The picture above shows my problem. As predicted by the Murata data sheet and as warned by R.A. Penfold, that nice 455 kc ceramic filter has a significant response at around 640 kc. This caused me to hear Brother Stair twice as often as I should have. Clearly a spur exorcism was called for.
Click on the picture for a better image. As noted last time, my first idea was to build a series 640 kc trap LC circuit and put it ahead of the ceramic filter. But I had trouble getting the desired high Q. So I then thought about putting a wider 455 kc filter ahead of my 12 kc filter. I would, of course, need one that did NOT have the 640 kc spur. I found a 455B filter in my junk box. I used the NanoVNA to look at its response. No spur at 640 kc! Yea!
Next I put the two filters together, 455B first, then the 12 kc filter. Success! You can see on the NanoVNA that there is no spur at 640 kc.
With close to the desired termination impedances, the passband at 455 kc looked pretty good. I just put 1500 ohm resistors in series at the input and output of the dual filter combo.
It worked. Spur exorcised! I no longer hear each SW broadcast stations at two spots on my dial.
