[time-nuts] HP10811 vs 00105 OCXO

John Miles john at miles.io
Fri Aug 7 17:21:27 EDT 2015

> Hi
> If that data is correct, then the 10811 you have is defective.
> Bob

Well... some of the data is reasonable for a scenario where a counter is being used to measure OCXOs.  

Looking at the ADEV plot, I'd say the blue trace (HP105B vs 5065A) is the most questionable one if it came from a standalone TIC or frequency counter, because 7E-12 @ t=1s isn't achievable with most counters under most circumstances.  A Wavecrest box can measure at that level if it's set up _perfectly_ to take bursts of 100+ wrap-free averages within a small fraction of the t0 interval.  It might also be doable with an HP 5370A/B under similar conditions, but I'd have less confidence that the averaging isn't distorting the measurement.  So while It looks like a valid measurement of an OCXO with some minor crosstalk or other external interference, that may just be a coincidence.  

Luciano, how was the blue trace taken?  Is this from your DMTD project?  If so, it's looking promising.

The green and magenta traces are definitely in the right ballpark for measurements on a 5370-class counter.  At 3E-11 @ t=12s the magenta trace is optimistic but not outrageously so, while the green trace looks exactly like I'd expect for a typical 10811 measured on a 5370.  The observed noise is due entirely to the counter until about t=200s.  We see a glimpse of the 10811's typical  ADEV at about 250 seconds, just before either drift or ADEV uncertainty causes the trace to turn upwards.  A longer run would be needed to distinguish between these two situations.

-- john, KE5FX
Miles Design LLC

More information about the time-nuts mailing list