A quick snapshot of the state of the GPSDO project. I'm finally getting the kind of stability figures I'd like to see.
By the way, next time I start out on a project can someone
please stop me if it looks like every debugging or verification step will take 24 hours minimum to do. The progess on this has been
glacial becuase of how long you have to wait to see if what you've changed is good/bad or if you need an extra bit of information. Basically what might be a two minute test on the bench for anything sane becomes a 24 hour test for this.
The results of the latest overnight run:
Looking good. This is the first time I've put fine grained results out there (64 seconds samples). Earlier efforts have only been for long averaging times.
RMS frequency error (aka standard deviation), excluding the startup transients, is 30 x 10
-12 (or 3 x 10
-11 if you prefer it in that form). Mean frequency error on the same basis is 3 x 10
-12.
A frequency stability of 3 x 10
-11 represents a 'good' OCXO. I've done some rather sketchy comparisons of the 4 OCXOs in my batch of them and this particular one seems to be one of the two 'good' ones in this batch, the other two are merely 'OK'. I don't have stability figures on the exact same basis for the other three becuase, you know, "
24 hour tests".
Just to give you a feel the results of the 'sketchy' tests were 69 ppt (i.e. 69 x 10
-12) for this OCXO, 51 ppt for the other 'good' one and 161 ppt and 162ppt for the 'OK' ones.
Preliminary figures suggest that the innate drift for this OCXO is about 0.64 ppb/day.