The cavity filter will tune from approximately 900 MHz to 1050 MHz and has a 3 dB bandwidth of about 2 MHz. Its purpose is to attenuate the fundamental (1024 MHz) and image frequency (1034.7 MHz) of local oscillator 2. For that purpose it will be tuned to 1013.3 MHz once it's ready. The rejection at 1034.7 MHz will hopefully be greater than -100 dB.
Mechanical drawing taken from http://scottyspectrumanalyzer.com/protocavfilt.gif
I have seen something similar in the RF laboratory of Motorola's product support engineering department back when I was an RF-technician apprentice. This cavity filter doesn't use the traditional rectangular shields with aperture coupling though, and it is somewhat difficult to find good information on the construction of such filters.
It turned out to be quite difficult to acquire copper tube here. Bauhaus was selling 3 m (10 feet) pieces of 28 mm (1 inch) diameter tube for 50 EUR, which I found rather unreasonable. After about two weeks of research I found a source in southern Germany that was able to provide me with the copper parts I needed for not even half of what Bauhaus was asking.
I have two options when it comes to the centre resonator. I either use solid copper pins and thread them in order to be able to tune the cavity, or to use 6 mm (1/4 inch) hollow copper tube and implement brass tuning screws as seen in the mechanical drawing. I don't have any tapper here so I decided to go with tubes for the prototype.
I used a hand saw to cut the four cavity tubes. This was really a pain in the neck and I recommend anyone who tries to build this kind of filter to buy a pipe cutter. Otherwise, you will spend a considerable amount of time trying to file the tube ends so that they may be reasonably parallel once placed on the bottom plate.
Everything is polished with steel wool. The copper pins in front of the cavity bodies are actually pieces of RG-141 semi-rigid coax.
Drilling holes for the resonator pins was the easy part. I gave up trying to drill 1 mm holes for the hairpin couplers with a hand drill. I may point out that I use 5 mm copper flat bar since I was undecided whether to use threaded solid pins for the resonators or hollow pipes. I guess a 3 mm sheet would have done the trick too, with less hassle. If you have access to a drill press I absolutely recommend to use it.