[time-nuts] Can somebody give me a lesson in OCXOs?

David G. Andersen dga+ at cs.cmu.edu
Sat Feb 25 20:20:11 EST 2006


A couple of things:

a)  Your box is already NTP-stabalized, right?

b)  Have you looked at just going for a TCXO + a frequency synthesizer
chip and overriding the current clock on your motherboard?

My guess is that such a combination will have your computer keeping
time about as well as you'll get without also retooling the OS a bit
to ensure that, for instance, timing information doesn't get lost
under heavy interrupt load.

The advantage to that approach is that it's probably the cheapest and
"just works" to make your existing setup better with no additional
configuration.  It probably gives good bang for the buck, since you can
get a 1ppm TCXO for a couple of bucks.  Far better than the ~20ppm
range you could see with a crappy PC XO.

Or do you just want a local time source board that you can have NTP
bang on all the time to correct itself, hence the PCI bit?  If that's
the case, I might do something different:  a serial or parallel
accessable external clock that you can convince NTP to talk to
as a better frequency (but not time) source.  Taking the OCXO 134
that you mentioned, building a divider to get it to 1PPS at RS232
voltages (or parallel voltages, even easier) is really easy.

  -Dave

On Sat, Feb 25, 2006 at 05:01:17PM -0800, Hal Murray scribed:
> 
> I'm primarily interested in temperature stability.  I want to make my PC keep 
> good time rather than wander around when the box warms up because I started 
> reading my mail.
> 
> I've been thinking of something for a PCI card. I'm willing to cheat and 
> leave the adjacent slot empty if the OCXO is a bit too tall.  So 12V or 5V is 
> more interesting than higher voltages.  But I'm also generally curious and 
> might do something else if I get a good idea that uses a different part.
> 
> 
> Is there some obvious inexpensive surplus/recycled part I should be looking 
> for?
> 
> EBay often has Isotemp OCXO134-10 at 10 MHz.  Currently they are $75 
> buy-it-now, none for auction.  The data sheet is available at Isotemp's web 
> site.  It says 5ppb over -30C to +60C.  (Anybody know what sort of shape that 
> curve has?)
> 
> I've also seen smaller 3 MHz units, currently buy-it-now at 4 for $25.  I 
> don't have data sheets.  Description says 27V.  I can read the part number 
> off one of the pictures, but Isotemp doesn't have anything on their web site.
> 
> Is anybody familiar with either of those?
> 
> 
> Any vendors known to be hobbyist friendly?  I'm not in a hurry.  I'd be 
> willing to wait until their next batch if that saves significant $.
> 
> Any sweet spots on the price/performance curve?  I expect commercial units 
> are more expensive than I want to pay for but I might do something a bit 
> crazy if I get in the right mood.
> 
> Digikey has a few small Conner-Winfield units at $100-$200.
> 
> 
> 
> 
> -- 
> The suespammers.org mail server is located in California.  So are all my
> other mailboxes.  Please do not send unsolicited bulk e-mail or unsolicited
> commercial e-mail to my suespammers.org address or any of my other addresses.
> These are my opinions, not necessarily my employer's.  I hate spam.
> 
> 
> 
> 
> _______________________________________________
> time-nuts mailing list
> time-nuts at febo.com
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> 
> 

-- 
Dave Andersen                                 dga at cs.cmu.edu
Assistant Professor                           412.268.3064
Carnegie Mellon University                    http://www.cs.cmu.edu/~dga




More information about the time-nuts mailing list