[time-nuts] Cheap Rubidium

[Bob Camp]

Hi

The gotcha with temperature compensation is that there are a lot of different time constants involved. Put another way, when the temperature does a step change, the "results" are still happening >2K seconds later, and they are related to the speed of the change. Simply coming up with a full set of coefficients is likely to be a mess. Once you do, you will find out if they are repeatable or not ...

The computer water cool systems look like a fairly simple way to go. You could run a water loop and simply heat (or cool) the water well away from the enclosure. It's not going to be high speed control, but with big enough mass in the enclosure it probably does not have to be. 

I'm not really convinced that you buy enough with a platinum RTD over a good glass thermistor to make it worth the trouble. The TC of the thermistor is 10X that of the RTD and that helps in a fixed point system. A Hart RTD will hold < 0.001 C in a triple point cell for as long as the ice lasts. A good thermistor will do the same. The stability of either one is "good enough" in this application. 

Bob




On Dec 24, 2009, at 1:56 PM, Bill Hawkins wrote:

> Well, perhaps it is time to think outside of the sand box.
> 
> We are trying to remove heat, not cool it to low temperatures. Who
> else worries about heat from electronics? Gamers, who buy water-cooled
> rigs for their over-cooked (clocked) computers. No magnetic fields
> nearby, with enough tubing. Look around in a DIY computer store.
> Now all you've got to do is control the temperature where the fan
> and radiator are, and extreme insulate the tubing.
> 
> Back into the box, microprocessors have been used to calculate
> temperature compensation almost since they were introduced. If you
> are measuring frequency with a counter, it does no harm to add or
> drop a cycle to adjust the frequency.
> 
> You'll need a high-stability temperature probe, such as platinum wire
> from Rosemount (now part of Emerson), and equally stable conversion
> electronics. The accuracy doesn't matter as long as the sensor doesn't
> change with anything but sensed temperature. The micro is simply a
> matter of programming (SMOP).
> 
> Good luck finding a converted temperature reading that is stable to
> 10E-13. But, hey, it's the magic season when the days start to grow
> longer.
> 
> Wish all problems were this easy. :-^
> 
> Bill Hawkins
> 
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>