Optical switch testing spans the ocean
Engineers at Polatis test optical switch modules and systems on both sides of the Atlantic.
Martin Rowe, Senior Technical Editor -- Test & Measurement World, 8/1/2006
 |
|
READ OTHER AUGUST ARTICLES: Beam steering |
BILLERICA, MA—Optical switches are fast becoming a component of choice for telecom engineers who develop reconfigurable networks. With optical switching, you don't have to convert optical signals to electrical signals and back just to configure a network. Skipping the conversion reduces network costs and maintains data rates.
Like any optical device or system, optical switches must be tested for various forms of power loss. Engineers at Polatis (www.polatis.com), which has a location in the UK as well as one in Massachusetts, measure parameters such as optical power insertion loss (IL), return loss (RL), polarization-dependent loss (PDL), and wavelength-dependent loss (WDL), all over a range of temperatures. The engineers make these measurements on identical systems in both locations, thus providing for consistency across the Atlantic.
A Polatis switch system consists of a switch-fabric core manufactured in Cambridge, UK. Inside the core, piezoelectric devices steer beams of light from one port to another. (See "Beam steering," for a description of how the Polatis switch works.) Figure 1 shows a map of how the components come together to form a complete switch system. Testing takes place at the module level and at the system level.
 |
| Figure 1. An optical switch system uses a core switch module to switch signals to optical ports. |
Following assembly, a core switch module requires extensive testing, even in production. After testing, a core module may ship to the Billerica, MA, facility for integration into a complete switch system or it may stay in Cambridge. The Polatis staff assembles a complete system by inserting a switch module into a case, adding a power supply, and in some instances, adding optical power meters into the system to provide power detection and optical attenuation.
To learn about how Polatis tests both its core switch modules and complete switch systems, I visited the Billerica facility where I met Tim Glenn, senior lead engineer, and Aaron Bent, VP of marketing. To learn about the capabilities in Cambridge, I spoke with production ATE manager Mike Grant by phone.
Testing the coreAt the Cambridge facility, Grant oversees testing of the optical core switch module, which consists of the piezoelectric actuators, optical collimators, and control electronics. A core switch module typically has 16x16 optical ports (256 possible crosspoints), or 32x32 optical ports (1024 possible crosspoints).
Optical testing begins with an automated inspection of the piezoelectric subassemblies, which move the collimators and steer beams of light. An inspection system consisting of a Pulnix CCD camera and a National Instruments PCI frame-grabber card measures the lateral displacement of the light beam when a voltage excites the piezoelectric device. LabView software captures and stores the characteristics of each actuator subassembly.
Following inspection, the beam-steering assemblies, control electronics, and collimators are assembled into a core switch module. The completed module goes through a series of calibrations to ensure that it properly steers light to the correct ports.
 |
| Figure 2. An optical power meter measures the switch module’s output power, while a peak-power-search algorithm calibrates the switch’s optical paths. |
Software automates the testing. A peak-power-search algorithm finds the position of maximum optical power. The software then records the position of the collimator as measured by an analog-to-digital converter (ADC) attached to the position-sensing electronics. The sense value is stored in the controller's memory, pending adjustments from thermal cycling.
Once operational, a core switch module moves to a thermal calibration station. A Sharetree temperature chamber that lets engineers calibrate the actuators from 0°C to 60°C. The engineers use the chamber to rerun the peak-power-search algorithm at the temperature extremes. By locating the beam position and measuring the output power, the system adds temperature-compensating digital offsets to the control board. Once calibrated, the switch module loses less than 1.0 dB of optical power from the entering light beam.
 |
| Mike Grant is the production ATE manager at Polatis’ Cambridge, UK, facility. |
Repeatability tests take several hours to perform, so the Polatis engineers usually run them overnight. "We run tens of thousands of switching cycles over a range of 0 to 60°C to check the operation of the switch, which generates tens of megabytes of data," explained Grant. Switch repeatability is measured with 100 scans of every input port/output port combination. The test uses an Agilent Technologies or Yokogawa amplified spontaneous emission (ASE) light source operating in the 1500-nm wavelength region to minimize amplitude noise in the measurement system. The test lets engineers verify that a switch module has a repeatability of ±0.05 dB.
The final acceptance test for a switch module involves measuring insertion loss, optical return loss, and crosstalk. In this test, engineers use the three-patchcord technique that complies with ANSI/TIA/EIA-568-B.3-2000 and other standards (Ref. 1). The patchcord is a reference for insertion loss by which Polatis engineers can compare a connectorized switch module's insertion loss.
To make the measurement, engineers first insert the reference patchcord into their test network and measure insertion loss with either a 16-channel or 32-channel Agilent or dBm Optics power meter. They also use an ASE source (an Agilent or a New Focus tunable laser) to drive the patchcord. Then, they substitute the switch module for the patchcord and measure again. Each of the switch outputs connects to a separate channel on the power meter, a setup that permits simultaneous measurement of insertion loss on the illuminated output and of leakage optical power, or crosstalk, on the "dark" outputs.
Polatis engineers also measure polarization-dependent loss (PDL) and wavelength-dependent loss (WDL) in the C-band (1530–1560 nm) and L-band (1565–1625 nm). To make the PDL measurement, engineers use an Agilent polarization controller and dBm Optics component spectrum analyzer. The Agilent 8169A polarization controller is under the control of the dBm Optics component spectrum analyzer for this test, and the system automatically measures loss over a matrix of polarization states. To make the PDL measurement, engineers use the patchcord to measure PDL of the test network, then they replace the patchcord with a switch module. The difference in PDL represents the switch module's PDL.
