Timer slave trigger polarity for input capture

[Yuan1]

I want to use TIM1_CH1 to measure pulse width. TIM1 works under slave reset mode. The input trigger is TI1FP1. Ideally, TIM1 resets upon rising input, then TIM1_CH1 captures the falling edge. The CCR1 value would be the pulse width. The problem is that there is no option to set TriggerPolarity. In the code generated by cubeMX, TriggerPolarity is the same as ICPolarity. I manually set TriggerPolarity to RISING. The code compiles, but the measured pulse width does not seem right. I must have missed something here.

sSlaveConfig.SlaveMode = TIM_SLAVEMODE_RESET;
 sSlaveConfig.InputTrigger = TIM_TS_TI1FP1;
 sSlaveConfig.TriggerPolarity = TIM_INPUTCHANNELPOLARITY_FALLING;
 sSlaveConfig.TriggerFilter = 0;
 if (HAL_TIM_SlaveConfigSynchro(&htim1, &sSlaveConfig) != HAL_OK)
 {
 Error_Handler();
 }
 sMasterConfig.MasterOutputTrigger = TIM_TRGO_RESET;
 sMasterConfig.MasterSlaveMode = TIM_MASTERSLAVEMODE_DISABLE;
 if (HAL_TIMEx_MasterConfigSynchronization(&htim1, &sMasterConfig) != HAL_OK)
 {
 Error_Handler();
 }
 sConfigIC.ICPolarity = TIM_INPUTCHANNELPOLARITY_FALLING;
 sConfigIC.ICSelection = TIM_ICSELECTION_DIRECTTI;
 sConfigIC.ICPrescaler = TIM_ICPSC_DIV1;
 sConfigIC.ICFilter = 0;
 if (HAL_TIM_IC_ConfigChannel(&htim1, &sConfigIC, TIM_CHANNEL_1) != HAL_OK)
 {
 Error_Handler();
 }

[waclawek.jan]

Which STM32?

When it comes to anything beyond "usual", Cube is not helpful but gets into way. Read the TIM chapter in manual and act accordingly.

In particular, for one channel, there's only one place where polarity is set: in TIMx_CCER.CCxP/CCxNP. This is regardless of whether this channel is used in slave-mode controller, or for input capture.

What you want to achieve is described in RM as "input PWM mode", and you need both CH1 and CH2 for that - CH1 for slave-mode controller to reset, and CH2 to capture - CH2 set to opposite edge in TIMx_CCER.CC2P as compared to CC1P, and CH2 set to capture on the CH1 signal by setting TIMx_CCMR1.CC2S to 0b10.

JW

PS. I can't find HAL_TIM_SlaveConfigSynchro() in older CubeF4 v1.21 which I pesonally use as a reference; could find it in current CubeF4 - this in itself should raise flags as per stability of Cube's API

[Yuan1]

The chip is F4CCU6 with the latest hal firmware (1.26.2).

I tried the "input PWM mode" example. It works fine. In that example, two channels are used to measure both frequency and duty cycle. A rising TI2FP2 triggers timer into reset, then channel 1 captures the falling edge, and channel 2 captures the rising edge. The code in the official example does not set trigger polarity.

STM32CubeF4/main.c at master · STMicroelectronics/STM32CubeF4 (github.com)

/*##-2- Configure the Input Capture channels ###############################*/ 
 /* Common configuration */
 sConfig.ICPrescaler = TIM_ICPSC_DIV1;
 sConfig.ICFilter = 0; 
 
 /* Configure the Input Capture of channel 1 */
 sConfig.ICPolarity = TIM_ICPOLARITY_FALLING;
 sConfig.ICSelection = TIM_ICSELECTION_INDIRECTTI; 
 if(HAL_TIM_IC_ConfigChannel(&TimHandle, &sConfig, TIM_CHANNEL_1) != HAL_OK)
 {
 /* Configuration Error */
 Error_Handler();
 }
 
 /* Configure the Input Capture of channel 2 */
 sConfig.ICPolarity = TIM_ICPOLARITY_RISING;
 sConfig.ICSelection = TIM_ICSELECTION_DIRECTTI;
 if(HAL_TIM_IC_ConfigChannel(&TimHandle, &sConfig, TIM_CHANNEL_2) != HAL_OK)
 {
 /* Configuration Error */
 Error_Handler();
 }
 /*##-3- Configure the slave mode ###########################################*/
 /* Select the slave Mode: Reset Mode */
 sSlaveConfig.SlaveMode = TIM_SLAVEMODE_RESET;
 sSlaveConfig.InputTrigger = TIM_TS_TI2FP2;
 if(HAL_TIM_SlaveConfigSynchronization(&TimHandle, &sSlaveConfig) != HAL_OK)
 {
 /* Configuration Error */
 Error_Handler();
 }

Manually configure the timer in cubeMX, the generated code assigns RISING to TriggerPolarity.

