|
|
Posted by Moe Trin on 06/28/07 00:45
On Tue, 26 Jun 2007, in the Usenet newsgroup comp.os.linux, in article
<DOGdnZ68FraQCxzbnZ2dnUVZ_vOlnZ2d@comcast.com>, Jerry Stuckle wrote:
>First of all, uncompensated clocks may or may not be 100ppm. I've got
>one server, for instance, which is off more than 500 ppm. Several
>others run in the 200-400ppm range.
Not enough detail - the original PC design had (as noted) a simple
crystal controlled oscillator on the motherboard. To reduce costs
(someone has to adjust the trimmer on the motherboard). manufacturers
went to oscillators in a can - a hermetically sealed can shaped like
a 14 or 16 pin DIP - the oscillator vendor sets the internal trimmer
in manufacturing, and the motherboard manufacturer merely plus in this
device. These devices, available from vendors like Dale, Raltron,
Vectron, and US Crystal are quite cheap (under US$4 in unit quantities.
less than a third that when buying a thousand at a time) and nasty, but
are normally spec'ed at +/-100 ppm over 0-70C. Finally, to reduce
costs still more, the oscillator may be part of the "real time clock"
such as the Dallas Semiconductor family. I don't have spec sheets on
these, but expect their accuracy to be comparable.
>You've come up with a lot of arguments. So you are saying you can
>guarantee - WITHOUT QUALIFICATION - that the clock will NEVER be set
>back while things are running?
Of course not - I have absolutely no control over what your users may
be doing to your systems. One would hope that if you are subject to
various legal requirements with respect to dinking with the clock, that
you have trained your users appropriately. If not, it really is your
problem, not anyone else's.
Old guy
Navigation:
[Reply to this message]
|