[Elecraft] More ideas for K2 modification

[Lyle Johnson]

Hello Sverre!

Well, maybe they need to be on the lsit for the K3 :-)
> >I'd like to see the BFO, and the PLL reference, generated from a DDS,
> so the "steps" at the PLL A
> >reference frequency wouldn't occur.  It would also eliminate the
> offsets you often get between Rx
> >and Tx when running PSK31 or other digital modes whtough the SSB
> section.
>
> Good idea - but a quite fundamental change to the design of the K2! I
> believe that smaller steps could be achieved with the PLL also, and that
> it possibly is a matter of software setup.

As I see it, it can all be done on the control board.  It takes the
following:

1) Instead of a DAC that "pulls" the PLL reference oscillator and the BFO,
you put in a pair of DDSes (could be one chip, certainly can be done with
two chips).

2) For the PLL, you change the R divider in the PLL chip to accept a lower
input frequency, say 100kHz or 500 kHz nominal.  This does not change the
PLL dynamics, the PLL reference frequency, etc.  I did this on an old ICOM
dual-band FM radio several years ago so I could tune it in 10 Hz steps
instead of 5 kHz steps for an OSCAR satellite uplink/downlink radio.  Worked
great, but I had to replace the microprocessor in the control head to make
it work!

3) You then set up a DDS for the new lowered output frequency.  Now, as you
tune this DDS, the PLL follows (same as now).  But the steps are precise, so
there is no error when the PLL changes its division ratio and the reference
oscillator is also adjusted.

4) The line running to the RF board is now a low frequency RF signal instead
of a DC control signal.

5) The BFO is just a DDS output run through an LC bandpass filter (or a
simple crystal filter if you are seriously worried about spurs and whatnot).
Now it, too, is precise.  You will notice when you do a CAL FIL for the BFO
that you can not get precise steps, which is OK for what it does, but less
OK if you want to implement passband tuning and so forth.

6) Now the line running from the control board to the SA602 Product Detector
chip is a low-level BFO signal rather than a DC control signal to a crystal
oscillator.

7) To make this all easy, you'd want to replace the RIT/XIT pot with a cheap
encoder so that you are inputting rotation rather then position to the CPU.
Then this knob becomes multi-purpose: RIT, XIT, passband tuning, even filter
variable bandwidth (why have only 4 selections if you can precisely control
the BFO's relationship to the filter from the front panel?).

> >And it would make passband tuning a breeze :-)
>
> Doesn't this require a dual-conversion receiver where the second
> oscillator is offset from its normal frequency?

No, you just move the HFO the same amount, but in the opposite direction, as
you move the BFO.  The net result is that the receiver stays tuned to the
same incoming frequency, but the passband is shifted. No extra conversion
necessary.  And with the CPU controlling both oscillators precisely, it
would work quite well.  As has been pointed out, you can do this crudely by
setting up more filter positions in the K2 to achieve the same end - but
then you lose the ability to select a narrower filter...

Oh well, I'm sure I've beaten this to death.

72/73,

Lyle KK7P