I have a Multiplus 2 / Smartshunt / RS 450-100 Charge Controller..
I hve 3x JK inverter BMS units patched into my Cerbo. All Cmmunicates and works well.
Battery banks are 3x 16S EV MB31 cells.
no grid - all solar(Heaps of Solar)
I have a issue where all 3x battery banks stop charging at a certain point - which seems to be full - yet the battery packs show 90% / 96% / 86% (Each one) - all differing levels ?!?!.
The batteries seem to tell the Victron to stop charging – even though its not 100 perecent or atleast the BMS on JK doesnt say so..
For the life of me, i cant sem to work out why.
JK internal SoC calculations are notoriously inaccurate for Li-FePO4. Doesn’t even come close that of a SmartBMV.
In our 12S Li-NCM system the SmartBMV settings are simply:
Charged voltage 50.3V
Discharge floor 20%
Tail current 2%
Charge detection time 15m
Peukert exponent 1.00
Charge efficiency factor 100%
Current threshold 0.00A
…
A typical full cycle runs from 50.4V to 41.8V (20% SoC) and back to 50.4V
At the end of charging, the BMV correctly measures back up to 99.9 or 100.0% SoC right before synchronizing (if given the time before charging ends). The consumed Ah’s track even more accurately.
That is near perfect accuracy and reflects the lithiums optimal cycle efficiency of >= 99.9x%. But only works if any cable/busbar losses between battery terminals and shunt are fully eliminated (big copper mm^2).
Not sure if Li-FePo4 has same near perfect cycle efficiency but that can be measured with the BMV.
…
To address your issue: only trust the BMV coulomb meter and if possible try to set the JK’s to synchronize to 100% more often.
…
Also, try to avoid obsessive balancing, with well balanced good quality cells should not need nearly as much balancing as seems to be expected by many.
Your BMS start balancing the batteries at 3.44V, what I think could be a little less (e.g.: 3.4V). However, it has a trigger voltage difference of 0.010V, right?
If so, I think it triggers too much re-balancing and, if not kept close to 100% SOC for balancing for a long time, they will never get balanced. If this is the case I think you will always face different SOC as the cells will not have, from each other, a close internal resistance.
By other hand, I use to have the same problem in my set that is made of 2 x 16kW batteries. I could get some improvements when I close the DC in a loop by connecting an extra cable from the second battery back to the first one. Look:
The two cables you see at the left side did not exist before. Now the current in between the two batteries flows via these new cables that is a better current path then going up to the Multiplus to reach the other one.
I concur the best place to start is the wiring. Make it as symmetrical and short (at least up to BMV) as possible. Next you could charge all batteries to full and make sure they are well balanced, <10mV difference at > 4.45V cell. Takes time but worth it.
Then do a full cycle to <3.0V and back to 4.45V without balancing to see if the JK’s track SoC correctly. This might not be the case, especially at low current charge/discharge where JK just does not count coulombs, cutoff around 0.2A I believe (and BMV simply counts them all).
Once you got to know how that all adds up over that cycle, you can try changing JK params to synchronize to 100% sooner. Or alternatively, if the batteries behave well in parallel, make the BMV your primary SoC source, not the BMS’s. The BMS’s will still protect the batteries (in a good system never need to turn them off) but will not tell the MP/Cerbo? when to start /stop charging anymore. It’s unfortunate but without accurate coulomb counting by the BMS, all the fancy extra functionality becomes pretty useless IMHO.
Did you build your own battery by any chance? In that case, and assuming a single JK carries enough current in total, did you consider building a single 3P16S battery instead?
Nevermind, I see those 3 rackmounted batteries. Do consider beefing up those cables and make it all equal length. Or go wild and make two ridiculously oversized fixed busbars to fit directly to positive and negative battery poles respectively. Especially with Li-FePo4’s flat Voltage/SoC characteristics you want to minimize resistance differences in the parallelized power paths of all connected batteries to an absolute minimum. Think short length 95mm^2 minimum, or 20mm x 50mm busbar. Nothing to do with current, only to drive any voltage differences under load way below a single mV.
This also helps with long-term durability as it will ensure all batteries see the exact same use over there lifespan, and age equally.
See also this excellent paper from orionbms (I have no ties with them): https://www.orionbms.com/manuals/pdf/parallel_strings.pdf
There are many, many people “talking about it”, like me. However, we can learn about some key points about LiFePO4 batteries:
Then make sure your balancing is ok. Your sets may force a long balancing time that, if your PV does not have, it may hold your batteries always unbalanced.
Also, I would make sure the cable sizes are correct and that you follow the recommendations on the first link above.
