Hi

For most OCXO’s most certainly not. The typical OCXO improves on ADEV as it runs. That said, a good OCXO should have a 1 second ADEV
of at least parts in 10^-12 rather than 10^-11. A 5370 should have a floor at 1 second of around 2x10^-11. You may be measuring your counter
doing something stupid ( = needs alignment) rather than the OCXO or Cs.

Bob

Thanks Bob.  That makes sense. -----------------------------------------------------------------------------------
GFS GPSDO list:
groups.yahoo.com/neo/groups/GFS-GPSDOs/info

  From: Bob Camp <kb8tq@n1k.org>

To: Bob Stewart bob@evoria.net; Discussion of precise time and frequency measurement time-nuts@febo.com
Sent: Monday, June 27, 2016 7:06 PM
Subject: Re: [time-nuts] OCXO noise as they retrace

Hi

For most OCXO’s most certainly not. The typical OCXO improves on ADEV as it runs. That said, a good OCXO should have a 1 second ADEV
of at least parts in 10^-12 rather than 10^-11. A 5370 should have a floor at 1 second of around 2x10^-11. You may be measuring your counter
doing something stupid ( = needs alignment) rather than the OCXO or Cs.

Bob

Are you sure it's not the PRS-45A?  Cs standards get noisy as the tube ages, typically becoming much noisier before failing entirely.  This is usually pretty easy to diagnose by watching the beam current fluctuate.

You're also bumping up against the limits of what your counter can do.  Readings in the 4E-11 to 8E-11 range at t=1s are common with 5370s, depending on any number of factors.

-- john, KE5FX
Miles Design LLC

Hi John,
That is a thought.  The 1s ADEV had been very stable across units until just now.  But, I don't know how to look at the beam current.  There is a "Monitor3" program that I can use (see link below) that can plot the Clock Servo values, as well as Zeeman Servo, Gain Servo, CBT Supplies, and Power Supplies.  It collects values each 10 seconds.  Which one(s) of these would have the most relevance? 

http://ae6rv.com/PRS-45A/Monitor3.png

I'll go ahead and run an ADEV of the 10811 in the 5370 against the PRS-45A.
Bob -----------------------------------------------------------------------------------
GFS GPSDO list:
groups.yahoo.com/neo/groups/GFS-GPSDOs/info

  From: John Miles <john@miles.io>

To: 'Discussion of precise time and frequency measurement' time-nuts@febo.com
Sent: Monday, June 27, 2016 7:50 PM
Subject: Re: [time-nuts] OCXO noise as they retrace

Are you sure it's not the PRS-45A?  Cs standards get noisy as the tube ages, typically becoming much noisier before failing entirely.  This is usually pretty easy to diagnose by watching the beam current fluctuate.

You're also bumping up against the limits of what your counter can do.  Readings in the 4E-11 to 8E-11 range at t=1s are common with 5370s, depending on any number of factors. 

-- john, KE5FX
Miles Design LLC

time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

On Mon, 27 Jun 2016 21:55:55 +0000 (UTC)
Bob Stewart bob@evoria.net wrote:

As the change is just about a factor of 2, my first guess would be that
you see numerical effects due to the finite tuning resolution of the GPSDO.
You can see it as a similar effect as in delta-sigma modulators. As long
as the output value is changing, the noise power (which can be assumed
to be constant for a first order approximation) will be spread over a
wide frequency range. But, if you output a constant value, the noise power
will get concentrated in a couple of spurs, which will then stick out quite
a bit.

A simple test for this hypothesis would be to send the GPSDO into hold-over
mode and measure the ADEV of the OCXO again with constant EFC voltage.
Another indication would be, if the ADEV increased only in a narrow range,
while slightly decreasing overall (though, this is a much weaker argument
as it can be confunded by other phenomena).

		Attila Kinali

--
It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
-- Miss Matheson, The Diamond Age, Neil Stephenson

Hi Attila et al,
I think I've finally got a handle on the problem.  A few months ago I posted that I was getting phase pops on my 5370A.  At least I thought it was on the 5370.  So, I bought another one on ebay, and the problem disappeared.  Then it came back, and rapidly got worse.  Unfortunately, the OCXO I had put into one of my GPSDOs was also bad.  So, when I did a three-way comparison, no comparison looked good.  Thus my conclusion that the new 5370 had somehow gone bad in the same way that the old one did.
So, last night, I plugged in an older unit that had always tested good and let it cook overnight.  This morning, comparing that to the PRS got me about 1E-10 --- not very good.  Then I compared the old one to what I had considered the best of the new units I have on hand, and it's down in the 4E-11 or so range at 1 second.  So, I swapped out the OCXO in the other unit I had been testing, and the ADEV at 1s is just under 4E-11, which is where I would expect with my test equipment.

So, my conclusion is that the PRS was the cause of the popping I reported back around February, but it improved for awhile before getting worse: about 1E-10 at 1 second tau.  At the same time, one of the units I was testing not only had a noisy OCXO, but the previous one I had in that unit was also bad.  So, essentially this was a comedy of errors.
I'm left with using one of my GPSDOs as a reference.  But, the difference between good and bad seems to be large enough that I think it will be good enough, even if it eventually were to become noisy.

I do use ebay OCXOs in my GPSDOs to keep the cost way down.  This experience underlines how important it is to test them carefully before depending on them.  Out of about 30 tested, I've had maybe 10 which are unacceptably noisy, which is more or less what I had planned on.  Like Bob Camp says, you never know what you're getting when you get one of these Chinese "recycled" OCXOs.  But, if you buy in some quantity, and carefully test them, enough are good to make them viable.
Bob
 -----------------------------------------------------------------------------------
GFS GPSDO list:
groups.yahoo.com/neo/groups/GFS-GPSDOs/info

  From: Attila Kinali <attila@kinali.ch>

To: Bob Stewart bob@evoria.net; Discussion of precise time and frequency measurement time-nuts@febo.com
Sent: Wednesday, June 29, 2016 10:49 AM
Subject: Re: [time-nuts] OCXO noise as they retrace

On Mon, 27 Jun 2016 21:55:55 +0000 (UTC)
Bob Stewart bob@evoria.net wrote:

As the change is just about a factor of 2, my first guess would be that
you see numerical effects due to the finite tuning resolution of the GPSDO.
You can see it as a similar effect as in delta-sigma modulators. As long
as the output value is changing, the noise power (which can be assumed
to be constant for a first order approximation) will be spread over a
wide frequency range. But, if you output a constant value, the noise power
will get concentrated in a couple of spurs, which will then stick out quite
a bit.

A simple test for this hypothesis would be to send the GPSDO into hold-over
mode and measure the ADEV of the OCXO again with constant EFC voltage.
Another indication would be, if the ADEV increased only in a narrow range,
while slightly decreasing overall (though, this is a much weaker argument
as it can be confunded by other phenomena).

            Attila Kinali

It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
                -- Miss Matheson, The Diamond Age, Neil Stephenson