[Elecraft] VCO Relationship with 4 MHz Oscillator

[Ron D'Eau Claire]

In a Don, W3FPR, points to a very good reason not to try to use an external
4 MHz clock for the counter. There's an issue with the accuracy available
too. 

The tuning voltages are converted to digital numbers for storage in memory,
and the digital numbers are converted back to analogue voltages as required
when tuning across the band. The conversion is done by DAC's - Digital to
Analogue Converters. During CAL PLL the DAC's sample the tuning voltages
applied to the oscillators and convert them to digital numbers for storage.
In operation the DAC's convert the digital number back to the tuning
voltage. The DAC's used by Elecraft are a trade-off between cost, complexity
and accuracy. From what I've read, they provide a tuning accuracy of about
30 Hz. Even if the K2 was able to measure the oscillator frequency with
absolute accuracy while running CAL PLL, the DAC's can't reproduce the
tuning voltage accurately enough to put that oscillator closer than about 30
Hz of the original frequency in operation. 

So getting the 4 MHz clock closer than is needed for an accuracy of about 30
Hz is an exercise in futility. It can happen accidentally that the value of
the tuning voltage returned for a specific frequency is exactly on, but
that's not predictable. 

Also, there is some drift in the voltage tuned oscillators even though the
PLL reference oscillator now has an active temperature compensation circuit
and the beat frequency oscillator has been improved. Of course *all*
oscillators drift somewhat, and those designed to be tuned over a range of
frequencies tend to drift more. 

The way designers have held the drift of other rigs to an absolute minimum
is to not tune the PLL reference oscillator. Instead they use a very highly
stabilized design that may even include putting the crystal in a
temperature-controlled oven. That does produce a dial accuracy within a
couple of Hz in some cases, but it does it at the cost of having the
phase-locked loop tuning in very small steps of a few Hz at a time instead
of the 5 kHz steps the K2 uses. That produces more phase noise in the local
oscillator that affects the receiver performance. (To the operator, the
phase noise usually sounds just like a somewhat higher "band noise" level.) 

So the oscillator design boils down to a trade-off between phase noise and
stability. Elecraft went for lower noise and a stability and accuracy of
some tens of Hz, instead of a stability and accuracy of a few Hz and more
noise. 

What really impresses me about really good engineering is not that someone
does things perfectly. Everything we do is a compromise of some sort; we can
always imagine it being more 'perfect' than it is. What impresses me about
really good engineering is the ability to choose the best compromises for
the challenge. The performance of the K2 is a good example. 

There is indeed a real 'art' to engineering, whether it's an airplane, or a
bridge, or a Ham rig. 

Ron AC7AC

-----Original Message-----


Hi Ron,

Thanks for this explanation.  Is there an implication here that if you had a
more accurate external 4MHz source you could switch it in when you run CAL
PLL to get a more accurate result in the look-up table?

Mike
VK1KCK