Large series/Parallel array of LiPO cells for robot

[opampsmoker]

Hello,
We have quite a large robot with a large series parallel array of  LiPO cells.
How can we sense the charge and discharge current in all cells so that we can see if any cells are hogging too much current? I am thinking we need loads of Hall Sensors? However, I hear that when  you initially  take the Hall sensor measurements of current…then if you can simultaneously take cell voltage measurements…….then in future , you can just take the battery voltage measurement, and infer the cell currents  from that. Is this true?

We want to be able to monitor  all the cells, so that we can see when any cell needs replacing, etc, as part of battery care.

[Rerouter]

Break your design into monitoring the parallel levels, then duplicate that design for the series levels.

Pretty easy constraint is every single cell needs a current shunt and a means to measure it, depending on the scale of the batteries, a fixed length of copper to your bus bar could be enough for that measurement.

If it is a bus bar, the other way would be measuring between cells down a bus bar, using the length of copper as a shunt, calibration would be a bit weird, but would give a heads up if 1 cell was taking more or less than the rest.

Other things would be just divide an conquere, if you don't need to locate the exact cell, just reduce to what part of the battery is an issue, then perhaps only groups of 4 are monitored, another good health indication is voltage vs mAh for charge / discharge, as if the capacities start off similar, your more looking for where the cell is becoming unbalanced

If this is for some idea of a protection against a shorted cell, other circuitry come in the mix,

[Siwastaja]

Paralleling cells and monitoring their currents individually is an economical and complexity disaster, no one in their right mind does that, ever, unless it's about academic interest or something similar.

Just buy decent cells, parallel them, and treat them as one big cell. The cells are designed for this, this is the intended use case.

A "weaker" cell naturally takes smaller part of the current due to having higher resistance.

Similarly, you can mentally model a larger cell of being multiple internally paralleled small cells. If there is a certain failure rate per cubic meter of active material, then this risk is the same regardless using one large cell or multiple small parallel cells of same total capacity. It's just that you can actually get top-quality cells in small form factors, but have to resort to dubious manufacturers with large sizes, so paralleling the small ones ends up being a more robust choice.

Monitor voltages in the taps of the series chain, and then monitor the total pack current.

[Rerouter]

The weakest cell ends up with less charge per mV of charge / discharge, so a lower current,

Somewhere in there is a trade off point where the self heating from capacity imbalance starts to degrade the other cells by one or more not sharing the load.

[NiHaoMike]

It's just that you can actually get top-quality cells in small form factors, but have to resort to dubious manufacturers with large sizes, so paralleling the small ones ends up being a more robust choice.

Not so if you use EV batteries, which nowadays are quite cheap per kWh.