is this correct way to current sense?

[3roomlab]

trying to convert some spare MC33072 into current sense. the residual floating voltage (in simulation) seems to be a problem ...

while i think i have got the general idea right, could someone hint if there is a way to remove the residual 3mV?

[T3sl4co1l]

Looks like input offset?

SPICE models are often made with a "typical max" amount of offset, so that you are reminded that it's there, even though it's not representative of an average part (which has an expected offset of zero..).

Your error due to all the resistor pairs being real resistors will be far in excess, even for very pricey ones.

Tim

[3roomlab]

what could i do to correct the resistor error? the few mA going thru the divider + op amp feeder?

(it reminds me of the video control "pedestal" ... lol adjust the entire setup up/down a few mV)

doesnt look like a simple add 1 resistor thingy ... an entire "pedestal" adjustment from another output op amp fix? it looks like i need to add a negative rail if i were to shift "pedestal" wise

come to think of it, it should be fine if im not going for 1mA accuracy. but was wondering ...

[T3sl4co1l]

You need three adjustments in total: one to vary the sense/divider ratio, to null CMRR; one to null the DC offset voltage (due to op-amp error and unbalanced leakage currents); and one to adjust overall gain (to calibrate slope).

Do your DC sweep thingy in a Monte Carlo simulation, where the resistors all have some specified tolerance.  You'll see the problem.

Since you have two sets of divider resistors, the CMRR ratio adjustment can be placed in any of them, as long as the adjustable range is give or take the resistor tolerance.  You might make one side intentionally "long", then make the other adjustable between normal, long, and twice as long (so, for 1% resistors, make the one side longer by 1%, and give the other an adjustable range of 0-2% longer).

Tim