When we designed the crystal oscillators in question (KIO2, KPA100), we
did extensive testing using dozens of crystal frequencies to find the
optimal ones. The frequencies presently in use are as good as anything
else that might be tried, I believe, unless you have a specific issue
that can be dealt with by moving an oscillator a short distance.
If someone wants to experiment with this, they should keep in mind that
spurious signal search can start on paper, but then must be done on the
K2 itself, in both RX and TX modes, across all bands. In TX mode this
is cumbersome: you have to move the VFO a kHz or so, press
TUNE+DISPLAY, check the spectrum analyzer, cancel TUNE, then repeat.
(You can't just spin the VFO knob during TX, unless I write some new
firmware :)
You have to also be prepared to accept a compromise solution (as we
did), because the mixing process involves harmonics up through high
VHF, making it nearly impossible to eliminate every last ham-band
birdie of any amplitude. To achieve that would require extensive
internal shielding, which would have made the K2 too expensive to be
profitable (for us) and affordable (for you).
Regarding custom crystals: Expect to pay about $10-20 for very small
quantities, depending on how quickly you want them. Cal Crystal is a
good choice. If a very promising alternative frequency is found,
crystals can be bought in large quantities for under $1.
I should mention that there's another, perhaps cleaner way to generate
negative DC supply voltages: use a high-amplitude AF sine wave
oscillator rather than an RF oscillator. This takes a lot more parts,
assuming you want a very pure sinewave that doesn't droop or become
distorted when loaded. I'd suggest a twin-T RC oscillator or a
carefully trimmed op-amp Wein bridge oscillator, followed by a
high-input-impedance buffer and a power amp running from 12 V. You then
have to add a detector similar to the ones used in the KIO2 and KPA100,
but of course the capacitors will be much larger due to the lower drive
frequency. This would certainly not fit on the existing KIO3 PCB; it
would have to be redesign using all SMD parts.
Note that RS232 ICs with on-chip charge-pumps use square waves, not
sine waves. They will create loud, broadband noise across the lower HF
bands unless completely shielded or implemented using very small SMD
components, ideally on a 4-layer board, with extensive I/O and power
supply decoupling.
73,
Wayne
N6KR
---
http://www.elecraft.com
_______________________________________________
Elecraft mailing list
Post to: Elecraft@mailman.qth.net
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com
