Inaccurate SOC for JK-PB2A16S-20P using 16 x EVE LF280K cells

[spikebyte]

Hello all,
First time poster! This has been a great forum and I've learned a ton.

I have two 48V LFP packs configured as 2PS16 using EVE LF280K cells, each configured with JK BMS (jk-bd6a20s10p). I had been running them as independent batteries tied together in parallel, but without any communication between them, connected to 2 x Sungoldpower SP6548 6.5kW inverters configured in parallel split-phase configuration (2P1/2).

I've been running this solar setup since 2019. It's worked fairly well, but since I've had no communication between the packs, and the inverters, I had to charge/discharge them using voltage, which is obviously not ideal with the flat-curve the LVP cells have. It's been tough to change the depth of discharge seasonally, hence why I finally was ready to start getting SOC average of both packs.

I recently upgraded my BMS on both of my packs to the JK-PB2A16S-20P. Before install, thanks to @Off-Grid-Garage 's absolutely awesome tutorials and settings, upgraded my BMS firmware to V15.41. I installed them over the weekend, but immediately noticed that the SOC calculation is way off. One of the packs was around 96% prior to replacement of the BMS, but after using the latest settings (under JK Inverter BMS JK-PBxx) on @Off-Grid-Garage https://off-grid-garage.com/my-settings/ page. After applying those settings, the first pack read 70%. Similarly after upgrading the second pack, got a similar reading. I had hoped that after a couple of days, it would charge fully, but they didn't.

So here's my issues:
1. SOC is reporting totally inaccurately
2. BMS communication between packs via RS485 shows "Bulk charging", but only 1 pack at address 0 is showup up in the JK Monitor software (v2.7).
3. My inverters, which uses RS485 for BMS communication, using the PYLON settings, times out after 3 minutes and gives a error code 61.

Let me detail my setup for clarity.


Using the included cables with the inverter (INV -> BAT) with the appropriate pins (3,5 -> 1,2) connected to the right set of ports, in port 1, or RS485-2 in the above diagram. I was able to do the firmware upgrade, so I know that I'm getting good communication with each BMS.

Both BMSs are configured as such: Master -> 0 (DIP 0000), Slave -> 1 (1000). The Inverter is plugged in with the above cable in RS485-1 with the cable mentioned above. The laptop to monitor things is connected to RS485-2, then the slave is connected from the master's RS4852 (right-port) to the slave's RS485-2 (left-port)

For inverter communication, the (INV -> BAT) is connected, as mentioned above, but only connected to the 2P1 inverter in the Battery RS485 port on the left. The black cable is connected to an OrangePI running Solar assistant.

The batteries negatives are connected from the negative to the BMS, BMS is connected to a bus connection, then to a Victron SmartShunt then to each inverter. The positives are connected from the positive to 200A breakers (one for each pack) then direct then a 1+2,1,2 DC switch, then split to each inverter.

Here's the master:

Slave:

PC connection:

Battery that was ~90% IIRC, but this was SOC after upgrade:


This is the error code and settings:



I tried today to validate that the communication between batteries were working, but the JK Monitor software is only able to see the first battery, no second.

As you can see in my software, only Device 0 shows.


I've attached PDFs for my inverter and EVE cells.

Any help or advice would be much appreciated! I can provide my old BMS configurations if needed. Thanks in advance!

 

[AshleyL]

......quick question....have you calibrate both JKBMS voltage and amperage yet?

 

[LakeHouse]

I've been running this solar setup since 2019. It's worked fairly well, but since I've had no communication between the packs, and the inverters, I had to charge/discharge them using voltage, which is obviously not ideal with the flat-curve the LVP cells have. It's been tough to change the depth of discharge seasonally, hence why I finally was ready to start getting SOC average of both packs.

What exactly was not working about this for the last six years? Why do you think you need to change the depth of discharge?
I'm becoming more and more convinced every day that no one selling solar equipment has ever bothered to put any thought at all into whether there's any benefit to having batteries communicate with inverters.

I can't really help with your issue of being able to communicate with only one BMS, but can help with the SoC issue: The BMS's are displaying an inaccurate SoC because when you first hook them up they just take a guess (you said it was 90%, they guessed 70%, so they didn't do a terrible job), and then track SoC based on current in and current out. All you need to do is charge them to 100% (actually to 100%, based on voltage, not tracked SoC) and they should be fine. If your battery-inverter comms are making this impossible (i.e., the inverter won't keep charging because the battery tells it not to), just unplug the comms, fully charge the batteries using the old-school open-loop method, then plug the comms back in.

 

[spikebyte]

......quick question....have you calibrate both JKBMS voltage and amperage yet?

I calibrated voltage directly after install. I did not calibrate amps as I figured 0A might work, but I guess it would make sense to adjust it. I could try that. Good idea!

 

[spikebyte]

What exactly was not working about this for the last six years?

Fair point. I tend to trust Andy's numbers better than mine usually. Sorry, I've had my built solar setup in all forms for 6 years. I've been rocking this particular setup for the last 3 or so years. But totally reasonable points...

