Using statistical analysis methods to predict switching stability

Switching reliability, repeatability and stability are crucial in Automated Test Equipment (ATE). There is an inherent variability in a mechanical switch and after an extended number of cycles, will cause an out tolerance condition. This could present itself as an intermittent test failure that will require isolation to the Unit Under Test (UUT) or the ATE. A method that could detect anomalous switching behavior and predict potential switching issues would be very beneficial. It could potentially preclude failures due to deviations in ATE paths caused by switching variability. This paper will discuss a method to test switches within the ATE system and identify suspect switches. The testing methodology is based on an existing system self-test. It can be implemented as an enhancement to the existing code or as a standalone test. Each group of tests within the switching self-test is repeated multiple times and data is captured for all iterations. Statistical analysis is then applied to the acquired data. The test data is assessed for two parametric characteristics; the average value and standard deviation of the repetitive measurements taken for each switch path. The data can be exported to a spreadsheet or calculated within the self-test code. Maximum limits are determined based on the resistive path and number of switches. These limits are then compared to the statistical results. The results can be displayed numerically and also be displayed graphically. Charts can be generated if the results are exported to a spreadsheet. This expanded testing can be performed on a periodic basis but less frequently than the existing self-test. The testing can be expanded to include an adapter built for the specific application. It would include wrap-around wiring and connectors to mate to the front panels of the switch modules. This would isolate the switches from the system and permit testing of just the switches without the system wiring. The approach and method of utilizing statistical analysis of self-test data provides a means to identify early signs of variability and initiate a proactive repair of degrading switches.