[time-nuts] Cycle jumps in Z3801A PPS?

Magnus Danielson cfmd at bredband.net
Mon Jan 29 19:29:00 EST 2007


From: John Ackermann N8UR <jra at febo.com>
Subject: [time-nuts] Cycle jumps in Z3801A PPS?
Date: Mon, 29 Jan 2007 17:04:58 -0500
Message-ID: <45BE6F8A.3020006 at febo.com>

> I'm looking at the PPS output of my pair of Z3801As.  One of them
> generates a nice plot with about 50ns peak to peak noise over 1 day
> (measured against a 5061A).  You can see that data at
> http://www.febo.com/time-freq/plots/cs1-z3801a_1.html.
> 
> However, the second unit seems to have random jumps, usually for just
> one PPS sample and occurring once every hour or two, of very close to
> 100 nanoseconds.  You can see what I'm talking about at
> http://www.febo.com/time-freq/plots/cs1-z3801a_2.html.
> 
> Since 10 MHz has a period of 100 ns, I'm guessing that this jump
> indicates some sort of one cycle slip in the PPS generator.
> 
> Has anyone else seen this behavior?

To me it looks like trigger jitter rather than cycle slips. You get a +100 ns
jump followed by a -100 ns jump. There could be several different sources to
this, but in effect I see it as if the 5370 misses to trigg on one cycle and
triggs (incorrectly) on the following edge instead.

Naturally, you should try to figure out where the fault is at, but you should
be able to correct for it in the meanwhile by just subtracting 100 ns on the
necessary samples (i.e. if a sample deviates more than 50 ns from the previous,
correct it). Such correction would make me feel a bit dirty, but the second
degree error is really small so in practice it is OK.

Did you use the same 5370 for both measure or different onces?
Trigger points? Can you make it better or worse as you fool around with the
trigger points? Signal levels?

Cheers,
Magnus



More information about the time-nuts mailing list