Victron Mp2 with Pytes battery charging behavior

Been having a nosey at my charging behaviour lately and it's got me thinking — curious if anyone else has dug into this with a Victron + lithium combo.

Running a Multiplus-II in the motorhome (single unit, not three like some of you lucky lot 😄) paired with Fogstar Drift cells and a third-party BMS talking to the MP2 via DVCC. What I've noticed is the CVL the BMS broadcasts can shift around a bit near the top of charge, and the MP2 seems to follow it pretty obediently — which is the whole point of DVCC I suppose, but it's interesting to actually watch it happen in VRM.

Specifically wondering about the absorption/float transition. My BMS will nudge the CVL down slightly once it's happy the cells are balanced, and the MP2 backs off accordingly. Seems sensible but I wasn't expecting it to be that dynamic.

For those running Pytes or similar rack batteries with the built-in BMS — are you seeing the CVL fluctuate much? Does the MP2 chase it cleanly or is there any lag/hunting going on?

Also — anyone had issues where the BMS sends a lower CVL than expected and the batteries never quite reach full SOC? Had a weird week where mine was topping out at around 95% and couldn't work out if it was the BMS being cautious or something misconfigured in VictronConnect.

Would be good to compare notes, especially from anyone on a bigger multi-inverter setup where the behaviour might differ.

@Davo83 Pytes talk CANBUS to the MP2 so if you've got the cable sorted and DVCC enabled in VictronConnect, the battery should be bossing the charge parameters itself — your MP2 is essentially just a well-paid lackey at that point. 🔋

Check VRM portal if you're logging data; the DVCC screen will show you whether the BMS is actually handing over CCL/DCL values or whether your MP2 is still freelancing with its own settings like it owns the place.

Classic symptom of it not working properly is absorption voltage sitting stubbornly at whatever you've manually keyed in rather than what the Pytes wants — usually 53.2V for a 16s pack.

@ROW_OffGrid has the DVCC side covered well 👍

One thing worth checking if you're seeing odd charge curves — make sure your ESS assistant (if you're running one) isn't fighting with the BMS CVL/CCL values. Had this exact headache in my van build last year.

Also pop into VRM portal if you're logging data, the DVCC page shows exactly what the battery is requesting vs what the MP2 is actually doing. Really handy for spotting where things go sideways.

The Pytes batteries seem pretty solid from what I've seen on here — should play nicely once the handshake is dialled in properly. Don't be tempted to override with manual charge settings, let the BMS do its thing!

Good thread this. One thing I've been curious about with the Pytes + MP2 combo — does the battery ever request a lower charge current via CANBUS when it's balancing cells, or does it just cap voltage and let the MP2 figure it out?

I ask because on my setup (different chemistry, Fogstar Drift) I can watch the current requests fluctuate quite noticeably in VRM when cells are diverging. Wondering if the Pytes BMS is similarly granular with its CANBUS messaging or whether it's more blunt about it.

Also @Davo83 — are you monitoring this through VRM or just VictronConnect locally? VRM's advanced widgets make it much easier to spot patterns over a few charge cycles rather than trying to catch it in real time.

@WattLiz yes! The Pytes will throttle charge current down via CANBUS when it's feeling thermally grumpy or approaching full — watched mine do exactly that on a warm day, MP2 just obediently backed off like a good lad. 🫡

Worth keeping an eye on the DVCC charge current limit readout in VRM if you want to catch it happening in real time — proper satisfying when you see the battery actually bossing the charger around rather than the other way round.

Had a few head-scratching moments on the narrowboat before I realised the battery was essentially whispering "oi, slow down" and the MP2 was listening perfectly — system working as intended, not a fault.