[time-nuts] Setting Rubidium to match GPS source

[SAIDJACK at aol.com]

In a message dated 9/28/2007 17:28:15 Pacific Daylight Time,  
hmurray at megapathdsl.net writes:

>How  stable is a typical well-designed DAC output over temperature and  
>whatever?

Very stable. Our firmware analyzes this stability statistically, and  adjusts 
for it. Can't get into too much details, suffice it to say that  when the 
Fury is locking an external Rubidium prototype from Quartzlock  there is very 
little thermal sensitivity of the combined unit when running  on the lab bench 
with Airco going periodically etc.


>What about a couple of extra op-amps?

Different story. Never tried that, only one Opamp is needed to do this  
though. Linear Technology yesterday announced a new "super duper" CMOS opamp  that 
would probably work quite well. Better to use an OCXO with 0V to 5V range,  or 
mechanically tune it away from 0V to an offset of say 2.5V or so.
 
Opamps will always introduce noise, drift, etc etc, so we do not typically  
use them.

>Do any OCXO vendors offer the DAC inside the oven with a digital  interface? 
 
>That seems like the obvious way to reduce temperature  dependencies.



Yes, but you have to include the DAC voltage reference as well. The  drawback 
is that you need to run digital wires into the sensitive OCXO case.  Second 
drawback: you have to run the chips at the Crystal temp, or near that; so  that 
reduces the component lifetime. Third drawback: you need a big can for  that. 
Fourth drawback: it's non-standard, so you don't get to pick-and-choose  from 
a wide variety of OCXO manufacturers. Also, this requires the OCXO to have  
more than the usual number of (expensive) pins. Lastly: what format to support? 
 I2C or SPI? According to Murphy you will always choose the wrong one...