At night when electricity price is very low, DESS decides it is more beneficial to use the power from the grid and save batteries for morning price increase. When such mode is active, energy goes from grid to consumption, but at the same time there is a 90W draw from the batteries. I tried deactivating inverters by switching mode to “Charger only”. Also tried changing from “On” to “Charger only” in Multi RS settings using VictronConnect. But nothing changes. Still the same 90W draw. Only if I set the mode to Passthrough, relays click and batteries go to Idle with 0W draw. Where do these 90W go? This is more than the whole house uses at night (base load). I have a 3 phase setup using Multi RS Solar. If inverters need power to react quickly to grid loss or to be able to shave peaks, why couldn’t they be powered from the grid? Or maybe charger could be used to offset the current on DC bus?
When DESS is active and ‘idling’ (e.g., passing power from grid to loads), the system is trying to match the AC Input and Output wattage. The circuitry inside the inverter/charger is still active, and that requires additional power which is supplied from the DC side of the system.
Technically you’re not “losing” power, because that ~90W would otherwise be taken from the grid.
For many systems which don’t require peak shaving it would be better for the inverter/charger circuit to just go to sleep (reducing that idle draw to a much smaller amount). I’ve spoken to a couple of Victron engineers on this forum about that, and I think it’s somewhere on the longer term roadmap (I believe its a fairly complex change).
My system has 4x 10kVA Quattros, so my idle draw is ~400-500W, so I feel your pain!
Yes, I understand inverters need power to stay active. Question is why Charger only mode does nothing to idle power? I guess DESS could automatically put inverters in Passthrough mode if there is a long Grid to Consumption period. But then peak shaving would not work which in some cases might be necessary. I don’t even know if it is possible to disable it in Multi RS systems. At least I couldn’t find where the settings are.
Could it be that DESS does not take into account this inverter idle draw? Now what happens is that in the evening DESS discharges batteries to grid and leaves just enough to make it to next day when solar charger kicks in. But it always falls short and importing from grid starts in the middle of the night. In many cases that costs more than the money earned from discharging to grid. For example, right before midnight it discharges 1kWh of what it thinks is excess battery to grid at a sell price of 0.18Eur, but in the middle of the night buy price is 0.25Eur and I end up buying the same 1kWh back at a higher price than I earned.
You are correct. It’s a “DC load” which DESS does not account for. It’s something Victron were looking into a few months ago for me as well but not sure what came of it.
VRM now lets you adjust the forecasted consumption. What I’ve done to work around this “issue” is to inject fake loads, so that DESS can plan around the additional draw. However, this won’t work when your system is idling / passing power from the grid — so you’ll still see SOC drift down in that scenario. But when its self consuming, this should allow you to bake in some buffer.
The latest Venus OS beta allows you to adjust the VRM forecast programmatically via Node RED.
You can also play around with dynamically adjusting your ESS minimum SOC.
Hope that helps.
I thought I will be smart and do exactly that before the night, but, DESS outsmarted me. After I decreased minimum SOC, it figured that now there is excess energy and dumped everything to grid.
What about the battery? How is it connected?
Have you tried entering a slightly lower capacity?
In my case the battery drain stopped when peak shaving is disabled due to low soc.
Actually in my case I connected a multimeter between the battery and the multiplus to measure the draw which was 0A, even when peak shaving was active.
It seems like the ripple created on the DC wires by the peak shaving electronics causes my JK-BMS to think there is a current flowing (about 1.2 amps) while actually there is no current flowing.
So I don’t use the soc of the JKBMS anymore but the SOC calculated by the multiplus itself. When I use the BMS for soc I have a SOC drift which is incorrect. (the best solution would to be using a better BMS or a Victron Smart Shunt).
According to Gemini, this behaviour is to be expected with some BMS modules:
When a MultiPlus performs peak shaving or handles dynamic loads, it creates high-frequency switching and low-frequency current/voltage ripples (at 100Hz/120Hz and PWM switching frequencies) on the DC lines.
Why this happens:
-
Shunt Sampling vs. Ripple: A JK-BMS calculates current by measuring the voltage drop across a small internal shunt resistor. Cheaper or unshielded sampling firmware on certain BMS models can misinterpret AC ripple components or noise superimposed on the DC bus as actual net DC flow, leading to “ghost” current readings (like 1.2A) when the actual net draw is zero.
-
The Resulting SOC Drift: Because the BMS integrates these false currents over time, its internal Coulomb counting (State of Charge tracking) drifts completely out of sync, often triggering premature low-SoC shutdowns or erratic behavior.
That’s quite interesting… I had been complaining for a year to my dealer about this ‘idle draw’ and he also said a Victron engineer said it was not a real draw but rather power being push/pulled at a very high frequency.
In my situation, I have fairly high quality batteries (well, for what meets the building code here). The batteries report a 10-15A draw when idling, while the Quattros report a 0-4A draw. Big difference.
I don’t think I can rely on the Quattro reported SOC for the same reason you highlighted - it could result in the BMS shutting the battery down if the SOC goes to low. My batteries are mounted outside so they also have internal heaters, which the Quattros can’t measure, nor would having a Victron shunt.
So while technically, energy may not be being ‘lost’, it’s still causing DESS to charge ‘extra’ in the mornings thinking the SOC dropped more than expected overnight (when my system typically idles).
I have not measured it like you did, but I’m not fully convinced that energy isn’t being lost because the Quattros stay quite warm when idling, so that energy must be coming from somewhere…