[time-nuts] Thunderbolt oven / non-stable operating temperature
Tom Van Baak
tvb at LeapSecond.com
Wed Dec 12 13:02:30 UTC 2012
Nice, clean plots. Thanks for sharing.
Ah, but does correlation imply causation? Note that crystal resonators, even inside an oven, are also sensitive ambient temperature sensors. As temperature changes the OCXO drifts off-frequency -- the TBolt then sees the average PPSns diverge and adjusts the DAC in order to reduce this accumulated time error and to keep the OCXO on-frequency.
Try computing the tempco of your oscillator using this raw data (and look up the EFC gain setting).
I'm surprised to see the PPSns error got so large (over 100 ns peak to peak). Is your TBolt using a factory default time constant? The plots suggest it has been set too high given the environment.
----- Original Message -----
From: Michael Perrett
To: Tom Van Baak ; Discussion of precise time and frequency measurement
Sent: Wednesday, December 12, 2012 1:06 AM
Subject: Re: [time-nuts] Thunderbolt oven / non-stable operating temperature
Although ignorant of why, I have a pretty good of idea of what is happening.
There is a 1:1 correlation between the TBolt temperature output reading and the rest of the TBolt reported values. In particular the estimate of the TBolt accuracy (10 MHz ppb), the PPS Offset and the controlling DAC voltage.
It "acts" like the environment temperature is sensed by the tmep chip, which, I believe, changes the DAC voltage and the self estimates of the goodness of the TBolt changes proportionally.
More information about the time-nuts