A decision system for electrolytic capacitors diagnosis

Electrolytic filter capacitors are frequently responsible for static converters breakdowns. The rise of temperature within capacitor is the most significant factor upon ageing and so lifespan. To predict the ageing of the capacitor, a method of predictive maintenance has been developed. It consists in monitoring the voltage ripple of the capacitor. This ripple is the best indicator of its worn state. It is function of parameters such as input voltage, output current and ambient temperature. A reference system taking into account all these parameters is determined for sound capacitors. We deducted that the dispersion of input voltage had a weak influence upon voltage ripple compared to other parameters such as output current and ambient temperature. And precisely the temperature is being an influent parameter in the diagnosis, measurement must be very relevant. That is why we measured the internal temperature of capacitor which takes into account the ambient temperature but also the internal heating of capacitor. This measurement is done in an environment not confined for not which is distorted. We then checked the relevance of our modifications by comparing our results with the actual values of the state of capacitors deduced using a bridge RLC.