Pylontech US3000C not balancing properly after sitting idle for 3 weeks

Had almost the exact same thing happen with my US3000C stack last autumn — left the motorhome sitting on the drive for about three weeks while we sorted some damp issues, came back to find a 47mV spread between the highest and lowest cell groups. Victron Cerbo was throwing all sorts of fits about it.

What sorted it for me was just a few slow, full charge cycles. I let the Multiplus push it up to 100% at a reduced current (dropped absorption down to about 20A rather than letting it rip), held it there for a good couple of hours, then did a full discharge down to around 20% and repeated. By the third cycle the spread had dropped to under 5mV and it's been rock solid since.

The Pylontech BMS is fairly conservative with balancing — it only really works hard at the top of charge, so if the pack's been sitting at 60% or so it just hasn't had the opportunity to do its thing.

A few things worth checking:

• Are all your inter-battery cables the same length and gauge? Uneven resistance causes uneven charging
• Check the RS485/CAN comms — sometimes a dodgy connection means the BMS isn't reporting correctly and the charger backs off too early
• Firmware on the US3000C — there were some known balancing quirks on older versions

Fogstar and a few others have written decent guides on this if you search around.

What does your cell spread actually look like at rest versus under load? That might point to whether it's a genuine balance issue or something else going on with one of the modules.

@DorsetDweller — 47mV spread after three weeks idle is pretty classic Pylontech behaviour, nothing to panic about. The cells self-discharge at slightly different rates when the BMS goes into sleep mode, so you'll always see some divergence after extended storage.

The fix I've used on my own stack (three US3000Cs in parallel feeding a Victron Multiplus-II) is simply running a slow, controlled charge up to around 53.2V and holding it there. The BMS will handle the cell-level balancing passively — just takes a few charge cycles to tighten back up.

Key things to check:

• Ensure your Victron absorption voltage isn't set too low (under 52V will hamper balancing)
• Confirm BMS comms are active via the CAN bus — passive balancing won't engage properly otherwise
• Give it 3–5 full cycles before judging the spread

What's your charge controller reporting for absorption hold time?

Good shout from @DodgyRoamer — worth adding that the recovery process can be a bit slow if your cells have drifted significantly. I'd recommend doing a few shallow charge/discharge cycles rather than just leaving it to sort itself out passively. On my own US3000C stack I found that running it down to around 20% and back up to 100% a couple of times got the spread back under 10mV within a day or so. Also make sure your Victron (or whatever inverter/charger you're running) isn't cutting absorption short — if it's set too conservatively the BMS never gets a proper chance to top-balance the cells. Check your absorption time settings if you haven't already.

One thing I'd add to what @DodgyRoamer and @Squib30 have covered — if you're finding the balancing is taking longer than expected, double-check your charge voltage is actually reaching the full absorption voltage rather than cutting off early. I've seen Victron setups in particular where the DVCC settings interact oddly with Pylontech's own BMS communications and the pack never quite gets to the top of charge where active balancing kicks in properly. Worth having a look at your charge logs if your inverter/charger supports it. A few full charge cycles back-to-back usually sorts a 47mV spread out nicely.

Just done three weeks idle on my boat with a similar Pylontech stack and honestly the passive balancing at top of charge is the real secret sauce — park it at 100% on shore power for a lazy afternoon and let the BMS do its thing while you drink tea and pretend you're being productive.

Had this exact drama with my US3000C in the motorhome last spring. Three weeks idle and it came back looking like a toddler who'd been left unsupervised with a biscuit tin — complete chaos, cells all over the place.

Quick question nobody's answered yet though: how are you actually monitoring the cell-level spread? Are you using