a standardization document last month that approves a spacing of 100 GHz (about 0.8 nm) or integer multiples thereof. Some wdm system providers have already announced systems with even smaller spacing (50 GHz).
Due to a dwdm network`s narrow spectral region, there is an ever-increasing need for a new type of test equipment that is acutely sensitive to wavelength: optical spectrum analyzers (osas) and multiwavelength meters (mwms). However, because dwdm technology is rapidly progressing, osas and mwms represent "uncharted waters" for many members of the industry. Although these instruments are somewhat similar, their core technologies, characteristics, and main applications create a niche for each one. It is therefore important to understand the differences between the two units before you decide which to use.
Optical spectrum analyzers
An osa is a piece of equipment that can divide a light signal into its constituent wavelengths and measure the power of each of these wavelengths. Measurement results are displayed graphically, with wavelength on the horizontal axis and power on the vertical axis. To obtain these results, two technologies are available: diffraction gratings and tunable filters. Because of the high cost and limited capability of tunable filters, diffraction gratings are the preferred solution.
By definition, a diffraction grating is a dispersive element that, thanks to the multitude of fine parallel lines on its surface, breaks up or "diffracts" a light signal into its optical spectrum. Once the signal is diffracted, it is possible to measure the power of any given wavelength. This is done by aligning a detector with a specific wavelength. To measure another wavelength, the detector must be realigned with that wavelength, and so forth.
A typical osa design will include either a rotatable grating or a mobile detector. Both features will change the spectrum-detector alignment in a continuous fashion. This "sweep" is what allows the osa to measure the power of all wavelengths of a given light signal and generate its characteristic graphs. Figure 1 shows the simplest osa setup with a fixed detector: the single-pass monochromator.
No matter what setup is used for an osa, there are always three parameters that combine to determine the unit`s wavelength resolution: the density of the lines on the diffraction grating`s surface, distance from the detector, and configuration of the optical circuit. Together, these parameters influence how much of the spectrum reaches the detector. Consequently, the parameters also determine the wavelength resolution to a few thousandths of a nanometer.
The parameters also determine the optical rejection ratio (orr), also known as the optical dynamic range. Because the spectrum is not perfect, each individual wavelength overlaps a little onto adjacent wavelengths. Therefore, a very strong signal can overlap its weak neighbors and smother them. The orr describes how robust the adjacent signals have to be in order to be distinguishable from the strong signal; the longer the distance, the weaker the signal can be. The dynamic range is therefore a crucial specification when you test a dwdm system. Since high-power channels are closely packed, the orr will determine how easily the osa can distinguish between channels and measure inter-channel noise. A good inter-channel noise ensures an accurate snr measurement.
Single- versus double-pass osas
A single-pass design (as in Fig. 1) can only improve the orr to a limited degree. However, the orr can be improved by multiplying the number of times that the input signal is diffracted. This is done by reflecting a section of the diffracted spectrum onto a second diffraction grating. Such an osa, with a double-pass design, further separates the signal into its components and makes a single wavelength easier to isolate for accurate power measurement. A double-pass osa therefore has better wavelength accuracy and resolution as well as a greater orr (by about 10 dB) than a single-pass osa.
The advantages of the double-pass osa, although impressive, are somewhat overshadowed by its size and fragility. Because the design includes more optical components, the double-pass osa is larger than a single-pass osa. In fact, the double-pass osa can be as large as a desktop computer, making it difficult for technicians to use the unit in the field. In addition, most double-pass designs require extremely precise alignment of the optical parts. Often, the alignment of these components can be deviated simply by moving the unit from one place to the next. As a result, few double-pass osas are considered rugged and durable due to their cumbersome size and frail design.
Wavelength power measurements with a detector have been mastered by the industry; since the osa`s power readings are based on this simple detector technology, the accuracy of an osa is very high. The display range of the osa--its ability to measure power levels from as high as +20 dB to as low as -70 dB--also reflects its power meter ancestry.
Multiwavelength meters
The mwm generates results that are similar but complementary to those of the osa. As with the osa, the results are displayed as power versus wavelength. mwm measurement results are most often displayed in tabular form, but graphical output is optional with some equipment. However, despite the similarity to the osa, the technology used to achieve results in the mwm is fundamentally different: the osa uses diffraction gratings, whereas the mwm uses interferometry.
With interferometry, the mwm segments the light signal into two equal parts and monitors the interference patterns that are created between the two signals (see Fig. 2). These interference patterns cannot be directly interpreted, so they must undergo signal treatment before becoming the recognizable results shown in Fig. 3. The mathematical tool used to treat the interference patterns is called a fast Fourier transform (fft). Although the mathematics behind it are quite complex, the fft basically changes an interference pattern from a function of time (i.e., variable optical delay in one arm of the interferometer) to a list of wavelengths (channels) that make up the optical signal being measured.
