BMS Communication goes offline after a few hours: VE.CAN TEC 128 Error-Passive

Been wrestling with something similar on my shepherd's hut build and thought I'd throw it out here for the collective brain.

Running a modest 20kWh Fogstar Drift LiFePO4 bank with a Victron Cerbo GX as the hub, and after a few hours of normal operation the BMS communication just... drops. The Cerbo loses sight of the batteries entirely, and I'm seeing VE.CAN error-passive (TEC 128) flagged in the logs.

From what I can piece together, it looks like the CAN bus is essentially throwing its hands up — error count climbs past a threshold, the node goes error-passive, and suddenly the Cerbo thinks it's flying blind. Charge/discharge decisions then fall back to defaults rather than proper BMS-governed limits, which on a lithium bank is... not ideal.

A few things I've tried:

• Swapping termination resistors at both ends of the CAN run
• Reducing cable length between the BMS and Cerbo
• Lowering the CAN bus speed

Marginal improvement, but the fault still creeps back in after a few hours.

My suspicion is it's either noise on the CAN line from the inverter switching, or possibly a grounding issue between components. The hut is on a metal chassis which I'm wondering might be introducing a ground loop.

Has anyone else seen this on a Victron + third-party BMS setup, particularly where you're aggregating data from multiple battery units? I've seen mentions of people using Node-RED to repack battery telemetry into Pylontech-style frames over CAN — curious whether that approach introduces its own timing quirks that could cause this.

Would love to know what finally solved it for others.

@CliveKnight that shepherd's hut build sounds lovely by the way — but yes, the VE.CAN Error-Passive rabbit hole is a deep one.

Had almost identical grief when I set up my garden office last spring. Turned out my CAN bus termination resistors were the culprit — both ends of the bus must have 120Ω termination, and I'd missed one. The Cerbo GX has a built-in terminator you enable via the dip switch inside the unit (easy to overlook).

Also worth checking:

• Cable length — keep VE.CAN runs under 100m and use proper twisted-pair
• Baud rate mismatch between the BMS and Cerbo (should be 250kbit/s for most LiFePO4 setups)
• Fogstar Drift BMS firmware — worth emailing their support as there were some known CAN stability fixes pushed out

What firmware version is your Cerbo running?

@CliveKnight classic TEC 128 horror story — I've lived this nightmare on the narrowboat.

Nine times out of ten it's termination resistors. VE.CAN is a proper CAN bus and needs 120Ω resistors at both ends of the network. Most people fit one, feel smug, and wonder why everything falls apart after a few hours when the bus goes Error-Passive under load.

Check your Cerbo GX settings — there's a termination toggle in the device list. Also measure the actual resistance across CANH/CANL with everything unpowered: should read ~60Ω (two 120Ω resistors in parallel).

Secondary culprit: cable length or quality. I used some bargain-bin twisted pair on my first attempt. Swapped to proper Belden and the ghost disappeared overnight.

Also worth checking your Fogstar BMS firmware — some earlier Drift batches had CAN timing quirks that Fogstar quietly patched.

@BoxerCamper curious what you found on the narrowboat — I had the exact same issue on mine last winter.

What sorted it for me was a dodgy termination resistor at one end of the VE.CAN bus. Classic symptom is it works fine when everything's cool, then as the boat warms up the connection gets flaky and the BMS drops off. Check your 120Ω termination at both ends with a multimeter — mine was reading around 85Ω which was nowhere near right.

Also worth checking the cable run length. VE.CAN doesn't love long runs without proper shielding, especially on a narrowboat where you've got interference from the engine and inverter nearby.

Sorted mine with a proper Victron VE.CAN RJ45 terminator rather than a bodged resistor someone had stuck in previously. Night and day difference.

@CliveKnight had almost identical grief on the static caravan setup last summer — worth checking whether your termination resistors are actually in place at both ends of the CAN bus. Sounds obvious but mine were missing from the BMS end entirely, which I only spotted with a multimeter across the CANH/CANL pins (should read ~60Ω with both terminators fitted).

Also, is your Cerbo on the latest firmware? There were some known VE.CAN stability improvements in recent releases that genuinely made a difference here.

One more thing — cable quality matters more than people think on CAN bus. Had borrowed a dodgy patch cable initially and it caused all sorts of intermittent weirdness before I swapped it out for a proper shielded twisted pair.

Would be curious whether @BoxerCamper's narrowboat fix was termination-related or something else entirely.