> When I say the notch presents a "short" it is not really a
> short but a very low impedance of say a few ohms. But by
> having the unwanted source impedance high rather than at 50
> ohms it is much easier to pull the high impedance down with
> the "few ohms" short circuit than it would be if we were
> working at 50 ohms for the unwanted.
>
> It works like a voltage divider between the two impedances.
> The higher the source is (from previous cavity) to the short
> the more loss there will be which is just what we are looking for.
Your use of the voltage divider description is probably the best way to
explain the effect, and shows why having the correct intra-cavity cable
length is important for getting the maximum rejection.
Using exactly the "right" cable length between two cavity filters will give
somewhere around 5 or 6 dB of additional rejection (i.e. if the cavities
individually afforded 40 dB of isolation, when cascaded with the right cable
length you'll get about 85 dB total). If you use exactly the "wrong" cable
length (i.e. if you're off by a quarter-wave), the combined isolation will
be LESS than that of the cavities individually; a good indication that you
have the "wrong" cable length is when you can't get the notches to overlay
at the same frequency (often the notches will look like they're "chasing
each other" on the VNA as you adjust them). Between the "right" and "wrong"
cable lengths you'll end up with a net isolation somewhere between the two
extremes. Assuming the cavities individually present a good match at the
pass frequency, varying the interconnecting cable length between two
cascaded filters will not affect the insertion loss or return loss.
In contrast, the cable length from the tee to the first cavity on each side
of the duplexer primarily affects only the insertion loss through the
duplexer, and the return loss from antenna to/from either Tx or Rx port,
unlike the cable between adjacent cavities which affects the isolation
(rejection) afforded by that half of the duplexer.
A while back I posted a blurb that demonstrates the effects of using the
"wrong" cable length between cascaded filters. I don't know if that made it
to the web site?
--- Jeff WN3A
