[Elecraft] K2 PLL Temperature Compensation Adjustment

[Lamb, Dick & Judy]

I'm trying without success to optimize resistor RA on the K2  
thermistor board.

I should first say that the temperature compensation with the  
provided 33K resistor is quite good--I just wanted to optimize it.   
With RA at  33K, I'm measuring drift on the 40 and 80 meter bands  
after transmitting to be about 40 Hz.  I'm measuring this by zero  
beating my K2 warmed up on receive only against a  standard, then  
transmitting continuously at 20 W for about 8 min. (resulting in a  
heat sink that's quite warm), then zeroing again against the  
standard.   For example, the measurement before transmitting (as  
displayed on the K2 readout) might be  4000.00, and the measurement  
after transmitting and zero beating again might be 3999.96.

I expected that by adjusting the value of RA between the limits of  
15K and 56K noted in the "K2 Temperature-Compensated PLL Reference  
Upgrade" instructions, I could move from a point of under  
compensation to over compensation.  For example,  given the same test  
conditions, one extreme might be 4000.00 from receive only to 3999.90  
after transmit, and the other extreme might be 4000.00 to 4000.10.   
Someplace between would be optimum compensation with almost no change  
during warm up due to transmitting.

I realize, of course, my experiment isn't precisely repeatable, and  
there will always be some random drift, and that different components  
will affect the drift at different points in the cycle.  However, I  
still expected to find that changing RA made a noticeable difference  
between the extremes, and I'm not seeing that.  At neither extreme  
can I get the zero beat display on the K2 to be higher than it was  
before transmitting.  At most, the extreme resistance values seem to  
change the outcome about 20 Hz.   I am running CAL PLL after each  
change in RA.

Squirting component cooler on the thermistor does cause a rapid and  
extreme change in frequency, so it appears the thermistor circuit is  
at least doing something.

Is my experience typical?  Might there be something wrong in my  
circuit?   Would changing values other than RA help?

Thanks.

Dick, K0KK