Timer input capture interrupt fires over and over - Not a late flag reset issue (probably)

[Tobe]

My input capture interrupt is fired over and over again. But when i put a breakpoint in the interrupt, and resume from it, it does not happen anymore.

Timer setup:

	// TIM 3 Inputs (Pedal rot - PB5 / Torc PWM - PB1)
	my_GPIO_InitStruct.Pin = GPIO_PIN_5 | GPIO_PIN_1;
	my_GPIO_InitStruct.Mode = GPIO_MODE_AF_PP;
	my_GPIO_InitStruct.Pull = GPIO_PULLUP;
	HAL_GPIO_Init(GPIOB, &my_GPIO_InitStruct);
	GPIOB->AFR[0] |= (	GPIO_AFRL_AFSEL1_1 |  	// AF2 (all)
						GPIO_AFRL_AFSEL5_1);
	__disable_irq();
	__HAL_RCC_TIM3_CLK_ENABLE();

	SET_BIT(  TIM3->CCMR1,  TIM_CCMR1_CC2S_0);
	SET_BIT(  TIM3->CCMR2, 	TIM_CCMR2_CC4S_0);	// channel is configured as input, tim_ic1 is mapped on tim_ti1, etc...
	// TODO: filter??
	SET_BIT(  TIM3->CCER,   TIM_CCER_CC2E |
							TIM_CCER_CC2NP |
							TIM_CCER_CC2P |
							TIM_CCER_CC4E |
							TIM_CCER_CC4NP |
							TIM_CCER_CC4P);		//  capture enable and Rising and falling edge

	// TISEL default input channels

	SET_BIT(  TIM3->DIER,   TIM_DIER_CC2IE | TIM_DIER_CC4IE); 	// enable interrupt on signal
	SET_BIT(  TIM3->CR1,    TIM_CR1_CEN); 		// enable counter timer 3

	NVIC_EnableIRQ(TIM3_IRQn);				// enable global interrupt
	SET_BIT(TIM3->EGR, TIM_EGR_CC2G);	//force interrupt
	__enable_irq();

Interrupt:

void TIM3_IRQHandler(){
	GPIOC->BSRR = (uint32_t) GPIO_PIN_3;
	if(READ_BIT(TIM3->SR, TIM_SR_CC4IF)){	// Torc PWM interrupt on CH4
		unsigned long captValue = TIM3->CCR4;	// read Capture register (resets interrupt flag)
		unsigned long lastRisingEdge = 0;
		unsigned long lastFallingEdge = 0;

		if(READ_BIT(GPIOB->IDR, GPIO_PIN_1)){	// Rising edge
			pwmDuty = (lastFallingEdge - lastRisingEdge) / (captValue - lastRisingEdge);
			lastRisingEdge = captValue;
		}
		else{	// Falling edge
			lastFallingEdge = captValue;
		}
	}

//	if(READ_BIT(TIM3->SR, TIM_SR_CC2IF)){	// Pedal rotation
//
//
//	}


	SET_BIT(TIM3->SR, 0xFFFFFFFF);//TIM_SR_CC2IF | TIM_SR_CC4IF); 	// Clear any int flags
	delayMicroseconds(1);
	GPIOC->BRR = (uint32_t) GPIO_PIN_3;
}

.I checked the pin, which is not floating nor receiving any relevant noise.

[AScha.3]

Maybe - you just need little delay in INT... read:

http://www.efton.sk/STM32/gotcha/g7.html

[Tobe]

I was thinking about what i did wrong the last time i had such issue. Alltough i am almost doing exactly the same thing, that could lead to this cause, it is not the cause with the current issue.

I added 1us delay after the reset of the flags, and it stil does not help.

[Tobe]

I have edited the post, as it now also happens without the other interrupt setup.

[Tobe]

Interesting:

If i let the pins setup as outputs, the interrupt is still fired.

[Tobe]

This is now the minimum setup, with which i can provoke the error:

	__HAL_RCC_TIM3_CLK_ENABLE();

	SET_BIT(  TIM3->DIER,   TIM_DIER_CC2IE | TIM_DIER_CC4IE); 	// enable interrupt on signal
	SET_BIT(  TIM3->CR1,    TIM_CR1_CEN); 		// enable counter timer 3

	NVIC_EnableIRQ(TIM3_IRQn);				// enable global interrupt

BUT this time, the problem persists after hitting resume!

[Tobe]

After a few more tries i think it has something to do with

SET_BIT(TIM3->EGR, TIM_EGR_CC2G);	//force interrupt

The repeated execution of the interrupt happens every time the interrupt is forced. But still it after resuming in debug, the problem dissappears.

Without forcing the interrupt, there is no issue!

[Tobe]

These are the registers just before the force of the interrupt, and at the beginning of the interrupt:

[waclawek.jan · Accepted solution]

TIMx_SR bits are cleared by writing 0 to them, not 1.

JW

[Tesla DeLorean]

+1 This

if (TIM3->SR & TIM_SR_CC4IF){ // Torc PWM interrupt on CH4
  ...
  TIM3->SR =  ~TIM_SR_CC4IF; // Send the inverse pattern, DO NOT RMW with AND
  ...

}

Do it reasonably early too, so the TIM/NVIC has a chance to clear prior to tail-chaining on exit