[ntp:questions] Is 24PPM an Excessive Real-Time Clock Correction?
root at localhost.invalid
Thu Jul 12 13:36:22 UTC 2007
David Woolley wrote:
> On most motherboards, there are only two crystal oscillators, a slow
> one servicing the RTC and a fast one serving all, or most, of the
> others. HPET may be an exception on some machines, but I would note
> that high precision doesn't actually mean high accuracy, so there is
> no need to use a different oscillator, unless HPET is a misnomer.
Thanks for pointing that out. (I was mistaken.)
On a related note, I've read the following:
"Care must be taken to use only one crystal oscillator source when
designing circuits to avoid subtle failure modes of metastability in
electronics. If this is not possible, the number of distinct crystal
oscillators, PLLs, and their associated clock domains should be
rigorously minimized, through techniques such as using a subdivision of
an existing clock instead of a new crystal source. Each new distinct
crystal source needs to be rigorously justified, since each one
introduces new, difficult to debug probabilistic failure modes, due to
multiple crystal interactions, into equipment."
More information about the questions