w6nct wrote: > The other (not-so-intuitive) part to this is that it is often > difficult to do the tuning of a duplexer piece-meal; not impossible, > just difficult. What I mean by this is trying to tune one stage > (cavity), then another, then another; and combining it all together > into the overall duplexer system. The problem is that the > interconnecting pieces of coax become part of the tuned circuit. Once > combined together, one cavity's tuning can impact the adjacent > cavities. When I pre-cut the interconnecting cables to the specific > resonant lengths, I could get much closer to having the combined > system pretty close; but I've always had to adjust things just a > little after it was all connected together as a duplexer system.
Your observation is correct, and... your method started off accurate. Tune each cavity separately and when you hook them up together, if you're seeing "double peaks" and other odd things... think about what's wrong. The LENGTHS of the interconnect cables are incorrect. (Technically including the loop lengths inside the cavities.) Why? The duplexer's cables were factory-cut for a frequency too far away from your desired frequencies. At the point where you hook everything up after individually tuning cavities and things don't look right -- you need to adjust the lengths of the cables to make the duplexer "behave". NOT the tuning rods. Retuning the cavities is NOT the right way to fix it if the cavities when hooked together. (The key here is to remember that is is NOT possible for the frequency of a quarter wave stub to change. It doesn't. But it's possible that the cable lengths are wrong, thus the filters can't work properly together to "combine" and make a better filter for the frequency in question.) If the duplexer as a full-set isn't producing a proper pattern on the test gear after tuning each can individually with a proper dummy load on the other side... the cable lengths are not right to couple everything together properly. > I am also told that a service monitor or spectrum analyzer with a > "tracking generator" built-in is also a preferred method; but I have > never been able to afford either. They make it real easy to see the above effects and fix them properly. After having used them, I'd beg, borrow or steal to never have to tune a duplexer any other way, ever again. Nate WY0X
