STM32 ADC 3V3 measurement

Question

I'm using Nucleo-32 board with STM32F031K6 MCU. I have some weird problem with ADC measurement. When ADC pin (PA4 in my case) is connected directly to VREF or 3V3 the value is always in range of 4031-4032 instead if 4095... For measuring low voltages, the ADC value has higher offset from the correct one (20% lower then it should be for 0.5V measuring).

Could anybody please explain this behavior?

ADC configuration:

  ADC_ChannelConfTypeDef sConfig;

    /**Configure the global features of the ADC (Clock, Resolution, Data Alignment and number of conversion) 
    */
  hadc.Instance = ADC1;
  hadc.Init.ClockPrescaler = ADC_CLOCK_ASYNC_DIV1;
  hadc.Init.Resolution = ADC_RESOLUTION_12B;
  hadc.Init.DataAlign = ADC_DATAALIGN_RIGHT;
  hadc.Init.ScanConvMode = DISABLE;
  hadc.Init.EOCSelection = ADC_EOC_SEQ_CONV;
  hadc.Init.LowPowerAutoWait = DISABLE;
  hadc.Init.LowPowerAutoPowerOff = DISABLE;
  hadc.Init.ContinuousConvMode = DISABLE;
  hadc.Init.DiscontinuousConvMode = DISABLE;
  hadc.Init.ExternalTrigConv = ADC_SOFTWARE_START;
  hadc.Init.ExternalTrigConvEdge = ADC_EXTERNALTRIGCONVEDGE_NONE;
  hadc.Init.DMAContinuousRequests = DISABLE;
  hadc.Init.Overrun = ADC_OVR_DATA_PRESERVED;

  if (HAL_ADC_Init(&hadc) != HAL_OK)
  {
    Error_Handler();
  }

    HAL_NVIC_EnableIRQ(ADC1_IRQn);

  /**Configure for the selected ADC regular channel to be converted. 
  */
  sConfig.Channel = ADC_CHANNEL_4;
  sConfig.Rank = ADC_RANK_CHANNEL_NUMBER;
  sConfig.SamplingTime = ADC_SAMPLETIME_239CYCLES_5;
  if (HAL_ADC_ConfigChannel(&hadc, &sConfig) != HAL_OK)
  {
    Error_Handler();
  }

Pin configuration:

__HAL_RCC_ADC1_CLK_ENABLE();
__HAL_RCC_GPIOA_CLK_ENABLE();

GPIO_InitStruct.Pin = GPIO_PIN_4;
GPIO_InitStruct.Mode = GPIO_MODE_ANALOG;
GPIO_InitStruct.Pull = GPIO_NOPULL;
HAL_GPIO_Init(GPIOA, &GPIO_InitStruct);

Answer 1

Thank everyone for help. Finally I found solution myself

This function call after ADC initialization makes all the magic to solve the issue

HAL_ADCEx_Calibration_Start(&hadc)

Answer 2

A couple of thoughts:

1. I wonder if your VDDA or VREF voltage has an offset. I took a look at the Nucleo-32 board schematic here and saw these pins connected to a bead which could have some DC resistance that matters if you are drawing a lot of current from that node which sometimes is used on ADCs for measurement reference.

Maybe try measuring the voltage at pin 5 of the MCU (VDDA/VREF+)?

2. How much time are you allowing for settling? It seems that the ADCs are 5Msps? It should be pretty fast but you might have some capacitance slowing your signal down more than you think.

3. Do you see the same problem on another analog input (e.g. PA1)? If so, this may indicate a VREF issue as well.