Help understanding Timer0 HFINTOSC interrupts and it not keeping up.

[Roger Jönsson]

Trying out and trying to understand Timer0 HFINTOSC interrupts.
When making it fire fast (HFINTOSC/64) it seems to go into a wall and the output of the subroutine does not fire any quicker. I'm guessing it has to do with either the interrupt-call or resetting things?
Is it GCbasic having to to a bunch of stuff on and after each reset or is it hardware related?

Prescaler divisions:
PRE0_256 fires each 16us
PRE0_128 fires each 8us
PRE0_64 fires each 8us
PRE0_32 fires each 7us

Interrupt flag reset TMR0IF = 0 has no effect, so I'm guessing gcbasic handles it in some other way.

For my current project PRE0_128 is enough, so not much of a thing at this time other than trying to understand, but Is there a more efficitent way?

#CHIP 16F15376,32
#OPTION EXPLICIT

DIR PORTA.7 out

'1.InitTimer0 source (osc or ext), prescaler, clocksource, postscaler
'2.SetTimer ( 0, byte_value, value), or
'  SetTimer ( 0, word_value (where the high byte sets the timer), value)
'3.StartTimer 0
'  (optionally use ClearTimer 0)

InitTimer0 OSC, PRE0_256 + TMR0_HFINTOSC, POST0_1 '32MHz/256/1=125KHz
Settimer (0, 0x0100)'125/2 = 62,5KHz
StartTimer 0

On Interrupt Timer0Overflow  call MyISR

Do
Loop

Sub MyISR
PORTA.7 = !PORTA.7 
PORTA.7 = !PORTA.7 
'TMR0IF = 0  
End Sub

[Anobium]

For the Timer0 it is a 16bit timer. So, the program needs to have added.

'Firstly tell the compiler to use a 16bit timer for this program
#DEFINE TMR0_16BIT

---

There are some demos for this chip - see C:\GCstudio\gcbasic\demos\Vendor_Boards\MPLAB_Xpress_Board_PIC16F15376\09_Interrupt_using_Timer0_to_LED.gcb and all the files in the same folder.

---

You are correct. TMR0IF is handled by GCBASIC. See PICInfo ( Interrupt Tab ) for the enable and event bits that are managed by GCBASIC.

[Roger Jönsson]

I tried the 16-bit timer yesterday without any improvement on these quick repetitions. Also (if I understand it correctly) I have to set the 16-bit SETtimer value (FFFF = shortest time) after the code in the subroutine called by the interrupt (at least I couldn't make it run the code inside otherwise) so I guess it skips out of that subroutine immediately and prepares for a new interrupt) and so the longer the length of the code in the subroutine the more it hurts the cycle time.

At the moment I am generating more than one square wave simultaneously (loops with counters adding 1 for each interrupt) and it works beautifully in the 8bit mode and the frequencies stays the same with more generator loops added. I have no idea how the flag reset (tex) is done, but it doesn't seem to come in conflict with my tone/voice generators generated.

In 16 bit mode (the only way I got it to work, setting the timer after the "voices" generated at the end of the subroutine), the more "voices" added the lower the cycle time. I couldn't get the interrupt cycle time much faster than in 8-bit mode unless the subroutine called was practically empty.

I will have a look at the demos and see if there is something useful.

[Anobium]

May be I do not understand. What time frequency are you requiring? This is a fact I do not know.9

I think the sub routine you are referring to is the interrupt handler, If yes, then, you will have STEPS_TO_DETERMINE_IT_IS_THE_CORRECT_ISR + YOUR_ISR. This is the latency.

If TEX is TMR0IF then this is completed automatically in the ASM. See the generated ASM file for this logic,

[Roger Jönsson]

It's about few KHz output tops from the code in the interrupt handler subroutine. I don't really require much faster at this time, but a faster cycle time yeilds better frequency resolution (more counter laps to set the required frequency) so I tested the limits. The steady stream of interupts is like a master clock for the tone generators I described. I stumbled on the fact that it stalled at 8us figuring that there was place for many instructions during that time and just wondered why it stalled. -Well all is relative. :)

So what I am doing is producing more than one square wave with independent frequencies simultantously and I can run some other code in between interrupts. I haven't decided what to use it for yet. -A one chip music box maybe? Controlling a big bank of PWM servos? Not quite sure yet, but it is like multitasking so it may yeild something useful.

Here I make two independent stable frequency squarewaves which I turn on and off "by remote" in the eternal do-loop, by starting and stopping the timer. For fun I mixed the tow squarewaves together so I can hear both with a small speaker connected to RA3.

#CHIP 16F15376,32
#OPTION EXPLICIT

DIR PORTA.7 out
DIR PORTA.6 out
DIR PORTA.5 out
DIR PORTA.3 out
dim ton1 as Byte
dim ton2 as Byte
dim mixer as byte

InitTimer0 OSC, PRE0_256 + TMR0_HFINTOSC, POST0_1 ''32MHz/256/1=125KHz
Settimer (0, 0x0100) '125/2 = 62,5KHz
StartTimer 0

On Interrupt Timer0Overflow  call MyISR


Do            'beep, beep, beep....
StopTimer 0
wait 500 ms
InitTimer0 OSC, PRE0_256 + TMR0_HFINTOSC, POST0_1 ''32MHz/256/1=125KHz
Settimer (0, 0x0100) '125/2 = 62,5KHz
StartTimer 0
On Interrupt Timer0Overflow  call MyISR
wait 500 ms
Loop



Sub MyISR   'Play two (or more) simultaneous squarewaves.
PORTA.7 = 1 'output to monitor how often the interrupt fires
PORTA.7 = 0

ton1 = ton1 + 1
if ton1 = 127 then 'time between polarity shifts = about 246Hz 
PORTA.6 = !PORTA.6
ton1 = 0
end if

ton2 = ton2 + 1
if ton2 = 100 then '= about 312Hz squarewave output on PORTA.5
PORTA.5 = !PORTA.5
ton2 = 0
end if

mixer = mixer + 1
if mixer = 1 then  'multiplexes between the tones
PORTA.3 = PORTA.6
end if
if mixer = 2 then  
PORTA.3 = PORTA.5
mixer = 0          'resets/nulls the mixer counter on last ton
end if

End Sub

[Roger Jönsson]

There is no Vendor_Boards folder inside my C:\GCstudio\gcbasic\demos
just two GCB Files, a read me and "this_is_useful_list_of_tools_for_the_ide"

[Anobium]

Have you installed the Demos? You can download from GCSTUDIO front page.

[Roger Jönsson]

I installed the GCStudio IDE complete package couple of months ago and got it updated a week or so ago. I looked at both gcbasic.sourceforge.io and gcbasic.com but I could not find a demos package, but I now did found the files in Denomstration_sources behind "Learn From Code" at gcbasic.com.

[Anobium]

Here is the link from GCSTUDIO,

[Roger Jönsson]

Sigh.

:)

-Thanks!