The first argument to the adev1 program is the sampling interval t0.
The program doesn't know how far apart the input file samples are
taken so it is your job to specify this. The default is 1 second.

If you have data taken one second apart then t0 = 1.
If you have data taken two seconds apart then t0 = 2.
If you have data taken 60 seconds apart then t0 = 60, etc.

If, as in your case, you take raw one second data and remove
every other sample (a perfectly valid thing to do), then t0 = 2.

Make sense now? It's still "continuous data" in the sense that all
measurements are a fixed interval apart. But in any ADEV
calculation you have to specify the raw data interval.

/tvb

Bruce,

2009/4/9 Bruce Griffiths bruce.griffiths@xtra.co.nz:

OK, I'm ready to be shot down on this but from what I can see right
now the measurement period of 2 sec should be maintained to satisfy
the measurement of drift which would otherwise be incorrectly
interpreted if I processed 400000 sec of data as only 200000 sec. I
can see that noise on the data can be broken down into two major
groups, drift and what I would really see as noise, IE PN, flicker,
random, etc. I guess I have been ignoring the whole drift component
with my missing data used for the ADEV plots. The point to me though
is that, even with the reduced data, an ADEV plot should be able to
characterise 'noise' for the actual sampling duration of the data, IE.
1 sec. What would obviously be incorrect is the affects of drift which
should logically show up as being twice as great. Maybe my idea of
using the 1 sec sampling period would work out better with HDEV.

73,
Steve

Steve Rooke - ZL3TUV & G8KVD & JAKDTTNW
Omnium finis imminet

Tom,

2009/4/9 Tom Van Baak tvb@leapsecond.com:

I think the penny has dropped now, thanks. It's interesting that the
ADEV calculation still works even without continuous data as all the
reading I have done has led me to belive this was sacrosanct.

What I now believe is that it's possible to measure oscillator
performance with less than optimal test gear. This will enable me to
see the effects of any experiments I make in the future. If you can't
measure it, how can you know that what your doing is good or bad.

73,
Steve

Steve Rooke - ZL3TUV & G8KVD & JAKDTTNW
Omnium finis imminet

We need to be careful about what you mean by "continuous".
Let me probe a bit further to make sure you or others understand.

The data that you first mentioned, some GPS and OCXO data at:
http://www.leapsecond.com/pages/gpsdo-sim
was recorded once per second, for 400,000 samples without any
interruption; that's over 4 days of continuous data.

As you see it is very possible to extract every other, or every 10th,
every 60th, or every Nth point from this large data set to create a
smaller data set.

Is it as if you had several counters all connected to the same DUT.
Perhaps one makes a new phase measurement each second,
another makes a measurement every 10 seconds; maybe a third
counter just measures once a minute.

The key here is not how often they make measurements, but that
they all keep running at their particular rate.

The data sets you get from these counters all represent 4 days
of measurement; what changes is the measurement interval, the
tau0, or whatever your ADEV tool calls it.

Now the ADEV plots you get from these counters will all match
perfectly with the only exception being that the every-60 second
counter cannot give you any ADEV points for tau less than 60;
the every-10 second counter cannot give you points for tau less
than 10 seconds; and for that matter; the every 1-second counter
cannot give you points for tau less than 1 second.

So what makes all these "continuous" is that the runs were not
interrupted and that the data points were taken at regular intervals.

The x-axis of an ADEV plot spans a logarithmic range of tau. The
farthest point on the right is limited by how long your run was. If
you collect data for 4 or 5 days you can compute and plot points
out to around 1 day or 10^5 seconds.

On the other hand, the farthest point on the left is limited by how
fast you collect data. If you collect one point every 10 seconds,
then tau=10 is your left-most point. Yes, it's common to collect data
every second; in this case you can plot down to tau=1s. Some of
my instruments can collect phase data at 1000 points per second
(huge files!) and this means my leftmost ADEV point is 1 millisecond.

Here's an example of collecting data at 10 Hz:
http://www.leapsecond.com/pages/gpsdo/
You can see this allows me to plot from ADEV tau = 0.1 s.

Does all this make sense now?

Very true. So what one or several performance measurements
are you after?

/tvb

Steve,

The penny may be falling but it is not completely dropped: Of course you can
feed your ADEV calculation with every second sample removed and setting Tau0
= 2. And of course you receive a result that now is in "harmony" with your
all samples / Tau0 = 1 s computation. Had you done frequency measurements
the reason for this appearant "harmony" is that your counter does not show
significant different behaviour whether set to 1 s gate time or alternate 2
second gate time.

Nevertheless leaving every second sample out is NOT exactly the same as
continous data with Tau0 = 2 s. Instead it is data with Tau0 = 1 s and a
DEAD TIME of 1s. There are dead time correction schemes available in the
literature.

Best regards
Ulrich Bangert

Ulrich,

2009/4/10 Ulrich Bangert df6jb@ulrich-bangert.de:

So why would my counter show any significant differences between a 1
sec or 2 sec gate time?

