ADC Fluctuation or Oscilation

[ytshin0877]

Hello

I am using 

NUCLEO-G474RE board to sample voltages from the the Battery management board. In the PCB board, I have opamp voltage divider to step down voltages around 24Volts to 2.5V level to be read from the STM. Please refer to the voltage divider network below. I have the divider node feeding into ADC. At node A, I am not seeing fluctuations, but at node B(Nucleo board side) I see the 400mV fluctuations. Please see the attached graphs. My setting are 96MHz clock prescaler divided by 10, sampling time is 42.5cycles, and buffer size is 4000. Grounds are connected by jumper wires with IDC connector and adc pins are connected by IDC+Jumper wires. I am not sure how this can be fixed. I tried slowing down the clock by increasing the pre-scaler number, but this didn't help. Could this be the ground bouncing case? I have been debugging this for weeks, and I have no clue what this is caused by.

Thank. you very much in advance

 

[TDK]

How are you taking the node B measurement exactly?

I suspect you have some ground loops in the setup.

[Ozone]

The large heatsink and the power transistor on board A suggest you are speaking about a kind "switched power supply" as battery management unit here. And the cabling to board B (the STM32) are relatively long.

I would suspect a noise issue, both for GND level and EMI reasons.

[ytshin0877]

I am using the Osciloscope to probe it.

[ytshin0877]

do you think that my biasing to the ADC pins are correct? Can this be the ADC charge injection effect? I am not so sure. Could you elaborate on how this fluctuation can be generated physically?

[TDK]

Kind of figured that much out already. Where are you probing and where is the probe ground? Long wires act as antennas. Probe ground should be close to the probe to minimize EMI induced noise.

[ytshin0877]

do you think that just integrating just STM microcontroller into just one board will solve the issue? since I am only getting fluctuation at the input of the ADC. If I do this, I think I can eliminate long wires and ground loop issues. Before this, I want to make sure that the long wires are the root cause of this issue.

[Ozone]

As user TDK already pointed out, your wiring most probably serves as antennas.
The total energy (charge) transferred to the S&H capacitor is relatively small, and nearby EMI field easily affect it.

Shielded connection cables for the ADC signal between boards A and B might help.

Or you could change the signal conditioning on board A to use a relatively high current across the cabling, and divide it down on board B for actual measurement. This would energetically reduce the impact of injected EMI on the result.

A third option would be to measure it on the spot, either by moving the MCU to board A, or by using an external ADC. There are relatively small and cheap ADC with serial interface available.