At 4/12/2008 01:54, you wrote: >By experimenting we found cable lengths for the "tee" that gave the >right picture on the HP service monitor. They ended up being slightly >longer than the original cables. > >The individual halves of the duplexer (two cans, each side, one high- >pass, one low-pass) looked great individually, but the "tee" section >was no longer the right length (and thus 50 ohms) at the lower >frequency.
Actually, the idea is to get an "open" at the T on the reject frequency so the pass of the other side will go through the T without any impedance bump there. Since you say the T lengths are ~1/2 wavelength, the cavity reject will already "look" like an open so you're just carrying that open up to the T without transformation, hence a 1/2 wavelength multiple. >(Special thanks to Jeff DePolo for his comments when I was first >messing with that first 4076!!! That and some local Elmering got me >on the right track to use the low-pass and high-pass as "advertised" >on the labels, no matter what the original pair was, or if it was >"upside-down" from what I was using it for. That was a key piece of >information because I had tuned it ALL wrong at first, resulting in >one of the strangest looking patterns I've ever seen, trying to drag >the low/high-passes the "wrong" way!) The fun duplexers to tune are the ones that are simply marked "transmitter" & "receiver". You have to look at the original frequency label (& hope it wasn't removed!) to figure out whether the TX side now becomes the RX side for your application. If they do end up being reversed, I cover up the original labels with the correct designation. Connecting a duplexer backwards on site can be a bad thing :( Bob NO6B
