Post Go back to editing

Using the LT6015 to monitor a stack of four 3.2V 100Ah LiFePO4 cells?

Yes, I know that there are dedicated battery stack monitors available, but in this case is the stack small and voltages reasonable. Four 3.2V cells give 12.8V. What I would like to do is to monitor each cell and disconnect the pack in case the voltage of one individual cell is outside the allowable range, being 2.5 ..4V.

The cells will be sitting in a boat. Used over summer and removed and top-balanced every winter. They will be charged from a solar cell, using a MPPT charger.

I did a quick simulation in LTSpice and also read the LT6015 datasheer. I have attached that file and I apologise for the messy schematic showing the opamps for two cells. LTSPice absolutely refused to play ball when I turned the opamp around 

I realise that the resistors need to be matched for the circuit to work properly. The idea is to have the measued voltages VC1, VC2 etc readable from a MCU as well as fed into comparators for the disconnect logic.

Comments? Would this be a workable solution, given the design criteria outlined?

Parents
  • Hi,

    It seems to be workable. If you need well matched resistors, you may use LT5400-2 which can give you up to 0.01% matching. You could also add some capacitors for filtering in parallel with the feedback resistors R2 and R6 and R4 and R8 to ground. This could give more protection against stability issues, ringing, noise and oscillations.

    Another possibility is to use LT1991, which already have integrated resistors well matched. Also, this part already have feedback capacitance integrated as well.

  • Thanks! Yes, I left out the caps in the simulation for clarity. I will look into your other suggestions and collect parts for a prototype!

  • LT1991 indeed seems to be a far better choice. Less components, less cost, less that can go wrong :) It is specified for 40V and rail-to-rail and it looks good in the simulation. Add one LT1017 pair per cell and a 4.096V reference, suitably divided down to 2.5V and 3.9V for the comparators, then combine the outputs of the "high" comparators in an OR-gate and same for the "low" comparators and I will have the required logical signals to isolate the battery pack in case one cell goes marginal. 2.5 and 4V are the absolute min/max values from the manufacturer, Max charging voltage 3.65V

Reply
  • LT1991 indeed seems to be a far better choice. Less components, less cost, less that can go wrong :) It is specified for 40V and rail-to-rail and it looks good in the simulation. Add one LT1017 pair per cell and a 4.096V reference, suitably divided down to 2.5V and 3.9V for the comparators, then combine the outputs of the "high" comparators in an OR-gate and same for the "low" comparators and I will have the required logical signals to isolate the battery pack in case one cell goes marginal. 2.5 and 4V are the absolute min/max values from the manufacturer, Max charging voltage 3.65V

Children