On the Repeater Builder Yahoo group, I ask: Has anybody built or tuned a duplexer with these coupling loops?


 An email response from Jeff WN3A:
Yes, many. Duplexers, combiners, custom filters of all kinds, you name it, using both with Kevin's loops as well as "factory" Varinotch loops.
  > My set of two in 6" cavities for pass at 146.9MHz and notch
  > at 146.3MHz work, a treat.
  > But the 2nd set of two, pass at 146.3MHz and notch at
  > 146.9MHz, has a double pass peak but is 10db down.
  >
  > A sign of unwanted coupling.
Not necessarily.
First, how did you tune the cavities? They should be tuned individually, with the coupling adjusted to get a nominal insertion loss of 0.7 dB per cavity at the desired pass/reject spacing. The pass must be tuned for best match, *not* least insertion loss. Iteratively adjust the pass frequency, notch frequency, and coupling to get the target performance values of 0.7 dB insertion loss, approximately 40 dB notch depth (at 600 kHz spacing), and the return loss should be > 20 dB. Granted, these values are based on real TX-RX cavities; if you're using something else I can't tell you what kind of performance to expect, but if you have the coupling right, and the cavity itself is good, they should not vary appreciably from the values I've given.
Once each cavity is tuned individually, THEN connect them in cascade with the correct-length cable.
  > My two cavities are connected by a 1/4 wave line of Times
  > Microwave RG-223,
Here we go again with the black magic quarter wave cable length issue.
  > I followed tx-rx-instruction-manual-vari-notch-duplexers-with-6-inch-cavities.pdf From Repeater Builder, where did I go wrong?
The correct and optimized cable length for a pair of TX-RX Varinotch VHF cavities using these loops at 147 MHz is 11-15/16" (303.2 mm) measured tip-to-tip of the type N connectors using PTFE-dielectric cable such as RG142B or RG400 with a 69.5% velocity factor.
If you do the math in reverse, you'll find that this length is appreciably shorter than a quarter wavelength; it's more like 77 degrees BEFORE mating the connectors to the loop assemblies, and around 72 degrees after subtracting the mating interface distances.
With the correct cable length given above, you should see a 5 to 6 dB "bonus" in notch depth with two cavities in cascade. That is, instead of 40 + 40 = 80 dB total notch depth, you should see 40 + 40 + 5 = 85 dB notch depth.

--- Jeff WN3A