Anyone else struggling to get accurate SOC readings with a cheap shunt on a 200Ah LiFePO4 bank?

So I picked up a Daly BMS (100A, 4S) and a no-name 500A/75mV shunt off eBay a few weeks back to monitor my 200Ah LiFePO4 build in the van. The shunt feeds into a Victron BMV-712, which I know is a decent bit of kit, but I'm getting some really inconsistent state of charge readings. It'll say 80% then drop to 65% within an hour of light use — a 12V compressor fridge and a couple of USB loads, nothing mad.

I've set the charged voltage to 14.2V (I'm using a Renogy 40A MPPT and a B2B charger from the alternator), Peukert exponent at 1.05, and charge efficiency at 99% as recommended for lithium. The battery hits 14.2V fine and the BMV syncs, but then the drift starts. I'm wondering if the issue is the cheap shunt itself — maybe the resistance isn't accurate enough, or there's some resistance in the cable connections throwing the coulomb counting off.

Has anyone actually measured the resistance of a budget shunt against the spec, or compared readings against a known-good meter? Is it even worth swapping it for a proper Victron SmartShunt, or am I barking up the wrong tree and this is more likely a wiring/connection issue?

JasonPhillips | 847 posts

@Brummie77 The flat voltage curve on LiFePO4 is the main culprit here — SOC estimation via coulomb counting lives or dies by your calibration points. A few things worth checking: first, that no-name shunt may have a tolerance way off the stated 75mV. Grab a decent multimeter and verify the actual shunt resistance against spec. Even a small deviation compounds badly over time.

Also, are you certain all your loads and charge sources are wired after the shunt? Any parasitic draw bypassing it will throw the BMV's counting off completely. The Daly BMS itself can sometimes draw a sneaky few milliamps that goes unaccounted.

Set your Peukert exponent to around 1.05 for LiFePO4 in the BMV settings if you haven't already — the default is calibrated more for lead-acid.

MultiPlus_Queen | 203 posts

@Brummie77 Had similar grief on my narrowboat with a cheap shunt before switching to a proper Victron one. The issue I found was the cheap shunt's resistance wasn't actually the rated value, so the BMV-712 was calculating current slightly wrong — tiny errors that compound massively over time on LiFePO4's flat curve.

Worth checking: is your shunt actually 75mV at 500A? A lot of eBay ones are wildly out of spec. A decent multimeter and a known load can give you a rough sense.

Also, are you setting the Peukert exponent close to 1.05 in the BMV settings? LiFePO4 behaves very differently to lead-acid there. The default BMV values are tuned for AGM/gel, so it's worth dialling that in properly.

HappyBodger | 412 posts

@Brummie77 One thing worth checking that nobody's mentioned yet — are you certain your shunt is properly isolated from the chassis? A rogue earth path can bleed current that completely bypasses the shunt, making your BMV-712 think you're losing less than you actually are. Also, that 500A/75mV shunt might be slightly out of spec if it's a no-name unit — the BMV needs accurate millivolt readings to calculate correctly. Worth borrowing a decent multimeter and measuring the actual voltage drop across the shunt under a known load to verify it's reading what it should be. A small calibration error compounds horribly over time with coulomb counting. The BMV has a shunt current setting in the menus — double-check you've got it set to match your actual shunt rating rather than leaving it on the default 500A/50mV.

BordersSolar | 156 posts

@Brummie77 Worth double-checking your BMV-712's charged voltage and tail current settings — these are what tell it "right, that's 100%, reset the counter." If those are off even slightly, your SOC drift will compound over every charge cycle. For LiFePO4 I'd suggest something like 14.2V charged voltage threshold with a tail current around 2-4% of your bank capacity (so roughly 4-8A on a 200Ah bank). Also, have you verified the shunt is the only path for current? Any direct connections bypassing it — even a small earth somewhere — will throw your readings right off. @HappyBodger sounds like they might be heading somewhere similar on that last point!

HelenThompson | 89 posts

Had exactly this on my narrowboat build. The issue for me was the shunt placement — everything must go through it, literally everything. I had my bilge pump wired direct to the negative busbar, bypassing the shunt entirely, so the BMV-712 never knew it was running. SOC was drifting by 15-20% over a week.

Also worth knowing — cheap shunts often aren't actually 500A/75mV as labelled. Resistance can be way off spec. Fogstar sell decent pre-calibrated ones for not much more than eBay tat, and at least you know what you're getting.

If all else fails, the BMV-712 has a Peukert exponent setting that's often overlooked on LiFePO4 — should be set to 1.05, not the default.

Derek1977 | 847 posts

@Brummie77 Something I haven't seen mentioned yet — LiFePO4's notoriously flat voltage curve makes coulomb counting absolutely critical, so any resistance in your shunt connections will throw things off badly. Those cheap eBay shunts often have poorly finished surfaces on the copper bars. I'd get a bit of fine wet-and-dry on the contact faces and make sure your sense wire terminals are torqued properly rather than just finger-tight.

Also worth checking: is everything drawing from the battery running through that shunt? Fridge, lighting, any trickle loads — if anything bypasses it the BMV-712 simply can't account for those electrons leaving the bank. Classic cause of drift that gradually gets worse over time.

What percentage error are you typically seeing, and does it get worse the longer you go between full charges?