ADC channel 12 less accurate than the other ?

[jean]

Hello,

Since two year, I am experiencing an issue with the ADC_CHANNEL_12 (pin PC12) of my STM32F767. I suspect it's a hardware issue coming from the MCU.

• a stable offset is added (randomly change each time I reboot the STM)

• this read value (via DMA) is way less accurate than the other inputs (I also use 3 other ADC pins that are working great), look like a lot of noise is disturbing the ADC (read value randomly change +2/-2% between the "real" voltage value)

I use a proper custom PCB circuit, no noise / EMI should be added. Again, the 3 other ADC input pins (PC3, PA0, PA1) are very accurate. Also tested with an oscilloscope, the signals are very clean.

What I've tried:

• changing ADC_RESOLUTION_12B to ADC_RESOLUTION_10B = it fixes the offset problem!
• reducing ADC_SAMPLETIME = some values are improving a bit the accuracy problem (in fact, changing the ADC_SAMPLETIME of the other ADC inputs also change the accuracy of the PC12 ADC !)
• Using ADC2 instead of ADC1 = does not change anything
• Using the tips of the STM32F7 errata sheet (setting the bit SYSCFG_PMC_ADC2DC2) = does not change anything
• Every other ADC configurations does not change anything

Maybe some of you came across this issue?

Thanks a lot for your help!

Cheers,

[jean]

void Adc::Init() 
{
	ADC_ChannelConfTypeDef sConfig;
 
  hadc2.Instance = ADC2;
  hadc2.Init.ClockPrescaler = ADC_CLOCK_SYNC_PCLK_DIV8;
  hadc2.Init.Resolution = ADC_RESOLUTION_10B; // 10B = PC2 is not offseted anymore
  hadc2.Init.ScanConvMode = ENABLE;
  hadc2.Init.ContinuousConvMode = ENABLE;
  hadc2.Init.DiscontinuousConvMode = DISABLE;
  hadc2.Init.ExternalTrigConvEdge = ADC_EXTERNALTRIGCONVEDGE_NONE;
  hadc2.Init.ExternalTrigConv = ADC_SOFTWARE_START;
  hadc2.Init.DataAlign = ADC_DATAALIGN_LEFT;
  hadc2.Init.NbrOfConversion = 2;
  hadc2.Init.DMAContinuousRequests = ENABLE;
  hadc2.Init.EOCSelection = DISABLE;
  HAL_ADC_Init(&hadc2);
 
  sConfig.Channel = ADC_CHANNEL_12; // PC2
  sConfig.Rank = 1;
  sConfig.SamplingTime = ADC_SAMPLETIME_480CYCLES;
  sConfig.Offset = 0;
  HAL_ADC_ConfigChannel(&hadc2, &sConfig);
 
  sConfig.Channel = ADC_CHANNEL_13; // PC3
  sConfig.Rank = 2;
  sConfig.SamplingTime = ADC_SAMPLETIME_144CYCLES;
  HAL_ADC_ConfigChannel(&hadc2, &sConfig);
 
  if(HAL_ADC_Start_DMA(&hadc2, (uint32_t*)&DMAvalues[0], 2) != HAL_OK) __DEBUG_BKPT();
 
 
 
  hadc1.Instance = ADC1;
  hadc1.Init.ClockPrescaler = ADC_CLOCK_SYNC_PCLK_DIV8;
  hadc1.Init.Resolution = ADC_RESOLUTION_12B;
  hadc1.Init.ScanConvMode = ENABLE;
  hadc1.Init.ContinuousConvMode = ENABLE;
  hadc1.Init.DiscontinuousConvMode = DISABLE;
  hadc1.Init.ExternalTrigConvEdge = ADC_EXTERNALTRIGCONVEDGE_NONE;
  hadc1.Init.ExternalTrigConv = ADC_SOFTWARE_START;
  hadc1.Init.DataAlign = ADC_DATAALIGN_LEFT;
  hadc1.Init.NbrOfConversion = 2;
  hadc1.Init.DMAContinuousRequests = ENABLE;
  hadc1.Init.EOCSelection = DISABLE;
  HAL_ADC_Init(&hadc1);
 
  sConfig.Channel = ADC_CHANNEL_0; // PA0
  sConfig.Rank = 1;
  sConfig.SamplingTime = ADC_SAMPLETIME_480CYCLES;
  sConfig.Offset = 0;
  HAL_ADC_ConfigChannel(&hadc1, &sConfig);
 
  sConfig.Channel = ADC_CHANNEL_1; // PA1
  sConfig.Rank = 2;
  sConfig.SamplingTime = ADC_SAMPLETIME_480CYCLES;
  HAL_ADC_ConfigChannel(&hadc1, &sConfig);
 
  if(HAL_ADC_Start_DMA(&hadc1, (uint32_t*)&DMAvalues[2], 2) != HAL_OK) __DEBUG_BKPT();
 
  /* reduce the ADC noise ? STM32F7 errata sheet */
  //SET_BIT(PWR->CR1, PWR_CR1_ADCDC1_Pos);
  SET_BIT(SYSCFG->PMC, SYSCFG_PMC_ADC1DC2);
  SET_BIT(SYSCFG->PMC, SYSCFG_PMC_ADC2DC2);
}

[waclawek.jan]

Write a minimal program which does nothing just reads this one ADC channel, to memory, and observe the result with debugger.

JW

[ingwmeier]

you could try to change Ranks of ADC_CHANNEL_12 and ADC_CHANNEL_13 to check if anything changes..

Another thing: did you test with HAL_ADCEx_Calibration_Start if this changes anything?

[jean]

Hi, thanks a lot for your help!

1. a minimal program (just the ADC values monitored with debug) = same result
2. changing the ranks does not change anything, ADC_CHANNEL_12 (PC2) is still not accurate
3. I also update all the HAL librairies to the most recent one, no changes

[Uwe Bonnes]

Does this happen on different layouts? If you didn't test with a totally different layout:

What signal run near to PC12? Maybe there is excessive noise . Or s there perhaps any signal > VDDIO on a TTa pin?

Perhaps try to reproduce on a Nucleo or Disco board.

[jean]

No, indeed the next step is to test it on a new PCB, will do it in the following months!

There is no noisy or high-speed signals next to PC12 ...

Or s there perhaps any signal > VDDIO on a TTa pin?

A TTa pin is a pin that can be declared as an analog input (connected to ADC) ?

If yes, is that possible that having for example a +5V on a TTa pin (even if this pin is not declared as an analog input) can influence the ADC_CHANNEL_12 ?

[waclawek.jan]

> having for example a +5V on a TTa pin 

That's definitively a no-no which may damage the part, not just influence the ADC readings.

Cut the offending signal and retry with the ADC.

JW

[jean]

Thanks for this tip, didn't really aware of that! After double checking my design, no pins of my STM is connected to something more that 3.3V, so I guess that's not the problem.

The next step is to test ADC_CHANNEL_12 on a new design, or a demo board. But the fact that I was able to remove the ADC "random offset" of PC2 by changing ADC_RESOLUTION_12B to ADC_RESOLUTION_10B still seems like a STM32 issue for me :)