PWM function incorrectly timing

[mikikimi]

Hello, i have question about PWM function. I set my PWM function on 30KHz:
PWM2_Init(30000);
PWM2_Set_Duty(127);
And this is part of my code which doesn't work well.
When i write this:

PWM2_Start();
delay_us(500);
PWM2_Stop();
delay_us(500);
PWM2_Start();
delay_us(500);
PWM2_Stop();

On my ossciloscope i see that my PWM function and delay doesn't working in this timing.
Before first delay my PWM are working 500-600us then delay is about 650 and second PWM is about 600-700us. Why?
Thanks.

3.jpg (41.67 KiB) Viewed 944 times

2.jpg (41.72 KiB) Viewed 944 times

1.jpg (40.34 KiB) Viewed 944 times

[paulfjujo]

hello,


Wich timer is associated to your PWM ?
Dely_ms normally doesn't use a timer .. just count somme loops.

30 Khz means T=33µS
you Start and Stop PWM without being synchronised with Delay
so maybe + - 33µS minima shift..
and
When you Stop PWM , does PRx (on Timer x used by PWM ) just roll over ? or not reach his normal value ...

somme PIC have ehanced function with PWM like 18F27K42
and Synchro facilities...

[mikikimi]

Hello, thank you for reply.
I am using pic16f1825 , so PWM2 function use TIMER1.
I will create some Delay function based on timer0 for better precision. ( thank you for idea).
Do i need some extra definition for TIMER1 to be synchronised with PWM2?

timer1 sinhronizacija.PNG (52.95 KiB) Viewed 920 times

[hexreader]

Attached is my attempt at example code to do what I think you want to do.

I noticed the following strange features shown in your scope traces:

1) A 500us burst of 30kHz should show 14 to 15 cycles of 30kHz. Your pictures only show 3 or 4 cycles

2) The third picture (1.jpg) shows you measuring only 2.5 cycles, not 3 cycles. A cycle of a square wave consists of 50% high AND 50% low.

3) Your scope shows something closer to triangle waves than square waves - no idea why that might be. My scope shows square waves.

4) I find that a 10k Ohm pull-down resistor gives better wave shape between bursts. Don't know the reason, but PWM2_Stop(); leaves RC3 as input.

[mikikimi]

Thank you for reply.
I did wrong scope trigger. I will show you now the same signal with better trigger. I saw your code, and is similar like mine. Could you show me pictures of your scope?
And why you wrote this line : TIMER2=0; Is this necessary?
Thing that bothering me is why i have longer cycles about 17-18 when my delay is 500us, but i saw in last reply ,that delay doesn't use timer, so that might be the problem.

1.jpg (47.62 KiB) Viewed 909 times

2.jpg (51.49 KiB) Viewed 909 times

[hexreader]

TIMER2=0; Is this necessary?

Try commenting out these two lines - you will see that pulses wander (roll).

For most applications this "roll" will not matter. If you need exact timing and total predictability, then synchronise the pulses to the Start_PWM().
There may be other ways to synchronise, but TIMER2=0 seemed an easy way for me.

Scope trace attached

Note that 500us is not an exact number of cycles.

[hexreader]

By the way....

If I wanted accurate, predictable timing, I would bit-bang the pulses.

The down side of bit-banging is that code is blocking, and the code will be no good if interrupts are enabled.

All depends on the application....