[Elecraft] KSB2-Filter Measurements

[Stewart Baker]

Having had to today off work because of the snow, I had a chance to look at my 
KSB2 SSB filter characteristics. My biggest interest was whether the filter had 
blow-by and if so, how much.

First I modeled the filter input and output switching networks using 150 ohms as 
the source and filter load impedance.  I expected more leakage from the 
switching, and was pleasantly surprised when I got 110 dB of isolation. 
Of course the real thing will have greater leakage, but at least it is in the 
Ball Park.

To get a rough idea of the crosstalk between the input and output connectors on 
the RF board and also the tracking I used the CW filter. Using a signal 
generator I injected a low level signal and peaked it in the filter. I then 
removed the CW filter and increased the signal generator level until I got the 
same S meter reading. Although rather crude this measurement gave me a 
measurement of around -95dB. Again not bad.

Using Spectrum Lab set to 24bit sampling I got a K2 OFF on screen noise floor of 
about -115dB. Switching the K2 ON raised this to -95dB.
I connected a noise generator and set my K2 to 28MHz, AGC OFF, and PreAmp ON.
I made the following measurements:-
0dB      1kHz
-6dB     2.7kHz
-60dB   3.5kHz
-90dB   3.9kHz  
Shape factor 1.3
The noise floor was intercepted at about 4.5kHz. Both sidebands were nearly 
identical. Switching off the PreAmp improved the reading slightly.
This performance was very much better than I expected. The passband has, however 
about a 2dB ripple in it. Any suggestions on a cure for this ?

I tried to improve the filter response by adding brass shim as screening on the 
top and bottom of the board, but it made no noticeable difference. No blow-by 
here...
I think that the filter is working just about as well as the design permits and 
I am very pleased with the results.

73
Stewart G3RXQ