[time-nuts] Updated Divider Jitter Results

[Poul-Henning Kamp]

In message <49D740D1.1030600 at febo.com>, John Ackermann N8UR writes:
>Sure.  I didn't write down the numbers because it was an interim step, 
>but I earlier did that and IIRC the std dev was around 20-25 ps.
>
>Would error in the cal circuit tend to falsely increase, or decrease, 
>the result?  Or both?

In my experience it generally skews your result downwards, but I have
seen it move upwards also, it all depends on the phase.

By running the 5370 from a unlocked source makes the input signals
"precess" over the input windows and averages stuff out.

I only lock my 5370 if I need to measure long timeintervals, for
short intervals the timebase error is lost in the noise.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.