Inversion of trigger events for the ADC and HRTIM

Context

While debugging the current measurements of the Twist board, we have noticed that the HRTIM triggered events on the wrong leg (HRTIMA on leg 2 and HRTIMC on leg 1).

Description

The HRTIM and ADC are synchronized by a system of triggers. These triggers are set on both sides with adc1 and adc2 reading triggers from event1 and event3 respectively, as can be seen by this code extract from adc_config.h

void _initialize()
{
	if (initialized == 0)
	{
		// Initialize the ADCs
		adc_init();
		initialized = 1;

		// Perform default configration
		configure_adc_trigger_source(1, hrtim_ev1);
		configure_adc_trigger_source(2, hrtim_ev3);
		configure_adc_trigger_source(3, software);

		adc_configure_discontinuous_mode(1, 1);
		adc_configure_discontinuous_mode(2, 1);
	}
}

From the HRTIM however, the trigger is set at the hrtim_common.c file. Here we can see this being set for the center aligned application.

void _hrtim_init_events_center_aligned(hrtim_tu_t leg1_tu, hrtim_tu_t leg2_tu)
{
	if(leg1_tu == TIMA && leg2_tu == TIMB){
	// setting the adc roll-over mode on period event
	LL_HRTIM_TIM_SetADCRollOverMode(HRTIM1, LL_HRTIM_TIMER_A, LL_HRTIM_ROLLOVER_MODE_PER);
	LL_HRTIM_TIM_SetADCRollOverMode(HRTIM1, LL_HRTIM_TIMER_B, LL_HRTIM_ROLLOVER_MODE_PER);
	
	// setting adc trigger
	hrtim_adc_trigger_en(1, 1, LL_HRTIM_ADCTRIG_SRC13_TIMAPER);
	hrtim_adc_trigger_en(3, 2, LL_HRTIM_ADCTRIG_SRC13_TIMBPER);
	}else if(leg1_tu == TIMA && leg2_tu == TIMC){
	// setting the adc roll-over mode on period event
	LL_HRTIM_TIM_SetADCRollOverMode(HRTIM1, LL_HRTIM_TIMER_A, LL_HRTIM_ROLLOVER_MODE_PER);
	LL_HRTIM_TIM_SetADCRollOverMode(HRTIM1, LL_HRTIM_TIMER_C, LL_HRTIM_ROLLOVER_MODE_PER);
	
	// setting adc trigger
	hrtim_adc_trigger_en(1, 1, LL_HRTIM_ADCTRIG_SRC13_TIMAPER);
	hrtim_adc_trigger_en(3, 3, LL_HRTIM_ADCTRIG_SRC13_TIMCPER);
	}
}

However, this does not work as is because the HRTIMA triggers the adc2 and HRTIMC triggers the adc1.

Solution

We have to find the source of the inversion. Is it an inversion? Or is it so that HRTIMA will always trigger the event 3 and HRTIM C will always trigger the event 1?

assigned to @cfoucher

Hello @lflavado,

(very quick investigation, not sure of the outcome)

I can see that in commit #e1fb48bf, changes were made to the initialization. Notably, as you can see here, event generation is now conditional.

However, no change were made to the _initialize() function of the ADC module, which always triggers on the same source. I suspect this could be the source of the error.

Actually, the _initialize() function only does initial configuration with default values.

For now, you may want to call hwConfig.configureAdcTriggerSource() to overcome the issue. In a more generic way, the trigger source should be auto-detected, as it currently depends on the hardware version :

void HardwareConfiguration::setBoardVersion(hardware_version_t hardware_version)
{
	HardwareConfiguration::hardware_version = hardware_version;

	if (hardware_version == O2_v_1_1_2 || hardware_version == O2_v_0_9)
	{
		uart_lpuart1_swap_rx_tx();
		hrtim_leg_tu(TIMA, TIMB);
	}else if(hardware_version == SPIN_v_0_1){
		uart_lpuart1_swap_rx_tx();
		hrtim_leg_tu(TIMA, TIMC);
	}else if(hardware_version == SPIN_v_0_9 || hardware_version == TWIST_v_1_1_2){
		hrtim_leg_tu(TIMA, TIMC);
	}else if(hardware_version == nucleo_G474RE){
		hrtim_leg_tu(TIMA, TIMB);
	}

}

I guess the issue lies here: depending on the hardware version, we use different timer units, and thus trigger different events.

closed