[time-nuts] A first MDEV plot...

[Bob Camp]

Hi

Is there any chance the plotting tool is doing "drift removed" on the ADEV
but not on the MDEV? Having the OCXO start to wander a bit due to room
temperature is not entirely unexpected at >=1K seconds.

I agree that the scaling looks a bit strange, but I'm ignoring that for now.

Bob

-----Original Message-----
From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On
Behalf Of John Miles
Sent: Tuesday, March 16, 2010 9:41 AM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] A first MDEV plot...


>
> This is my first Allan Deviation plot. It consists of the PPS
> output of an Oncore GT+ (Navigation-grade) GPS vs. an Ovenaire
> 10MHz OCXO divided down to 1Hz via an HP 5328A counter. The HP
> 5370A is free running on its 10811-60101 OCXO. The
> control/plotting is done by KE5FX's T.I. utility software.
>
> The curve clearly shows the limitations of the GPS at Tau=1s...
> But what else can be said of the GPS, or of the test setup when
> looking at the plots?

It sounds like there might be a scaling problem.  Do you have the 1-Hz input
connected to START and the 10-MHz output from the GPS connected to STOP?
That's really the best way to go, unless the DUT is drifting so much that it
covers the whole 100-ns period range in only a couple of measurement cycles.

If you are connecting 1-pps sources to both the START and STOP channels, try
entering 1 in the Nominal Frequency field.  Otherwise, if 10 MHz is
connected to the STOP channel, enter 10E6 as the frequency.

If you're running a new(ish) build of TI.EXE, you can hit the 'f' key and
get a chart of averaged frequency counts.  If those don't reflect the actual
frequency of your STOP-channel source, it's a further indication that your
setup isn't right.

Ordinarily a 5370-based ADEV measurement of two decent sources will come in
at about 2E-11 to 6E-11 at t=1s.  It looks like you're a factor of 10,000
too high, which is probably a result of incorrect phase unwrapping, again
due to an incorrect Nominal Frequency.

You can view the file c:\program files\ke5fx\gpib\time interval
plots\HP_5370B_TADD2_residual_floor.tim if you want to see a near-worst-case
measurement floor for the 5370.

> One thing I see is that the bottom valley is much more pronounced
> when plotting MDEV, whereas with ADEV the curve barely breaks.
> How much more revealing is the MDEV curve in this case, and what
> effect is shown here? What dominant factor causes the valley. Is
> it hitting a noise floor, a true GPS performance deterioration or
> is is simply a calculation artefact?

MDEV is doing a better job showing the flicker floor of the measurement.
Standard ADEV plots show both white and flicker PM as a 1:1 slope on a
log-log graph.  MDEV displays white PM noise as a 1.5:1 slope that reaches
its floor sooner than ADEV does.

I usually look at plain old ADEV on stable sources, or HDEV in cases where
linear drift needs to be disregarded.

> I use the HP 5328A as a divider by 10 million. This is a very
> useful feature if you own one of these counters. I presume that a
> straight line plot gives a warm feeling about the divider
> performance, but it would be great if someone on this list who
> owns a HP 5328A could characterise the Timebase Output when
> dividing by 10 million. I will try some measurements using the
> 5370A and one of my OCXOs, but I am not sure I own what is
> necessary to make valuable measurements.

You might try duplicating the residual 5370 measurement that I mentioned
above, to make sure everything is working OK.  This was done by feeding a 10
MHz signal through a splitter to the input of a TADD-2 divider and the
5370's STOP input, with the 1-pps output from the TADD-2 connected to the
START input.  It will likely be possible to do better if you optimize the
trigger levels, input amplitudes, and such, but you shouldn't see anything
worse.  External arming may also help, but I wouldn't mess with that until
you're seeing the right order of magnitude on the graphs.

> So much to learn! So much to try!

It takes a brave soul to use that app with no documentation. :)

-- john, KE5FX


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