[S32K]PWM disturbed by SPI transfer

cancel
Showing results for 
Search instead for 
Did you mean: 

[S32K]PWM disturbed by SPI transfer

Jump to solution
533 Views
carlos_kan
Contributor II

Hi

When I want to generate a 20KHz,50% duty cycle PWM by FTM Module(FTM0,CH0,PTB12),and in the main loop there is a 50ms slave watchdog-feeding SPI transfer, the PWM wave is abnormal as below:

(The wave is a MOSFET driver's output(12V) that follows the MCU FTM output)

受SPI干扰PWM波形.jpg

(I guess once the SPI is transmitting, the PWM is stopped or is pulled high)

However, when I comment the SPI transfer code, as that there is no code in for(;;), the PWM becomes normal:

正常PWM波形.jpg

So could someone tell me the reason or how to send SPI frames without disturbing the PWM output?

I use functions from the S32 DS example projects, following is partial code of the project.

main loop:

	for(;;)
	{
		LPSPI0_transmit_24bits( GENSTAT_read_command);
		GENSTAT_read_result = LPSPI0_receive_24bits();
		printf("GENSTAT :0x%06x \n\r",GENSTAT_read_result);
		GENSTAT_read_result = 0;

	}

SPI init:

void LPSPI0_init_master(void) {
  PCC->PCCn[PCC_LPSPI0_INDEX] = 0;          /* Disable clocks to modify PCS ( default) */
  PCC->PCCn[PCC_LPSPI0_INDEX] = 0xC6000000; /* Enable PCS=SPLL_DIV2 (40 MHz func'l clock) */

  LPSPI0->CR    = 0x00000000;   
  LPSPI0->IER   = 0x00000000;   
//  LPSPI0->IER   = 0x00000003;  
  LPSPI0->DER   = 0x00000000;   
  LPSPI0->CFGR0 = 0x00000000;  
  LPSPI0->CFGR1 = 0x00000001;   
  LPSPI0->TCR   = 0x51000017;   /* 16-bit data */
  LPSPI0->CCR   = 0x04090808; 
  LPSPI0->FCR   = 0x00000003;  
  LPSPI0->CR    = 0x00000009;   
}

FTM init:

void FTM0_init(void)
{

	PCC->PCCn[PCC_FTM0_INDEX] &= ~PCC_PCCn_CGC_MASK; 	/* Ensure clk disabled for config 	*/
	PCC->PCCn[PCC_FTM0_INDEX] |= PCC_PCCn_PCS(0b010)	/* Clock src=1, 8 MHz SIRCDIV1_CLK */
                              |  PCC_PCCn_CGC_MASK;  	/* Enable clock for FTM regs 		*/

	/*!
	 * FTM0 Initialization:
	 * ===================================================
	 */
	FTM0->MODE |= FTM_MODE_WPDIS_MASK;  /* Write protect to registers disabled (default) 				*/

	FTM0->SC	=	FTM_SC_PWMEN0_MASK|FTM_SC_PS(2);     	/* TOIE (Timer Overflow Interrupt Ena) = 0 (default) 			*/
										/* CPWMS (Center aligned PWM Select) = 0 (default, up count) 	*/
										/* CLKS (Clock source) = 0 (default, no clock; FTM disabled) 	*/
										/* PS (Prescaler factor) = 7. Prescaler = 128 					*/
	FTM0->COMBINE = 0x00000000;	/* FTM mode settings used: DECAPENx, MCOMBINEx, COMBINEx=0 */
	FTM0->POL = 0x00000000;    	/* Polarity for all channels is active high (default) 		*/
	FTM0->MOD = 100-1 ;     	
}

FTM configuration and start:

void FTM0_CH0_PWM_init(void)
{
	FTM0->CONTROLS[0].CnSC = FTM_CnSC_MSB_MASK
							|FTM_CnSC_ELSB_MASK;  		
    FTM0->CONTROLS[0].CnV =  50;	/* FTM0 ch1 compare value (~75% duty cycle) */
}
void start_FTM0_counter (void)
{
	FTM0->SC |= FTM_SC_CLKS(3);
}

Great thanks.

1 Solution
505 Views
carlos_kan
Contributor II

I  find that the problem is from the PRINTF function. As I comment this line,  the PWM output is always normal and is not influenced by SPI transfer. The project is configurated of ewl_c Console mode, so I can see the information printed by printf in the S32DS console window. I think the printf occupies CPU core, but I do not know the reason. Perhaps someone can tell me why.

 

View solution in original post

4 Replies
318 Views
Lauch_Spargel
Contributor II

I had the same problem with LPUART and PWM.

Removing printf also worked, thanks!

Best regads,

Lauch_Spargel

0 Kudos
517 Views
Robin_Shen
NXP TechSupport
NXP TechSupport

Hi Carlos,

Have you try to debug the project and find any issue? LPSPI should not interfere with the PWM waveform. Would you please attached a simple test project here? So that I can check it on S32K EVB.

Best Regards,
Robin

0 Kudos
503 Views
carlos_kan
Contributor II

1.png

506 Views
carlos_kan
Contributor II

I  find that the problem is from the PRINTF function. As I comment this line,  the PWM output is always normal and is not influenced by SPI transfer. The project is configurated of ewl_c Console mode, so I can see the information printed by printf in the S32DS console window. I think the printf occupies CPU core, but I do not know the reason. Perhaps someone can tell me why.