Pytes V5 SOC stalls at 20% and 90% on every cycle (~20 Ah / ~30 Ah), "15%" is almost empty

Really interesting timing on this thread — I've been pulling my hair out over almost identical behaviour on my van build for the past couple of months.

Running a single Pytes V5 (280Ah) through a Victron SmartShunt and a Cerbo GX, and the SOC readout is frankly useless for capacity planning. The 20% floor is the worst offender — the BMS seems to throttle current draw heavily before the GX even registers low SOC warnings, so you lose a meaningful chunk of usable capacity without any useful heads-up. My 90% ceiling stall is less pronounced but still visible on the VRM graphs as a distinct plateau, usually lasting 20–30 minutes before it climbs through.

A few things worth checking if others are seeing this:

• Charge voltage calibration — I had mine set to 56.8V initially; dropping to 56.4V seemed to reduce the top-end stall slightly
• DVCC settings in Venus OS — make sure the Pytes is actually controlling charge limits via CAN rather than fighting the MultiPlus-II's own absorption settings
• Firmware on the BMS — there's been at least one update floating around that supposedly tightens SOC accuracy, though I haven't verified it improved the stall behaviour specifically

Has anyone with a parallel bank of these (two or more V5s) found the stalling is worse or better than single-unit setups? I'd expect cell balancing across units to smooth it out but I genuinely don't know.

Would be good to hear from anyone running these long-term — are the stall points drifting at all as the cells settle in?

@PaulaFisher91 Classic Pytes V5 quirk — those SOC "sticky points" at 20% and 90% are the BMS doing its internal cell balancing dance, and until the cells are properly top-balanced after a few proper full cycles, your Victron's SOC readings will be absolutely lying to your face 🎭

Key fixes:

• Force a proper absorption cycle — let it sit at 56.8V until tail current drops below ~2A
• Set your Victron charged voltage to 56.0V minimum for SOC sync
• Don't trust SOC% — trust voltage until you've got 10+ cycles through it

The "15% = empty" thing is your Cerbo/SmartShunt not having a calibrated baseline yet — do a full discharge/charge and let it relearn.

Mine sorted itself after about a fortnight of proper cycling in the van!

@PaulaFisher91 mine did exactly the same in the garden office setup until I let it do a full slow absorption cycle — basically the Victron and the Pytes BMS needed a proper "getting to know you" date before the SOC maths stopped being absolutely delusional 😄

Those SOC "stalls" are the Pytes V5 BMS doing coulomb counting recalibration at its internal reference points — perfectly normal, if maddening.

The real fix: set your Victron absorption voltage to 55.2V (3.45V/cell) and hold it there for a proper 2–3 hour absorption rather than letting ESS cut away early — the BMS needs that sustained top-end to anchor its 100% SOC reference point, which cascades down and corrects the bogus 20% floor.

Also worth checking: Victron → Pytes communication via CAN-bus — if the DVCC managed charge current is fighting the BMS's own limits, you'll get exactly those plateau artefacts @BoatPaddy mentions.

Once the BMS has done 3–4 properly completed full cycles with accurate tail-current cutoff, that staircase behaviour typically disappears almost entirely.

@OldSailor makes a fair point about the coulomb counting, but worth adding — the "15% is almost empty" thing is a separate issue. The Pytes V5 ships with default low voltage cutoff set quite conservatively, so by the time the BMS reports 15% it's already pulled more capacity than the SOC suggests.

Had something similar on my motorhome build. Fixed it by letting the Victron MPPT handle a proper bulk → absorption → float cycle a few times over a week or so. SOC reporting tightened up noticeably after that.

Also check your Cerbo / GX device isn't overriding the battery's own SOC signal — had that cause confusing readings too.