During the late part of winter 2012, I spent a great deal of time reading up on duplexer theory and construction. I really wanted to understand what I was doing, not just blindly follow a recipe to build something. The more I read, the more I believed it would be possible to build a properly functioning unit using parts from the former TV diplexer and sideband filter. If the telescoping resonator stub sections could tune low enough, or be modified to tune low enough, then we should be in good shape.
By this time, I had brought John (WB2DVE) on board to help with machining tasks needed for duplexer construction. His first assignment had three parts: 1) Coupling loops – create discs with a BNC and an N connector mounted on them; 2) Top plate – holes with shoulders to receive the coupling loop discs; 3) Inside partition – cut two pieces to length to make one complete partition. A few holes would need to be drilled in the outside panels to match the nuts on the partition; I decided to do this on-site with a handheld drill.
I also used some strap pieces removed from the diplexer to create coupling loops. The shape and dimensions of the coupling loops were copied directly from this web page, by Gary (NZ5V). John provided the BNC connectors, I recovered the N connectors from obsolete equipment in the “junk pile.” Two identical coupling loops were built, and two holes were cut in the top plate. John cut an extra, blank disc to fill one of the holes so that I could test both bandpass and bandpass-bandreject (BpBr) configurations.
Due to the size of the 6-cavity cabinet, all of the assembly and testing of the prototype had to take place at the transmitter site, which is a 20 minute drive from the office. As soon as I had all of the machined and cut pieces from John, I eagerly arranged a long lunch break to assemble them into the prototype cavity. John’s work was flawless, and everything fit together perfectly. You will recall from an earlier post that one of the resonator stubs was longer than the others; I made sure to use that one for the prototype, so as to provide the best chance for success at the lower frequency.
Testing the resonant frequency of the cavity is pretty straightforward: connect a signal generator to one coupling loop, a spectrum analyzer to the other, and see what comes out! The peak in cavity response is easy to spot. I kept pushing the tuning rod lower and lower, and reached resonance at 53.67 MHz (our transmit frequency) with less than an inch of the adjusting rod to spare! It was not possible to reach 52.67 MHz (the repeater’s receive frequency), but at least we now know how long a resonator stub needs to be for resonance at our repeater’s frequencies, and that it will indeed fit (barely) within the 51″ height of the cabinet.
Why does it fit, when the calculation of a 1/4 wavelength at our frequency comes out a little longer than 51″? Because the resonator stub is fat. This same principle applies to antenna design; for example, with a yagi (beam) or dipole, making the elements fatter decreases their resonant frequency, meaning that you end up making them shorter to compensate. It also makes them a little more broadbanded, which is usually great for antennas, but it reduces the “Q” of a resonant cavity. We could have achieved higher “Q” and better notch depth with a skinnier resonator stub, but at our frequency, it would have been too long to fit in the cabinet. There are always tradeoffs, it seems. Fortunately, this is one we can live with.
Resonance alone isn’t the whole story. How is the “Q”? What the heck is “Q” anyway? “Q” basically stands for Quality, and with regard to filters or other RF components, it refers to how sharp the response is. Think of how an antenna works. If it is broadband and covers a wide frequency range, it has a low Q. If it has a narrow frequency range, it has a higher Q. For duplexer cavities, we want a very high Q, because the distance between a cavity’s notch and passband is very small (only 1 MHz in our case). The high Q means that a single cavity can attenuate 52.67 MHz by 40 dB, yet 53.67 MHz escapes virtually unscathed.
I’m happy to say that the Q of the prototype cavity was pretty good for duplexer use. I didn’t think to make notes or take pictures of the spectrum analyzer while testing the prototype, so you’ll have to take my word for it. In the bandpass configuration, insertion loss (attenuation at the pass frequency) was around .4 or .5 dB with the coupling loops adjusted for maximum coupling, and the response rolled off quite sharply – something like 15 dB or more at 1 MHz out from the center frequency.
Reasonably high Q, and the resonator stubs will tune low enough while still fitting in the cabinet. Life is good! Time to go full steam ahead with construction, and that’s where we’ll pick up the story next time.