Failure prediction of electrolytic capacitors during operation of a switchmode power supply

Electrolytic filter capacitors are frequently responsible for static converter breakdowns. To predict these faults, a new method to set a predictive maintenance is presented and tested on two types of converters. The best indicator of fault of the output filter capacitors is the increase of ESR (equivalent series resistance). The output-voltage ripple ΔV_o of the converter increases with respect to ESR. In order to avoid errors due to load variations, ΔV_o is filtered at the switching frequency of the converter. The problem is that this filtered component is not only dependent on the aging of the capacitors, but also on the ambient temperature, output current, and input voltage of the converter. Thus, to predict the failure of the capacitors, this component is processed with these parameters and the remaining time before failure is deduced. Software was developed to establish predictive maintenance of the converter. The method developed is as follows. First, a reference system including all the converter parameters was built for the converter at its sound state, i.e., using sound electrolytic filter capacitors. Then, all these parameters were processed and compared on line to the reference system, thereby, the lifetime of these capacitors was computed