[Elecraft] 4 Mhz Oscillator calibration

[Ron D'Eau Claire]

Andreas wrote:

Interesting. You are saying that the reference signals (WWV or what
ever) should be close or even inside the amateur bands in order to get
reasonable results, aren't you?

If I do use the 10 MHz WWV, can I assume that a frequency on 14,100 is
almost dead on, even though WWV on 20Mhz is way off?

How about using one of the beacon frequencies on 15 m or 20 m instead of
using WWV?
------------------------------------------------------------------------
------------------------
Hi! 

According to the Elecraft K2 User's Manual (Rev C) , they stated that
when using a signal with a known frequency to adjust C22 to make the
display read accurately: "Because of the way CAL PLL works, you can only
use a signal source that is at one of the lower band edges, e.g. 7000.00
or 10000.00 kHz." 

I suspect that is because those frequencies are ones most accurately
measured by the CAL PLL function. CAL PLL builds a "lookup table" of
tuning voltages needed to produce the desired frequencies from the local
oscillator and BFO. This table does NOT include EVERY frequency you
might tune your K2 to. Frequencies in between the measured ones are
interpolated or "estimated" by the firmware, so the frequency display is
not as accurate for those. 

That's my understanding of the issue. ALSO, keep in mind that the
firmware has been improved to improve the overall frequency accuracy
since that procedure was written. I don't know whether that is as much
of an issue today as with earlier firmware. If you want to use an "on
air" signal on some other frequency, I'd query the Elecraft folks about
it first. 

Some months back several ops on the reflector here noted that their dial
accuracy was vastly improved by setting the 4MHz oscillator as close to
4.0000 MHz as possible.  Some of them had well-calibrated highly-stable
frequency standards they could use. I didn't, but I have long used WWV
at 20 MHz to set the 4MHz clock in my frequency counter, and it seemed a
logical way to set the 4MHz clock in my K2. I did so and reported
excellent results. 

I used an external receiver that I have in the shack. I tuned in WWV at
20 MHz in "AM" mode, and arranged a pickup wire so I could hear the 4
MHz oscillator beating with the WWV carrier. The adjustment was very
easy. Several ops noted that they did NOT have a second receiver and
used the K2 itself. Then I was asked to document the procedure, so I
tried it with the K2 and found that it was very easy to adjust C22 for
zero beat with one of the sidebands of WWV instead of the carrier. That
resulted in the 4 MHz clock being several hundred Hz off frequency. A
couple of other ops also noted that they had to "give it several tries"
to get it right. The two-step process of using the SPOT tone that I sent
you earlier seems to have fixed that ambiguity and still results in a
setting as close as the K2's circuits could use.

You mentioned that WWV at 20 MHz is "way off". I have noticed that my
"dial accuracy" on 20 MHz is not nearly as good as it is in the Ham
bands or at 10 or 15 MHz for that matter. So I assumed that the firmware
is not as good at estimating the correct control voltage that far
outside of a Ham band. 

Let me share a couple of other points about the "dial accuracy" that
bear on this issue. 

Using a frequency counter to spot check frequencies across the Ham
bands, I have found that my K2 is within 20 or 30 Hz of the displayed
frequency on all bands at MOST times. Please note that the oscillators
in the K2 are NOT temperature-compensated. They DO drift a bit with
temperature. They are variable crystal oscillators - VXOs - rather than
free-running oscillators. So they are reasonably stable but they still
change frequency somewhat when subjected to changing temperatures. And
when they move, the displayed frequency is thrown off. 

That surprises some ops used to fully-synthesized rigs. The K2 is NOT a
fully-synthesized rig with the oscillators rigidly "tied" to a master
clock oscillator. That gives it some very big advantages in the
"cleanness" of the oscillator signals that shows up in everyday use in
low noise in the receiver and higher than usual resistance to certain
types of interference, but it has the disadvantage that the oscillators
may "drift" a bit in frequency. 

And then you must consider the fact that the hardware in the K2 (the
digital to analog converters or "DACs") have a resolution limit too.
That resolution limit allows an additional frequency error of 20 or 30
Hz. 

Note that Elecraft says that the VFO stability is "<100 Hz total drift
typ. from cold start at 25 degrees C". Under most conditions, my K2 is
well inside that limit.  When we talk about "drift" under normal
operating conditions, we are talking about values too low for most ops
to detect by ear. 

If there is anything further that I can do, just ask! 

73,

Ron AC7AC
K2 # 1289