by Thomas Jensen, Oliver Funke, and Ulrich Wagemann
Swept homodyne interferometry can be combined with a power meter to measure loss, polarization-dependent loss, group delay, and differential group delay, offering a precise and thorough test of 40-Gbit/s passive components.
The deployment of 40-Gbit/s networks presents a challenge to network equipment and component manufacturers—and therefore to test and measurement systems. The challenges result from new techniques such as forward error correction, dispersion compensation, and digital post-processing, as well as the use of specialty fiber, which are all essential in 40-Gbit/s networks, because they allow additional margin for error-free data transmission.
This evolution of optical networks creates greater demands for testing passive components. Testing passive optical components for loss alone is no longer sufficient—precise characterization and control of the dispersion properties of all devices has become mandatory. The need for this "all-parameter" testing brings new challenges to the process of qualifying optical components for deployment.
At higher data speeds and narrower channel spacing, "run time" properties of the signal become important, in addition to loss. As a rule of thumb, a tenth of a bit period is an acceptable delay (or dispersion) for a system. For example, a bit period for 40 Gbit/s is 25 ps, so 2.5 ps is an acceptable system delay. But a system contains many components, so the delay contribution of each individual component must be smaller.
Most passive components route light or drop parts of it. Independent from their dispersion properties, compromises in loss performance of these components are unacceptable. In advanced networks, components can lead to system penalties or even failure by not meeting the specifications of four critical parameters: insertion loss, polarization-dependent loss, group delay, and differential group delay. The ability to test all parameters will be essential in passive optical devices such as fiber gratings, arrayed waveguide gratings (AWGs), and add/drop modules, particularly at 10 and 40 Gbit/s, and for narrow-band components.
The measurements of loss and delay are traditionally treated as separate tasks and addressed by separate solutions, or by solutions optimized for the testing of a single parameter.1 A fundamental shift in test instrumentation is needed to address changing test needs toward an all-parameter analysis. Test and measurement instruments must precisely characterize all relevant device parameters, rather than a limited subset or a single parameter. In addition, any test procedure must be as accurate and simple as possible to shrink test time and cut the cost of the test. Any new test methodology should therefore use a single connection to minimize test uncertainties.
TESTING PASSIVE COMPONENTS
Typical passive devices include fiber Bragg gratings (FBGs), thin-film filters, or AWGs. Because their first purpose is wavelength routing, they require precise specification for loss during the manufacturing process. Such dynamic, high-accuracy, and high-resolution testing is typically addressed by a tunable laser combined with a power meter.2, 3 Narrow-channel devices challenge current dispersion tests because narrow-channel loss characteristics go hand-in-hand with steep dispersion traces.4
To meet these testing requirements for loss measurement, our system setup uses a tunable laser source with low-noise output. For measurement of dispersion properties, the setup uses a technique called swept homodyne interferometry.
In swept homodyne interferometry, a laser source is wavelength-tuned while the arm lengths of the interferometer remain fixed (see Fig. 1). One arm measures the device under test (DUT), while the second is used as reference. The optical signals are combined and a fringe pattern is detected by a diode. The optical setup itself requires no moving parts. The setup can be extended easily to measure the transmission and reflection properties of devices.
This kind of test will result in an interferometric pattern of the kind
which will be observed in the detector plane, where ELO specifies the local oscillator, Edut is the field amplitude that passed the DUT, and n is the light frequency. The phase information of the DUT, φ, can be extracted by mathematical means and translated into group delay.
Swept homodyne interferometry gains dispersion information with high spectral resolution because the phase information of the DUT is retrieved from a single wavelength, not from modulated sidebands. We have used the group delay characteristics of an acetylene-gas-cell peak to check the capabilities of this method. Gas-cell peaks are based on molecular absorption lines and are useful for wavelength calibration because of their limited spectral width. The measured data from the acetylene gas peak using swept homodyne interferometry show very close agreement to a theoretical trace.
Swept homodyne interferometry is not well-suited to measure long devices such as fibers. Rather, it is best applied to wavelength-selective, narrow-band components. Swept homodyne interferometry gains its high sensitivity from the fact that it compares optical phases, not electrical phases. This technique requires care when setting up an experiment, but accuracy of less than 100 fs is easily achieved for both group delay and differential group delay.
By adding couplers to the swept homodyne interferometry setup, parts of the signal can be redirected to power meters to measure loss. A polarization controller determines polarization-dependent loss using the Mueller-Stokes method. Insertion loss and return loss are then calculated as an average over the polarization variance.
Because light signals travel as a group at many frequencies, group delay, τ, is the parameter of interest. Group delay is closely connected to phase delay, measuring how the phase changes with wavelength on a very small wavelength scale. The group delay unit of measurement is time and is denoted in picoseconds. Group delay can be calculated from the phase delay, using the formula
In some cases, chromatic dispersion characterizes a component. Chromatic dispersion describes the slope of group delay vs. wavelength and is measured in picoseconds per nanometer.
Group delay is a measure of how much a light pulse is "stretched" when passing through a component. Differential group delay is a measure of the polarization dependency of group delay (see Fig. 2). There is a strong correlation between average loss and polarization-dependent loss.
In our test setup, a polarization-diversifying receiver records two traces of group delay leading to eigenstates as a function of wavelength, TE and TM. These traces are averaged to determine group delay; subtracting them results in differential group delay. This procedure ensures that the recorded group delay trace is free of polarization effects that can result, for example, from a change in the input polarization state between two measurements.
For the phase measurement, the tunable laser sweep speed is set to 40 nm/sec. Measurement time of a single sweep including numerical analysis is in the range of a few seconds. Averaging of individual traces can improve the signal-to-noise ratio.
As representative DUTs, we chose a thin-film filter and an FBG. Thin-film filters in high-speed optical networks minimize the ripple of the spectral group delay response over the passband needed by the modulated signal (see Fig. 3).
Fiber Bragg gratings are a central building block of modern optical networks and are often built into add/drop modules or demultiplexers. In a nonapodized FBG measured in reflection mode, the grating period exhibits a slight polarization dependency in reflection, causing a shift of a few picometers between the loss spectra for the principal states (see Fig. 4). This birefringent behavior shows up as polarization-dependent loss with sharp double-dent-like structures.
Birefringence is a material property that is inherent in many FBGs and AWGs. As can be expected, birefringence also results in a shift in group delay spectra for two principal states of polarization. Group delay peaks are accompanied by similar sharp double-dent differential group delay structures as well. This double dent exhibits a centered minimum where the group delay value is identical for both principal states of polarization (see Fig. 5).
As we have shown, swept homodyne interferometry is capable of resolving the finest structures of group delay and differential group delay traces or of measuring flat traces with very little residual noise floor. In addition to its capability to measure with great accuracy and over a high dynamic range, the method provides advantages through its swept wavelength principle, which leads to shorter measurement times.
1. D. Derickson (ed.), Fiberoptic Test and Measurement (Prentice Hall, 1998).
2. E. Müller et al., Proc. OFC 2000, paper WB 2.
3. E. Leckel et al., Proc. OFC 2000, paper WB 4.
4. Guy Sauvé, WDM Solutions, 43 (September 2001).
Thomas Jensen is project engineer, R&D; Oliver Funke is project leader, R&D; and Ulrich Wagemann is product manager, tunable laser sources, at Agilent Technologies, Deutschland GmbH, Herrenberger Strasse 130, 71034 Böblingen, Germany. Ulrich Wagemann can be reached at firstname.lastname@example.org.