Dale Parfitt wrote:

While this approach looks good on paper, it often fails badly when the
desired frequency is so close in to the notch frequency. I just put a
quarter wave stub on our VNA and found that while it does diminish the 123
signal -33dB, it also attenuates the 120.6 signal by a whopping -22dB.
There is also an enormous VSWR upset -120:1 or so- this is perhaps not
important in your receive only application.
Each year we build hundreds of filters for this exact application-
AWOS/UNICOM separation. Typical insertion loss is under 1dB while the notch
is -40dB. The filter is about the size of a cigarette pack exclusive of the
N connectors.

I think you get narrower selectivity if you use an odd integer
multiple of quarter wavelengths for the stub. The longer the line,
the greater the phase change with frequency. A 1/4 wave shorted stub
goes from open to short in a 1:2 frequency ratio. A 3/4 wavelength
stub goes from open to short in 3:4 ratio of frequency. Etc. I think.