[sdiy] DsPIC timer issue when used as DCO!

[Colin f]

> Look at the slope of the wave , its slowely rising, if the 
> timer had been loaded with the correct number but still with 
> the wrong Shape voltage, (a high note triangle as we see on 
> the far right)this slowely rising ramp would instead have 
> been a low amplitude fast saw!

If the timer were at the correct rate, even this slow rising ramp would be
reset at the correct period.

> This becuse the PA397 current mirror are a rising ramp whos 
> shorted to AG by the NPN transistor!.

You're not looking at the output of a rising ramp.
The waveshaper on the 3396 (and presumably the 397) 'folds over' the top of
the wave once it reaches the upper limit.
If you want a ramp wave, you apply a timing control voltage that causes the
wave to just reach the fold over point at the end of a timer period, so you
only get a rising ramp.
If you want a triangle wave, you apply double the timing voltage for the
ramp.
That causes the wave to fold over and return almost to zero within the same
timer period.
Modulating between these two voltages gives you waveshape modulation between
ramp and tri, like on a Matrix 6.
I don't think the behaviour of the wave shaper if it doesn't get reset after
returning to zero is documented, but my guess is - you end up with a change
to the positive direction, and a slow rise, like you see in the diagram.
Which brings us back to a timer error.

Cheers,
Colin f