[Elecraft] Trying to improve the K2's IIP3; mission (almost...) impossible!

[D.W. Harms PA2DW]

Hi all,

I have asked before on this reflector if anyone knew why the K2 does not use
a high level mixer. Don W3FPR said "put one in and report your findings
here". So thats what I did, and here are my results sofar. By the way, leave
your soldering iron cold, as the endconclusion is that it may be a better
idea to leave the K2 as it is ;)

IIP3 of mixers at their lower frequency ranges, are something like 15dB
above the 1dB compression point. The TUF1 (or TOP-1 as sometimes used by
Elecraft, not sure what the maker is!?!) has +1dBm compresion point, so it
has a total IIP3 of +16dBm.
Adding the losses of the bandpassfilters and switches brings the IIP3 at the
claimed IIP3 of 21 dBm (preamp off).

The TUF-1H (or TFM-1H) has +14dBm of input compression point what turns out
to be +29dBm of IIP3 at HF bands! Adding the input filter loss - you can get
something like +34dBm of (compelete RX) IIP3, but only when the postmixer
amplifier and the XTAL filter are linear enough!
What value is good enough for the postmixer stage? With +29dBm of mixer-IIP3
and 5dB of conversion loss its OIP3 is +24dBm. Now if the postmixer
amplifier has an IIP3 of +24dBm, it will lower the overall IP3 by 3dB!!! So
you need a postmixer amplifier that can do at least +29dBm IIP3 to get the
promised +34dBm IIP3 of your RX!
I am not sure what the K2 postmixer amp is capable of, nor did I measure,
but perhaps someone knows that?
There are many great designs of linear IF amps, all consuming 5-10W, so that
almost equals the output of the barefoot K2. Not a good idea hi!

If the needed gain of the postmixer is lets say 10dB, then the XTAL filter
will have to have an IIP3 of more than +44dBm which turns out not to be so
trivial! The K2 may not reach that, but we have not measured that...

Final conclusion is that the IIP3 can not so easely be brought up to the
level that will beat that new Icom rig (IC7800). Perhaps it is possible, but
it would turn the whole design of it's frontend upside down and also, the
powerconsumption will drastically increase (due higher needed LO and the
required power of the postmixer stage).
Our conclusion is also, that the K2 is a very balanced design. Intelligent
in almost every detail. Compliments go to the designers!!

For those that want to know if I noticed any difference with the TFM-1H
installed (and of course raised LO to 17 dBm!!); I did not notice a thing at
all, but my antenna is probably not good enough to know for sure...

Finally I want to thank thank Robert S53WW, Douwe PA0DKO and Jos PA3ACJ in
assisting me with theory, parts (TFM-1H donated by S53WW) and measurements.

Any comments are very welcome!

Best 73, Dick/PA2DW