BMS SOC tracking and correcting

[JHdotDigital]

Firstly I'd just like to say great work, this project has allowed me to get my DIY battery up and running with out too much head scratching and effort on my part.
This is not a problem with this project, but I'm hoping we can incorporate something to solve my (and possibly others) issue.

What I've noticed is that my BMS SOC is drifting since my initial top balance. After less than a month the voltage at 100% SOC is lower than it should be. This is probably partly my own fault as I went for a 200A BMS (I guess this model will not be as accurate at coulomb counting due to the ADC scaling than say the 100A version), this is more than double what my inverter can even use, however I am where I am. Has anyone else seen anything similar? and any solutions would be great!
When 100% SOC is reported to my Solis inverter it stops charging and therefore does not allow taking the battery to its true 100% SOC.
I expect if it was the other way around the BMS would reset 100% SOC on the voltage limit.
What I'm thinking is if the battery voltage has not entered the absorption stage can we tell the inverter the SOC is 99% if the BMS says its 100%.
I can see what to do in the software and do it myself, but it would be nice to hear others thoughts on why this might be a bad idea etc.

Thanks Jonathan

[denveronly]

make a change in BMS Capacity (Ah). Make it bigger, it will recalculate to lower SOC, and after full charge self-adjust to the rythm

[JHdotDigital]

Thanks, I've done that and got back where I need to be. I'll continue to keep a close eye on things.

[Uksa007]

What I've noticed is that my BMS SOC is drifting since my initial top balance. After less than a month the voltage at 100% SOC is lower than it should be.

I think adjusting the SOC is probably not a good idea on a couple of fronts:

• It would essentially disable a safety feature of the inverter.
• It does not fix the root cause of the issue.

I suspect your issue maybe because you have not calibrated the BMS voltage and current.
Once these are calibrated it should not drift by any amount that causes issues.
I calibrated mine at 55.2V and 50A(my max charging current), eg when the inverter is charging at 50A you need set the calibration value in the BMS.
Voltage I calibrated when in absorption and the batteries are not taking any more current.

Give that a try.

[JHdotDigital]

I have checked the voltage and it was pretty much bang on against a calibrated meter. Current it not so simple though, I will try and devise something to check and calibrate.
Do you know if there are different calibration constants for charging and discharging as what I'm seeing would suggest a small imbalance in the charge/discharge measurements?
Or do I have an issue with the zero offset being wrong. I have noticed that when my inverter is in standby the measured current is zero! I suspect it is actually using some current. Also the BMS will also be drawing a small current which also appears to not be measured. Do you see the same?

[Uksa007]

Yep 0A current when inverter is idle, the BMS is probably not so good at measuring small currents.

There is only one current calibration, seems to do both directions, it's probably more important to do charging current.
Mine was out a bit (i used a clamp meter) my BMS never goes to 100% unless I hit the OVP limits which never happens, charging to 55.2v
Max SOC is at 99% when fully charged.
Do you have your OVP limit set too low? 3.6V(LiFeO4) is the default and works well.

[JHdotDigital]

Thanks for getting back so quick. That's very useful to know. Sounds you have the opposite to me, I always hit 100% but not my charge voltage limit.
OVP is 3.6V per cell and I have set charge limit to 3.45V per cell (total 55.2V) when it hits this voltage with some persuasion it does reset SOC to 100% and this occurs well before any cell over voltage so I'm happy with my balance state. However its looking like I'm drifting away from 55.2V at 100%. Maybe if I just keep letting it drift and the BMS will eventually correct it's SOC.

[denveronly]

Thanks for getting back so quick. That's very useful to know. Sounds you have the opposite to me, I always hit 100% but not my charge voltage limit.
OVP is 3.6V per cell and I have set charge limit to 3.45V per cell (total 55.2V) when it hits this voltage with some persuasion it does reset SOC to 100% and this occurs well before any cell over voltage so I'm happy with my balance state. However its looking like I'm drifting away from 55.2V at 100%. Maybe if I just keep letting it drift and the BMS will eventually correct it's SOC.

