[time-nuts] 5370A jitter - What to look for

Magnus Danielson magnus at rubidium.dyndns.org
Mon Jul 16 19:59:20 EDT 2007


From: Bill Janssen <billj at ieee.org>
Subject: [time-nuts] 5370A jitter - What to look for
Date: Mon, 16 Jul 2007 16:30:12 -0700
Message-ID: <469BFF84.3040104 at ieee.org>

Hi Bill,

> I need to fix my HP 5370A jitter to reduce it to something reasonable.
> It is now about 3 nano seconds and is partially periodic. It is not 
> totally random
> If I plot the data (using plotter) I see what looks like two oscillators 
> beating against
> each other with superimposed jitter

I got to ask, what is your signal?

What is the beating frequency, i.e. roughly at what frequency/period-time does
your peaks of jitter occur?

If you have certain rates of jitter, they will beat against the instruments
measuring rate and hence the beating. Changning integration time or varying the
hold period may be methods to investigate if this is the problem.

> Wondering if that has a known source or do have to find the problem my 
> self :-)

I am pondering about what could cause such a failure mode. Trigger levels is as
always a usual suspect. Trimming those can be necessary. Never use a counter
without a scope within armslength (and cables) reach since you want to make
sure you have your levels and clean transition secured. False triggers could
get similar results.

You could certainly use the counters statistics unit to get some aid, where
min, mean and max should be able to give you a good hint about what is
happening alongside the std dev.

If the interpolators didn't lock I'd think you would see the lock failure
indication.

> I do not have the two matched cables and the pulse generator to do the 
> complete alignment.
> I do have a spectrum analyzer to do the 200 MHz. alignment
> 
> I can send the data file if that is permitted.

As long as they are not too long that should be fine. If you can post them
separatly and provide a link that would be ideal. Tossing them offlist into my
inbox will do just fine too.

Cheers,
Magnus



More information about the time-nuts mailing list