[time-nuts] Best counter setting for ADEV?

[Tom Van Baak]

Hi Mark,

When you operate the 5370A in that configuration you are essentially just measuring the RMS sum of jitter of the 5087A output, 5370A input circuitry (ZCD), interpolators, and reference clock. I haven't measured this myself but I would expect jitter in the 5370A reference 10811 would have the same effect as jitter in the 5087A 10811. In other words, you can't tell which one is better; all you can tell is the RMS sum of their instabilities. This may be a lot to ask but try swapping the 10811's in the 5087A and 5370A and re-run the test to see if it's symmetrical. Bruce may also have some insights into the jitter contribution of the reference clock vs. the input clock(s).

On your simultaneous measurements -- yes, this is a good thing to do. I run my house 5 MHz this way; using two high-end TIC's to inter-compare three high-end standards.

Note if your samples are time-synced close enough you can create three data sets from any two: namely 1050 vs PRS10, PRS10 vs Fury, 1050 vs Fury. For added confidence measure all three pair with TI counters, and see how good the closure is if each one of the three counters didn't actually exist. Make sense, or shall I explain more?

/tvb

----- Original Message ----- 
From: "Mark Spencer" <mspencer12345 at yahoo.ca>
To: <time-nuts at febo.com>
Sent: Tuesday, October 02, 2012 10:19 AM
Subject: Re: [time-nuts] Best counter setting for ADEV?


On a somewhat related note over the weekend I spent a few hours characterizing the performance of two of my HP5370B's for making ADEV measurements by feeding the stop and start inputs with identical 10 Mhz signals from a HP5087a distribution amp. Not surprisingly they perform somewhat differently. More surprisingly the results vary depending on the oscillators used for the clock source for the HP5370B's. (I was under the impression the built in 10811's were more than adequate enough for this application.)

I've also found that when trying to characterize the performance of an oscillator (in this case a disciplined PRS10 Rb unit) making two simultaneous measurements using different references and overlaying the results can provide some additional confidence as to their accuracy. (I suppose if I had an H maser I probably wouldn't need to use both an OCXO and GPSDO at the same time to have confidence in the measurements (: ) 

Looking at the phase differences in this case gives me a reasonable degree of confidence that the OCXO has not significantly drifted during the first 20,000 seconds or so and that the HP5370B's are providing valid data.