Current sensing without using INA19X type amps

[langwadt]

This is the best I could do, nothing else made sense as clearly. I thought of a convoluted BJT based circuit which referenced the common mode and worked on the high side but that just wasn't very good.

I can get 0.1% resistors which should minimise the error. Its not great by any means but if I add a multipoint calibration in the code, I think it should work reasonably well...

I can only get a 1% 50mΩ shunt so that's a bummer but nothing manual calibration can't fix. I guess this will have to do. I'll have to adjust the quiescent current so that the pack doesn't discharge itself over time.

I developed almost the same circuit at one point, and never liked the performance of attenuators followed by an instrumentation amplifier.  If I did it now, I would use the precision version of the dual transistor circuit, shown below, and it could be adapted to micropower operation.  This is from Linear Technology application note AN105.

The advantage here is that resistor matching affects gain rather than common mode rejection.

by why when you can get INAxxx and similar that will do the job with a single cheap part?

[David Hess]

I developed almost the same circuit at one point, and never liked the performance of attenuators followed by an instrumentation amplifier.  If I did it now, I would use the precision version of the dual transistor circuit, shown below, and it could be adapted to micropower operation.  This is from Linear Technology application note AN105.

The advantage here is that resistor matching affects gain rather than common mode rejection.

by why when you can get INAxxx and similar that will do the job with a single cheap part?

The original poster wanted a solution using more common parts, which I can understand.

Some applications cannot be reduced to more common parts, like fast difference amplifiers, but high side current sensing is not one of them.