# non-inverting current sense op amp output unstable during the measurement

I would like some advice on how to solve the following problem. The following circuit is motor driver with current sense circuit. The circuit run motor at cca 30mV when is not under load, but when the load is applied to motor, measuring voltage rise up to 500mV, hold It for 2 second and shut down driver. When I want to measure voltage between "unstable point" and gnd there is a big problem with stability. When i measure voltage with FLUKE 179 multimeter and original test probes I get cca 500 mV under load which is good.

If I add an extra crocodile wire (10-20cm) to test probes there is cca +- 100mV difference in measurement. Only if I add 1M Ohm resistor between "unstable point" and crocodile wire and 1M Ohm resistor between gnd and second test probe, I can add an extra crocodile wire after resistors and the measurement is right (+- few mV difference).

If I add cca 20cm extra crocodile wire on "unstable point", then add 1M Ohm resistor between crocodile wire and test probe, there is again +- 50-100 mV difference in measurement (even if there is a 1M Ohm resistor connected after 20cm crocodile wire). If I add 1M ohm resistor only on "unstable point" and not to gnd probe, there is also cca 50mV difference. Also I notice that if I touch "unstable point" with my hand, voltage on that point collapses so I suspect that the noninverting op amp circuit is unstable or the voltage range of measuring (30mV - 500mV) is to low to handle external disturbances. It's funny that crocodile wire without any load connected between multimeter probes and "unstable point" effect on measurement.

It would be best to include your layout, but assuming your layout is good, I would try filtering your sense line.

Your present capacitor - only 33 pF - has a negligible effect on this circuit, so the output of your LM324 will have all of the PWM and motor switching frequency on it. The Fluke meter's DC input has to make some sense of this, and when you add a wire the capacitance changes, changing the signal.

I would change your sense circuit:

simulate this circuit – Schematic created using CircuitLab

You may have to adjust R_ADDED and C_ADDED to adjust how quickly shutdown will occur. This approach filters out the switching before it gets to the op amp, which won't handle the high frequency well. Also, I have lowered R22 because the ADC in the processor has a high input impedance.

If you are stuck with your layout, you could try just increasing the value of C16. The switching waveform would still have to get through your op amp, and it would be distorted, but it might suffice.

Good luck!

Do you have a scope?
Motors tend to draw current in spikes. This means that the voltage on R19 would be going up and down very rapidly. If so, the LM324 may be hitting its slew rate limit. An RC low pass filter could make the LM324 happier if that is the case.
Depending on the meter you are using to make your measurements, the large AC component on the "unstable point" may be making it inaccurate. Increasing C16 will help this.