16 bit timer1 formula?

[brandonb]

i was playing with tmr1 frequencies and was wondering is there a math formula to arrive at a desired freq? like in avr where with their 16 bit timer can say osc/prescaler/desired frequency = ocr1a
file below cycles 30hz to khz

[medelec35]

Hi Brandon,
The basic fromula for timer 1 Interrupt Frequency is:
InterruptFrequency = Fosc/(4 x Prescaler x (65536-PreloadedValue))

So If you want to calculate Preloaded Value then:
Preloadedvalue = 65536 - Fosc / (4 x InterruptFrequency x Prescaler)


Martin

[brandonb]

thanks martin, im amazed at how easy you come up with equasions, that doesnt come natural for me on this flowchart i was looking at the datasheet and seen those tmr1h and l registers so i gave it a go, is this the best way of doing this

[brandonb]

So If you want to calculate Preloaded Value then:
Preloadedvalue = 65563 - Fosc / (4 x InterruptFrequency x Prescaler)

is the 65563 ment as 65535?
can you break the math down, im not getting the right answers?

[Enamul]

Hi

is the 65563 ment as 65535?

I think he is trying to write 65536 (2^16) not 65563

[medelec35]

is the 65563 ment as 65535?
can you break the math down, im not getting the right answers?

Sorry its a typo. Yes your correct its wrong.
It should be 65536 (and not 65563 or 65535)
Since used copy and paste (The lazy way ) both ended up wrong

I have edited my post.
An example:

Fosc = 4MHz
Deisered Interrpupt frequency = 12.57Hz
Prescaler = 2

Preloadedvalue = 65536 - Fosc / (4 x 12.57 x 2)
= 65536 - 4000000/100.56
= 65536 - 39777.24741
=25759 (rounding up)


Edit: Enamul is correct. In my mind was 36, but my fingers had a diffent idea

[brandonb]

ok, i wanted to make sure, i get a slightly different frequency, it might be the way i did the flowchart, i havent slept since yesterday, so i wanted to make sure

[medelec35]

I just altered your flowchart to work on 16F877A with Clock Speed set to 19660800Hz
I'm aiming for 100Hz so trying with prescaler of 2
So

Preloaded value= 65536-19660800/(4 * 100 *2)
=40960

So count was set to 40960 then compiled to chip.

If all is working then frequency measured should be 1/2 the actual frequency since o/p state is toggled on each interrupt.
Result using E-blocks test pod:

timer1 100Hz.png

(140.25 KiB) Downloaded 4275 times

As you can see frequency measured = 49.950Hz
x 2 = 99.9Hz
Which is very close indeed!

Martin

[brandonb]

martin, i dont know whats wrong, hypotheticly speaking i enter 1+1 in calulator and get three, will try again tommorrow

[brandonb]

martin, are there speed limitations on the timer1 or should i change the code... here is a calculator that i created from your equasion, from top to bottom prescaler 1,2,4,8

tmr1 calculator.fcf

(12.53 KiB) Downloaded 341 times

running th calculations for 50Khz i come up with the numbers on the top line when simulate, to conferm this is correct i checked on this site as well for match http://eng-serve.com/pic/pic_timer.html which is does
here is flowchart that should cycle 50Khz

tmr1 test.fcf

(12.5 KiB) Downloaded 350 times

here is the logic scope shot where it only does 44.5Khz , is this a limit of tmr1 or do i code it differently

[brandonb]

i rewrote the code in another compiler just for kicks and its the same

Code: Select all

unsigned int i;

 void interrupt()
  {
   if(pir1.TMR1IF && pie1.TMR1IE)
   {
    tmr1h=255;
    tmr1l=96;
    lata|=1<<5;
    for(i=0;i<2;i++){}
    lata&=~(1<<5);
    pir1. TMR1IF=0;
   }
  }

void main()
{
  osccon=0xf0;//32Mhz
  ansela=anselb=0;
  trisb=trisc=255;
  trisa=0;
  lata=0;
  
  intcon.PEIE=1;
  t1con.T1OSCEN=1;
  t1con.TMR1ON=1;
  t1con.TMR1CS1=0;
  t1con = t1con & 0x0F;
  intcon.GIE=1;
  pie1. TMR1IE=1;

  while(1)
  {

  }

}

[Gary Freegard]

Hi Brandon

I too have had similar problems in the past. All I end up doing is adjusting the timer value unto I get what I want.

I use this program http://pictimer.picbingo.com/.

I ran a version of your program on a 18F2455, using the program above I set tmr1h to 255 and tmr1l to 16 and the output was 45.6 KHz.
I then adjusted the timer1 values in the program(above), so that the interrupt frequency was 45.6KHz , the new value for timer1 was 23 lower.
I then added that onto tmr1l to make it 39, and the result is 50.0KHz output.

With high frequency timer interrupts I know you have to take the interrupt latency into account which is 3/4 cycles. Plus it takes 2
cycles to write to the timer. This explains a change of 6 for the timer but not the additional 17

I also tried a low frequency (200Hz), tmr1h = 21 and tmr1l = 183 and the output was 100.00Hz

Hope this helps

Gary

[brandonb]

gary, thanks for conferming this is normal... i thought i was missing something because it was so far off, figured the whole point of a timer interrupt was that it was solid, the 8 bit versions are

[Gary Freegard]

Hi Brandon

Glad to be of help to you.

Just carried out some other tests, Actual Freq should be half of desired:

• tmr1h___tmr1l____Desired Freq____Actual Freq
  255_____244_____1MHz__________No output
  255_____232_____500KHz ________127.7KHz
  255_____208_____250KHz_________84.5KHz
  255_____16______50KHz _________22.8Khz
  253_____179_____ 20KHz _________9.79Khz

When 23 is added to the last three the output is what would be expected.

Gary