Update — a second freeze the same evening, and some findings on what triggers it
Following up on my earlier posts: the battery froze twice more this evening, about 55 minutes apart, both during the same cooking session with an electric oven. Same failure each time — AC, battery and EPS/off-grid outputs all drop to 0 W at once, comms stay up, and it only recovers when my supervisor re-pushes the operation mode (smart → off → smart).
The usual trigger is a fast load step. I went back through 10 days of history and cross-checked every freeze against my house-power log. Both of tonight's lined up with the oven's heating element switching:
- 19:03 — element ON: house load ramped 90 W → 2.2 kW in 6 s; froze right as the battery was asked to ramp its discharge up to follow.
- 19:59 — element OFF: discharge dropped 2.2 kW → 80 W in 6 s; froze 30 s later.
And this isn't a one-off pattern: 8 of the last 9 genuine freezes were preceded, within 2 minutes, by a fast load step of ≥1 kW — often the oven's rapid ON/OFF 2 kW cycling, sometimes on top of other loads. It trips on a sudden step in either direction (a sharp turn-ON or a sharp drop), and not on the state of charge — in every case the previous identical step had been followed with no problem.
But it's not exclusively load-driven. One of those 9 freezes happened with a completely quiet house (80–180 W) and no fast step at all in the preceding 12 minutes — it simply froze while idling. So fast load steps are the dominant trigger, but not the only one.
No protection flag is raised. At the instant of each freeze:
- battery voltage healthy (49.6–49.8 V), SoC 78–82 %, device temp 36–39 °C
- discharge was 2.2 kW — under the 2400 W rating
- is_error = 0, SoC-limit flag = 0, grid state normal — nothing latched
Everything electrical is nominal; the device just silently drops all outputs and stays there until a mode is re-sent. To me this looks like a control / restart-logic hang rather than a genuine protection trip — a real protection would have a measurable cause and would clear on a defined reset, not by re-sending the same operation mode. The fact that it can also freeze while idling (not just on a transient) points the same way.
Question for Zendure: does the device log any internal fault or protection code at these events? From HA I see no error flag at all (is_error = 0, no SoC/grid limit), so if there's an internal protection or anti-oscillation routine reacting to fast load changes, it isn't surfaced anywhere I can read.
Is there a way to read the device's fault log?
Happy to share second-resolution logs.