I've just done a Google search for "dead time correction scheme" and I
just turn up results relating to particle physics where it seems
measurements are unable to keep up with the flow of data, hence the
need to factor in the dead time of system. This form of application
does not appear to correlate with the measurement of plain
oscillators. Yes there is dead time, per say, but I fail to see how
this can detract significantly from continuous data given a sufficient
data set size (as for a total measurement time).

I guess what we need is a real data set which would show that this
form of ADEV calculation produces incorrect results, IE. the proof of
the pudding is in the eating.

73,
Steve

--
Steve Rooke - ZL3TUV & G8KVD & JAKDTTNW
Omnium finis imminet

Tom,

2009/4/10 Tom Van Baak tvb@leapsecond.com:

My reference to "continuous" data would be defined as measurements
over a specific sampling period with each sample following directly
after the previous. This seems to be what is generally required for
the calculation of ADEV in the literature and postings on this group.
Such that techniques like the picket fence are suggested as a way to
deduce "continuous" data when using instruments that are unable to
measure sequential cycles of the input.

Agreed.

It is certainly true that 1 second sampled data collected at 60 second
intervals cannot be fed into an ADEV calculation as having a tau of 1
sec as the resultant calculation will show incorrect results when
noise like drift is a factor. If the data set is pre-processed and
corrected for such effects as drift, I believe it should be possible
to feed this discontinuous data as "continuous" data for the
measurement of short tau with reasonable accuracy.

I guess it really depends on what level your measurement system is
able to work. For, say, the output of a 10MHz OCXO it would be
desirable to measure the source frequency although that would require
a fast measurement system and significant storage. The benefits of
this is that the input source is not degraded in the process of
division down to a more manageable frequency. We are currently
discussing the effects of the introduction of noise into frequency
standards just with distribution amplifiers and dividers. The ability
to measure such close in noise effects would indeed be a great bonus
and I envy your abilty to perform that.

Yes, I understand.

Well there are a number of them. The selection of best free-running
OCXOs. The effects of locking an OCXO to GPS and the "tuning" of this.
Running a OCXO in active holdover mode. I'd like to separate the
effects of temperature, rate of change of temperature, aging,
humidity, atmospheric pressure and, possibly, gravity on a
free-running OCXO. By changing just one variable at a time, I'd like
to measure the effects of each one with respect to determining the
correction required from a holdover circuit. Agreed, some of these are
simply defined as frequency change in the oscillator but I will wish
to measure the full system performance and need some form of
yard-stick to work to. Now this may not give the most politically
correct and acceptable results for such measurements as ADEV but it
will give me something for comparison. If I can make it match the
politically correct results of others, I also can make comparisons
with their results too.

Does all this make sense now?

You see I need to understand things in my own mind and have problems
when things do not make perfect sense to me. So if I as stupid
questions or make ridiculous suggestions, it's just my way of learning
by probing for the limits. There was a UK TV science program called
Take Nobodies Word For It, a paradigm I have always lived by so I tend
to prove things to my own satisfaction and probably re-invent the
wheel a lot of times. I never went to uni so my classical education is
limited but I have spent all my life learning off my own back.

Kind regards,
Steve

--
Steve Rooke - ZL3TUV & G8KVD & JAKDTTNW
Omnium finis imminet

Steve,

suppose your source has a 0.5 Hz frequency modulation. Would you see it with
2 s gate time or a integer multiple of it? Would you notice it with 1 s gate
time or an odd integer of it?

Google for the STABLE32 manual. THIS literature will bring you a lot
further, many well documented source examples in Forth and PL/1, hi. F.e.
you may look here:

http://www.wriley.com/

Best regards
Ulrich Bangert

Ulrich, and Steve,

Wait, are we talking phase measurements here or frequency
measurements? My assumption with this thread is that Steve
is simply taking phase (time error) measurements, as in my
GPS raw data page, in which case there is no such thing as
dead time.

/tvb

Tom Van Baak skrev:

I agree. I was also considering this earlier but put my mind to rest by
assuming phase/time samples.

Dead time is when the counter looses track of time in between two
consecutive measurements. A zero dead-time counter uses the stop of one
measure as the start of the next measure.

If you have a series of time-error values taken each second and then
drop every other sample and just recall that the time between the
samples is now 2 seconds, then the tau0 has become 2s without causing
dead-time. However, if the original data would have been kept, better
statistical properties would be given, unless there is a strong
repetitive disturbance at 2 s period, in which case it would be filtered
out.

An example when one does get dead-time, consider a frequency counter
which measures frequency with a gate-time of say 2 s. However, before it
re-arms and start the next measures is takes 300 ms. The two samples
will have 2,3 s between its start and actually spans 4,3 seconds rather
than 4 seconds. When doing Allan Deviation calculations on such a
measurement series, it will be biased and the bias may be compensated,
but these days counters with zero dead-time is readily available or the
problem can be avoided by careful consideration.

I believe Grenhall made some extensive analysis of the biasing of
dead-time, so it should be available from NIST F&T online library.

Before zero dead-time counters was available, a setup of two counters
was used so that they where interleaved so the dead-time was the measure
time of the other.

I can collect some references to dead-time articles if anyone need them.
I'd happy to.

Cheers,
Magnus