DESS keeps draining battery overnight (and even charges it) when house is using grid

Q&A and troubleshooting
, , , , ,
1

I’m having for sometime a ‘high’ idle usage on my Victron system around 200+ watts which make me wondering what is going on… It drains over 1 kWh over like 5 hours making it RTE really inefficient.

DESS configured in Green mode, minimum SOC set to 10%

Here is a good example of last night:

  • Midnight: Dishwasher using the battery (the peak, green line);
  • 1.00: Battery reached it’s 10% SOC (empty) so grid kicks-in (blue line)
  • 2.00-3.30: grid is used for the house but still battery is drained with 220watt
  • 3.30-5.15: finally no battery drain
  • 5.15: charging up 0.35kwh from grid to battery
  • 5.15-7.00: draining the battery again when not delivering power to the house.
  • 7.00: solar starts producing so hard to tell what’s exactly going where (but battery seems idle again)

Home Assistant view:

Screenshot from 2026-06-15 10-35-47
Screenshot from 2026-06-15 10-35-47445×422 20.6 KB

Advanced view:

image
image402×912 33.6 KB

DESS:

image
image761×1110 34.1 KB

Maybe related… but a lot of times the L1 Multiplus throws Low battery alarm, even though it’s well above the 50v that is configured:

image
image493×448 11.4 KB

  • 3x Multiplus II 3000VA (in 3 phase mode) with Cerbo GX (using stable wifi)
  • VM-3P75CT Grid meter (using ethernet)
  • JK-BMS (16kWh yixiang battery)
  • Gridpoint set to 0W
2

The inverters have a certain self-consumption during idling, but for 3 MP 3000 that should not be above 50W.

what is your grid feedin / consumption at that time? Any inverters efficiency is quite low for small power values, so having a 50W consumption (and a -20 setpoint?) plus the 50W for the inverters themself may well need a 200W DC Discharge.

Looks something like this usually.

image
image570×386 75.1 KB

and where that value would be for a 1% load… you may guess it :slight_smile:

Also you should check the DC Currents of the Multiplus logged in VRM. The BMS itself may just “overreport” the small current there.

3

Consumption around that time is around 120 watts so pretty low indeed, but it’s taking all that from the grid according to the VM-3P75CT.

I also pulled the DC metrics from the Multiplus itself which shows a different number so maybe I’m just looking at vanity metrics from the JK-BMS (and the percentage SOC going down being inaccurate) which the DESS also looks at?

As you can see below it’s reporting (now charging from solar) 37.9A on the Multiplus and 43.0A from the JK-BMS resulting in a difference of 116 watts; where 2315-84+20 = 2211 watts (pretty close to the 2202 watts)

image
image705×721 41.5 KB

4

Both the Multiplus and the JK BMS have a very bad reading on DC amps, use a smartshunt or lynx shunt for more precise measurements. I have a lynx shunt and also JK BMS, the lynx shunt is always spot on (measured with DC clamp meter), the JK BMS is always off.

5

Soc is even more guessing by the BMS. Educated guessing based on measured current, yes, but sometimes can be quite inaccurate as well.

A shunt would be the best choice, but it’s a question if the “increased precision” (or it’s impact) is worth the additional costs and complexity. Configuration on that is not easy and miss-configured it may be even worse.

Eventually the jk-bms offers some ways to calibrate things or influence measurements and fine-tuning that a bit may be sufficient?