[Elecraft] K3EXREF and Trimble Thunderbolts

Tue Apr 5 13:35:49 EDT 2011

Julian, G4ILO wrote:
> 
> What are the advantages / disadvantages of this type of frequency standard
> over the Efratom LPRO-101 which is a rubidium standard?
> 
> Julian, G4ILO
> 

I'm a duffer at this time/frequency stuff, but I've scored well enough at
the ARRL FMT to get my call in QST, and I gave a presentation on it at the
Bay Area Maker Faire last year, and a NIST scientist told me I didn't say
anything egregiously wrong, so I'll try to stick to what I understand here. 
If you want to know more, go to Febo.com or time-nuts and read a bit (before
you ask questions) and you'll find people with literally orders of magnitude
more experience (and interest) than you ever thought possible.

Both an Rb oscillator and a GPS-DO (and for that matter, your own Cesium
standard) can provide you with a frequency output, but they're different
types of things.  A GPS-DO is a managed device with an output, like WWV or
the grid 60 Hz (mains 50Hz)  frequency.  A Rubidium oscillator is like a
quartz crystal, but more stable over the long term (modulo a well-defined
linear drift), so once it's set, it stays set exceptionally well, though
usually the phase noise isn't as good as a temperature-controlled quartz
crystal.  A Cesium standard is called a standard because it's how we define
time; when properly adjusted, it will remain in calibration quite well.

A GPS-DO usually contains a crystal oscillator which is controlled
(disciplined) to match the data coming in from GSP satellites, which
themselves contains Cesium standards which themselves are actively managed
by the US government.   So it's like listening to WWV or one of the other
time signals on HF, but it uses the 1.4 GHz range and averages multiple
signals, so you get better results with smaller propagation-induced errors. 
There are some Rubidium GPS-DO devices as well, some commercial and some
done by smart hams, but even those, as far as I know, contain a quartz
crystal oscillator which itself is disciplined by the Rubidium bulb
microwave resonance.

There's arguments about different ways of disciplining all of these
oscillators and if you're interested, look for the time-nuts mailing list or
read on febo.com, or read about Allan Deviation.  But briefly: there are
multiple sources of noise, and if you measure the noise on different time
scales, you get different answers about how much noise there is.  So,
roughly, phase noise is at one end of the chart and how far your wristwatch
is off at the end of the month is at the other end.  What time scale you
care about determines what type of oscillator you use and how you measure
and take care of it.

A Rubidium oscillator does quite well on many of these time scales, better
than Cesium in some of them, but worse than a plain quartz crystal in
others.  

Unfortunately, my guess is that buying an Rb standard and using it instead
of the quartz TCXO in the K3 would result in worse phase noise which
translates into a noisier band RX and noisier TX.  Quartz is pretty much the
way to go for transceivers given the areas of concern that hams have. 
(I.e., would you rather have 20dB less band noise or be able to get within
0.1 millihertz of the band edge instead of just 1 millihertz of the band
edge?)

More philosophically, we define time in terms of what Cesium does.  Both Cs
and Rb physics packages have knobs to adjust for external magnetic field
("C" Field), so both can be set wrong, but Rubidium bulbs "age" and change
frequency (though at an fairly well characterized rate), so you can't
substitute a Rubidium oscillator for a Cesium one.  According to NIST,
gravity is a big effect as well, so if you ever move one of your
oscillators, its calibration field is toast.  And finally, temperature
control is of the utmost importance, and those who are serious use multiple
nested containers for temperature isolation.

For ham use as a frequency standard, in terms of phase noise performance,
the algorithm used and the oscillator that's being disciplined (and how
often, as part of the algorithm) has the strongest influence. 

For example, a GPS-DO is additionally affected by what satellites are in
view (the "constellation"), and the Trimble units are affected by the
changes more so than more expensive units which are designed as frequency
references, because they make abrupt changes when the constellation changes,
which is pretty much every few minutes as sats pass over.

The eBay Trimble units in particular have trouble with this.  They were
designed for E-911 installations in cell sites, where they were used to
calculate distance via speed of light to a mobile unit. My (unfounded,
personal) belief is that it was more important that nearby units all had the
same idea of the time than that they have reduced noise on the shorter-term. 
So they make more frequent adjustments to their internal quartz oscillators
that are comfortable for a frequency standard, though a number of smart
folks have figured out algorithm tweaks to mitigate this effect, making the
temperature the dominant factor, for time scales of 1hr - 24 hrs.

The K3XREF product that Elecraft is developing is, to my understanding, an
external disciplining interface for the TCXO in the K3. It's probably a
frequency counter / microprocessor which reads the internal oscillator and
the external 10 MHz reference, and when the internal oscillator doesn't
produce the right number of cycles in 10 million of the external
oscillator's cycles, it adjusts the voltage on the TCXO to bring it back
into spec.  But if you do this too often, you'll introduce phase noise into
the K3 (think of it as FM-ing).  Wayne N6KR has said it does this a few
times a second an has achieved a trade off between accuracy and phase noise. 
(I presume he doesn't adjust during TX, for example.)

I leave my K3 on most of the time, and I've found that it is seldom off more
than +/- 3 Hz.  But the K3XREF would let it off +/- 1Hz as soon as you turn
it on, provided it's hooked up to your external reference.  The actual
received frequency is only valid for one mode and one filter, once you
calibrate it, since when you shift modes or filters the offets of the
various internal IF stages varies.  (Keep that in mind if you use the K3 for
the ARRL FMT.) 

Leigh/WA5ZNU

--
View this message in context: http://elecraft.365791.n2.nabble.com/K3EXREF-and-Trimble-Thunderbolts-tp6242158p6243171.html
Sent from the Elecraft mailing list archive at Nabble.com.