> The way it works is that the upper three bits of an 8-bit register are
> allocated for the PWM's. These bits are "incremented" by adding 0x20 to
> the 8-bit register. When the pwm rolls over, then the carry gets set and
> the shadow register picks this up with the rlf.
>
> Now to adjust the PWM you only need to change the relative counts between
> the pwmset's and the pwmcount.
>
>
> One of the benefits with this approach is that it can scale quite well to
> 4, 5 or whatever number of bits. Or if you prefer, a non-2^N rollover rate
> can be accomodated with phase accumulators. (For example, you can load W
> with say 5 and roll over about every 51 iterations).
>
> -----
>
> If you only need 8-bits, then you might consider putting the 8 states into
> one register. E.g.:
>
> two_to_the_N_minus_one:
>      andlw   7
>      addwf   pcl,f
>      retlw   1
>      retlw   3
>      retlw   7
>      retlw   0xf
>      retlw   0x1f
>      retlw   0x3f
>      retlw   0x7f
>      retlw   0xff
>
>
> start:
>      movf    pw1,w
>      call    two_to_the_N_minus_one
>      movwf   pwmset1
>
>
>    ; etc for N PWM's
>
> loop
>
>      rlf     pwmset1,w
>      rlf     pwmset1,f
>      rlf     output,f
>
>   ; etc for N PWM's
>
>      goto    loop
>
>
> Scott
>
> PS. Feel free to post this to the list if you want.
>