Michael P. McArthur
In the quest for more bandwidth, everyone wants tighter spacing on WDM channels. Whereas 100 GHz between channels was the spacing to acquire last year, 50 GHz is the spacing this year. With the introduction of interleaver technology, there is a push for passive components to work at 25-GHz spacing, and manufacturers are already looking at 12.5-GHz components. This means a need for improved testing techniques to speed up manufacturing.
New technology is being employed to speed up measurements of spectral attenuation, return loss, and polarization-dependent loss on multiple-port devices, such as multiplexers, demultiplexers, fiber Bragg gratings, filters, and isolators. Utilizing a high-performance tunable laser and multiple-power sensors, these new testing systems synchronously and simultaneously measure optical power across a range of wavelengths, on all optical outputs. These sweep systems are more rapid than the traditional method that measures each output individually with an optical spectrum analyzer (OSA).
Component manufacturers need to connect the testing environment to the corporate local area network (LAN) system for downloading the vital statistics of a particular component to their central tracking database. Some systems provide this, as well as the individual component test reports needed for shipping the component to the network system manufacturers. Component manufacturers also need the ability to connect bare fibers to the sensor units and adjust the wavelength range to look at the specific wavelengths of interest. A good sweep system should cover both the C band (1535 to 1565 nm) and the L band (1570 to 1630 nm).
One of the bigger concerns in testing wavelength-dependent devices like multiplexers and interleavers is wavelength accuracy and linearity. As the spacing on channels gets closer, there is a need to decrease the margin of error of the test equipment. New sweep systems have wavelength accuracies of ±3 pm. This is an order of magnitude better accuracy then the current generation of OSAs, which have accuracies of 20 to 50 pm.
Spectral attenuation, or wavelength-specific insertion loss, is another attribute that is very important for component manufacturers to measure. When measuring spectral attenuation on a notch filter, the instrument needs to have a dynamic range of greater than 60 dB. Since the slope of the ideal notch filter goes from full transmission to complete filtering of the light very quickly, the sweep system requires fast detection of and response to changes in optical-power intensity. Because sweep systems take measurements directly from an optical-power sensor, there is no Gaussian effect as seen on an OSA, and the sweep systems can detect very rapid changes in optical power, on the order of 60 dB/pm.
To increase the accuracy of the spectral-attenuation measurement, optical-power sensors should be highly accurate and the tunable laser should have very tight control of output power. Sweep systems maintain accuracy that closely follows that of an OSA (see Fig. 1).
When performing filter measurements, component manufacturers measure parameters such as center wavelength, spectral width at a particular threshold, ripple, and crosstalk. When it comes to these measurements, component manufacturers must state the parameters according to customers` requirements. The center wavelength is commonly calculated by averaging the wavelength on both sides of a 3-dB peak threshold. Others define the center wavelength as that corresponding to the International Telecommunication Union (ITU) grid of standard center wavelengths intended for WDM signals. Some component manufacturers specify the shape of the filter by taking several spectral-width measurements at differing thresholds. Filters have ripple on the peak caused by interference that occurs inside of the filtering mechanism.
Since no component is perfect, there will always be crosstalk on multichannel devices. Sweep systems measure these parameters more accurately because the sides of the filter lack the Gaussian effect. Since different network manufacturers require a different set of measured parameters, the reports for the devices differ from batch to batch. Reporting features should be flexible to enable the component manufacturers to produce reports with multiple parameters.
Polarization dependent loss
As the acceptable margin of error for components decreases, polarization dependant loss (PDL) becomes a major concern for building passive components. A network component can attenuate light depending on its polarization state. In the high-loss region of a passive component, PDL is of major concern. Sweep systems account for PDL by using a polarization controller and a Mueller matrix calculation, which can then quickly and easily be added to the measurements taken. The Mueller matrix, M, is a 4×4 real matrix describing the transmission and polarization characteristics of light through an optical device under test (DUT). Using the Stokes polarization vector, S = (S0, S1, S2, S3), where S0 represents total intensity, S1 represents the amount of linear horizontal or vertical polarization, S2 represents the amount of linear +45° or -45° polarization, and S3 represents the amount of right-hand or left-hand circular polarization, the following matrix equation can be written:
With a sweep system that includes a polarization controller, the wavelength calibration of all channels on the sensor unit is taken and then used as a reference to see any fluctuations in power that occur due to the light source or PDL in the polarization controller (see Fig. 2). The above calculations are performed automatically within the sensor unit added to the reporting function of a sweep system.
Optical return loss
Until recently, sweep systems could not handle taking optical-return-loss (ORL) measurements. A DUT may contain multiple reflecting surfaces, resulting in an interference pattern and making the total return loss for a DUT wavelength dependant. In this situation, a narrow-band source, such as a distributed feedback (DFB) or external cavity laser (ECL), will produce a large or small reflectivity, depending on its wavelength. A broadband source, such as an LED or even a Fabry-Perot laser, will return an average reflectivity.
Traditionally, ORL measurements are taken via broadband sources, which average the ORL, ignoring the wavelength dependence of reflectivity. However, for best accuracy, ORL should be measured with a light source with spectral and coherence properties close to the actual sources that will be used on the DUT. For WDM systems, this means that multiple wavelengths need to be measured with a narrow-banded source. If the laser of the sweep system, which has similar characteristics to the WDM sources, is tuned across all of the wavelengths of interest, a graph of reflectivity vs. wavelength can be produced.
With the increase in bandwidth requirements, this return-loss information is much more important to component manufacturers. The ability to take ORL measurement with the same integrated package as the spectral attenuation and PDL measurements can speed up testing and increase production of the much needed passive components. It can also allow for integration of the ORL test results into the reporting function.
One problem with a sweep system is that it takes time to set up and learn how to use it. Systems with a user interface similar to instruments such as an OSA have a clear advantage over component test equipment in which ease of use is more intuitive. There are sweep systems versatile enough to run from the front panel as well as controlled by a computer with a graphic user interface. Easy set-up and calibration features requiring as little user input as possible will save the user time.
1. S. Schmidt and C. Hentschel, PDL measurements using the HP8169A polarization controller. Böblingen, Germany: Hewlett Packard, PN 5964-9937E, 1997.
Michael P. McArthur is a technical support engineer at Ando Corporation, 4504 Longacres Court, Arlington, TX 76016. He can be reached at email@example.com
FIGURE 1: One channel output trace of a WDM multiplexer shows that sweep systems maintain an accuracy of ± 0.2 dB across the entire power range and better than ± 0.1 dB in the lower loss regions compared to an OSA.
FIGURE 2: The signal for PDL measurement is sent through the polarization controller and then split off to the device under test (DUT) and the sensor unit.