On to integration
 |
| Tim Glenn is senior lead engineer at Polatis’ Billerica, MA, facility. |
Calibration of the core switch modules in Cambridge ensures that the maximum amount of light reaches the intended output. Sometimes, though, users need to reduce optical power to a controlled level. By deliberately detuning a switch—by misdirecting the light beam—users can control the amount of optical power at an output port. This detuning lets a switch operate as a variable optical attenuator (VOA), an option on Polatis switch systems.
Users can configure the switch system for constant output power or for constant attenuation. To attain a constant output power, the switch system monitors the output power and adjusts attenuation in real time. For constant attenuation, the system measures both incoming and outgoing power, adjusts the amount of attenuation, and then holds the attenuation. Power monitors installed into a switch system make controlled attenuation possible, but these power monitors need calibration. "We need to perform the power-monitor calibration because each optical detector's responsivity differs slightly," said Glenn.
Polatis calls an integrated switch system with power monitors a VOA switch tray (VST). After a VST is assembled in the integration room, Glenn and others test it with a rack of instruments consisting of a dBm Optics component spectrum analyzer, a New Focus tunable laser source, and an Agilent polarization controller. "After integration, we repeat many of the measurements done in Cambridge," said Bent. "We compare RL, PDL, and WDL measurements of the integrated system against those done in Cambridge on the core switch module before we calibrate the optical monitors." Figure 3 shows the test setup in the integration lab.
 |
| Figure 3. Engineers use this system to measure IL, PDL, RL, and WDL in an integrated switch system. |
Following integration and initial testing, a completed VST moves from the integration room to the calibration and final-test room. The room contains two test stations, each using a Thermotron environmental chamber. One station performs calibration of the optical power monitors, and the other performs final test.
 |
| An optical switch under test in a thermal chamber. |
The calibration station uses a rack of instruments that include laser sources and power sensors. Three Agilent tunable laser sources and one Agilent fixed laser source generate the light. The fixed laser source (1550 nm), along with an EDFA and JDSU attenuator for controlling power over a large range, lets engineers perform initial functional test and calibration. The tunable lasers vary the wavelength during a wavelength-dependency calibration. A polarization controller minimizes uncertainties caused by PDL of the test system and polarization-dependent responsivity (PDR) of the integrated power monitors.
An Agilent 8166A lightwave multichannel system populated with 33 optical power sensors measures optical power as it enters and exits a VST under test (Figure 4). Up to 32 power sensors connect to the output ports for output power measurement. The additional power sensor is used as a reference for measuring input power. A standard Polatis switch connects the laser sources to the polarization controller. Test signals then travel to the VST under test.
Light that exits the VST under test travels to the calibration station's power sensors, which serve as references for the power monitors in the VST. Once calibrated, the power monitors achieve accuracy of ±0.25 dB, or ±5% of range. "During a specific calibration measurement," explained Glenn, "we hold all variables such as temperature, power, and wavelength constant except one, namely the one variable whose impact on detector responsivity is being characterized during the measurement. For example, we hold wavelength constant at 1550 nm and temperature at 25°C, while we calibrate over a wide range of input power: –30 dBm to +24 dBm."
 |
| Figure 4. Calibration of power detectors takes place with the switch (VST) in a thermal chamber. |
To perform a calibration, the system software written in LabView compares the readings from the VST's optical detectors to those from the calibration station's optical sensors and records the differences. The system repeats the measurements over the full ranges of temperature, wavelength, and optical power, changing only one variable at a time. From the measurements, the software creates a look-up table and stores the table's values in the VST. During normal operation, the VST applies the table values to its power measurements, which brings them into line with those of the external optical sensors. The optical detectors become NIST-traceable through their calibration with the external power meters.
The optical cables used in the calibration and test systems are critical to producing a switch that meets its specifications for loss and power-monitor accuracy. Losses in the fibers between the test instruments and the VST under test are negligible, but insertion losses from the connectors must be minimal and consistent. Cleaning of the optical connectors and the quality of the connection are critical to proper performance. "If someone else cleans the connector bulkheads and makes the connections," Glenn emphasized, "I run an initial set of measurements. I know from experience how well the system performs. If it doesn't meet my requirements, I'll remake the connections."
Alignment of optical fibers in connectors is also critical. To minimize loss uncertainty, Glenn designed the tester to use multimode fibers on the VST's output ports—other paths use single-mode fibers. Multimode fibers have a 50-µm core diameter, whereas the single-mode fibers used inside the VST have 8-µm core diameters. That difference in diameter essentially eliminates the chance of misalignment, which dramatically improves measurement accuracy. Each multimode cable essentially becomes an extension of the corresponding reference power sensor. In effect, the power sensor's "large-area detector" now resides at the front panel of the switch, where calibration of the internal power monitors is specified.
Following calibration, VSTs move to a final-test station where engineers measure IL, RL, PDL, WDL, and repeatability. They make all of these measurements over the VST's operating temperature range. Repeated testing at the module and system levels assure users that a Polatis switch meets its tight specifications for power loss and repeatability.
| Reference |
|
|
