[time-nuts] Help needed with frequency standard
Luiz Alberto Saba
las at intercat.com.br
Wed Sep 30 14:41:09 UTC 2020
Em 30/09/2020 11:06, Dana Whitlow escreveu:
> Hello Luis and Mark,
>
> Luis, am I correct in supposing that the 6685 is basically a GPSDO
> with a 10MHz output
> that you are using as an external reference for a counter (the CNT-90)?
Nope. It'a counter/calibrator originally developed by Pendulum for Philips.
Look at:
http://www.testequipmenthq.com/datasheets/FLUKE-PM6685R-Datasheet.pdf
> My experience with a not-locked-to-PPS PRS-10 seems to suggest that
> a 2-hour measurement run is not sufficient to show the whole story. A day
> (or even several days) is more appropriate.
>
> Also see Bob's (KB8TQ's) comments on the Rb's stability, submitted to
> the group just now on 9/30/2020. I concur fully with his observations.
>
> I have the PRS-10 (a used telecom variety which does not have the PPS
> locking capability), and a Symmetricom L-Pro.
>
> I also have 2 or 3 ancient T'bolts and a fairly new CNS Clock II. All of
> these are GPSDOs but with PPS outputs as well. The CNS Clock II's
> PPS output is largely corrected for sawtooth error by an internal HW
> mechanism. I'm not so sure what, if any. sawtooth mitigation the T'bolts
> may have.
>
> My interest is almost exclusively in phase, not frequency per se. I view
> a source as being a phase ramp generator whose phase slope is simply
> what most folks would call "the frequency". The question that I always
> want to ask about a source is: "What is the rms variation in phase between
> the DUT and an ideal source putting out the same average frequency, when
> measured over a specified frequency band?".
>
> I have never been enthusiastic about frequency counters or time interval
> counters because of my belief that these devices only sample the phase
> at a few selected zero crossings of the input signal ("selected" by
> the pulse
> rate), and ignore everything that goes on in between. Instead I prefer
> direct
> phase measurements of a reasonably high frequency CW signal such as
> the common 10MHz "reference" frequency.
>
> So I tend to record IQ signals derived from quadrature demodulators in one
> form or another. For about two years my favorite was a home-brew affair
> built from Mini-Circuits components. This certainly had its warts,
> principally
> phase measurement errors due to quadrature phase error in the demod,
> drifty DC offsets in the I & Q signals, etc, but it was good enough
> for a lot of
> learning and development of improved techniques. Then, much more
> recently,
> I "discovered" the Signal Hound family of USB spectrum analyzers. The
> model
> I selected was the lowest bandwidth model they sell, the SA44B. I had
> ignored
> them for years, thinking that the baseband phase noise of such inexpensive
> units would be really awful. But then I got a chance to do hands-on
> evaluation
> of one, and discovered that things were not bad at all /provided that
> the external/
> /reference was being driven by a stable source./ The /internal/
> reference, however,
> was about as bad as I had expected. Anyway, the SA4B has quite a
> variety of
> tricks up its sleeve, including an excellent IR recorder (when
> operating in ZERO-
> SPAN mode). This recorder can stream IQ to disk with no particular
> limit to the
> file size as long as you don't fill up the O/S partition with data
> thereby crashing
> your computer. I always record to an external SSD to avoid this risk.
>
> Then, with an IQ recording on file, I can leisurely post process to
> put a narrow-
> band filter around the signal of interest, convert to unwrapped phase
> versus
> time, and make a plot. I can measure slope value(s) on that plot to
> get localized
> frequency information, or whatever I want to see.
>
> Oops! Gotta run now.
>
> Dana
>
>
>
>
>
>
>
>
> On Wed, Sep 30, 2020 at 7:11 AM Luiz Alberto Saba <las at intercat.com.br
> <mailto:las at intercat.com.br>> wrote:
>
> Dana & Mark:
>
> I have a PM-6685R counter and a CNT-90.
> I "calibrated" the rubidium of thr 6685 the following way:
> Use the output of 6685 to feed the timebase of CNT-90 and measure,
> using
> statistical mode, 1 measure/s, for 7200 sec the output of my
> thunderbolt
> in a "calm" period of the day. Then I adjust the rubidium utnil
> CNT-90
> reads 10.000.000.000.000 after 2 hours. Time consuming and a patience
> test, but I could not think in other way to calibrate my 6685 and
> CNT-90.
> Well... Using the same CNT-90 with the same rubidium timebase from
> 6685
> to read the output of my PRS-10 connected to a 1 pps source,
> during the
> same 2 hours, the p-p variance of the reading was about 3 mHz.
> It's low?
