STM32F446 inverse PWM duty cycle?!?

[Matt Blessinger]

Posted on March 17, 2017 at 02:59The original post was too long to process during our migration. Please click on the attachment to read the original post.

[nichtgedacht]

Posted on March 17, 2017 at 16:04

Hi

GB_H_Pin GA_L_Pin  GA_H_Pin  GC_L_Pin  GC_H_Pin  GB_L_Pin

GA_H_GPIO_Port GC_L_GPIO_Port

are looking special.

Was it generated with CubeMX?

The names do not match for one member each.

Dieter

[Matt Blessinger]

Posted on March 17, 2017 at 16:37

 ,

 ,

Hi,

Yes, I generated the initial code with CubeMX. I've since then modified the timer initialization functions to what I previously tested on the Nucleo. The names are defined in main.h and shown below.

♯ define

GC_L_Pin GPIO_PIN_7

♯ define

GC_L_GPIO_Port GPIOC

♯ define

GC_L TIM_CHANNEL_2

♯ define

GC_H_Pin GPIO_PIN_8

♯ define

GC_H_GPIO_Port GPIOC

♯ define

GC_H TIM_CHANNEL_3

♯ define

GB_L_Pin GPIO_PIN_9

♯ define

GB_L_GPIO_Port GPIOC

♯ define

GB_L TIM_CHANNEL_4

♯ define

GB_H_Pin GPIO_PIN_9

♯ define

GB_H_GPIO_Port GPIOA

♯ define

GB_H TIM_CHANNEL_2

♯ define

GA_L_Pin GPIO_PIN_10

♯ define

GA_L_GPIO_Port GPIOA

♯ define

GA_L TIM_CHANNEL_3

♯ define

GA_H_Pin GPIO_PIN_11

♯ define

GA_H_GPIO_Port GPIOA

♯ define

GA_H TIM_CHANNEL_4

[S.Ma]

Posted on March 17, 2017 at 18:35

In an old code, the PWM generation was setup like this:

void SetTimerOutputCC_SingleEdge(Timer_t* Timer, u32 n, u32 Value_lsb) { Timer->EdgesSize[n] = 0; // Non zero triggers the DMA mode for multiple pulses Timer->EdgesTableAdr[n] = 0;

TIM_OCInitTypeDef OC;

TIM_OCStructInit(&OC); OC.TIM_OCMode = TIM_OCMode_PWM2; OC.TIM_OutputState = TIM_OutputState_Enable; OC.TIM_Pulse = Value_lsb; OC.TIM_OCPolarity = TIM_OCPolarity_Low; TIM_OCnInit(Timer->TIM, n, &OC);}

Or when starting the code, after enabling the timer clock, force a

__HAL_RCC_TIM1_FORCE_RESET();

__HAL_RCC_TIM1_RELEASE_RESET();

This should make the peripheral closer to the HW Power on reset.

Or the PWM depends on the initial pin level?

[waclawek.jan]

Posted on March 17, 2017 at 23:27TIM_OC_InitTypeDef

sConfigOC;

[...]

  sConfigOC.

OCMode

= TIM_OCMODE_PWM1;

  sConfigOC.

Pulse

= 10;

  sConfigOC.

OCPolarity

= TIM_OCPOLARITY_HIGH;

  sConfigOC.

OCFastMode

= TIM_OCFAST_DISABLE;

if

(HAL_TIM_PWM_ConfigChannel(&htim, &sConfigOC, TIM_CHANNEL_2)

You have to have initialized *all* members of

TIM_OC_InitTypeDef struct before calling HAL_TIM_PWM_ConfigChannel. This for Advanced timers (i.e. TIM1 and TIM8) among others includes the polarity of the inverted output. You surely already know this from the concise and detailed 'library' manual.

JW

[Matt Blessinger]

Posted on March 18, 2017 at 20:06

I tried the force reset and release to no success. In fact, it made it so channels 2 & 3 were always inverted instead of randomly inverting.

[Matt Blessinger]

Posted on March 18, 2017 at 20:09

Well, that did it! The CubeMX code has the inverted output config calls as well, but I deleted them since I didn't have them during my testing. Thank you!

This is the code that works:

  sConfigOC.

OCMode

= TIM_OCMODE_PWM1;

  sConfigOC.

Pulse

= 100;

  sConfigOC.

OCPolarity

= TIM_OCPOLARITY_HIGH;

  sConfigOC.

OCNPolarity

= TIM_OCNPOLARITY_HIGH;

  sConfigOC.

OCFastMode

= TIM_OCFAST_DISABLE;

  sConfigOC.

OCIdleState

= TIM_OCIDLESTATE_RESET;

  sConfigOC.

OCNIdleState

= TIM_OCNIDLESTATE_RESET;

[Bumsik Kim]

Posted on April 07, 2017 at 03:30

I had the exactly same problem as OP, and I fixed the problem too. OCNPolarity was problem. I only missed that field in the previous code.