Featurerequest: Temperature depending maxu current

[endurance1968]

Maybe it is already there or could be achieved via other measures.

I am looking for something like a temp based max charge current.
0Celsius: 0c (c-rate)
5C:0.05c
10C:0.7c
20C:1c
30C:1c
40C:0.75c
45C:0.5c
50C:0c
means mac current based on temp & capacity of connected batteries. Current curve between given point could be linear interpolation.
Always minimum of configured algo and this curve (via config file) should be used.

Asking for it as my battery is located in a garage and in winter it could go down to 5Celsius maybe even ~0C. Did never check so far if the resistence based algo will also result in far lower max current at low temps. currently (at 10C) I see 210A CCL and 314A DCL. Which is so far OK (1800Ah connected) but what happens if only one pack would be connected due to maintenance...

[pulquero]

In principal, the resistance based algorithm should work fine as cold will increase the internal resistance, leading to a higher V/I ratio. Disconnecting a pack should also be ok - lack of current will be detected.

[endurance1968]

thanks for quick reply - I will monitor. Currently I have insulated the area with the batteriepacks. Next two days minus -8 outside. Lets see what happens with the temps.

[endurance1968]

As promised I monitored it a while. Graph shows six packs conected to BA. Trend is korrekt - you can in principle see the lower the temp the lower are the limits. THere are some exeptions and peaks. Especially when no current (idle) the limits are going sometimes up I assume due to lack off measurement accurracy?

In total it seems no extra temp based compensation needed.

Charge absolute max is set to 210A - Discharge 400A (but anyway limited by the 3x Victron MPII 5k)

[pulquero]

That's good news! If you like you can try to improve the behaviour by tweaking the following parameters:
VOLTAGE_HISTORY_SIZE = 10
MIN_VOLTAGE_DELTA = 0.02
MAX_IR_ERROR_PERCENTAGE = 0.1
in the python file /data/BatteryAggregator/battery_service.py. You will need to use svc to restart the service after any changes. If you find any particularly useful, I can expose them in the json config file for you.

[endurance1968]

I will play with that values, is there any description in which direction what parameter can influence results?

Happy New Year! and thanks for your efforts.

[pulquero]

So, it basically tries to fit a line to a graph. The history size is how many points to use. A larger history size potentially means better precision but if there's a rapid change then some points will be old and out of date.
The voltage delta is the minimum voltage change required before a point is recorded. So v needs to change by at least this amount before I-V is recorded. A higher value will make it less sensitive to change.
When IR is calculated, the error in the calculation has to be less than the error percentage else the value is discarded. Too small and the IR will never update as you never meet the error percentage. If your history is too large and your delta too small, this will act as a filter to throw out any junk results.