Clock drift. Not strictly a Linux question.

[Tony Alfrey]

Roger Oberholtzer wrote:
> On Mon, 2007-02-19 at 03:40 -0800, Tony Alfrey wrote:
> 
> [relevant text deleted]
> 
>> I see this link which may be relevant:
>>
>> http://www.gpsclock.com/nextlevel.html
>>
>> in which they describe in broad terms how to lock the clock in your box 
>> to a decent temperature-controlled "real" oscillator.
> 
> We have considered a counter card solution. However, all our systems are
> poor when it comes to free PCI slots. So that solution is problematic
> from an implementation POV. I was disappointed to see that the guy in
> your link did not report his findings on replacing the MB crystal. In
> practice this would be problematic as well as it would void any
> warranties, and make field replacement less easy. But it would be nice
> to know the effect of doing this.

A simple test of all of this would be to simply stick a 1 watt resistor 
on the side of the crystal and power it with a simple lab supply, 
thereby heating up the crystal even further, and see if your offset 
problem become even more severe than it is now.  If this turns out to me 
the case, then it might be possible to devise a small clip-on 
thermoelectric cooler to stick on the crystal.

> 
> I will see what SuperMicro have to say about all this. And I will have
> this tested in a wider range of MBs.
> 


-- 
Tony Alfrey
tonyalfrey at earthlink.net
"I'd Rather Be Sailing"