Dan Michaels wrote:

> Probably 20 pF in the worst case. So, for instance, a pin connected to
> an equiv R of 10K [basically, the parallel combination of all Rs on
> the pin] and assumed stray cap of 10 pF would have a 3dB low-pass near
> 1.6 Mhz.

  Not questioning the maths, that's not a low-pass, that's the
"relaxation" time of an un-driven line.  The driver has a far lower
impedance, and will happily drive that to the dozens or even hundreds
of Megahertz.

> Regarding resistor arrays and stray cap, per se, it would appear
> that neither Johnson's book nor Paul B.

  I'm just an amateur - and a theorist.  Be warned!

> seem to think this is too very important, except in the context of
> overall considerations, ie, general R values, track size/ length/
> spacing, etc.

  As the discussion related to crosstalk, it involves the mutual
impedance between lines as a series component, versus the shunt
impedance to ground (just discussed, including the source impedance) of
the receiving line.  It only gets significant when the former becomes
sufficiently low to be comparable to the latter.  Factors affecting both
tend to counterbalance.
--
  Cheers,
        Paul B.