I’ve now disabled all the other meters apart from the EM24 grid meter. I’m still seeing peaks of nearly 3000kWh. It’s great that they aren’t being added together, but it obscures the others.
Does anyone have any other ideas?
Hi,
this is the Shelly for your heat pump — it sometimes reports a “0 meter reading”, which triggers a reset in the analysis. Then, as soon as it reports the correct 2666 kWh again, the analysis interprets that as 2666 kWh of “consumption” in that minute.
How have you integrated it? (Shelly integration, Node-RED, third-party script)?
That’s right. I do have a Shelly in front of the heat pump, but they haven’t been able to be integrated via the console for very long. I had the Cerbo search for the Shellys, and the one for the heat pump was still enabled too. I’ve disabled it now. Otherwise, they’re not integrated into the Victron system. I only use it via the Shelly app. The Shelly has also been there longer than the peaks have been occurring.
But surely Victron should be able to sort that out?
I’ve encountered similar problems with Volkszähler in the past: a cumulative consumption meter can’t report zero or values that differ significantly
An alternative might be to send the power reading in watts instead of absolute meter readings?
Yes, that sounds less accurate at first, but in practice it works surprisingly well
If the problem is on our side, certainly 
In general, though, there are meters that deliberately reset to 0 — after a restart, at regular intervals, when the maximum value of the register is reached, and so on. The energy calculation handles that as intended — only unintended “zeroes” are false positives and should be avoided.
So we’d need to identify and “fix” the exact cause, which is why I asked how the Shelly is connected. Third-party scripts are unfortunately sometimes not protected against “zeroes” if, for example, there’s an MQTT reconnect, or some part of the system turns null into a 0, etc.
It does get tricky, of course, if the Shelly itself is the cause. The EM3 has cropped up in a few threads, whereas, as far as I remember, the PM Shellies never have.
I do that too in some cases where there’s no other option, but if the device sends meter readings, it’s not really necessary. It also offers no protection against temporary communication outages. The device’s internal meter keeps counting, and the reading is then at least recorded “with a delay” once the fault is fixed.
and “Cerbo GXC00A HQ2215ZJYVG v3.73v3.73 Up to date”
I’d almost bet that Victron doesn’t test this combination of firmware.
For cost reasons, you have to focus on current versions when testing.
Since I come from a software development background, I’ve got into the habit of installing new firmware about two months after its release.
That saves a lot of hassle with computers, and so far it’s worked well for me with Victron too.
It was the Shelly. Ever since it was disabled in the Shelly device class, the peaks have stopped! 