OMAP L137 timer plus 32-bit Chained mode with period reload

Hirofumi Fujita

Intellectual 1780 points

Other Parts Discussed in Thread: OMAPL138

Hi,

I posted below, but let me also post here.

http://e2e.ti.com/support/dsp/tms320c6000_high_performance_dsps/f/115/t/138467.aspx

I faced the trouble to use timer plus on OMAPL137, in 32bit chained mode with period reload.

As I tried, like below, timer never started counting. But,

- 32bit Unchained mode with period reload was OK

- 32bit Chained mode with continuous operation was OK

So I'm suspicious that timer plus on OMAPL147 supports 32bit chained mode with period reload.

Do you have any idea?

-----------------------------------------------------

    tmr0Regs->TGCR = CSL_FMKT(TMR_TGCR_TIMMODE,32_CHAIN)
                     | CSL_FMKT(TMR_TGCR_PLUSEN,ENABLE)
                     | CSL_FMKT(TMR_TGCR_TIM12RS,NO_RESET)
                     | CSL_FMKT(TMR_TGCR_TIM34RS,NO_RESET)
                     | CSL_FMK(TMR_TGCR_PSC34,0x1);

    CSL_FINS(tmr0Regs->PRD12,TMR_PRD12_PRD12,TIMER_PERIOD);

    CSL_FINS(tmr0Regs->PRD34,TMR_PRD34_PRD34,TIMER_PERIOD);

    CSL_FINS(tmr0Regs->REL12,TMR_REL12_REL12,TIMER_PERIOD);
    CSL_FINS(tmr0Regs->REL34,TMR_REL34_REL34,TIMER_PERIOD);

CSL_FINST(tmr0Regs->TCR,TMR_TCR_ENAMODE12,EN_CONTRELOAD);

------------------------------------------------------

thanks & best rgards,

Hirofumi Fujita

over 13 years ago

0 Miguel A. Alanis over 13 years ago

TI__Genius 9490 points

Hirofumi,

it is interesting to hear this.

OMAPL137 only has 2 timer instances.

Can you please try the following:

Just to double check, are you able to see the actual TCR Register offset (in CCS) and check to make sure that the TCR enamode12 bits are properly getting set?

Also, have you tried this on the TMR1 and see if you see any changes?

regards,

miguel

0 Hirofumi Fujita over 13 years ago in reply to Miguel A. Alanis

TI__Intellectual 1780 points

Hi Miguel,

Thank you for the comment. I checked your suggestion, but unfortunately, both timer 0 & timer 1 didn't work as I expected.

But I found that, I could get the expected result when I use simulator. That means, my software should work properly, I think.

And also, on the OMAPL138, the same problem were there, i.e.

32bit chained with continuous works,

32bit chained with continuous reload doesn't work

best regards,

Hirofumi Fujita

0 Miguel A. Alanis over 13 years ago in reply to Hirofumi Fujita

TI__Genius 9490 points

Fujita-san,

Did this feature work on omapl138 timer3 instance?

This smells like a potential issue either w/silicon or with documentation. Can you please try on timer3 instance? also can you please fwd me your ccs project to see if I can reproduce on our local environment.

regards,

miguel

0 Hirofumi Fujita over 13 years ago in reply to Miguel A. Alanis

TI__Intellectual 1780 points

Hi Miguel,

I tried timer3 in OMAPL138, but unfortunately, it didn't work.

And I sent my projects to you by separate mail, so could you try those?

best regards,

Hirofumi Fujita

0 Hirofumi Fujita over 13 years ago in reply to Hirofumi Fujita

TI__Intellectual 1780 points

It looks like that, Miguel is out of office for a while..

Does anyone have any idea about this issue?

I put the test project I use on this thread.

5554.test_projects.zip

The help from the expert will be highly appreciated.

thanks in advance & best regards,

Hirofumi Fujita

0 Miguel A. Alanis over 13 years ago in reply to Hirofumi Fujita

TI__Genius 9490 points

Hirofumi,

Sorry for the delay, but I have not had a chance to look at this yet until now, as I have been very busy after my return from timeoff. Today I have reproduced your observations on my OMAPL138 EVM with a sample code we had from our validation guys and I ran on all 4 timer instantiations and I observe the same results as you.

-TIMER CHAINED continuous w/reload mode did not even started counting.

-Timer chained continous w/out reload mode worked ok.

-Timer UNCHAINED continuous w/reload mode worked ok

I am currently trying to get this confirmed by the design team (which is working on other projects for other business unit so we might not get a quick response). Once the design team can re-confirm we will probably update the errata or the timer documentation to remove this feature.

regards,

miguel