We have designed a metering system for three-phase electric appliances based upon the metering IC ADE77880. After producing the prototype board, during the first tests, we realized that the measurements of the rms voltages on the three-phases were affected by a slow, but constant drift: soon after the turn on, the system works well and it measures correctly about 220V rms on all the phases. After some minutes the measured values go slowly towards lower values: for instance, after about 15 minutes we measure 216V rms. In order to understand the phenomenon, we tried to cool the system by means of a fan and we noticed that, after some minutes, the measured values of the voltages came back to the initial, good ones. Considering that the voltages at the input of the ADE77880, measured with a precision voltmeter, do not change at all during all the described tests, we conclude that the problem is due to a strange thermal drift of the ADE. We have also monitored the value of the 1.2V internal reference that we use for the ADE, but we did not notice any drift for this voltage. Moreover the current measurements do not exhibit any sign of drift and the temperature of the ADE seems absolutely normal. Do you have any suggestion or hint to give us in order to solve the problem ?
Thank you for your attention and best regards.