GPSDO usage during frequency measurements

92 views
Skip to first unread message

sp6...@gmail.com

unread,
Jun 10, 2021, 7:11:36 AM6/10/21
to HamSCI
Hi there,

I am wondering how GPSDO frequency should be used (other than calibrating radio clock just before measurements) when performing long-term frequency measurements?

I am thinking about radios that doesn't accept external synchronization signal.

I suspect that 1[kHz] signal monitored by FLDIGI should be somehow compared to the signal GPSDO output or both signals should be recorded at the same time?
When radio clock is drifting away how we compensate for that having GPSDO signal accessible on the bench?

If that maters I am using FT-891 but I think wider audience is interested how to utilize GPSDO in long-term frequency measurements.

Thank you for your feedback!

Grzegorz SP6HFE

Black Michael

unread,
Jun 10, 2021, 7:56:32 AM6/10/21
to HamSCI
Depends on the receiver.

I run the ppFMT and I can tell you that the vast majority of receivers have drift that can be corrected with a 2nd GPSDO.

For rigs that don't accept an external reference the GPSDO provides a known frequency that can be used to correct any rig variations.

FLDigi has this capability built in to its FMT mode and I can measure a 3rd GPSDO to around 1 milliHertz long term.

Mike W9MDB


--
Please follow the HamSCI Community Participation Guidelines at http://hamsci.org/hamsci-community-participation-guidelines.
---
You received this message because you are subscribed to the Google Groups "HamSCI" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hamsci+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hamsci/a3bccc1a-b3a8-4926-b66d-b567b0491612n%40googlegroups.com.

Dana Whitlow

unread,
Jun 10, 2021, 8:40:46 AM6/10/21
to ham...@googlegroups.com
I keep running into "things" that use inexpensive (i.e., affordable) synthesizers,
that cannot in general produce convenient frequencies such as at integer multiples
of 1 Hz.  Such devices may appear to be doing so, but on careful checking they
do not; instead, they produce something nearby, which are not even predictable
over the frequency range of the device without additional information.  

Does anybody in this group know if this is a risk with any of the usual ham radio
gear?  If so, then frequency measurements made with such gear may not be
nearly as accurate as one might suppose.

The Bodnar GPSDOs having user-adjustable reference output frequencies do warn
of this issue in product data, but without additional work on the part of the user the
actual frequency remains a mystery.  I do not presently have access to one of these
units and so cannot report on actuality.

The specifications for the SRS SG38x signal generators also mention this topic,
and the spec sheet says: "Frequency error < (10^–18 + timebase error) × Fc".   
This is certainly good enough for me, but my pocketbook has objections  :-)

I thought I'd toss this out as food for thought.

Dana     K8YUM
Kerrville, TX




Black Michael

unread,
Jun 10, 2021, 8:49:30 AM6/10/21
to ham...@googlegroups.com
We haven't identified any rigs with such problems yet....though nobody is using cheap synthesizers in our group that I know of.

Everybody that uses FLDigi has been able to get sub-Hz accuracy with a GPSDO reference signal (even without GPSDO feeding the rig).

You can pick a local AM station to measure too and you'll get a very stable signal...albeit not necessarily very accurate (they are allowed to be off frequency a bit).

Mike W9MDB




Dana Whitlow

unread,
Jun 10, 2021, 9:01:03 AM6/10/21
to ham...@googlegroups.com
Thanks, Mike.  

If I can ever get my hands on a ham rig or two having external reference inputs, I'll 
do some measurements (once I manage to lick my own system limitations).

In my book, sub-Hz accuracy is pretty gross.  With care I can keep my Rb within
about 1E-11 for several hours at a time, which would be 100 micro-Hz at 10 MHz.

I'll tell you, that SRS generator gets more tempting by the moment.

Dana


Dana


Phil Erickson

unread,
Jun 10, 2021, 9:07:05 AM6/10/21
to Unknown
Hi Mike,

  FYI, I'm not sure AM carriers are always a good choice.  It depends very much on the hardware and engineers at the transmitter; David McGaw (Dartmouth) is on this list and can tell you that many carriers are not necessarily stable even in a relative sense.  I'm attaching a quick plot of a northeast region AM station carrier frequency as a function of universal time, recorded on a GPSDO locked system; it comes from one of Dartmouth's AM Doppler network receivers in Schenectady, NY.

  There are two stacked traces; ignore the bottom one. The upper station has a clear carrier modulation every 15 minutes with a strange waveform.  Amplitude is at the 0.1 Hz level (between 1.0 and 2.0 on the Y axis is 1 Hz).  I would probably not want to use that as a reference.

  Note that I am deliberately not naming the station as it is not relevant here.  Of course, no listener would ever notice this is going on because of the nature of AM detection, so it probably just doesn't get addressed.

