Telecom Equipment Manufacturer Test engineers must often include a wavelength characterization procedure to qualify a product under test. Such test can be performed by different equipment ranging from a tunable filter with an adjunct power meter, an Optical Spectrum Analyzer (OSA) or a Wavemeter. An OSA is indeed an assembly of a Diffraction Grating based tunable filter section, a chopping motor and slits, then a power detector (see our article for more in depth info on OSAs). OSAs being optimized for the specific application of wavelength versus power measurement at a specific resolution, there is not much value for a test engineer to build his own tunable filter based measurement system and write all the software required. The other test product common on the market is the Wavelength meter. All Telecom Wavelength Meters found on the market are based on a Michelson interferometer. Although both the OSA and the Wavemeter will deliver wavelength and power readings for a given WDM or other light signal, the attributes of both products must be considered and the selection must be made based on the measurement specifications needed.
The main differences between the OSA and the Wavemeter are the power reading dynamic range, the wavelength selectivity, the wavelength range, the acquisition speed, the feature richness and the price. In most cases, a Wavemeter are cheaper than OSAs. The reason is quite simple: the construction of a Michelson based Wavemeter is not as complex as the construction of an OSA monochromator (the filtering part). But some Multi-Wavelength Meters (MWM) are now priced higher than OSAs; mostly because of their feature richness. When involved in determining both a wavelength and its associated power level, the OSA is the best general purpose optical measurement tool you can buy. OSAs cover a very large wavelength range sometimes from 350 to 1700nm making them a great choice for R&D groups. Throughout that range, the reading accuracy will vary as the internal error correction algorithms to compensate for linearity issues, wavelength dependencies and gain blocs. In the case of a Wavemeter, the interferometer is a single mechanical sweeping mechanism that travels through the entire span whatever the range you query; only the values displayed will change. There is only one detector. Most Wavemeter have a InGaAs detector well suited to cover the traditional O,S,C and L bands used in Telecom. Multi Wavelength Meters behave much like an OSA and can display many peeks of wavelength in one sweep. Such Wavemeters still operate on Michelson’s interferometry but include Fourier transforms algorithms and much more signal processing hence their higher price.
Lets talk specs:
The best Wavelength Meters (WM) offer wavelength measurement accuracy of ±0.2 parts per million (ppm) which, at 1550 nano-meter (nm), translates to ±0.3 pico-meter (pm). The Ando AQ6317B OSA offers ±20pm at 1550 and the Agilent 86142B is at ±10pm. Lets be clear; OSAs can guaranty such level of accuracy after a calibration using their internal calibration source (if equipped) and only using a high resolution mode. Wavelength Meters will deliver their accuracy across all wavelength range.
Where the OSA shines compared to the WM is with dynamic range. A Burleigh WM will typically read a signal down to -35dBm then only see noise. An Agilent OSA will read down to -90dBm. In general, WM and MWM saturate if the incoming signal is above +10dBm. Most of the OSA can handle powers up to +23dBm. In many telecom applications, if the source goes through an amplifier before hitting the network (or the test station) and although the selection of a WM might be best to read wavelengths accurately, the addition of an attenuator, a splitter and power meter before the WM will increase the station cost largely and the OSA might then become the best choice.
Although the WM has better capability to resolve and display on a wavelength or many wavelengths in the case of a MWM, it has a limitation that cannot be overlooked. This limitation comes with the spec named Selectivity. Typically, if 2 peaks are closer than 50GHz with power levels differences of 25dB or more, the WM might not be able to resolve the difference between those 2 peaks correctly and treat them as one “blob of light”. The measurement will then be flawed if your use of a WM was to look at each individual wavelength and not sum them as one.
Neither of the OSA or the MWM offers accurate power reading in the C band. Specs will vary from a typical accuracy of 0.2 to 0.5dB. The OSA could be considered more accurate on power measurement as it is less sensitive to Polarization.
Then which one should I choose?:
We are really in the heart of test engineering here where the measurement requirements rule on which test product to select. The first question to answer is if the data to be acquired is a single wavelength or multiple. If many, the choice becomes limited to either a Multi Wavelength Meter or an OSA.
Then, it becomes a question of application. Are you testing systems or characterizing components? If working on components, what aspect of the said component is critical? If you are working on a power equalizer dedicated to ensure that all light channels are transmitted at equal power then the power reading accuracy might be more important than the wavelength measurement accuracy. In that case, an OSA seems more suitable. If many wavelengths are to be found and displayed, what are the criteria’s for a pass or fail test? If you need to display and qualify that two wavelength peeks are in line with the DWDM ITU grid then the MWM will give you the most accurate wavelength measurement. Instead, if you are qualifying incoming wavelengths in adjacent channels at very different power levels then you might face the wavelength selectivity issue that we exemplify in the “what is wavelength selectivity?” video that you can view here then the OSA will be your safe choice.
When working on a Test Station with only one wavelength to be measured at a time, with power between +10dBm and -40dBm, the decision becomes an easy one: a cheaper and more precise Wavemeter. In many instances, technology as old as the Agilent/HP 86120B will do just fine, at the fraction of the price of today’s newest units.
If your test strategy involves the use of pre-programmed test features such as EDFA noise measurement then the choice might be limited to the OSA. Before selecting a model from another, you must then verify which features are offered with each Optical Spectrum Analyzer model.
If one of your measurements involves the reading of a broad spectrum light source then again, the OSA is the only choice being the only unit designed to filter and display a large spectra of light at a time.
If you want to read in more depth about Michelson interferometer, I suggest http://en.wikipedia.org/wiki/Michelson_interferometer