TMS320F28335: When PWM Interrupt is triggered from TBCTR = Zero, on every 10th count there's delay in PWM Interrupt trigger

Chandana B V

Part Number: TMS320F28335

I am using EPWM1 module to trigger interrupt at valley(TBCTR = 0) for Every 100us.

My EPWM Period is 10KHz and PWM Base Clock is 150 MHz, I am using triangle PWM(UPDOWN Count mode)

In this case I am toggling the GPIO inside the ISR routine, Every 10th pulse(Every 1ms ) I am getting a 20us delay in the ISR trigger.

I made sure that I am not running any other higher priority interrupt at every 1ms (10th pulse ).

This issue is there only when I trigger the Interrupt @TBCTR = 0, When I trigger the Interrupt at peak ie TBCTR = PRD , I don't see any delay in the ISR trigger, ISR gets executed every 100us properly.

I tried different PWM module other than EPWM1, the issue still persists when I trigger ISR at TBCTR= 0.

When I trigger the Interrupt at TBCTR = some count (count for 1us, count for 2us ) the drift time on every 10th pulse reduces. As I move the ISR trigger far away from TBCTR = 0, the drift eventually disappers.

over 5 years ago

0 Cody Watkins over 5 years ago

TI__Mastermind 42195 points

Chandana,

This is a very interesting effect.

What happens in you try to toggle your GPIO every 50us using the TBCTR= 0 event? Does the delay cut in half, does it remain the same?

Just thinking through the problem, I think this could come from some sort of interrupt preempting or it could be caused by a weird effect inside of the PWM module. I think the first one is more likely. What other code are you running? Do you have other interrupts enabled? Sure there are not higher priority interrupts, but a low priority interrupt could be finishing up around your TBCLK = PRD ISR and is keeping the PWM ISR from starting because interrupts are not nested by default.

Regards,
Cody

0 Chandana B V over 5 years ago in reply to Cody Watkins

Intellectual 340 points

I try running the ISR at 50us and 200us, for 50us the delay occurs at every 20th pulse and for 200us it occurs at every 5th pulse( every 1ms )

I am using DSP/BIOS 5.1 and I have disabled all other interrupts.

If I trigger PWM Interrupt at TBCTR = COMPB, and as I slowly increase the value of COMPB the delay disappears.

0 Cody Watkins over 5 years ago in reply to Chandana B V

TI__Mastermind 42195 points

Chandana B V said:
I try running the ISR at 50us and 200us, for 50us the delay occurs at every 20th pulse and for 200us it occurs at every 5th pulse( every 1ms )

What is the length of the delay at each ISR frequency?

50us delay =?
100us delay =?
200us delay = ?

Chandana B V said:
If I trigger PWM Interrupt at TBCTR = COMPB, and as I slowly increase the value of COMPB the delay disappears.

What CMPB value does the delay completely disappear? As you increase CMPB what is the rate of decay of the delay? (is it linear, logarithmic, or something else ?)

Regards,
Cody

0 Chandana B V over 5 years ago in reply to Cody Watkins

Intellectual 340 points

Cody Watkins said:
What is the length of the delay at each ISR frequency?

50us delay =?

100us delay =?

200us delay = ?

50us delay = ~7us

100us delay = 20us

200us delay = ~8us

What CMPB value does the delay completely disappear? As you increase CMPB what is the rate of decay of the delay? (is it linear, logarithmic, or something else ?)

The delay progression is not consistent, I used COMPB = 105(which is 1.4us) to get the ISR at exactly 100us.

below is the delay for different COMPB values.

COMPB = 105, delay = 0

COMPB = 10 t0 40, delay = 20us

COMPB = 50 t0 90, delay = 7us

COMPB = 90 to 101, delay = ~1us