Why do you think you need to change the depth of discharge?

I've been playing with those numbers here and there. I've been messing with the tradeoffs between $$ savings and longevity of the batteries. I think I'd want to charge to 100%, then discharge down to 10% max.

I'm becoming more and more convinced every day that no one selling solar equipment has ever bothered to put any thought at all into whether there's any benefit to having batteries communicate with inverters.

100%. It's extremely frustrating... Considering how common Li-ion + LFP are these days, you think they'd work a little harder to get CAN/RS485 standardized...

I can't really help with your issue of being able to communicate with only one BMS, but can help with the SoC issue: The BMS's are displaying an inaccurate SoC because when you first hook them up they just take a guess (you said it was 90%, they guessed 70%, so they didn't do a terrible job), and then track SoC based on current in and current out. All you need to do is charge them to 100% (actually to 100%, based on voltage, not tracked SoC) and they should be fine. If your battery-inverter comms are making this impossible (i.e., the inverter won't keep charging because the battery tells it not to), just unplug the comms, fully charge the batteries using the old-school open-loop method, then plug the comms back in.

So that's actually an issue I didn't really clearly mention. I've been attempting to get them to charge up under voltage-based ("USER" settings in the inverter), but they won't charge past around 80%, so I'm not sure what's going on there. I've set the float charge voltage up to 57.6V and it still won't get them to 100%. Any advice here? You think maybe calibrating Amps may fix?

 

[kommando]

The amps calibration is fairly useless, the shunt on the BMS board is a cheap low quality one so calibrate at 10A and its only accurate at 10A. Couple that with its inability to measure low amps at all, which results in SOC drift, and you are better with an external shunt.

SOC drift is partly fixed in Firmware 15.35 onwards, but its a software fudge.

 

[spikebyte]

In the late afternoon, each pack draws 30+ amps, so I could try to calibrate it at that point. But that was my understanding as well that it likely wouldn't make a huge different to calibrate that. Voltage however should be fine to calibrate, and I did that before connecting the load. Values were stable too at the time.

Like I said, I have the smart-shunt in-line to track overall SOC, but TMK there's no way to hook that up to my inverters. Since I'm not using Victron (I'd love to get them, but dont have the funds right now), I don't think I can use that with my inverters currently.

 

[LakeHouse]

You think maybe calibrating Amps may fix?

No, this won't change anything.

Any advice here?

I use the non-inverter JK BMS and I've never had a problem with it resetting to 100% on full charge. My best guess would be to check this setting, and make sure it's low enough that you're exceeding it when you charge to (what should be) 100%:

 

[spikebyte]

Hrm. That's what I have set...

 

[frankrudis65]

Hi there,as it was said before changing something which used to work is always a gamble!I have deliberately gone without any communication between bms and inverter for two reasons: to keep things simple,especially in parallel situations and 2: To have two independent setup controlling and protecting things.The inverter does the charging and discharging and the bms thinks about the protection of the cells and balancing them.
Calculating the SOC is by far not a simple task and at its best will give you always an idea what's happening,unless your equipment gets calibrated each time before measuring.Unless you have got some sort of smart shunts(the victron shunts are very precise) and occasionally go to a complete charge and a reasonable discharge,you will never get some reliable statistics.
Back to your actual problem: Try with some different bms settings: first of all charging to 3.45V anything above is stressing the batteries without any margin to have much more capacity! Discharging to less then 2.95V or 3.00 V the same here: there is not much to gain going lower!
Therefore Cell UVP 2.95V Cell UVPR3.00V SOC 0%= 3V SOC 100%= 3.45V and cell balancing start at 3.44-3.45 V depend how long your bulk charge is...Float I would keep at 3.40V max if possible a little lower....Now you can play a little with the battery capacity settings: you might set this value a little bit higher,until you get the SOC fairly correct. Don't be afraid of trying things out until you keep the safe side of the game(charging and discharging).Also be aware of the Ampere you're charging :there's no need to have your bank charged at lunch!Give everything some time...
Think about getting at least 1 smart shunt to be able to compair and calibrate the other equipment....
I've got my system after months of trial and error so far that 1: The Soc stated is very precise 2: all equipment show the same Soc.
Keep in mind that an occasional shutdown of the bms's is necessary(at least the jk once) to reset the balance lead's resistances .This helps you to keep an eye on the quality of the connections...Best of luck Frank
PS you need to get your cells quite good balanced,otherwise the highest voltage cell ,or the lowest voltage cell command the Soc!

 

[Orlando]

@spikebyte

Your problem is the BMS Voltage calibration. Calibrate the BMS voltage to match what the inverter is supplying and what the BMS is reading. Ex. if you set the RCV to 3.45V calibrate the BMS Voltage to at least 55.2V when the battery is full and amps are very low. Or better yet a littte higher, like 55.22V

Then set your 100% SOC Voltage to 0.015V below RCV in this case set SOC 100% to 3.435V. Once calibrated it will work great!

 

[hardtop]

Revert the firmware upgrade.

 

[spikebyte]

Revert the firmware upgrade.

Which version do you suggest I revert to?