> > So I went to making the clock frequency variable.  I didn't try the RC
> > oscillator; it may work at 12 MHz but I doubt it.
>
> The C part does work at 12 MHz RC, although the data sheet says 4 MHz is
> the limit (for RC, it was a -10 part).  However, the frequency is _very_
> "variable" as the C must be very small (near the capacity of the copper
> traces).  I watched a continous square output on the oscilloscope while
> touching the traces.  Wheee, factor 10-100!!  Not exactly a reference for
> VGA signals!

The PIC's internal RC oscillator is limitted to about 4-6MHz if you want any
real stability; getting a higher frequency using a different chip for an
RC time source shouldn't be a problem, though.  The key is to use as small
an "R" as possible so you can then maximize C.  In addition, it may be worth-
while to design the oscillator so that its period is less than an RC time
constant (e.g. use a +5-powered relaxation oscillator with threshholds at
2v and 3v; the cap will then only have to swing by a volt.  If you reduce the
swing too much you risk noise problems, but a relaxation oscillator with a
1 volt swing should be able to output 10-20Mhz or so quite nicely.