Hi Victron staff,
I’m also affected by the “charged” bug and since I see several separate topics describing the same behavior there might be something more structural going on (I see a trend).
Symptom (for me):
When I start a charge session, the EVCS begins charging normally, but after a random period (typically >5 min, <30 min) it the EVCS switches to the “Charged” status while the car is clearly not full and still wants to charge. The session then effectively stalls until manually stopped and started again.
Environment:
- Affected car: Mercedes EQS — reproducible
- Second EV (Jeep Avenger): no issues on the same EVCS
- A separate, simpler Mode 2 charger (no load sharing) at the office: no issues with the affected car
- EVCS firmware: reproducible on most current builds (I tested myself starting from 2.05 and could reproduce on every version next).
To me, this strongly suggests an interaction between the affected car(s) CP signaling and the EVCS’s control pilot (CP) state detection mechanism, rather than a fault in the car alone.
My poor-mans hypothesis:
I understand the IEC61851 standard a bit, during an active charge the car holds the CP line at 6V (state C) via its internal resistor network. When charging completes, the car switches the resistor network so the CP line goes to 9V (state B) also known as “connected but not charging”. This C-to-B transition is exactly what the EVCS should interpret as “Charged”.
My best guess here is the EVCS is intermittently misreading the CP voltage as ~9V (state B) while it’s actually ~6V (state C).
This guess is enforced by the fact that we saw some rework on the CP calibration/detector side in the latest firmwares.
Why it’s only happening once in X time and not every second is a big unknown to me.
My best guess here (so many guesses…) is that it might be related to some timing/load issue on EVCS side.
What I’ve tried:
- Disabled auto CP calibration (Modbus register 5133 = 0) — no change.
- Reverted EVCS firmware to 2.05 (latest release before auto cp calibration was introduced) — no change.
- Dealer investigation — they applied a car-side software update — no change.
Request to Victron (@Lpopescu):
- A debug build with additional CP/state logging (raw CP voltage samples, detected state transitions, measured current per phase, and the exact trigger that flips the status to “Charged”) would let affected users self-diagnose and provide you with real data. Right now there’s no visibility at all into why the EVCS decided the car was charged. *Even if this would be a one-off version, it would greatly help people (read: installers) in the self-diagnose of possible problems.
- If useful, I’m happy to capture data (Modbus polling of CP state / current registers, or a scope on the CP line) if you tell me what would help most.
Supporting topics: