[time-nuts] ADEV vs MDEV

[Bruce Griffiths]

Pete

Not sure how you implemented the 4ms holdoff.

With a 5370A/B I would have cascaded a pair of 74HC4059's to divide down 
f1(~10MHz) produce a 200Hz (or whatever frequency desired eg 10Hz, 1Hz 
etc) signal to drive the external ARM input.
Connected f1 to the START input and f2 (~ 10MHz but can be any frequency 
within the counter input bandwidth) to the STOP input.
The jitter of the 200Hz arming signal isn't critical. The counter then 
starts on the next zero crossing transition following the ARM signal 
transition and stops on the next STOp input signal transition.
Unwrap the phase measurements and calculate ADEV, MDEV etc.

If one uses a 74AHC74 to resynchronise the output of each HC4059 then 
the divider should be usable to 20MHz or so without any ambiguity in 
identifying the next clock transition after each divider output transition.

Bruce

Pete Rawson wrote:
> Bruce,
>
> THANKS.
>
> The paper you provided the last link for is excellent.
> I haven't seen it before&  it provides a marvelous
> explanation of the various stability tools.
>
> The specs on the CNT81 don't reveal that the 50ps
> single shot TI capability is what the built-in display
> shows. The GPIB data provides 25ps resolution,
> which is the internal data.
>
> My measurement setup begins with a Wenzel
> Sprinter OCXO @ 100MHz. This feeds an Analog
> Devices AD9513 evaluation board providing 2
> 10MHz outputs (norm&  complement). One of
> these outputs clocks a 74HC4059 in divide by
> 1E4 mode. That 1KHz pulse starts the CNT81.
> The other 10MHz output stops the CNT81.
> The TI measurement is paced at 4ms delay.
> The CNT81 reference is supplied by a Tbolt
> 10MHz output.
>
> The resulting series of TI measurements are
> nominally 77ns, taken every 5ms, back-to-back.
>
> I use "Plotter" to analyze and plot the ADEV&
> MDEV results (Thank you to Ulrich for such a
> fine tool). The MDEV line crosses the 1E-13
> level at, or very near, 10s. As expected, the MDEV
> slope is -3/2 on the log-log plot vs. the ADEV
> slope of -1.
>
> I would be happy to send a typical data file to
> you, if it would be of interest.
>
> Pete
>
>
>