Watt measurement of VM-3P75CT jumping up and down in a 1kW range

I have mounted a VM-3P75CT but when the solar panels are active, the watt measurement of the VM-3P75CT almost seems random.

Solar panels are enphase and connected to AC out.

What’s wrong?

When the multiplus has finished charging, the measurement of the VM-3P75CT is reasonable stable within a 100 watt range with 2000 watt grid feed in from the solar panels. It almost seems that the grid measurement control loop creates a a lot of fluctuation.

I have the same problem.

I have 3x 3P75CT meters. Grid, Enphase connected to AC-IN and Enphase connected to AC-OUT.

  • If there is no sun or a lot of sun, all 3 meters are working as expected
  • If there is little to no sun, only Enphase AC-IN is working as expected. Both Grid and Enphase AC-OUT are going up and down and it also seems to be moving across L1, L2 and L3. Sometimes even by as much as 200W

Some observations:

  • if I keep the stats for AC-OUT open in remote console, the numbers are also going up and down but are far far more evened out. They seem still incorrect and more or less confirm the changes jumping over L1,L2 and L3. This makes it hard to believe it’s just measurement error?
  • If I pull the breaker on Enphase AC-OUT and put the power back on, the behaviour does not start immediately. The Micro’s (IQ7a) are generating but the odd behaviour starts only after a few minutes
  • I checked the assistant, it’s configured for s50.2 - m51.7 - d52.0 and if I emulate that by pulling the AC-IN breaker and use DVCC to manipulate the charger the Micro’s are throttling as expected
  • I changed all three assistant to not “have PV”. And again using DVCC confirming frequency shifting is not applied. But the readings are still heavily fluctuating

Exhausted all options and have now submitted a support case with Enphase asking them for release notes. For me all that remains is that the 12 IQ7a Micro’s at AC-IN are on an earlier firmware than the 8 IQ7a Micro’s connected to AC-OUT.

In the attached video’s you can see the system running in early morning with and without the AC-OUT (PV2) micro’s connected. You can clearly see the high resolution of the grid meter and PV2 picking up on the way the micro’s inject the power to the grid. With those micro’s disconnected, the 12 remaining micro’s on the older firmware do not seem to trigger that behaviour on grid and PV1.
As mentioned earlier, the first 5 mins after power-on, everything is stable. Not sure how that fits in.

Odd behaviour with PV2 powered on:

No problem once PV2 is powered off:

No response from Enphase yet but it’s the only option that remains. The most likely cause is that Enphase are using their burst mode more aggressively and during longer periods of low lights instead of just during shading.

I have seen the behaviour before but never as extreme as this and most certainly not with such a huge difference between 2 firmware versions. Burst mode is known to cause this behaviour. Specially if you have a large amount of panels, the reading can go up and down with a few 100s of W during low light and shading.

I know Victron is working on slowing down their sample rate for meters set for PV. Think the latest beta already includes this if connected via can instead of ethernet. With ethernet soon to follow. Let’s hope the result will be smoothening or averaging not simply a lower resolution to Cerbo.

But that does not smoothen the grid readings. The difficult thing is that it seems there is nothing to fix here. At the time of the reading the sample actually is correct. It just looks a bit broken due to the use of capacitors.

Installed a victron current sensor on the pv panels today. The watt reading of the victron current sensor is pretty stable, but the VM-3P75CT watt reading is still going all over the place.

I doubt if the victron can keep a good 0 watt grid setpoint.