stm32cubewl_v1-2-0 HAL_ADCEx_Calibration_Start does not work

[John1960]

In the middle of function returns HAL_ERROR.

stm32cubewl_v1-1-0 - the function works well.

[John1960]

Sorry, MCU STM32WLE5

[Louis AUDOLY]

Hello and welcome @John1960​ ,

You can have a look at the example ADC_AnalogWatchdog that you can find here :

STM32Cube_FW_WL_V1.2.0\Projects\NUCLEO-WL55JC\Examples\ADC\ADC_AnalogWatchdog

In the main function there is a calibration that don't return an error.

Which code are you running ?

Best regards

[John1960]

I use my own code, which use ADC. First I want to calibrate it, so I call HAL_ADCEx_Calibration_Start. 

I began from stm32cubewl_v1-1-0, and it worked well.

Then I upgaded library to stm32cubewl_v1-2-0, and the function began to return HAL_ERROR.

Functions HAL_ADCEx_Calibration_Start  from version 1.1 and from version 1.2 are extreemly different.

Part of my code - Green LED is ON:

void ADC_Init ()

{

 RCC_PeriphCLKInitTypeDef PeriphClkInitStruct = {0};

 PeriphClkInitStruct.PeriphClockSelection = RCC_PERIPHCLK_ADC;

 PeriphClkInitStruct.AdcClockSelection = RCC_ADCCLKSOURCE_HSI;

 if ( HAL_RCCEx_PeriphCLKConfig ( &PeriphClkInitStruct ) != HAL_OK )

  {

   Error_Handler();

  }

 /* Peripheral clock enable */

 __HAL_RCC_ADC_CLK_ENABLE();

 hadc.Instance = ADC;

 hadc.Init.ClockPrescaler = ADC_CLOCK_ASYNC_DIV64;

 hadc.Init.Resolution = ADC_RESOLUTION_12B;

 hadc.Init.DataAlign = ADC_DATAALIGN_RIGHT;

 hadc.Init.ScanConvMode = ADC_SCAN_DISABLE;

 hadc.Init.EOCSelection = ADC_EOC_SINGLE_CONV;

 hadc.Init.LowPowerAutoWait = DISABLE;

 hadc.Init.LowPowerAutoPowerOff = DISABLE;

 hadc.Init.ContinuousConvMode = DISABLE;

 hadc.Init.NbrOfConversion = 1;

 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_OVERWRITTEN;

 hadc.Init.SamplingTimeCommon1 = ADC_SAMPLETIME_160CYCLES_5;

 hadc.Init.SamplingTimeCommon2 = ADC_SAMPLETIME_160CYCLES_5;

 hadc.Init.OversamplingMode = DISABLE;

 hadc.Init.TriggerFrequencyMode = ADC_TRIGGER_FREQ_HIGH;

 if (HAL_ADC_Init(&hadc) != HAL_OK)

 {

  Error_Handler();

 }

 /* ADC interrupt Init */

 HAL_NVIC_SetPriority(ADC_IRQn, 0, 0);

 HAL_NVIC_EnableIRQ(ADC_IRQn);

// return;

 if (HAL_ADCEx_Calibration_Start(&hadc) != HAL_OK)

 {

  /* Calibration Error */

 GreenLED_On ();

  Error_Handler();

 }

 } // ADC_Init

[Philippe Cherbonnel]

Hello,

A regression has been introduced in release v1.2.0.

Issue:

In case of high frequency ratio between CPU and ADC (CPU high frequency, ADC low frequency), function HAL_ADCEx_Calibration_Start() returns an error.

Root cause:

A delay must be fullfilled between ADC enable and ADC disable.

Fix:

In function HAL_ADCEx_Calibration_Start(), a delay must be added:

  LL_ADC_Enable(hadc->Instance);

<delay here>

  LL_ADC_SetCalibrationFactor(hadc->Instance, calibration_factor_accumulated);

  LL_ADC_Disable(hadc->Instance);

The fix will be available in next STM32WL FW package release.

Documentation is being updated to mention this constraint.

Best regards

Philippe

[Dana Myers]

 /* Apply calibration factor */
 LL_ADC_Enable(hadc->Instance);
 
 while (!LL_ADC_IsActiveFlag_ADRDY(hadc->Instance))
 	/* spin */;
 LL_ADC_ClearFlag_ADRDY(hadc->Instance);
 
 LL_ADC_SetCalibrationFactor(hadc->Instance, calibration_factor_accumulated);
 LL_ADC_Disable(hadc->Instance);

As above, I spin on ADRDY, no need for a timer

[Motla1]

I reported the issue for STM32CubeWB as well.

The library already have some macros that wait for ADRDY the flag, just remove the LL_ prefix:

 /* Apply calibration factor */
 ADC_Enable(hadc);
 LL_ADC_SetCalibrationFactor(hadc->Instance, calibration_factor_accumulated);
 ADC_Disable(hadc);

https://github.com/STMicroelectronics/STM32CubeWB/issues/59

[John1960]

OK

[MechMan]

Hi. I have this problem with stm32g07, but only -Og compile flag. The solution the same, Delay after Enable or use ADC_Enable

[Dana Myers]

is there a plan / what is the plan to roll out a fix for this? Using FW 1.3 I still see this.

Thanks -

Dana