73
Phil W1PJE




--
----
Phil Erickson
phil.e...@gmail.com
Screen Shot 2021-06-10 at 8.59.50 AM.png

Black Michael

unread,
Jun 10, 2021, 9:21:55 AM6/10/21
to ham...@googlegroups.com
I think you'll find the Bodnar Mini GPS does that with a lot less $$ involved.
Similar phase noise at 10MHz carrier above 100Hz.
Found the SG380 spec sheet here

Inline image

Inline image

SG380 shows 1e-11 1s Allen variation
Bodnar shows about 1.5e-11
Isn't that good enough?


Inline image

Mike W9MDB

Black Michael

unread,
Jun 10, 2021, 9:35:23 AM6/10/21
to ham...@googlegroups.com
At least the deviations are repetitive and known and stable.
Main reason for using local AM is to see your own rig drift....and then to use a GPSDO reference to see how you can correct that drift (and rig measuring error).

Mike W9MDB




Phil Erickson

unread,
Jun 10, 2021, 10:37:11 AM6/10/21
to Unknown
Hi Mike,

  Just to be clear: most of them are not nearly that stable in my experience - I just yanked out the first one I had handy.  Variability is high.  I guess enforcement isn't what it used to be.

73
Phil

Robert McGwier

unread,
Jun 10, 2021, 11:35:14 AM6/10/21
to ham...@googlegroups.com
And all of them have phase noise and frequency drift incompatible with serious precision coherent measurement attempts.  They don't care about phase noise as much as they care about frequency stability, and even that is "barely meet the standard and cheat until caught and save a few bucks".

Bob




--
Dr. Robert W McGwier, Ph.D.
Adjunct Faculty, Virginia Tech
ARDC Member of Board
N4HY: ARRL, TAPR, AMSAT, EARC
Sky: AAVSO, SkyHub, Auburn AS, Skyscrapers

Dana Whitlow

unread,
Jun 10, 2021, 12:19:53 PM6/10/21
to ham...@googlegroups.com
Hi Mike,

I'm well familiar with the phase noise plots for the SG38x series.  But this only tells
me the "sideband phase noise" at offsets of >= 10Hz.  What I'm more interested in
is what I'll term "baseband phase noise", which is some measure of the noise
coming out of a perfect phase detector measuring the phase difference between
the "unknown" signal and a perfectly quiet reference signal at the same average
frequency.  I'm most interested in the rms baseband phase noise over the band
of roughly 0.2 Hz to 10 Hz.  This could perhaps be best expressed in the form of
PSD from DC to some upper limit > 10 Hz.

I've asked SRS for some information on this, but they don't even specify or
(apparently) even bother to measure this property on their generators.  Interesting
that you found the 1s Allen Deviation.  Either I'm blind, or SRS has improved
their datasheet since I last asked a couple of years ago.  Perhaps my enquiry
back then has spurred them to include that measurement.

So, if the SGS is good enough, there is a fair likelihood that the Bodnar would
be, too.  But I presently don't know how to tell whether either meets my particular
need.  I've asked around "time-nuts", but so far nobody has ventured how to
convert from an ADEV plot to what I want to see.  So at present I'm inclined to
regard ADEV as a sometimes-useful way to compare competing instruments,
but otherwise of no practical value to me.

Do you happen to know the secret of converting ADEV numbers to something
more useful like phase deviations over a specified band?  If so, I'm definitely all ears.

Thanks much,

Dana


John Ackermann N8UR

unread,
Jun 10, 2021, 12:36:28 PM6/10/21
to ham...@googlegroups.com
Hi Dana --

I strongly suspect that most of the rigs out there will have uHz (and
maybe mHz) offsets from nominal frequency. It seems to be in the nature
of the common DDS and PLL systems that you simply can't generate an
absolutely precise frequency. I think it's in the nature of any digital
process applied to an analog signal -- there is always just so much
resolution, either in a DDS lookup table, or in the fractional-N divider
in a PLL. (Remember, you can't exactly represent all fractions in
binary; that's why floating point value have only so many usable digits.)

However, if you know the details of the chip(s) involved, you can work
out what the offset is for a given dial frequency. I recall that
someone has created a calculator to do that for at least one rig, but I
can't recall who it was, or what radio.

