We have developed a 3 phase energy monitoring product using the ade7758. It works as expected in 3 phase applications. We were presented with the need to quickly come up with a product that could monitor split phase power, where V1 and V2 are 120V, 180 degrees out of phase with one another, and V1+V2=240V. Both legs are referenced to the same neutral. We thought that it should be possible to adapt the 3 phase design to work in this application simply by hooking the V1 and V2 phases to the ADE7758 phase A and C respectively. As mentioned, both legs are referenced to the same neutral. We use external processing to add the results that are read from the 7758. Both the A and C phase CT’s are configured exactly the same with respect to source and load.
Unfortunately we do not have split phase power available here to test with, so we tested with 208V bi Phase wall heater and even tried feeding the A and C phases with the same, single phase 120V source. Both of these tests worked as expected.
When attached to the aforementioned split phase configuration however, we notice something very strange. During the summer when an AC unit was providing load, we didn’t see anything too unusual. V and I made sense and the Watt and VA numbers seemed reasonable. One problem is that we no longer have that data at a suitable granularity to analyze, since we can’t store everything for more than a week.
However, now that the load has gone from an AC unit to a heater, here is the problem we are trying to sort out: given a load that is almost purely resistive (a basic heater), V and I read as expected, but both VA and WATT registers return negative values which are ~V*I. I have added some code to handle the fact that the 7758 could see another line crossing interrupt while the CPU is accessing it in the process of servicing the ISR for another line crossing. It handles A, B and C being fed by exactly the same phase, so I don’t believe that is an issue. The spec indicates that the 7758 should handle this case, regardless.
I realize this is probably not an ideal use for the 7758, but is there anything we are doing wrong or a reason that it simply wouldn’t work?