I'm posting this with all due respect to those who disagreed with an earlier post, and in the hopes of discovering any error I might be perpetuating.
A few weeks ago, a member of the group was asking for help with interference on the input of a 900-MHz ham repeater from a co-located FM broadcast station. Among the possible remedies discussed were coaxial stub filters on the receiver's transmission line. One of the initial proposals was an open, 1/4-wave stub tuned for the FM broadcast frequency, fed on a coaxial T-connector. This is, indeed, a common method to "trap" a particular frequency. I set forth that this wouldn't work, as the desired pass frequency was too near the 9th harmonic of the trap, which means it, too, would be attenuated. (These traps are VERY wide when fed on a T-connector, and work at all odd harmonics of the fundamental.) The open 1/4-wave coax trap, sometimes called a "suck-out trap," is best suited to a case in which the reject frequency is at least double the desired pass frequency, to avoid attenuation of the operating frequency itself. I proposed that better success might be achieved with a shorted, 1/2-wave stub tuned for the 900 MHz receive frequency, which would be nearly invisible at the 900 MHz pass frequency, but provide 20+ dB of attenuation at most frequencies below about 450 MHz. I did this based on experience not only using such shorted traps, but also after much past experimentation with my Wavetek sweep generator. Two subsequent posts took issue with my suggestion. One, from a member claiming engineering credentials, suggested my trap would appear as a "dead short" on the operating frequency, and that a shorted quarter-wave trap was the correct method. No supporting theory was offered. Another post suggested that a shorted 3/4-wave trap was correct, based on recollection of an instructor's comment. I've built and used several of these 1/2-wave traps, but it's been a few years, and I didn't want to dispute these comments until I'd gone back and made some actual measurements. I'd drop the matter, but this is too useful a technique to have it discredited unfairly. I still have the sweep generator, but not a scope, so I put my MFJ 259B analyzer, a 50-ohm dummy load, and a 41-inch piece of RG-8M (1/2-wave cut for 2m) on a T connector and look at SWR and impedance. Here are the resulting measurements of resistance, reactance, and SWR: 146.15 MHz (Shorted 1/2-wave): R=47, X=2, SWR=1.0 (Virtually unchanged from the dummy load alone) 73.08 MHz (Shorted 1/4-wave): R=23, X=20, SWR=2.4 (Z= about 31 ohms) 73.08 MHz (Open 1/4-wave): R=25, X=26, SWR=2.7 (Z= about 38 ohms) 146.15 MHz (Open 1/2-wave): R=3, X=1, SWR=12.0 (Z= about 3 ohms) 152.8 MHz (Open 1/2-wave): R=2, X=8, SWR=21.1 (Z= about 9 ohms. Note: This was the SWR peak, higher in frequency than the "shorted" frequency in part because the braid was folded back, instead of connected to the tip of the center conductor.) The readings at 146.15 MHz, coax shorted, were nearly identical with my 2m ground plane attached in place of the dummy load. Note that the only arrangement which looks like a "dead short" is the open 1/2-wave stub. The bandwidth of the shorted 1/2-wave trap on the dummy load was about 12 MHz for and SWR of 1.2 or less on 2m. Note that this trap would be plenty wide for use on the antenna side of a duplexer. (The corresponding 900 MHz version would theoretically be 45+ MHz wide given the same Q.) It also puts the feedline at DC ground, and serves as a crude high-pass filter below its fundamental frequency. It was explained to me by a cavity guru who first showed me this trick that the reflected energy in the shorted section returns to the T-connector at near-equal amplitude, and in phase, with the original signal. If the coax was lossless and the connectors perfect, the impedance of the stub at the pass frequency would be infinite, making it truly invisible in the system. In short, (no pun intended,) these measurements look just like what I've seen for years on my sweep gen. If you can demonstrate where I'm wrong here, based on actual data, please elaborate. If you're not sure, please cut one yourself and measure it. 73, Paul, AE4KR
