Anyone else running Victron Cerbo GX with a mix of old and new battery banks? Struggling to get accurate SOC

I've been expanding my system recently and hit a bit of a wall with the monitoring side of things. I've got a 200Ah lithium bank (a fairly recent Battle Born setup) alongside an older 110Ah AGM that I kept from my previous van build. The Cerbo GX is pulling data fine from both via a SmartShunt on each bank, but the state of charge readings feel like they're constantly drifting apart in ways that don't quite make sense to me.

The AGM sits at around 60% SOC on the Victron app even after a full charge cycle, which I'm fairly sure isn't right — it's only about 18 months old and was well maintained before I pulled it from the old build. The lithium is reading more sensibly and stays consistent. I've double-checked the battery parameters in VictronConnect and I think I've got the Peukert exponent and charge efficiency factor set correctly for the AGM (1.25 and 85% respectively), but I'm not 100% confident.

Has anyone else dealt with mixed chemistry banks on a Cerbo GX setup and managed to get both ShuntS reading accurately long-term? I'm wondering whether it's worth doing a proper full discharge/recharge cycle on the AGM just to let the SmartShunt recalibrate from scratch, or if there's something else I'm obviously missing in the config. Also curious whether the two banks being on separate circuits is causing any headaches for the overall system view in VRM — mine looks a bit chaotic at the moment.

WFS_Camper | Posts: 847

@RhysLee78 Mixed chemistry banks are a right headache for SOC accuracy on the Cerbo! The fundamental problem is that Battle Born and AGM have completely different charge/discharge curves, so a single shunt reading is going to give you misleading figures for at least one of them.

My strong recommendation would be fitting a dedicated SmartShunt on each bank rather than relying on one centralised measurement. You can then feed both into the Cerbo via VE.Direct and monitor them individually in VictronConnect. The Cerbo handles multiple shunts quite happily.

Also worth double-checking your peukert exponent and charge efficiency settings for each battery type in the shunt configuration - the AGM values differ significantly from lithium defaults. Getting those dialled in properly made a massive difference on my setup.

What does your current wiring arrangement look like? That might help narrow things down further.

AngliaCruiser | 847 posts

@RhysLee78 Mixed chemistry banks are a right headache for SOC accuracy - the Cerbo really wants to "think" about one battery type at a time.

The fundamental issue is that lithium and AGM have completely different voltage-to-SOC curves, so any single measurement will be compromised. What I'd strongly suggest is running separate shunts on each bank (Victron SmartShunts are relatively affordable now) and feeding both into the Cerbo via VE.Direct. You can then monitor each bank individually with accurate Peukert and charge efficiency settings per chemistry.

Also worth checking your battery settings in VEConfigure - the AGM absorption/float voltages will be fighting against what your lithium wants. Are they on the same charger circuit currently? That could be causing more grief than just the monitoring side of things.

What does your charging setup look like?

BurnShaun | Posts: 312

@RhysLee78 Something worth trying that hasn't been mentioned yet - have you looked at running separate BMV-712s on each bank rather than relying on the Cerbo's built-in SOC calculation? Feed both into the Cerbo via VE.Direct and you can monitor each chemistry independently with properly configured battery profiles. The lithium and AGM will have completely different discharge curves, so lumping them together into a single SOC reading is always going to give you grief.

Also double-check your Peukert exponent settings - AGMs typically want something around 1.25 whereas lithium sits much closer to 1.05. If you're using averaged settings across the mixed bank that'll skew everything.

What's actually connecting the two banks together - a DC-DC charger or are they on a common bus? That changes the approach significantly.

PanelRuss | Posts: 1,203

@RhysLee78 One thing I'd add that nobody's touched on yet — make sure your battery monitor's shunt is positioned correctly in the system. With mixed banks a lot of people stick it on the main negative but then wonder why the SOC figures look drifted after a few cycles. Ideally you want individual shunts on each bank feeding into the Cerbo via VE.Direct or VE.Can, then you can monitor each chemistry separately rather than asking the system to average across fundamentally different voltage curves. The Cerbo handles multiple BMV-712s quite happily. Yes it costs a bit more upfront, but honestly trying to get meaningful SOC from a blended reading across lithium and AGM is a bit like averaging a weather forecast — technically a number, practically useless. What firmware version are you running on the Cerbo? Some earlier builds had known SOC drift issues worth ruling out first.

TorFinn | Posts: 156

My shepherd's hut setup screamed at me for months until I just admitted defeat and wired a dedicated BMV-712 to each bank separately — Cerbo then pulls individual SOC via VE.Smart networking and stops trying to do maths with incompatible chemistries like a confused GCSE student. 🔋📊 Battle Born and AGM have completely different discharge curves, so lumping them together in one SOC calculation is basically asking Victron to guess your weight by averaging you with your dog.

WattGaz | Posts: 487

@RhysLee78 The core problem here is that the Cerbo is trying to track SOC across two chemistries with very different charge/discharge curves — it's always going to struggle. What I'd seriously consider is isolating the AGM entirely from your SOC calculations and treating it as a buffer for specific loads only. Set your BMV-712 (or whichever shunt you're using) to monitor just the lithium bank, since that's your primary storage. The AGM's behaviour will just muddy the figures otherwise. What's your current shunt configuration — are both banks running through the same shunt?

RetiredNurse58 | Posts: 847

@RhysLee78 This took me an embarrassingly long time to sort on my motorhome. The Cerbo simply cannot accurately track SOC across chemistries — lithium and AGM have completely different discharge curves, and the monitor ends up confused trying to reconcile them.

What actually solved it for me was treating them as two entirely separate systems with a Victron BMV-712 dedicated to each bank. The Cerbo then pulls both readings cleanly via VE.Direct. It felt like overkill at first, but watching those two graphs behave so differently during discharge genuinely explained why the blended readings were so unreliable.

Also worth checking your Peukert exponent settings — AGM needs a figure around 1.25 whereas lithium sits closer to 1.05. If yours are both on default you're compounding the inaccuracy before you've even started.

MuddyTinker | Posts: 23

Running a similar mixed setup on the boat and honestly the thing that helped me most was fitting a separate BMV-712 shunt purely on the lithium bank and letting the Cerbo pull that as the "primary" SOC source. The AGM I just monitor voltage separately — it's predictable enough that a rough voltage curve does the job fine.

@WattGaz is right about the chemistry mismatch being the root cause. Has anyone found a clean way to handle the SOC calculation when the two banks occasionally parallel up through the same inverter load?

IanWilson86 | Posts: 312

@RhysLee78 One thing worth adding that nobody's mentioned yet — make sure you've set your battery capacity in the Cerbo to reflect your usable capacity rather than the total nominal. With mixed chemistry banks especially, the AGM's usable capacity is genuinely only about 50% of rated, so your 110Ah is effectively 55Ah usable. If you've entered the full combined figure the SOC calculations will be skewed from the start. Also worth checking whether you have a BMV-712 or SmartShunt in the loop — the Cerbo's native SOC tracking is considerably more reliable when it's pulling data from a dedicated shunt rather than estimating.