Anyone else running Victron Cerbo GX with a mix of MPPT controllers — how are you managing state of charge drift?

I've got a 48V system with two SmartSolar MPPT 150/35s and a 200Ah lithium bank (Gobel cells, 16S). The Cerbo is pulling data from both controllers and my MultiPlus-II 48/3000 over VE.Can and VE.Direct, but after a few days of partial cycling I'm noticing the SOC readout on VRM creeping away from reality — sometimes 8-10% out by the end of the week.

I know the BMS should technically be the source of truth, but mine (a JK BMS) only connects via Bluetooth and there's no clean way to feed that into the Cerbo without something like a Raspberry Pi running a bridge script. I've had a poke around on the Victron community forum and there are a few RS485 adaptors people have bodged in, but nothing that looks properly reliable long-term. Wondering if anyone here has cracked it or just accepts the Cerbo's own coulomb counting with regular full-charge resets to keep things honest.

Has anyone swapped to a BMS that talks VE.Can natively, like a Batrium or REC-BMS, and actually found the SOC accuracy worth the extra cost? Or is tweaking the charge/discharge efficiency figures in the Cerbo settings enough to tighten things up without going down that rabbit hole?

Hey @BazKnight57, one thing that made a big difference for me was making sure the battery capacity and Peukert settings in VEConfigure are spot on for your specific cells — Gobel LFP behaves quite differently from the defaults. Also, are you using a BMV-712 or SmartShunt? The Cerbo's SoC estimate is far more reliable when it's slaved to a dedicated coulomb counter rather than relying on the MPPTs' own calculations. Without a shunt as the "trusted" source, you'll get drift creeping in, especially across two controllers pulling charge simultaneously. Set the shunt as your primary SoC source in the DVCC settings and let the Cerbo arbitrate from there. Sorted mine out completely once I did that.

@BazKnight57 the real culprit is usually having your DVCC settings not quite right — if Shared Voltage Sense and Shared Current Sense are both enabled on the Cerbo, it'll consolidate everything into one proper SoC calculation rather than each controller doing its own maths and confusing the poor thing 🔧 also worth checking your battery monitor is set to the MultiPlus-II rather than defaulting to one of the MPPTs, because the inverter/charger sees the full picture including loads whereas the solar controllers are just watching their own slice of the current.

@BazKnight57 had the same drift nonsense in my shepherd's hut until I wired a dedicated BMV-712 as the sole SoC source and told the Cerbo to trust that rather than trying to average across two MPPTs having a maths argument — Fogstar cells are dead honest once you've got a single source of truth doing the counting rather than a committee.

@BazKnight57 had the same drift issue on my 48V setup — what actually cracked it for me was forcing the system to do a proper full charge cycle to let the Cerbo resync its SoC. If your cells never quite hit absorption properly, the % just wanders further and further off over time.

Also worth checking: are both MPPTs on the same VE.Can network with identical charge profiles? Even a small voltage offset between them confuses the SoC calculation. I use VictronConnect to verify they're truly in sync rather than just assuming the Cerbo has sorted it.

One more thing — tail current threshold. Mine was set too high so it kept "finishing" charge early. Dropping it to around 1-2% of battery capacity made a noticeable difference to accuracy.

@BatteryPaula is on the right track with the BMV-712, but worth adding — make sure you've got the BMV set as the battery monitor in the Cerbo's Device List, not just installed and forgotten about. Caught me out for weeks on my narrowboat setup. The Cerbo will happily keep averaging SoC from the MPPTs instead if you don't explicitly tell it which device to trust.

Also check your battery capacity and Peukert settings in the BMV itself — if those are off, even a "correct" setup will drift over time. I'm running Fogstar Drift cells and had to tweak the charge efficiency factor down slightly from default to get stable readings.

DVCC with SVS enabled as @NaeClue mentions is still worth doing alongside, they complement each other rather than being alternatives.

Good shout from @GafferTapeKing19 and @BatteryPaula on the BMV-712 route. One thing I'd add — with Gobel cells specifically, make sure your battery capacity setting in the BMV matches your actual usable capacity rather than the nameplate figure. I found on my own 16S build that the Peukert exponent also needed tweaking; lithium is much flatter than lead, so you want that set close to 1.05 rather than the default. Also double-check your charged voltage and tail current thresholds in the BMV settings — if they're too conservative the unit never registers a proper full charge synchronisation point and the drift just accumulates over weeks. Once I nailed those two things the SoC on my Cerbo stayed rock solid. What BMS are you running with those Gobels? That can complicate things if it's also reporting SoC to the Cerbo and creating conflicts.

Something nobody's mentioned yet — double-check your battery capacity setting in the BMV-712 is accurate to your actual usable capacity rather than nominal. With 16S Gobel cells I found the real usable figure was slightly lower than spec once I'd accounted for my BMS cutoffs, and even a small discrepancy there compounds the drift over time. Also worth enabling synchronised charging in VRM so the Cerbo gets a consistent reference point when both MPPTs hit absorption simultaneously. @Nobby30 makes good points about Gobel-specific settings — what peukert exponent are you running? I dropped mine to 1.05 on lithium and it made a noticeable difference to accuracy.

Had the same drift issue on my static with a similar mixed MPPT setup. One thing worth checking that nobody's mentioned — your charge efficiency factor in the BMV settings. Default is 99% but with lithium it's often better nudging it up to 99.5% or even keeping a closer eye on it over a few full cycles and adjusting accordingly.

Also, are both your MPPTs on VE.Direct or VE.Can? I found one of mine wasn't being properly recognised until I checked the network priority settings in the Cerbo — easy to miss.