73,
John
----
> <mailto:sp6...@gmail.com> <sp6...@gmail.com
> <mailto:sp6...@gmail.com>> wrote:
>
>
> Hi there,
>
> I am wondering how GPSDO frequency should be used (other than
> calibrating radio clock just before measurements) when performing
> long-term frequency measurements?
>
> I am thinking about radios that doesn't accept external
> synchronization signal.
>
> I suspect that 1[kHz] signal monitored by FLDIGI should be somehow
> compared to the signal GPSDO output or both signals should be
> recorded at the same time?
> When radio clock is drifting away how we compensate for that having
> GPSDO signal accessible on the bench?
>
> If that maters I am using FT-891 but I think wider audience is
> interested how to utilize GPSDO in long-term frequency measurements.
>
> Thank you for your feedback!
>
> Grzegorz SP6HFE
>
> --
> Please follow the HamSCI Community Participation Guidelines at
> http://hamsci.org/hamsci-community-participation-guidelines
> <http://hamsci.org/hamsci-community-participation-guidelines>.
> ---
> You received this message because you are subscribed to the Google
> Groups "HamSCI" group.
> To unsubscribe from this group and stop receiving emails from it,
> send an email to hamsci+un...@googlegroups.com
> <mailto:hamsci+un...@googlegroups.com>.
> <https://groups.google.com/d/msgid/hamsci/a3bccc1a-b3a8-4926-b66d-b567b0491612n%40googlegroups.com?utm_medium=email&utm_source=footer>.
>
> --
> Please follow the HamSCI Community Participation Guidelines at
> http://hamsci.org/hamsci-community-participation-guidelines
> <http://hamsci.org/hamsci-community-participation-guidelines>.
> ---
> You received this message because you are subscribed to the Google
> Groups "HamSCI" group.
> To unsubscribe from this group and stop receiving emails from it,
> send an email to hamsci+un...@googlegroups.com
> <mailto:hamsci+un...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/hamsci/926136522.2769469.1623326189159%40mail.yahoo.com
> <https://groups.google.com/d/msgid/hamsci/926136522.2769469.1623326189159%40mail.yahoo.com?utm_medium=email&utm_source=footer>.
>
> --
> Please follow the HamSCI Community Participation Guidelines at
> http://hamsci.org/hamsci-community-participation-guidelines
> <http://hamsci.org/hamsci-community-participation-guidelines>.
> ---
> You received this message because you are subscribed to the Google
> Groups "HamSCI" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to hamsci+un...@googlegroups.com
> <mailto:hamsci+un...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/hamsci/CADHrwpe_FjTZWFitjdLyr089sUM_XgiF%2B_oQZ1znGhvp5cLHpA%40mail.gmail.com
> <https://groups.google.com/d/msgid/hamsci/CADHrwpe_FjTZWFitjdLyr089sUM_XgiF%2B_oQZ1znGhvp5cLHpA%40mail.gmail.com?utm_medium=email&utm_source=footer>.

Black Michael

unread,
Jun 10, 2021, 12:38:20 PM6/10/21
to ham...@googlegroups.com
The Bodnar plot does have Sigma(Tau) at 50Hz which is close to what you want.  You could probably extrapolate.

You would need a standard (e.g. Rubidium) to make that measurement,.

Mike W9MDB






John Ackermann N8UR

unread,
Jun 10, 2021, 1:02:49 PM6/10/21
to ham...@googlegroups.com

On 6/10/21 12:19 PM, Dana Whitlow wrote:
> Hi Mike,
>
> I'm well familiar with the phase noise plots for the SG38x series.  But
> this only tells
> me the "sideband phase noise" at offsets of >= 10Hz.  What I'm more
> interested in
> is what I'll term "baseband phase noise", which is some measure of the noise
> coming out of a perfect phase detector measuring the phase difference
> between
> the "unknown" signal and a perfectly quiet reference signal at the same
> average
> frequency.  I'm most interested in the rms baseband phase noise over the
> band
> of roughly 0.2 Hz to 10 Hz.  This could perhaps be best expressed in the
> form of
> PSD from DC to some upper limit > 10 Hz.

The minimum offset a manufacturer shows for their oscillator is a good
clue about what performance they expect it to have. If the data isn't
specified below 100 Hz, it's probably because it doesn't look very good
down there.

The lowest offset you typically see in a phase noise plot is 1 Hz.*
Below that ADEV tends to be a more useful value, I suspect because over
>1 second intervals environmental and other slow(er)-changing factors
tend to dominate over the oscillator's inherent noise (remember, a 0.01
Hz offset is equivalent to tau=100 seconds in an ADEV plot).

There's not a completely straight-forward conversion of ADEV to dBc/Hz
but you might take a look at the most excellent Stable32 software by
Bill Riley
(https://ieee-uffc.org/frequency-control/frequency-control-software/stable32/)
which can render raw phase or frequency data as a phase noise plot as
well as various flavors of ADEV. I don't know how many people actually
use that rather than the ADEV plot, but it's there if desired.

[ Stable32 is the industry standard tool for calculating frequency
stability, and thanks to Bill's generosity it's now available for free
download at the link above. The plan is for the source code to be
available, too, but it requires a lot of work to clean up and remove
some proprietary libraries before that can happen. That work is
proceeding, but slowly. ]

73,
John

* John Miles's 5330/3120A TimePod and 53100A PhaseStation analyzers can
measure offsets down to 0.01 and 0.0001 Hz respectively, if you want.
Reply all
Reply to author
Forward
0 new messages