AVR-Chat

>
>I have a project where I need to be able to adjust the crystal frequency
>to a specific "exact" value.  I haven't breadboarded it yet, but do you
>have any experience or advice along this line?  My thought was to just
>make one of the crystal caps variable (make any difference which one?).
>  For example, using a 10 MHz xtal, I wish to be able to zero beat WWV.
>(Long term stability not too important, so it would not be ovenized or
>otherwise temp-compensated.)

I would think you could do this with a trimcap, maybe half the 
nominal C value, across a fixed cap to make up the rest of the value.
I'd try the input side first, the driven side should mostly be about 
phase shift.  You could even use a Varactor diode, and feed it with a 
DAC, if that's interesting.

>I'm not too sure how close the cheap microcontroller crystals are to
>their specified frequency either -- imagine that would depend on
>manufacturer, lot, etc.  In my application some selection would be
>acceptable rather than go to closer tolerance xtals.

They are spec'd to a given PPM over temperature

>Other than parts count and cost, do you see any downside to using the
>microcontroller's oscillator in this manner versus using a separate
>crystal oscillator (with transistor) driving the microcontroller?

I'd be wary of using the on-chip circuit for this application.
It's likely to have a lot of noise on it's supply, and would probably 
be pretty broad.
uP crystals aren't too interested in stability in the <1S timeframe.