William K. Borsum wrote:
>
> >This has brought up an interesting thought though,  Would it be somehow
> >better to use FOUR capacitors, Each half normal C, two to VCC and two to
> >ground??
> >If the LOCAL symmetry can be preserved, then maybe there won't be as much
> >present on the VCC rail as in the conventional technique.   On uPs and other
> >chips with xtals, my habit is to use a Murata EMI filter (three lead cap)
> >rather than just a bypass cap. Maybe this is a way to cut the cost by using
> >two additional small caps instead of the Murata device.
> >
> >I may just do this in a proto. I have three in development at the moment,
> >and now that I see a possibility for improvement, it becomes interesting.
> >(and I can test the other case too)
>
> Please, I would like to hear the results of this test--on or off the list.
> Kelly

  I'm sort of intrigued as well - I'm wondering what the results of, (a)
caps just to Gnd, (b) Caps just to Vcc, and (c) caps split between Vcc
and Gnd, are.

  And what happens with smaller vs. larger bypass cap totals across the
PIC.

  I usually way OVER-bypass my PICs, as I do very small runs, those in
manufacturing situations might have worse results, but I'm wondering
about quantity of those results.  I don't have good EMI metering gear,
sadly.

  (Sort of think that we may see that with a 10uF tantalum & a .1 MilCer
from Vcc to Ground, right there on an F84, that the Vcc rail doesn't
have tons of EMI noise off the oscillator - But, solid numbers beat Wild
Guesses, any day <G>)

  Mark