Interferometry is a technique that inherently gives accurate measurements of power and wavelength peaks relative to each other. There is nevertheless no guarantee of absolute wavelength or power accuracy in interferometric measurements. Furthermore, if fft calculations are not a source of error in theory, they may be in practice. For example, in order to perform the fft, the mwm takes a number of samples on the interference pattern. If insufficient samples are taken, certain details of the interference pattern will be lost. This undersampling will generate errors, such as ghost peaks, in the results table. Current number processing techniques limit the sampling rate to 12 kbits/sec. A second sampling limitation occurs when signals are transmitted at rates similar to the sampling rate. In order to measure these signals, the average of several measurements will have to be calculated in order to eliminate the effects of signal modulation on the power at a given wavelength.
To minimize result errors and increase absolute wavelength accuracy, the mwm can contain a built-in helium-neon (HeNe) laser. HeNe lasers have a very important characteristic: No matter what happens, they emit one very specific wavelength with extreme accuracy. By sending the light of a HeNe laser through the interferometer at the same time as the signal being measured, the mwm can have a very accurate reference for all other measurements. Since the HeNe wavelength is invariable and its accuracy is known, all measured wavelengths also become accurate, as long as the small effects of air pressure have been taken into consideration. The HeNe laser therefore gives the mwm exceptional wavelength resolution and absolute accuracy, particularly when compared with an osa that does not have such a reference.
The interferometer and fft generate relative power results; for example, a result could read that 50% of the light signal is concentrated at 1553.455 nm. In order to assign an absolute value of power to each wavelength, the mwm uses a detector--separate from the interferometer--to measure the entire system power. Using the total measured power, the mwm assigns absolute power values at each wavelength according to a weighted average. This average weighting, combined with the fact that noise levels create very little, if any, coherent interference patterns, means that noise levels are very difficult, if not impossible, to measure. Consequently, a mwm yields a weaker orr than an osa--rarely better than 20 to 25 dB.
Making the right choice
To date, there are no industry standards for dwdm testing because the dwdm segment of the industry is so new. Still, there are a number of recognized test procedures from which system manufacturers, installers, and maintainers can choose.
Channel wavelength and power tests are used to verify each channel to ensure that the wavelength and power are within the allowable limits specified by the system designer. Drift measurements (of either power or wavelength) look at these same channel characteristics over time to verify their stability and/or to predict their future behavior.
System tests include signal-to-noise ratio (snr) and crosstalk tests. Both of these tests help to gauge the quality of the signal carried in a channel. Total system power may also be measured during system tests since it is directly linked to adverse phenomena, such as Brillouin and stimulated Raman scattering as well as four-wave mixing.
The osa and mwm indeed have different strengths and weaknesses that make each unit better suited for certain tests. The osa has better absolute power accuracy, which is particularly useful for such tests as crosstalk measurements, where power levels must be compared. The unit`s superior dynamic range makes measuring the snr easy because the osa is capable of detecting noise levels between the powerful, tightly spaced channels. The osa does have one drawback: Its wavelength resolution, although acceptable for most applications, is insufficient for measuring small drifts in wavelength over time.
Yet the strengths of the mwm cannot be ignored. The mwm`s better wavelength accuracy and resolution are ideal for pinpointing exact channel centers. Also, since the absolute wavelength accuracy is very good, several measurements can be taken over time in order to measure the drift in channel wavelengths. However, the mwm lacks the absolute power accuracy required to perform channel or component gain measurements within a system or over time (as is the case with crosstalk). The mwm has poor orr, making snr very difficult to determine using this device.
The mwm can also be sensitive to certain input signals, such as Synchronous Optical Network (sonet) transmissions. As discussed earlier, the sampling rate of the mwm is very close to the transmission rate for sonet frame information. The mwm must be able to find the average of several readings in order to minimize errors due to the similarity between these two rates.
The table summarizes the suitability of each unit to the most common dwdm system tests. Another touchstone is the location of the tests to be performed. If tests are to be conducted outside the laboratory during system installation and maintenance, portability and durability of the equipment is paramount. Unfortunately, very few osa models are designed to meet these requirements. Most become quickly decalibrated after being moved from one place to another. Physically, mwm units are also somewhat delicate pieces of equipment; however, they lend themselves to field use more than most osa models. As with the osa, special attention should still be given to durability before selecting an mwm.
Choosing between an osa and an mwm is no small feat. The fiber-optics community has yet to determine where the market is heading. The decision can seem all the more difficult when comparing products of different manufacturers, which also have certain advantages and drawbacks. Keep in mind, however, that neither the osa nor the mwm is ideal in all measurement situations and that it may be necessary either to make some compromises or to use both units. The decision can be made much easier if you take into account your immediate and potential needs. In addition, by surveying industry trends, you will be able to readily identify which tests weigh more heavily with the industry. u
Lynn Mason is outside-plant product manager at exfo Electro-Optical Engineering Inc., Vanier, QC, Canada.
Fig. 3. Once the interference patterns detected by the multiwavelength meter have been treated via a fast Fourier transform, they can be displayed as shown above for interpretation.