[Elecraft] K3 diversity phase drift

[Geert Jan de Groot]

[Seems that including the KSYN3A schematics is causing my message to get 
sniped by the mailing list software. Re-sending, w/o the schematics; you 
will need to look them up on the Elecraft website yourself]

Hello,

You may want to elaborate if the K3 you're using to measure, has the old 
KSYN3 or the new KSYN3A synthesizers.

The old KSYN3 is a PLL, *phase* locked loop and I expect both 
synthesizers to keep the phase constant.

The KSYN3A is a rather interesting design in this respect; Elecraft 
calls it "frequency loop".
I've included the KSYN3A schematics from the K3S schematics archive for 
convenience [not anymore - download from elecraft.com yourself]
Frequency generation is done using a SI570. It was a surprise to me that 
the noise specs are so much better, but there you are.

One problem though is that the Si570 can only use an internal frequency 
reference so things like the stable TCXO or K3EXTREF won't work.
To that end, the Si570 signal is mixed with the signal from a Si531, 
which *does* have an external frequency reference input, but supposedly 
less good noise specs, and the results from mixing these signals on the 
SA612 mixer is fed to a DSP PIC to create a control loop that will 
compensate for a drifting Si570 on-chip clock generator.
I'm not sure, but I think keeping exact phase reference is much more 
difficult here. Not a problem for the classic KSYN3A use, but this is 
what you may be measuring.
It would be one of the technical reasons for the "both synthesizers must 
be the same type, KSYN3A" as well as the additional cabling in case of 
upgrading from KSYN3 to KSYN3A. I'll leave details on that for you to 
explore.

Wayne made some cryptic comments on this design on the list some time 
ago. I spent an enjoyable evening studying these schematics alone and it 
is stunts like this that make Elecraft so enjoyable for me.

Perhaps also interesting, the KX3 uses the same SI570 signal generator 
but does not have this frequency compensation loop which makes sense 
given the low-power-usage requirements of the KX series.
Rather, the KX3 sports a temperature sensor and a frequency-drift vs 
temperature calibration to compensate for temperature frequency drift.
(I'm not sure if this is calibrated off factory since I bought my KX3 
second hand and the temp compensation thing is quite labor intensive, 
but temperature drift of my second-hand KX3 got rather better when I 
did/re-did the temperature calibration thing - details of the procedure 
are on the Elecraft website.)

The KX2 uses the simpler Si531 synthesizer, but does have the same 
temperature probe hardware as the KX3. I am not aware though if the KX2 
software has frequency temperature compensation functionality - never 
seen it.
But at least this shows a performance difference between the KX3 (Si570, 
same as K3s, high performance) and the KX2 (Si531 simpler generator)

It is unfortunate that the K4 schematics are not made available. I can 
understand the need to protect IP but I would happily sign an NDA to get 
them (and not post K4 schematic details on the list).

In any case, "K3 with KSYN3 is probably very different here than K3 with 
KSYN3A" for your diversity measurements. For the results of your study, 
you may want to include that parameter.

73, Geert Jan PE1HZG

PS: the schematics are the B version of the KSYN3A. A key difference is 
the inclusion of the U12 reset circuit. Without, the KSYN3A can possibly 
corrupt the programming of the DSP PIC it seems, and that would warrant 
a trip to the factory since this device is not field programmable. 
Elecraft sells retrofit mod boards to add U12 to existing A versions of 
the KSYN3A and as long as the PIC isn't corrupted you can fit the mod 
board yourself - I have.
Back to the diversity phase shift discussion....