New method improves detection of laser diodes

　　 There are several sources of error in the inspection of high-power laser diodes used in telecommunications. These error sources include coupling high current pulses, photodetector coupling, slow response of the detector itself, and errors. By properly handling these problems, the test time can be shortened, the test accuracy can be improved, and the failure rate can be reduced.

　　LIV curve

　　The basic test of laser diodes is the light-current-voltage (LIV) curve, which measures the electrical and optical output power characteristics simultaneously. This test can be performed at any stage of production, but is first used for laser diode selection, i.e. to eliminate bad diodes in advance.

　　Perform a current scan on the device under test and record the voltage at each scan step. At the same time, use an instrument to monitor the optical output power. This test is best performed in pulse mode early in production, before the laser diode is installed into the module. At this point, the diode is still in its original state and pulse detection is necessary because the component has no temperature control circuitry at this point. If tested with direct current it would at least change their characteristics and at worst damage them. In subsequent production, when they are installed in temperature-controlled modules, they can be tested with DC current and the results can be compared with the pulse test. Additionally, some diodes will pass the DC test but not the pulse test.

　　Analysis of LIV test data can determine the characteristics of the laser, including the critical current to produce laser light, quantum efficiency, and nonlinear characteristics of the output (Figure 1).

　　Testing laser diodes requires a properly formed current pulse. It should reach full current as quickly as possible and remain stable long enough to ensure accurate results. In the initial stages of testing, pulses with a width of 0.5ms to 1ms are generally used. Current variations range from tens of milliamperes to 5 amps.

　　　　　　　　　　　　　　?

　　　　　　　　　　　　　　　　　　　　Figure 1

　　Impedance matching

　　To deliver high-speed current pulses to a laser diode while avoiding reflection problems, it is generally believed that a transmission line—such as a length of coaxial cable—can be used. But the most commonly used coaxial cable has an impedance of 50Ω, and the impedance of the diode is about 2Ω, which is a very mismatch. Although a 48Ω resistor can be connected in series, this will create new problems; a 50Ω system will require a voltage of 250V to pass 5A current, which is very dangerous for both people and equipment. In addition, since the dynamic resistance of the laser decreases as the current increases, the test conditions change as the test progresses.

　　Using low-impedance coaxial cable may be an effective solution, but doing so will change the dynamic resistance of the laser diode. Another way is to connect the laser diode with two 10Ω coaxial cables and apply pulse current to both ends of the cable (Figure 2). In this way, applying a voltage of less than 10V to the diode can produce 5A of current. Because the system has a current source, problems caused by changes in the dynamic resistance of the diode are avoided.

　　Even the most careful impedance matching is unlikely to be perfect, so it is practical to use the shortest possible transmission lines. This is also to minimize the loop area connecting the laser diode.

　　　　　　　　　　　　　　　?

　　　　　　　　　　　　　　　　　　　　　　　　Figure II

　　electrical measurement

　　When high-speed pulses are applied to a laser diode, it is not easy to measure its voltage and current. Measuring voltage with a cathode ray tube detector can also cause problems, one of which is grounding. The frequency range of the detector must reach 1GHz.

　　Current measurement is a little simpler. The measurement can be made by placing a low value resistor (lower resistance than the laser diode) in series with the diode, but the capacitance and inductance of the resistor are required to be very low. Wirewound resistors have inductance losses, so they are not suitable for high frequency testing.

　　Choosing a Photodetector

　　 There are three commonly used detector materials: silicon, germanium, and indium gallium arsenide (InGaAs), each with its own advantages and disadvantages. As shown in Figure 3, the choice of detector depends largely on the wavelength it is adapted to. When the wavelength is less than 800nm, silicon is the only option. However, the commonly used wavelengths in the telecommunications field are between 1300nm and 1700nm. Indium gallium arsenide is the best at this time because its response characteristics are very stable in this range. However, there are problems with the response of InGaAs to pulses. To avoid overheating of laser diodes, it is best to test with short enough pulses, but InGaAs detectors need long enough to reach some kind of stable state.

