BMS percentages - how reliable are they?

[The herring]

I have three 12kWh LiFePo batteries with JKBMS in parallel. I monitor the reported SOC through Bluetooth continuously. Strange part is that the reported SOC is not the same on all three batteries. Two of them are always within 2% of eachother, but the third one drops faster than the other two. It is also slower to get to full charge.
The cells are the same 272Ah in 16S and the wire length is the same on all 3 batteries. The BMS's are a little different though - the third battery has a larger model.
Is it just the reported values from the BMS? Any suggestions?
I've attached a screenshot from HA...

Tia The Herring

 

[time2roll]

Under significant load does the amps drawn on each battery exactly match?

 

[The herring]

Under significant load does the amps drawn on each battery exactly match?

I added some sensors to monitor this and I attached a screenshot with the past two hours. It looks like it's discharging faster, but charging is the same? Will know more tomorrow...
I also added some sums to see how it stacks over over an entire day....

 

[Don B. Cilly]

Out of curiosity, how to you get the JK BMS to talk to HA?
-

 

[The herring]

Out of curiosity, how to you get the JK BMS to talk to HA?
-

I use https://github.com/jblance/mpp-solar
It's a python thing, that picks up the values with bluetooth and pushes them with mqtt to HA.
I have an rPi3 that only has this task....

 

[chrisski]

In my signature block, there is a document on how to parallel batteries. If both leads are on the near battery, it could do something like you’re describing.

 

[time2roll]

I added some sensors to monitor this and I attached a screenshot with the past two hours. It looks like it's discharging faster, but charging is the same? Will know more tomorrow...
I also added some sums to see how it stacks over over an entire day....

This could be normal. The very low resistance of these cells contributes to these variations.
How much of the capacity is getting utilized? 60%+? I believe the issue is worse with short cycles. Doubt there is any harm in it.

 

[Don B. Cilly]

I use https://github.com/jblance/mpp-solar

Thank you. It took a while, but I got it to work.
Except... it returns pretty useless data.
I'll play with it a little more...
-

 

[The herring]

I use this config:

[SETUP]
# Number of seconds to pause between loops of processing the sections
# i.e. the pause at the end of an entire run through the config file
# default is 60
pause=60
# ipaddress or hostname of the mqtt broker, default is 'localhost'
mqtt_broker=10.100.0.42
# mqtt broker port number, default is 1883
mqtt_port=1883
# username and password (if required) for mqtt broker, default to None and not used unless defined
mqtt_user=********
mqtt_pass=********

[JKBMS#1]
type=jkbms
protocol=JK02
port=C8:47:8C:F7:88:FE
command=getCellData
tag=22018
outputs=mqtt

 

[Don B. Cilly]

~$ jkbms -p C8:47:8C:E2:8D:26 -c getCellData
Command: getCellData - BLE Cell Data inquiry
------------------------------------------------------------
Parameter                       Value           Unit
header                          55aaeb90           
record_type                     02                 
record_counter                  213                 
voltage_cell01                  3.806417863259254e-31   V   
...

voltage_cell19                  2.1019476964872256e-43  V   
...
resistance_cell01               0.0             Ohm 
...
resistance_cell24               2.9626331329345703      Ohm 
...
average_cell_voltage            1.7516230804060213e-42  V   
delta_cell_voltage              2.350988701644575e-38   V   
highest_cell                    1                   
lowest_cell                     1                   
flags                           0000               
uptime                          0D0H0M0S           
checksum                        dc                 
highest_cell_voltage            3.806417863259254e-31   V   
lowest_cell_voltage             3.806417863259254e-31   V

I have 8 cells.
1.7516230804060213e-42 V?
And all... anyway, thanks.
-

 

[The herring]

Maybe try protocol JK04? Otherwise make a post in his gitlab comments? He was very helpful with my questions...

 

[Don B. Cilly]

I asked him. Specifying -P JK02 works. Thanks.
I wrote a one-liner (for bash) that returns the relevant info for 8 cells. Easy to modify for other numbers. Made an alias for it :·)

alias jk="jkbms -p C8:48:8C:E2:8D:26 -P JK02 -c getCellData | sed -n '7,14p ; 58,65p'"

(Obviously) substitute the MAC address for your BMS's.
-