Best approach to variable software PWM generator

[dgg91]

I need a pwm software generator with a frequency range from 10Hz to 500Hz and a progammable duty from 0% to 100%

Im happy with a duty jumps of even 5%(5%-10%-15%..)

Im using a PIC18F2550 with 20MHz external crystal and the goal is to control the frequency&duty sending the values via usb from a PC application, i already have the communication working so the only problem is the pwm part

the output will control a NPN transistor which will turn on/off a LED thus the output signal its not critical at all

So far my idea:
With tmr0 interrupts
The lowest period will be 2ms(1/500Hz) since i want at least a 5% resolution duty cycle i need to have at least 20 counts, so the clock must overload at a minimum of 2/20= 100uSec, then the rest of the program have to change the output_pin if needed

I assume that's reasonable? I have no idea if that's so fast or reachable taken into account that the usb communication have some interrupts aswell i think

[Ttelmah]

That is totally reasonable.

A search here will find examples of low frequency PWM's using this approach.

How fast are you clocking the CPU?.

With a 20MHz crystal on this chip, you have a choice of:

20MHz, 10MHz, 6.66MHz, 5MHz, 48MHz, 24MHz, 16MHz, 12MHz, 8MHz.....

48MHz obviously gives you the most 'time' (cycles) to handle everything.

[gjs_rsdi]

CCS have a built in function for PWM

Example:

[Ttelmah]

You are missing the point.

He wants a PWM, well _below_ the minimum frequency the standard hardware PWM offers. The CCS function only gives what the hardware supports.

There are enhanced PWM generators on some chips with wider ranges (all the PIC24's/30's and a few PIC16's). But on other chips you have to generate the PWM, by having a 'tick' interrupt at the shortest interval involved, and deciding in this whether each pin should be on or off. This is what he is asking. There is example code for this in the code library, and in quite a few posts here.

[joergn]

If a counter based PWM is done, you may want to avoid visible flicker of a PWM based LED driver, if other interrupts are may delaying such port changes.
You can disable such other interrupts when coming closer to port change action or you work with priority of interrupt. If the Timer interrupt routine is very fast and the other interrupts are also very fast there will be probably no interference.
My newest LED driver demo board for new US patented LED junction temp measurement is done by PIC16F688 20MHz running as a single channel PWM out of a timer with 350Hz.
TXRX is done by external USB bridge to keep the MCU cost low.