[time-nuts] Thunderbolt Loop Damping
holrum at hotmail.com
Tue Jul 8 14:01:22 EDT 2008
I do modify the OSC ppb value to morph it into nanoseconds of error before I feed it to the ADEV code (at least I think I do... the documentation is not really clear on what the value really is).
The constant 1 second offset that I added to the data makes absolutely no difference in the results (I tried it both ways). It was just there for consistency with another program that I have. The Trimble numbers are 32 bit floats, my 1 second offset number is a 64 bit double, it can handle the value just fine without losing any precision. The ADEV code subtracts pairs of values, so that constant value gets removed automatically (and to quell your concerns, I have removed it from the code).
As far as what the statistics are showing... well they show the ADEV values of the Thunderbolt's own measurements of the signal errors (auto ADEV?). How those compare to measurements against an external reference is left as an exercise to the reader.
The curves do seem to make some sense. I am seeing curves with distinct breaks in the slope at points that correspond to slope changes in your Thunderbolt ADEV plots. The main difference is below around 50 tau where a "normal" adev curve shows rising values then begins to fall. The plots on the Thunderbolt data show constantly falling adevs with distinct breaks in the slope. If the numbers were totally meaningless one would expect to see a line of constant slope as tau increases. Past 50 tau or so, the ADEV values seem to agree with what one would expect.
The i’m Talkaton. Can 30-days of conversation change the world?
More information about the time-nuts