Hello, I have the following problem with my Multiplus 2 3000. It doesn’t count the feed-in and consumption correctly. The feed-in is always several kWh lower than at the grid meter. On the other hand, consumption is always higher and does not correspond to realistic consumption. Of course that ruins the whole statistics.
If I read the values directly on the EM540, they are also correct and match the grid meter
My Victron system ran for a year without any problems as a battery charger via AC without its own MPPT. There was only a small PV system via microinverters.
The problem has occurred since the expansion by two MPPTs with 4 and 6 modules and a 10kW Fronius GEN24 inverter.
Does anyone have an idea how I can solve this?
What you can clearly see in the VRM is that at times when the PV generates a lot of electricity, the consumption is sporadically much higher than it actually should be. I have attached screenshots.
Thanks for your help
The PV system currently consists of the following components:
1x Multiplus 2 3000 at phase 1
1x Pylontech US2000C
1x Pylontech US5000C
1x Lynx Power IN
1x SmartSolar MPPT 150/45 with 6 modules and 3 strings East
1x SmartSolar MPPT 150/35 with 4 modules and two West strings
1x Fronius GEN24 10.0 3 phase with 32 modules, 16 per MPPT.
1x Hoymiles HM1500 with 4 modules
2x Hoymiles HM400 with one module
1x Hoymiles HM300 with one module
1x EM540
The Fronius inverter is only queried once every 5 seconds.
So, while the values of the meter and the multi are changing 50 times, the inverter value will update once. This will cause any sort of temporary errors, for example if the fronius reports “3 kW” on phase 1 (for 5000ms) and the meter drops from 3kW Feedin down to 1000 in that time - the system would conclude, that consumption has raised by 2kw - until the fronius is updated with reduced feedin values.
Not much you can do (for the momentary view), just a concurrency issue.
This also takes direct influence on the overall consumption “count”.
The values “To-Grid” and “From-Grid” however should be calculated out of the meters internal counters - and they should be correct beside any error in temporary display-values.
If you are good with python, you could disable the original victron-fronius readout and query the inverter faster on your own, like every 200ms is fine for the json-api, when using wired connections.
Thank you for your answer. I didn’t know that and I’m surprised that, as Victron worked with Fronius some time ago, they didn’t make sure that it was correct.
Unfortunately, the To-Grid value is often incorrect either. Sometimes there is a deviation of less than 1 kWh, sometimes a deviation of 5 kWh, for example in summer. With the number of inverters, the Hoymiles only update themselves every 5 seconds. One is queried every second, but because only one is queried per second, it’s similar to the Fronius with 5 seconds. When I only had the Hoymiles without the Fronius and the MPPT, there were no problems.
If I see it correctly, the only correct solution if you don’t want to program Python is to put an energy meter in front of the Fronius and all the other AC inverters, right?
EDIT: But if I go into the console and count how quickly the phase values from Fronius change, they are updated faster than 5 seconds. More like 1 second or less.
I’ve just checked my fronius, and the problem seems to be (time to time) on the fronius side:
The cerbo indeed seems to pull once per second, so replacing the scripting won’t help.
However, when I rapidly refresh the json api of the inverter, at some point it starts to deliver the same values for multiple consecutive refreshes.
May be a coincidence that values didn’t change, but also may be that the (tiny) microprocessor in the fronius gets to busy and can’t keep up it’s data refreshing with the poll rate.
Using a dedicated meter would ofc. Solve this in either case, but it’s really a expensive option just to get some readings right.
This shouldn’t be the case. This value is independent from the poll-rates and therefore any errors there.
How do you tell it’s wrong? What is your reference data? (Which you assume to be correct?)
I took a reading directly from the EM540 and compared it with the data from the official grid meter.
Both data are identical, only something different is displayed in the VRM.
Hmm, you should do the following:
Create a advanced graph for the Energy Meters Counter in VRM, for example:
In your case, you may want to use the “energy to net” parameter.
- Write down this number.
- Write down (the exact same value !) Read from the EM540 directly. If it’s not the same, we already have an issue here.
- Write down the Counter of your main grid meter as well. (Will be a different absolute value)
Repeat after 24h
Now, you can calculate the exact amount of energy drawn send to grid (second value - first value).
The numbers from the advanced graph should match the numbers from the EM540.
The original meter will have different absolute values, but the difference should be about the same, +/- a small error margin.
VRM should display as “to grid” the very same difference the EM 540 counter yields when calculating manually. (You then have to pick exactly the same times in the graph, for example 7am to 7am or whatever your readings are.)
If there are any differences, this may help to identify the issue.
Thank you very much for your help. I noted the values yesterday and today. The values read hardly differ, but what is noticeable is that the value of the installation data is not the same as under vrm advanced. These are the measured values:
October 7th 10:44 p.m.
EM540 5719.775
VRM 5719.7
GRID 5723.473
October 8th 10:44 p.m.
EM540 5739.428
VRM 5739.4
GRID 5743.116
That corresponds to approx. 19.7 kWh.
Here is the image from the installation data where only 19 kWh are displayed:
If I extrapolate this difference, I get exactly the same high differences that I had in the summer after the expansion, where I was always missing around 3-4 kWh to the grid.
So, these values look okay. EM540 is measuring about the same as your grid meter and vrm has the exact values the EM540 is reporting. No visible issue there, looks good.
For any other value, you would need to consider that the view “today” would now lack 10.44 - 23.59 of the day before - but since we are talking about grid feedin - that time should equal 0 anyway. (Unless you had some battery2grid at that time, equaling the missing 0.7 kWh? DESS is sometimes doing strange things 
)
I will check these numbers on my system tomorrow, tiny differences can run unnoticed for a long time.
Was curious, and… yeah, well:
So, either this value is - against all odds - not derived from the counter of the meter or it is a bug - whatever could go wrong when subtracting two values 
Maybe @Barbara could provide some insights how the “To Grid”-Value is actually calculated, if not based on the meters reported counter?
I’m glad that the problem was reproduced. I’m always very meticulous when it comes to things like this, which is why I notice things like this straight away. I didn’t have DESS active in the summer, so I can definitely rule it out in the summer.
The system didn’t detect Battery2Grid either.
How is the SetPoint set in your system?
Even if the difference in value per day isn’t that big, the incorrect value still adds up quite high over the whole year. When I think about the summer, it quickly adds up to 100 kWh per month that weren’t displayed correctly.
I hope Barbara can shed some light on this.
