K2 ssb filter, improving the shape factor and the ultimate rejection

Merlin W3ICT reported on the Elecraft mailing list that putting "houses" around the matching toroids on the ssb board improved the ultimate rejection of the ssb filter.
I have long had a problem of AGC activation from stations far from the operating frequency when using the OP1 ssb filter. I cured the problem by putting a 153 kHz BP filter on the AGC IF. This reduced the activation except when the offending signal is on the AGC IF mirror frequency 306 kHz above or below the operating frequency.

screen

Now I have put some screens on the ssb filter, using pieces of tin and inspired by W3ICT. The picture shows the middle screen and the top screens and one of the vertical screens.

The middle screen is vertical and goes from the ground connection (the inv U) near TP2 towards capacitors H and G. The height of the screen should be the same as the crystals. Solder the screen to the inv U and using a short piece of wire also connect the screen to the ground hole which is between capacitors C and M. This eliminates crosstalk between capacitors C and M. Also this screen supports the screens on the top.
The first top screen is horizontal and goes from the middle screen over T1 down to the circuit board, bend the screen over T1. The width of the screen should be so it extends from D14 to the solder connections on top of crystals X1 and X2. Solder the screen to the top of the crystals on the same spot as the ground connections on the crystals and also solder it to the middle screen.
The second top screen goes from the middle screen to T2 down towards the circuit board, bend it so it covers T2. Solder the screen to the middle screen and top of X6 and X7.
The first vertical screen goes from the middle screen towards D14. Solder the screen to the middle screen near the inv U ground connection and to the top screen.
The second vertical screen goes from the middle screen towards D4. Solder the screen to the vertical screen near the inv U ground connection and to the top screen.
Also put vertical screens on the other side of the top screens. The screens go from the crystals to the edge of the circuit board. Do not solder the screens to the crystals, only to the top screens.
On the bottom side of the SSB circuit board make a low (abt 5mm) vertical screen extending from the inv U ground connection towards the middle of X4. The screen should be bent near the inv U, and then it should follow all ground connections towards capacitors H and G. Solder the screen to all the ground connections it touches. Observe that there is a capacitor on the RF board which limits how far the screen can extend towards X4.
This screening attenuates radiation from T1 and lessens T2 pickup of unwanted signals.
Be very careful not to short any connections when installing the screens. Note the current used in RX mode before you start and if it increases you likely have shorted some connections.

The results

I made some observations before and after the mod. I use a crystal oscillator on 3587 kHz to generate a S9+40 signal. Putting a finger near T1 or T2 would change the AGC activation before the screening, now there is no change. I noted the ACG activation from the strong signal before and after.

At 40 kHz from the S9+40 signal AGC activation is reduced from 2 to 1 LEDs. 1 led is the noise level.
At 306 kHz from the S9+40 signal AGC activation is reduced from 4 to 2 LEDs. This is the AGC IF image.

My guess is that the ultimate rejection has increased at least by 10 dB, which is not bad. And the shape factor of the filter should also have improved, as now there is less coupling between stages.
Next generation of the SSB board should include measures to take care of crosstalk between input and output of the SSB crystal filter. Simple screening of the crystal filter should do the job, but the screens have to be on the top and the bottom of the circuit board. This should make the K2 SSB receiver performance in pair with the CW receiver.

Last modified: lau jan 15 10:44:40 GMT 2005
73 de TF3MA