>
> I am designing a product that generates some precisely timed signals.  I am
> using a PIC16C84-4 with a 4Mhz XTAL.  Testing the code in mpsim shows that
> my timing meets the spec. (min 55 usec max 61 usec 58 usec nominal)  MPSIM
> shows my timing to be 57 usec and 59 usec using a 1 usec clock cycle (4 Mhz).
> Analysis shows that +- 3% tolerance will work.  This should be no problem for
> a XTAL right?  Well I put the device on the scope and my actual numbers were
> min 60.6 usec max 62.2 usec.  This is about +- 5% and puts me out of spec.  I
> could lower my time in software (short term fix) but this will make the design
> difficult to produce.  I also have lots of stuff to do during the time between
> transitions and lowering the time will impact my time line.  Any ideas on how
> to adjust the frequency or why it would be 5% low?  I am using 30pf cap for
> C1 and C2 of the crystal and am running the PIC in XT XTAL  mode.  I do not
> have the resistor between the XTAL and OSC2 pin. (Ref data book p2-758)  The
> XTAL is one from a local surplus shop but I have never had problems with their
> crystals before.
>
> Thanks for any help
>
> Norm
>
Put your scope probe on the XTAL pins and look at the waveform. On the pin
which is an output, you should see 5v p-p swing and a decent 4MHz
signal. On the other pin, which is an input, you should see the same signal
at a reduced level, depending somewhat on the characteristics of your scope
probe. But if you set your timebase at, say, 10us/div and you see a ragged
waveform with chunks ripped out of it, you definitely have a crystal
problem. Also, a 1 meg across the crystal cannot hurt and will probably
improve startup reliability.

Good luck,
Reg