　　As shown in Figure 4, the InGaAs detector is very unstable even within 10 microsecond pulses. This problem will be more serious if the pulse width is reduced to 1 microsecond. The germanium detector does not have this problem, so it is better suited for short pulses.

　　　　　　　　　　　　　　　　　Figure 3

　　　　　　　　　　?
　　　　　　　　　　　　　　　　　Picture 4

　　detector coupling

　　There are several ways to couple the output of a laser diode to a detector. One approach is to shine the laser directly onto the detector, but this approach has several drawbacks. First, there is no guarantee that all light will hit the detector. For example, the cross-section of the emitted beam is elliptical, or the diameter of the beam is larger than the effective receiving area of ​​the detector, or the emitted beam is not aligned with the detector, which will cause part of the light to be lost. Second, some detectors are polarization sensitive, which can lead to more errors. Third, the output of some high-power laser diodes will cause many detectors to reach saturation and fail.

　　The integrating sphere is usually the best solution. It is a hollow sphere with an inner surface coated with highly reflective material and two ports. One end is fixed on the detector and the other end is used to input the measured light (Figure 5). The integrating sphere can receive all the light emitted from the light source and distribute the light evenly on the inner surface after scattering. The detector installed on the side of the integrating sphere can "see" a fixed proportion (about 1%) of the input light. In this way, the power of all incident light can be calculated, and very high-power light can be measured without worrying about the detector being damaged.

　　　　　　　　　　　　　　　　　　　　Figure 5

　　Detection speed

　　There was a time when demand for fiber optic communications equipment exceeded supply and manufacturer efficiency became a secondary concern. Today, however, testing must be as fast, accurate and cheap as anything else. This indicates that an optical power meter is not a good choice and the detection time of this instrument is too long.

　　To avoid this problem, standard practice is to use an instrumentation suite that includes a pulse source, a light measurement component (photodiode detector, etc.), a pair of high-speed current-to-voltage converters, and a high-speed multi-channel digital sampling oscilloscope (DSO). Pulsed sources generate pulses and other instruments measure electrical and optical responses.

　　This process may require thousands of pulses. Sometimes there are hundreds of pulses per level. This seems to improve sensitivity, accuracy and precision, but it masks the problem of waveform distortion. This is also a long process, taking tens of seconds to several minutes for each device under test. This system can test approximately 2,500 parts per day, and each set of inspection equipment costs approximately US$150,000.

　　A newer approach is to include all functions in a single instrument. This instrument is essentially a pulse source measurement unit whose output impedance and cable are closely matched to the impedance of the laser diode. The measurement part of the system integrates multi-channel data acquisition, specialized timing circuits, high-speed current-to-voltage converters, and a digital signal processor (DSP) that emulates the functionality of a DSO and controls the measurement program.

　　This instrument is programmed by the internal DSP according to the detection sequence given by the GPIB bus to determine the sequence of LIV scanning. Once programmed, the digital signal processor alone performs the pulsed LIV scan without the need for instructions or computer control from other devices. In fact, the instrument directly provides control signals to various components through digital I/O ports.

　　Quick analysis of pulse measurement results is achieved through DSP, which is no longer as time-consuming as before. This reduces pulse current and voltage detection time to a few seconds and minimizes software complexity.

　　Because it only takes a few seconds to test a single unit, 15,000 diodes can still be tested every day even when the system utilization rate is only 85%. Purchasing such a system costs only a fraction of the original price, but results in higher production capabilities.

　　Such systems can be designed in both pulsed and non-pulsed modes. Both functions can be used on the same platform to scan both types of LIV through the same detection channel. Comparing pulsed and non-pulsed test results can provide more complete information about the performance of the device under test.

　　The 3rd generation LIV detection system combines all relevant functions into one instrument to greatly increase detection capabilities.