[time-nuts] Low-Cost Rubidium Performance
lists at rtty.us
Fri Feb 10 17:11:55 UTC 2012
Have I seen system bandwidth in published data? - pretty much never. Kudos
to Symmetricom for putting it up on the display on the 5125. The system
bandwidth debate goes back well into the 1960's. Pretty much every paper
Dave Allan ever presented up through the 1980's, he got a question about
bandwidth from the floor. I don't ever remember there being a definitive
reply back in that era.
Confidence bounds - same thing. How many points before you even display the
data? If I'm using 3 points and you are using 100 the error bars will be a
bit different. The reply to that question was generally - you really should
have 100 or more points. I doubt we would see much 100,000 second data if
people got rigorous on that....
By no means am I saying that ADEV shows all issues on all devices. We are in
agreement there. My point is that people should know what you did to your
data before you put it up on display. I have indeed seen people use a *lot*
higher order fits than a simple linear estimator.
From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On
Behalf Of Magnus Danielson
Sent: Thursday, February 09, 2012 9:12 PM
To: time-nuts at febo.com
Subject: Re: [time-nuts] Low-Cost Rubidium Performance
On 02/10/2012 02:31 AM, Bob Camp wrote:
> To add another wrinkle to this.
> Correcting ADEV for systematic errors and then not mentioning you did so
strikes is something I find a bit problematic. If say you decide to take out
13th order drift, you should say you did so. The discussion of what to
correct and how is older than ADEV. Since there is no "standard" set of
corrections, one should be clear about what was done.
Agreed, but even if we know that system bandwidth affects low-tau
noise-forms, when did you last see the bandwidth given? Confidence
bounds is another issue.
But my main point is that you still look at the same damn plot rather
than looking at some other plot for the other effects. The point is to
separate them in order to clearer see each of them. For short-term,
noise dominates, and depending on where we are on the spectra the ADEV
or phase-noise plots serves us best. TIE curves is interesting, but MTIE
curves adds a certain aspect, drift analysis curves another aspect etc.
For jitter, separation of random noise and deterministic noise is am
important to properly engineer the system. MTIE is another of those.
I love my ADEVs, but as my suppliers know, it's not the only thing I ask
for in an oscillator.
How does my phase shift as the temperature cycles? ADEV won't tell me,
but I bet the guy in the next cubicle can enlighten you if needed.
This is all about understanding the limitations of the tools we have, so
we use them for what they do best. ADEV is a great tool, just don't over
use it. Whenever someone use their favorite tool (say ADEV, FFT or
Object Oriented programming) for everything and everything needs to be
answered by it, then the big warnings-bells goes off in the back of my
head. As great as it ever is, know the limits of what you use...
time-nuts mailing list -- time-nuts at febo.com
To unsubscribe, go to
and follow the instructions there.
More information about the time-nuts