sSlaveConfig.SlaveMode = TIM_SLAVEMODE_RESET;
 sSlaveConfig.InputTrigger = TIM_TS_TI2FP2;
 sSlaveConfig.TriggerPolarity = TIM_INPUTCHANNELPOLARITY_RISING;
 sSlaveConfig.TriggerFilter = 0;
 if (HAL_TIM_SlaveConfigSynchro(&htim4, &sSlaveConfig) != HAL_OK)
 {
 Error_Handler();
 }
 sMasterConfig.MasterOutputTrigger = TIM_TRGO_RESET;
 sMasterConfig.MasterSlaveMode = TIM_MASTERSLAVEMODE_DISABLE;
 if (HAL_TIMEx_MasterConfigSynchronization(&htim4, &sMasterConfig) != HAL_OK)
 {
 Error_Handler();
 }
 sConfigIC.ICPolarity = TIM_INPUTCHANNELPOLARITY_FALLING;
 sConfigIC.ICSelection = TIM_ICSELECTION_INDIRECTTI;
 sConfigIC.ICPrescaler = TIM_ICPSC_DIV1;
 sConfigIC.ICFilter = 0;
 if (HAL_TIM_IC_ConfigChannel(&htim4, &sConfigIC, TIM_CHANNEL_1) != HAL_OK)
 {
 Error_Handler();
 }
 sConfigIC.ICPolarity = TIM_INPUTCHANNELPOLARITY_RISING;
 sConfigIC.ICSelection = TIM_ICSELECTION_DIRECTTI;
 if (HAL_TIM_IC_ConfigChannel(&htim4, &sConfigIC, TIM_CHANNEL_2) != HAL_OK)
 {
 Error_Handler();
 }

In this presentation en.STM32L4_WDG_TIMERS_GPTIM.pdf, on page 30, it says "Reset mode: a rising edge on TRGI reinitializes the counter". Maybe this is why in cubeMX there is no option to set TriggerPolarity, because only a rising edge can reset the counter. However, if you set the input capture polarity of channel 2 to falling, cubeMX will assign falling to TriggerPolarity. Strange...

sSlaveConfig.SlaveMode = TIM_SLAVEMODE_RESET;
 sSlaveConfig.InputTrigger = TIM_TS_TI2FP2;
 sSlaveConfig.TriggerPolarity = TIM_INPUTCHANNELPOLARITY_FALLING;
 sSlaveConfig.TriggerFilter = 0;
 if (HAL_TIM_SlaveConfigSynchro(&htim4, &sSlaveConfig) != HAL_OK)
 {
 Error_Handler();
 }
 sMasterConfig.MasterOutputTrigger = TIM_TRGO_RESET;
 sMasterConfig.MasterSlaveMode = TIM_MASTERSLAVEMODE_DISABLE;
 if (HAL_TIMEx_MasterConfigSynchronization(&htim4, &sMasterConfig) != HAL_OK)
 {
 Error_Handler();
 }
 sConfigIC.ICPolarity = TIM_INPUTCHANNELPOLARITY_RISING;
 sConfigIC.ICSelection = TIM_ICSELECTION_INDIRECTTI;
 sConfigIC.ICPrescaler = TIM_ICPSC_DIV1;
 sConfigIC.ICFilter = 0;
 if (HAL_TIM_IC_ConfigChannel(&htim4, &sConfigIC, TIM_CHANNEL_1) != HAL_OK)
 {
 Error_Handler();
 }
 sConfigIC.ICPolarity = TIM_INPUTCHANNELPOLARITY_FALLING;
 sConfigIC.ICSelection = TIM_ICSELECTION_DIRECTTI;
 if (HAL_TIM_IC_ConfigChannel(&htim4, &sConfigIC, TIM_CHANNEL_2) != HAL_OK)
 {
 Error_Handler();
 }

[waclawek.jan]

Again, whatever is clicked in CubeMX is not that relevant; it's the registers content what controls the timer's behaviour. So read them out and check/post if in doubt.

If input to slave-mode controller is taken from CH1, polarity is controlled by CH1's setting, see description of TIMx_CCER.CC1P. TRGI is output from multiplexer which is after the CH1 polarity selector.

When testing on actual hardware, does it work otherwise?

JW

[Yuan1]

Yes, you are right. TI1FP1 is the same signal feeding both the slave controller and input capture module. That's why their polarity is always the same. Using one channel under reset mode can only capture the rising edge.

[Yuan1]

The problem was solved by selecting TI1_ED as the trigger source. TriggerPolarity is RISING, while ICPolarity is falling.

sSlaveConfig.SlaveMode = TIM_SLAVEMODE_RESET;
 sSlaveConfig.InputTrigger = TIM_TS_TI1F_ED;
 sSlaveConfig.TriggerPolarity = TIM_INPUTCHANNELPOLARITY_RISING;
 sSlaveConfig.TriggerFilter = 0;
 if (HAL_TIM_SlaveConfigSynchro(&htim1, &sSlaveConfig) != HAL_OK)
 {
 Error_Handler();
 }
 sMasterConfig.MasterOutputTrigger = TIM_TRGO_RESET;
 sMasterConfig.MasterSlaveMode = TIM_MASTERSLAVEMODE_DISABLE;
 if (HAL_TIMEx_MasterConfigSynchronization(&htim1, &sMasterConfig) != HAL_OK)
 {
 Error_Handler();
 }
 sConfigIC.ICPolarity = TIM_INPUTCHANNELPOLARITY_FALLING;
 sConfigIC.ICSelection = TIM_ICSELECTION_DIRECTTI;
 sConfigIC.ICPrescaler = TIM_ICPSC_DIV1;
 sConfigIC.ICFilter = 0;
 if (HAL_TIM_IC_ConfigChannel(&htim1, &sConfigIC, TIM_CHANNEL_1) != HAL_OK)
 {
 Error_Handler();
 }

[waclawek.jan]

For TI1F_ED, polarity is effectively ignored, as that triggers at *both* edges. But, if you are interested only in one pulse's length, the reset from falling edge is effectively ignored as capture happens before the reset, and there's no capture until after the subsequent rising edge - so you are right, this is a way to do it with only CH1.

Please select your post as Best so that the thread is marked as solved.

JW