Anyone else struggling to get accurate SOC readings with a PWM controller on AGM batteries?

Been having a nightmare lately with my setup — 200W of panels (two 100W units in parallel) feeding into a cheap 30A PWM controller, charging a pair of 100Ah AGM batteries wired in parallel. The controller's display reckons I'm at 100% SOC by about 11am on a decent day, but when I actually measure resting voltage after a couple of hours off load it's sitting around 12.5V, which to my understanding is more like 75-80% full. Something's clearly off.

I've read that PWM controllers are notoriously bad at accurate SOC estimation because they're just reading voltage during or just after a charge cycle, which is always going to be elevated. But I'm wondering if anyone has actually solved this rather than just accepting duff readings. I did look at adding a standalone battery monitor (something like a Victron BMV-712), but I want to make sure that'll genuinely give me useful data even if the controller itself is still a bit rubbish.

Has anyone run a BMV or similar shunt-based monitor alongside a basic PWM controller and found it worthwhile? And is there anything else I should be checking — battery health, wiring, anything like that — before I start throwing money at the problem?

@KeithClark71 this is exactly the problem I ran into with my shepherd's hut build. PWM controllers are notoriously poor at accurately reporting SOC because they're not actually measuring it properly — they're largely guessing based on voltage, which is unreliable especially with AGMs that have a fairly flat discharge curve.

The real fix for me was adding a dedicated battery monitor rather than relying on the controller's readout. A Victron BMV-712 transformed everything — it measures actual coulombs in and out rather than just voltage.

Worth checking also whether your AGMs are ever reaching a proper absorption stage — PWM controllers often struggle to fully charge AGMs, which means your batteries are perpetually in a partial state of charge, making SOC readings even more meaningless.

What controller are you running exactly? Some have worse shunt monitoring than others.

Hey @KeithClark71, the root issue is that PWM controllers use pulse-width modulation to "approximate" a full charge rather than properly topping the batteries off — your AGMs never quite reach 100% SOC, so any voltage-based reading is already working from a skewed baseline. A decent battery monitor with a shunt (Victron BMV-712 is the gold standard but pricier options exist) will give you far more accurate coulomb-counting than relying on the controller's display alone. Also worth checking your battery temperature — AGMs are particularly voltage-sensitive and a cold UK winter will throw readings right off. Long term you'd genuinely benefit from an MPPT upgrade, but a shunt monitor is the quickest win without replacing your whole setup.

Hey @KeithClark71, worth adding to what @AlanWard and @CraftyWelder have said — AGMs are particularly unforgiving here because they need that precise absorption stage to properly top up. What I'd suggest in the short term is grabbing a dedicated battery monitor like a Victron BMV-712 or even the cheaper Votronic units. They use coulomb counting rather than relying on the controller's voltage readings, so you'll get a far more honest picture of your actual SOC regardless of what the PWM is doing. Longer term, your setup would genuinely benefit from an MPPT upgrade — you'd also squeeze noticeably more out of those panels, especially on grey UK days. But the monitor alone will at least stop you flying blind in the meantime. What voltage readings are you currently seeing at the battery terminals during bulk charge?

Been through exactly this on the boat. The thing nobody mentions is that surface charge makes it worse — your BMS or controller sees a temporarily elevated voltage after a charge cycle and thinks you're at 95%+ when you're really nowhere near it.

Rest the batteries for a couple of hours with no load and no charge, then check voltage. You'll likely get a sobering reading.

Longer term though, even a decent standalone shunt-based monitor like a Victron BMV-712 will only give you accurate SOC if the charge quality is actually there in the first place. Rubbish in, rubbish out.

@CraftyWelder is right about the fundamental PWM limitation — I swapped to an MPPT on my narrowboat setup and the difference in how the batteries actually felt (and lasted) was immediate.

Great thread, this. One thing I'd add that nobody's touched on yet — if you want a more reliable SOC reading without upgrading to MPPT straight away, invest in a decent battery monitor with a shunt (Victron BMV-712 is the go-to, but the Renogy one is cheaper and does the job). It measures actual current in and out rather than relying on voltage alone, so you're not fooled by surface charge or PWM's choppy output.

Mount the shunt on the negative terminal and make sure all loads and charge sources go through it — people often miss connecting the solar controller's negative through it and then wonder why the readings are still off.

Won't fix the underlying charging limitations @CraftyWelder mentioned, but at least you'll know what's actually going on. 🔋

Great points from everyone here. One thing worth mentioning — have you checked your battery interconnects, @KeithClark71? With two 100Ah AGMs in parallel, poor or mismatched cable lengths between them can cause uneven charging, which throws your SOC readings out even further because one battery ends up doing more work than the other. Ideally both interconnect cables should be the same length and gauge. Also, your PWM controller almost certainly won't have proper temperature compensation unless you've got a remote sensor attached — AGM voltage thresholds shift noticeably with temperature, so if it's sitting in a cold garage or shed your controller could be cutting off too early and reporting a false full charge. Worth double-checking both of those before spending money on anything else.