As i remember BMS calculates it SOC value using OVP value either, so you will never see it 100% unless you lower your OVP to 3.5 i.e which should not happen because 3.45 is already set

i have my Li ion 4.1 volt set to charge and OVP 4.15

[JHdotDigital]

Quick update on this, I'm loosing about 10% charge per week with roughly 50% charge/discharge per day.
It got so bad the other day the BMS tripped on undervoltage but was still stating 31% charge at the time. Once it had tripped the SOC did reset to 0% So I'm back to roughly where I need to be.
BUT, and this is a rather large BUT, when my BMS tripped it took out the TJA1050, this is the second TJA1050 I've had fail, the first one blew when I turned the BMS off in the app. As the BMS cuts the Gnd connection this leads me to believe I'm seeing ground potential differences great enough to take out the TJA1050.
To fix it I'm using a serial isolator (Si8621) and a 5V-5V isolated DC-DC converter such as this multicomp one (MPB0505S-1W) between the ESP32 and the TJA1050, I'd highly recommend other so this too for piece of mind.

[Uksa007]

You need let the BMS hit the Low voltage and then High Voltage protection before it's SOC is calibrated.

RE TJA failing, you are the only one I know of that has this issue.
How are you powering the ESP32, You should never remove the GND from the ESP circuit, it should be connected to the battery side Negative, not the BMS P-
I have tripped my Over/under voltage protection/other alarms many many 10s to 100s of times, mostly for testing, and never had one fail.
The ESP32/TJA should be grounded via the JK-BMS GPS/RS485 connection, regardless of the power supply, assuming you have the GND connected there. Some inverters also have a GND on the CAN interface, could try with that connected.
In anycase the ESP32 power supply should remain on when the BMS is on regardless of BMS output.
Unless you can measure with an oscilloscope we are just guessing, could be induced voltage into the CAN connection due to faulty or poorly designed inverter.

Consider using my Interface Kit, it removes all the hacked together modules and is bulletproof. It uses a different(Ti) CAN bus chip that is native 3v3, more robust, designed for automotive use.

Please consider supporting my work so I can continue to develop new feature and provide help and assistance.
https://www.patreon.com/Uksa007Codedevelopment

[JHdotDigital]

Thanks, yes I deffo owe you a beer!
My BMS is: JK-B2A20S20P
I have connected my ESP32 Gnd to the GPS port Gnd and the ESP is powered by a USB adapter I had laying around (No Earth!)
I'm using a Solis battery inverter: RAI-3.0KW 48ES 5G - This seems to work fine on the "Pylon" setting but does need the battery name changing in the yaml file to match the inverter, so for pylon:
can_id: 0x35E data: [0x50, 0x59, 0x4C, 0x4F, 0x4E, 0x20, 0x20, 0x20]

The Solis inverter does have a CAN ground and I have connected it! This is probably part/all of my problem! Hence the idea to isolate the ESP - CAN link.
And when I've thought about it, if the CAN is not isolated in the inverter I'm actually bridging out the BMS FETs. This could lead to smoke on the GND connection at the weakest point! I don't know if it does, so have come to the conclusion I need to play it safe now as I think I've been lucky not to take out the BMS or Inverter!
I would have to say my combination of BMS and Inverter could lead to problems!
PS I have queried whether CAN needs a ground reference. The general consensus is that it will work without and is usually quite tolerant, but if you don't use one you run the risk popping transceivers, leaving me no better off should I disconnected it.

[Uksa007]

The latest code available on Patreon has much better support for strict pylon compliance.

I don't use a GND connection on the CAN bus, in fact my inverter does not even have one on the Inverter for the BMS interface RJ45.
It's a differential pair, so GND is not normally needed, I generally recommend that you don't use it unless you have a unique requirement or issues, generally as you found out, it causes more issues than it solves with inverters.

[Uksa007]

Closed due to inactivity.

Please consider supporting my work so I can continue to develop new feature and provide help and assistance.
https://www.patreon.com/Uksa007Codedevelopment