> yes, but i was expecting much less. Something in the uHz range.
> I did something wrong or that's it?
>
> Em 29/09/2020 23:23, Dana Whitlow escreveu:
> > Mark:
> >
> > I don't have a PPS syncable PRS-10 yet, but will probably be
> getting one
> > within a
> > year or so. My expectations are not great, however, because I
> already know
> > that
> > the received GPS time has diurnal wiggles to the tune of up to
> 10's of nsec
> > p-p.
> > And I also know that the PRS-10 (and probably all others in its
> price
> > class) make
> > seemingly random frequency "adjustments" on a time scale of just
> a few
> > hours.
> > For a GPS disciplining loop to fix both, it would have to have a
> loop time
> > constant
> > long compared to a day on the one hand, yet a loop time constant
> of well
> > under a
> > day on the other. It's kind of tough to have one loop with both
> properties
> > at once.
> >
> > Dana
> >
> >
> > On Tue, Sep 29, 2020 at 1:22 PM Mark Spencer
> <mark at alignedsolutions.com <mailto:mark at alignedsolutions.com>>
> > wrote:
> >
> >> I would be curious to hear from those who have a GPS
> disciplined PRS 10 in
> >> their home time labs. I was never that impressed with the
> performance of
> >> one of mine in that configuration from a 10 MHz ADEV
> perspective (I did try
> >> making a few parameter changes.) I ended up essentially free
> running mine
> >> but would occasionally apply a 1pps signal to mine as means of
> adjusting
> >> its output frequency on an as needed basis.
> >>
> >> I may try this again with a different PRS10 (the 1pps output on
> the unit
> >> in question failed a few years ago but I have a spare that has been
> >> recently put into service.) Perhaps the second unit will work
> better.
> >> Both units were obtained via the usual auction site.
> >>
> >> Mark Spencer
> >> mark at alignedsolutions.com <mailto:mark at alignedsolutions.com>
> >> 604 762 4099
> >>
> >>> On Sep 29, 2020, at 8:02 AM, Luiz Alberto Saba
> <las at intercat.com.br <mailto:las at intercat.com.br>>
> >> wrote:
> >>> Hi guys
> >>>
> >>>
> >>> I want a stable & reliable frequency standard.
> >>>
> >>> I do NOT have a cesium beam.
> >>>
> >>> I DO have a lot of gpsdos and some rubidiums.
> >>>
> >>> Since I have one PRS-10 and this device has a 1 pps input, you
> already
> >> know what I'm thinking...
> >>> What I have:
> >>>
> >>> Trimble Thunderbolt
> >>>
> >>> Trimble UCCM
> >>>
> >>> Symmetricom UCCM
> >>>
> >>> Samsung UCCM
> >>>
> >>> HP 58503A
> >>>
> >>> TruePosition
> >>>
> >>> Oscilloquartz Star 4+
> >>>
> >>> and a little board with a ublox LEA 5T and 1 pps output.
> >>>
> >>> Which device is more suitable for disciplining the PRS-10?
> >>>
> >>> My gps antenna setup is:
> >>>
> >>> Symmetricom 58532A on the top of the water tank of my building
> with a
> >> very good sky view.
> >>> A Raven Industries LA-21-1575-100-T. About 50 m of RG6.
> >>>
> >>> A Symmetricom antenna distributor (don't remember the code but
> is the 4
> >> outputs one)
> >>> By the way... Someone knows the delay of the inline amplifier
> and the
> >> antenna distributor? (for Lady Heather)
> >>> Mark Sims is out there reading this list? I'm having problems
> with LH
> >> 6.14 and the UCCM units.
> >>> Any help you can give me will be welcomed.
> >>>
> >>>
> >>> Thanks
> >>>
> >>>
> >>> Luiz Alberto Saba
> >>>
> >>> from Santos - São Paulo - Brazil
> >>>
> >>>
> >>>
> >>> _______________________________________________
> >>> time-nuts mailing list -- time-nuts at lists.febo.com
> <mailto:time-nuts at lists.febo.com>
> >>> To unsubscribe, go to
> >> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> >>> and follow the instructions there.
> >> _______________________________________________
> >> time-nuts mailing list -- time-nuts at lists.febo.com
> <mailto:time-nuts at lists.febo.com>
> >> To unsubscribe, go to
> >> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> >> and follow the instructions there.
> >>
> > _______________________________________________
> > time-nuts mailing list -- time-nuts at lists.febo.com
> <mailto:time-nuts at lists.febo.com>
> > To unsubscribe, go to
> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> > and follow the instructions there.
>
More information about the Time-nuts_lists.febo.com
mailing list