Grid Setpoint not tracking very well when running Parallel Multiplus

Experiencing this problem

Two Multiplus units in Parallel. Single phase, Grid wired up on AC IN, house on AC 2 which includes a solar inverter + panels (i.e. AC coupled)

Using a Victron CT on the grid connection to the Multiplus. The reading is accurate whether drawing from the grid or feeding.

HOWEVER….

Grid setpoint must be doubled to achieve an intended actual. For example, If I want to export 5kW, I must set it to 10 kW. This applies when wanting to export 8kw, setpoint must be 16kW!.

Similarly, Grid current limit, when set to 63 A, (the breaker is 63A), Victron only draws ~32 A through. This is resolved by setting grid current limit to double 63 A, which is 126 A.

Firmware of the Cerbo and both Multiplus are up do date, and identical on the MP.

Yes they appear consistent but I have to adjust home assistant to compensate for this. I am concerned that a fix will come down the line in the middle of the night and my system will try and draw immense currents, with hopes and dreams resting on a breaker. Perhaps I work in a check of the power draw, if it ever exceeds the actual limit, it pulls up any subsequent drawing on the grid and flags to me to take a look again….not ideal.

AI did tell me an energy meter such as ET112 should fix the issue, but why? the CT clamp works like a charm….that works flawlessly….. its just the values have a factor of 2 on them. This is easy or at least very possible to code around in the Victron “Matrix”.

I have not found a fix for this, using AI to trawl the wonderful internet for me.
A response from a human being, with a positive contribution would be fantastic.

I have not posted before (or recently and have forgotten). Please let me know if further information is needed. The Archived post is extremely detailed, my problem is virtually the same as it.

Thank you very much

You could try some Node-RED. These are some flows I have to monitor for the energy meter dropping out, and modify the grid export to allow a max feed of 5000W on a single phase, comparing the total of all phases. There is some good code I put together in here that you could modify. It has been running well for over 6 months. Code for the flows is in the text files. It sounds like the problems you have are due to parallel units, but if you are like me and dont trust the system to run without issues, you can put in some safeguards.

CHECK FOR 0 POWER.txt (9.8 KB)

ADJUST GRID EXPORT.txt (4.2 KB)

Not very sure but I seem to remember that concerning current limit, inverters in parallel divide the figure by two. I don’t know about the grid setpoint.

Hi Owen,

Thank you for the response. I am not familiar with Node Red use, it was my next avenue if I could not get Home Assistant to work, but I did. I have used AI extensively to both teach and do the heavy lifting of code generation, with me checking the logic is correct. I will use your code as an input for the consideration, so thank you very much for sharing it. You raise a good point, to install safeguards in the logic, very doable, and offers a backup. I’m rather impressed with Home Assistant, and Victron, aside from this one little doozy it seems has been missed deep in the innards of the coding for a Single phase Parallel system.

Hi Fideri,

Yes they do, when you log in with VE BUS Quick Config, you can see the value is halved.

When the system was first installed, both inverters had the internal CT’s in use and worked almost at a correct state. However there was a separate problem of phantom gain and draw. When you had consumption on the house side, the system would feed to the grid at a rate of roughly 7kW consumption = 1 kW grid feed. This appeared linear for both consumption and consumption (but opposite effect - grid draw). When the PV produced excess which should feed the house and go into the battery - 7kW excess would draw 1kW from the grid.

This could be overcome by adjusting the setpoint to the “scaled” amount, if it was drawing 1kW, changing the grid setpoint to -1000kW would bring it to zero net.

However this is a very dynamic figure in regular normal use, and this fix would need to be done in Home assistant and performed at least once every second - real time. This is more frequently than I have confidence in the capabilities of HA, and unnecessary as holding a grid set point is a core function of the Victron system.

I had a bit of help from a solar installer who’s also an engineer to get the system up and running, and then we worked together to improve it. Node-RED runs directly on the GX device, so it isn’t a separate service. You can also configure Node-RED to ping the Home Assistant server and send an alert or change a state if it goes offline.

Node Red seems the way to go. In my experience, Node Red can freeze which would leave your system exposed. I would suggest setting low, fallback figures in case nothing is forthcoming from Node Red but I don’t know how to do that.

@alexpescaru can you help with a scripting/registry approach? On the face of it, that would be more reliable than Node Red.

I’ve posted some time ago a list of available Victron registers. Here it is again: Victron Registers List
In the meantime, for sure additional registers were added.
Take into account that not all registers apply to a certain device. Each device has a certain set.
Also, each device can be accessed by different means (CAN, VE-Direct, etc) depending on how it’s connected and therefore the utility to send/get data is different. ( vreg, vregd, etc)

I had no idea it runs on the Cerbo! very interesting. I will consider this in a future adjustment, if only as a fail safe at first to act conservatively if the HA goes dead.

I have masked the problem completely using logic in Home Assistant. Also put in some failsafes to remove the double up if a fix comes in the night and actual current exceeds that of the breaker for more than a few minutes, functioning like a virtual breaker. Should this be detected, the double up is removed and the change flagged.