3 INA282 are replacing 3 Opamps typical current shunt monitors found in low side of three 1/2 bridge, this case a PWM motor controller.
The INA (V+ 2.975) Ref1 ground, Output pin 5 into Ref2 pushes up the pulse train floor. The .005 ohm shunt appears to be detecting high side current on the low side MOS near 80db at 12.5 kHz up to 20kHz. The high side is over 180 volts DC making it necessary to monitor 3 motor phases from the low side MOS shunts.
The interesting thing is the current detected below .2 amp is very low (73mv) INA signal yet a single pulse of the PWM period train reaches over (1.8v) near 12.6 amps. After that only numerous single higher pulses appear to make up the extrapolated current DSP wave form and signal integration goes flat.
There is a single current pulse at the end of each PWM cycle that is far higher than prior pulses spaced apart from the others. Still it for some reason it impossible to get 1MSPS ADC in 32x oversampling to establish a current ramp from INA output signal. Without integrating the PWM duty cycle into the ADC sample filtered data the analog current sample barely captures the signal valley let alone the peaks and is unable to build much of any ascending monotonic ADC order.
Anyone had same situation - seemingly impossible?