[ntp:questions] 500ppm - is it too small?

[David J Taylor]

David Woolley wrote:
> David J Taylor wrote:
[]
>>  http://www.maxim-ic.com/appnotes.cfm/an_pk/3566
>
> This seems to suggest crystals have quadratic temperature dependency.
> My understanding is that normal cuts have cubic dependency.  Of
> course, 32kHz crystals are often cut for wrist watch use, and using a
> cut with quadratic characteristics might give a flatter response
> close to body temperature, at the expense of poorer behaviour for
> signficant deviation.

Ideally, you choose the cut for the application.  Quadratic dependency 
might be good for oven-controlled or on-wrist, but what happens when you 
take the watch off?  Cubic for a wider temperature range, perhaps?

With the cost of computers being as it is, I suspect that every penny 
saved on every component is important.  I'd be interested to know what 
actual accuracies are seen for typical PCs (rather than more expensive 
workstations, minis or mainframes).

Cheers,
David