Ah; understood. Trying again:
Nerdvana_Cerbo_c0619abc1493_inst_ttyS4_Interfaces_Mk2_Tunnel.rvsc (4.5 KB)
Have you tried testing without using both the virtual switch and powerassist?
Cut off looks sufficiently low to not be an issue if the battery voltage is stable.
What is visible on your battery app logs?
I have not; I will test with virtual switch and powerassist disabled.
There’s a battery app that logs every 15 secs or so. On a test this afternoon it logged 25.77 V where the inverter tripped. For my next test I’ll increase it’s logging frequency to ~ 1 per sec.
BTW I just tested with a couple of resistive loads (water heater and oven) totalling 3.2 kW, which led to a fairly immediate trip, corresponding with the 25.77V measurement mentioned earlier.
Ok. Does the app keep a log for any BMS exceptions or does it only report voltages?
Is the BMS visible to the GX and on DVCC?
What is configured as the battery monitor in the GX system setup?
Can you pull advanced widgets from VRM that shows the BMS limits?
Each battery has its own independent BMS. Each BMS keeps a count of error conditions; they are all zero.
No.
Nothing explicitly configured. The GX is new, added mainly to investigate this issue. Battery monitoring is part of the vessel’s overall multivendor management system.
Er, not to my knowledge: I don’t think the battery’s internal BMSs integrate with VRM.
We do track the current and power in and out of each battery and their SOCs but that’s part of the multivendor platform independent of the internal BMSs.
Looking at the available data you have a battery or wiring issue.
This is based on a low resolution sample - VRM is updating every 15 minutes, it could be updating every minute which would help.
Your load is dropping your battery below the cutoff voltage. If the batteries believe their voltage is constant, then the loss is in the connections to the inverter.
This is ~2600W from the battery, so a pre-loss load of 2000-2300W.
This overlaps with high ripple warnings:
Thanks; very interesting.
I attempted to change the VRM logging interval from 15 to 1 minutes but couldn’t find the relevant setting.
There are individual shunts on all three batteries, before they are joined at a busbar. They are used for the primary power monitoring, coulomb counting and SOC tracking. That’s where I can speed up logging frequency of about a second, if needed. To go faster than that I use an oscilloscope.
I think I obviously need to repeat my test with the 'scope on the Quattro’s DC inputs; I’m now having doubts about the rigour of my previous tests. I don’t have a clamp meter on board right now but I will get one when I return to the UK next week.
BTW, the two events you saw yesterday were my resistive load test at ~16:00, and an aircon test with powerassist disabled (no discernable difference).
I made a quick attempt to disable Virtual Switch but could not find where it was enabled.
Thank you again for all the help with this 
Its on the GX. Settings → VRM.
From the charts I doubt settings will help.
There is an iffy connection, switch, fuse holder or the like between the battery and Quattro.
It will help to see what the builder installed and to check everything is tight and properly connected.
Sometimes a “cost effective” component gets used, some are known to be problematic.
A multimeter and measuring under load should show the culprit.
OK; I think the problem is resolved. And of course you were entirely correct: The problem was a resettable thermal circuit breaker adjacent to the inverter, protecting DC circuit.
The circuit breaker itself was not tripping, which I think is why I overlooked it. But testing with an oscilloscope showed a very short change of voltage, at which point the inverter tripped and current through the circuit breaker stopped. I can only guess that the circuit breaker was starting but did not complete its disconnect operation.
Anyway, it will be replaced by a high current fuse, as in the Victron documentation (probably just 250A rather than the suggested 400A). My local installers have a reputation for quality rather than “cost-effective” parts and I’ll make sure that applies in this case.
Thanks again for all the help, guidance, insights… and patience )
Cheers
Sean
Can I quote you for my wife? She resists the sentiment
Glad you’re sorted and happy to help.