[time-nuts] antennas

Magnus Danielson magnus at rubidium.dyndns.org
Tue Sep 28 17:19:19 UTC 2010

On 09/28/2010 09:38 AM, Hal Murray wrote:
>> What is the purpose of the temperature sensor chip on the PCB, anyway? Isn't
>> the temperature inside the OCXO much more important?
> The OCXO isn't perfcet.  The frequency will shift slightly with environmental
> temperature changes.

Let's consider a thermal gain of 1000, such that 1 K of external 
temperature difference becomes 1 mK internal. If we want to compensate 
the remaining error, putting a temperature sensor inside the OCXO, it 
needs 1 mK resolution for 1 K external temperature shift. Putting the 
temperature sensor externally will the OCXO thermal gain improvement in 
resolution. So a 1/32 K resolution external becomes a 1/32 mK resolution 

This model is oversimplified as the thermal response-time and that of 
the oven will smooth out temperature shifts, and an external sensor does 
not experience the same transient response.

> It may have been more important without the high-quality OCXO option that
> many recent TBolts seem to have.
> Here is some data:
> http://www.megapathdsl.net/~hmurray/time-nuts/TBolt-volt-temp-1.gif
> http://www.megapathdsl.net/~hmurray/time-nuts/TBolt-volt-temp-2.gif

Looks like thermal lag is the reason for the opening between the heat-up 
and chill-down periods.


More information about the time-nuts mailing list