LTC6803-3 top cell voltage inaccuracy with 3S battery


I've been developing a BMS for use with small scale EVs and I've been using the LTC6803 for the battery stack monitoring. I've been through a couple of board iterations of it and if you're more interested I've been keeping a project log forum (electric skateboard builders) you can check it out at:

The board is meant to support battery stack configurations between 3 to 12 cells in series, giving the user flexibility in their pack configuration with just one BMS.

Picture of 0.3 HW boards.

I've been very impressed by the cell voltage measuring accuracy overall, but with 3S pack configurations I get a voltage measurement error of 10 mV below the actual voltage for the top cell consistently. The lower 2 cells show accurate readings. This top cell measurement accuracy doesn't seem to be exhibited with 4-12S configurations based on my practical testing.

Would you have any idea/clue why the 3 cell series stack would give the top cell measurement inaccuracy of consistently ~10 mV.

The LTC6803 is powered from the battery pack terminals and then a dedicated cell sensing connector is used.

  • 0
    •  Analog Employees 
    on Sep 23, 2019 9:34 PM

    There are two scenarios that come to mind. One is that the total V+ applied needs to be at least 10V to ensure the reference and MUX are properly biased. With 3 cells, this may not be assured. The other is that V+ may be lower than the upper cell potential by enough to compromise the upper cell MUX switch performance. V+ should not be more than maybe 300mV below the upper cell or error can develop. If there is a separate power path for V+, see that there isn't significant voltage drop in that area.