Forums

All:

Just spotted a flaw in my logic - the PSCDIV does not allow for divide by 5, it works in powers of 2 from divide-by-2 to divide-by-8192...

Todd Anderson,

Have you seen the BIOS PRD module?  PRD is for "periodic" and will let you create a function that runs periodically.  You can configure how often you want it to run (period) as well as which function to run.

You can create a new PRD from within the BIOS config file (*.tcf).  Just right click on "PRD - Periodic Function Manager" and then select "insert new PRD".  Then, you can configure the PRD instances' properites:

Steve

Steve:

Thanks for your input!

Yes, I am aware of the PRD module, but to get the granularity needed to go 8.33 msec between periods would mean that the "usual" tick of 1 msec would suffer. I actually found another way of doing the precedure, so I do not need the timer. However, the question remains - if using a general purpose timer for the DSP-BIOS timer tick, are other GP timers available? OR is it assume that everything will be done under the Bios umbrella?

If I choose to use one of the other GP timers, how do I handle interrupts?

Todd Anderson,

Did you enable the GP timer interrupt?

If so, then you should be able to map that interrupt to a HWI in BIOS.  You would configure the HWI whose number matches the GP timer interrupt to run the function (ISR) you define.

Also, I not sure if you've already seen this but it came up in a Google search: http://www.ti.com/lit/ug/spru595c/spru595c.pdf

Steve

Thanks Steve.

However, I see some problems yet. In the DSP-BIOS GUI, I have a choice of 3 timers to use. If I pick one for DSP-BIOS (say Timer 0), BIOS creates the necessary interrupt service routine to use when servicing "timer ticks." It then looks like I have the choice to use the other 2 timers, if necessary.

If I use Timer 1, AND I also need to have associated interrupts, then I need to be able to create an interrupt service routine that will distinguish between Timer0 interrupts and Timer1 interrupts, because of the "timer interrupt aggregation." In that interrupt service routine, I would need to "Branch" to the Timer0 interrupt service routine if it was a Timer0-associated interrupt or "Branch" to the Timer1 interrupt service routine for the Timer1-associated interrupt.

Why is it necessary to use a general purpose timer if I can use PRDs inside of DSP-BIOS? Usually, BIOS events are clocked at a rate of around 1 msec timer ticks, and the PRDs can be set up for larger timings, around increments of milliseconds. There are times that a specific rate is needed - like 8.33 msec - to create a timer tick that would get to that would probably complicate larger timings, create more code, and add additional overhead in terms of interrupt timings.

All of this assumes that I can use Timer1 when Timer0 is being used for DSP-BIOS. The question I still need answered:

If Timer0 is being used by DSP-BIOS, do I have access to Timer1 and Timer2?

Steve:

Thanks for you past answer. I noticed that the DSP/BIOS 5.42.00.07 inserts a call to _CLK_dispatch into HWI_4.

Because it is grayed out, I cannot set up a different routine that would distinguish between the 3 interrupt sources. So, it looks like I cannot use timer interrupts for timer1 or timer2 if I am using timer0 as a BIOS timer tick.