[Elecraft] K2 Frequency Alignment

[Masleid, Michael A.]

>After doing the Temp comp mod, I very
>carefully aligned the K2 MCU oscillator by Zero Beating it against WWV =
at 20
>Mhz.  This is confirmed audibly, a very slowwwww beat, and visually, =
when
>the audio is piped to the MixW I get a fading track which goes solid =
when I
>disconnect the antenna. The two signals are within 1/2 to 1/4 of a Hz =
of
>each other.  I even fed the signal into the ant jack of the 706 and the =
two
>signals are right on.   BUT, WWV at 10 Mhz is ~ 30 Hz low and WWV at 20 =
Mhz
>is ~ 80 Hz low.  ( Upper and lower side band.) WHY?

>And yes, I did do a CAL PLL, etc.=20

Hello Rich

In my case (3430) and in yours too, I'd guess, the gating for the =
frequency
counter is not the same as others.  When I hooked the probe to C22 I =
didn't
get 4000.00 MHz.  (More like 3999.98 MHz).  This means that I have to =
adjust
the C22 high to get the counter calibrated.  And so do you.

I checked this rather carefully by generating a 4 MHz signal off of an =
oven
controlled oscillator (OCXO) checked against WWVB.  The D layer is =
stable
enough on the daytime path to be very exact.  The end result was WWV at =
10 MHz
being 30 Hz low.

To get out of trouble, I generated a 20 MHz signal from the OCXO, set =
voltage
to match TP1, ran CAL FCTR, and hooked the probe to the 20000.00 KHz =
signal.
The counter said 19999.90 KHz.  Of course.  So I adjusted C22 until the =
counter
showed 20000.00 KHz.  Ran CALL PLL and CAL FIL.  Checked, and now WWV at =
10 MHz
was about 20 Hz high.  Rats.  Of course, at this point it was obvious =
that I
should set CAL FCTR to show 19999.96 KHz to split the difference.  So I =
did, and
WWV at 10 and 15 MHz are within 4 Hz now, but W1AW on 40 meters shows =
about 20 Hz
high.  I prob. have better error distribution with CAL FCTR at 20000.00 =
KHz, but
I like WWV to "look right" so I buttoned it up at that point.

The bottom line, I think, is that if CAL FCTR doesn't show 4000.00 KHz =
when you
probe the 4 MHz oscillator that runs the MCU, the setting the 4 MHz =
oscillator
exactly to 4 MHz won't work too well.  It is better to attach the probe
to some other oscillator, with a known value, and adjust C22 so that CAL =
FCTR
matches the known value.

All this is splitting hairs of course.  Most crystal drift more than =
this in a
year.

(Equipment list:  K2 #3430, Textronics 453, Cushman CE-4, Tru Time =
60-TF, A-60FS)

73, Michael AB9GF