2005 Articles
O/E Component Testing Pitfalls
Inexpensive testing for expensive components is feasible, if done correctly.
Compared to electronics, optoelectronics (or optics and photonics) technology in telecom is still evolving and at a stage similar to the Optical Vacuum Tube Era. It is not yet up to the optical transistors or the optical IC stage. This is natural since optoelectronics in telecom was actively developed just after 1998. Handling light is much more delicate than electrons.
The transmission speed for optical communications rose from 52 MB/s to 10 GB/s and beyond, while the number of channels sent through a single strand of fiber increased from one or two channels to 240. Optical components evolved slowly due to the complexity in production, testing and reliability issues. Telecom components must guarantee 20 to 30 years of operation.
The primary challenges with high-speed optics are electronics and test. The real challenges are electronics and test. Although we are trying to measure optical parameters like optical power and temporal profile, optical power must be converted to an electronic signal for further analysis. When we go for a higher speed, the time interval of one bit is reduced and the number of photons in the bit is also reduced. That means we are handling a much lower signal level in high-speed optics. When there is a low signal level, electronic noise issues arise.
When testing optoelectronics, it is necessary to determine the true outputs. The true output is the electronic signal, which is converted using a photodiode from the optical signal. Optoelectronics has the highest level of electronics, so testing is critical. To ensure the intended output and optimal performance, a cross-checking routine should be instituted to compare the simulation and test results, or use two different test methods to compare those results. Simulation is performed because it is impossible to test all cases due to lack of time and the constraints of the test environment. All measurement procedures and preparation of the test equipment should be checked for consistency. Engineers experienced in testing these products need to be involved in the analysis. Review the input and output parameters of the design and compare them with the calculated results and the simulation and experiment results.
Designers can take certain steps to ensure a package is tested and working as designed. Based on the functionality in the design, all the input, output and interaction parameters within the package must be identified. Good methods to create and combine the input parameters should be proposed and checked before starting main testing. The design engineer needs to check test methods that will be used. Some tests may engage inevitable and unwanted interaction parameters. They must be recognized and considered by using simulation or other testing.
The designer might find that some high-speed and low-signal level testing can be difficult to measure because the extreme parameters may exceed the limits of the test equipment. For that reason, testability and test options should be considered in the design phase. Statistical variations in the parameters also need to be considered when creating the design. If designs are made without consideration of test methods, and the design cannot be fully tested because of constraints with the equipment, it may lead to poor yield.
Avoid certain pitfalls when testing optoelectronic components. Ultra-short and ultra-low optical or electric signals are always difficult to handle or measure. Testing high-speed, ultra-small opto components or packages requires careful preparation and analysis.
Consider a 40 Gb/s optical transceiver. The transceiver is a product integrating an optical transmitter, receiver and other relevant optics/electronics. It should generate a 40 Gb/s optical signal from a 40 Gb/s electronic signal and detect a 40 Gb/s optical signal that converts to a 40 Gb/s electronic signal. One bit in the 40 Gb/s signal is occupying 25 picoseconds (ps) of time interval. If timing jitter or perturbation is more the 30% of 25 ps, or 7.5 ps, it can create a problem. Measuring that type of package requires test equipment with a very high quality capability for testing timing jitter.
Polarization is generating statistical variation on this low signal level, so some method to average out or filter out the impact due to polarization should be designed. Testability means a stable and consistent measurement using regular test equipment. If some product can create yearly revenue of $1 million over the next five years but requires the company to purchase $10 million of test equipment, the company cannot make that product and be profitable, or it will have to forego measuring that parameter. Optoelectronics companies rarely offer outsourced test services or use of their test equipment because demand does not warrant the cost of the equipment.
Expensive testing using expensive test equipment for expensive components is easily designed. But, inexpensive testing using inexpensive test equipment for expensive components can also be designed by an experienced engineer. If a product cannot be tested, it is risky to introduce it to the commercial market.
Michael C. Shores is president of TXP, a design and manufacturing firm (texasprototypes.com); michael.shores@texasprototypes.com.
