timer interrupt rate not accurate (GPIO debug line)

Robert Wolfe

Other Parts Discussed in Thread: OMAP-L138

Hello,

I want to use a Timer 0, to generate an interrupt at 50 kHz. So I've set the appropriate PRD = ( ( 24 MHz / 50 KHz ) - 1 ), and put a GPIO toggle in the related interrupt that occurs, for that Timer 0, 12, and am checking it on a scope. But it comes out to 45.88 KHz on a scope, hooked up to that GPIO pin. I've confirmed the input AUXCLK used to drive the TIMER 0, 12 counting, as 24 MHz. For that, the PRD should be 479, which it is. So I'm not sure why the interrupt is not occurring at 50 KHz, as it should be, and so any help with trying to figure that out why would be appreciated. Is the 50 KHz interrupt rate too quick, for the processor to keep up? Would the act of toggling the GPIO slow it down enough to result in that?

Regards,

Robert

over 5 years ago

0 Rex Chang over 5 years ago

TI__Guru 50170 points

Hi, Robert,

I assume this is the same device of OMAP-L138/C6748 DSP as the other thread submitted by you.

Rex

0 Robert Wolfe over 5 years ago in reply to Rex Chang

Mastermind 7860 points

Hi Rex,

It is! Same effort for both ... trying to find out why the interrupt is occurring at less than the prescribed 50 KHz. At first, I wondered if it was because of drops, but maybe not ... I had instrumented the TIMER 0, 12 interrupt to look for that DROP bit in INTXSTAT, and incremented a global counter, but it only had a value of 1 ... probably some startup condition. So this thread was for any other ideas on what to look at.

Robert

0 Robert Wolfe over 5 years ago in reply to Robert Wolfe

Mastermind 7860 points

I took out the GPIO driver, from TIMER 0, 12 interrupt, but the rate is still less than expected, measured using TIMER 0, 34 (delta counts each time into the 12 interrupt). So it doesn't appear to be GPIO delay-related. I routed AUXCLK out the CLKOUT pin, and saw the 24 MHz expected, so I'm not sure why the ( 24 MHz / 50 KHz ) - 1 ) PRD value is not working. That is a value of 479, but TIMER 0, 34 delta counts each time into the interrupt are around 523.

0 Robert Wolfe over 5 years ago in reply to Robert Wolfe

Mastermind 7860 points

picture of the 24 MHz AUXCLK driving the timer.

0 Robert Wolfe over 5 years ago in reply to Robert Wolfe

Mastermind 7860 points

I had logic in the interrupt, that cleared the TIM12 ... holdover from a previous implementation. Once removing that, the interrupt appears to be occurring on time, as expected, confirmed via GPIO debug.

Processors

Processors forum

timer interrupt rate not accurate (GPIO debug line)