Using statistics to reduce the uncertainty in system level susceptibility testing

Products containing electronics must generally be tested for electromagnetic susceptibility, as part of their EMC certification process, at the end of their development cycle to ensure that the test is representative of any future use of the product. However, the earlier that the product as a whole, or the electronic systems that might form part of the product, can be tested for susceptibility, the more benefit can be gained from such a procedure. Both a saving in time and cost can be achieved in the product development process. This paper describes a statistical technique which has been developed for use in the automotive industry with a bulk current injection (BCI) test. It has been designed to help overcome the uncertainties in correlating results from such a system-level bench test with results from the final vehicle-level sign-off test carried out in a full-threat field within a semi-anechoic EMC chamber. The technique uses a probability distribution function with the mean transfer function and its standard deviation to estimate the risk of the test system, once installed in a vehicle, malfunctioning when that vehicle is exposed to radiation of a particular field strength and frequency. The effect of vehicle body dimensions on these risk figures has been examined and a simple mathematical model using statistical input data has been produced which shows some agreement with the measured data