Measurements on two well known 3cm waveguide filters

About six months ago, prior to building up another 3cm transverter, it seemed a sensible idea to spend a little time building up some band pass filters. The results were eventually graphed out and filed away, but they often get refered to, and it may be that they are of interest to others too.

Some unused WR-90 waveguide was donated by G4BAO and a Google search quickly found the 2 pole design that Zack Lau  (W1VT) published in QEX (July 1997) – but it’s also in Vol 2 of the ARRLs UHF/Microwave Projects Manual.
Using hand-tools and a pillar drill, three units were made up to see what reproduceability could be expected. All soldering (endplates, posts and sma connectors) was done with a hot-air gun. The three filters were pleasingly similar, and one is shown plotted on the accompanying graph. Used as a receive image filter for a 288 MHz, IF, this gives about 50dB rejection which was judged to be good enough. However, for transmit use, the LO rejection is less than 40dB and although two units could and have been used in tandem, it was a good excuse to try the more complicated five section post filter that Chip Angle (N6CA) designed.
(see http://www.ham-radio.com/n6ca/microwave/filters/10368-28bpf.html )
For this, the posts had to be turned on a lathe to get the correct diameter, but the results are quite impressive and more than adequate for the 288 MHz IF transverter. Insertion loss at 3cm was 3db – a little high compared to N6CA’s original, but good enough for the purpose in hand so no adjustments were made to improve this.

Added to the graph is the response of a typical, lightly loaded pipe-cap filter, as a comparison. These seem to appear in articles everywhere (see the ARRL UHF/Microwave Projects Manual mentioned earlier, for example). Although this looks to have good out of band rejection, the response of all the pipecap filters that I have seem eventually to come back up to only  20-30dB, so more than one is really needed for decent performance. They are very easy to make though and seem quite forgiving in their construction. The particular one graphed used a 15mm cap with a M2.5 brass tuning screw/resonator.

Non of the filters shown were adjusted for match or coupling, and the two pole item shown clearly suffers from overcoupling.
These particular filters are also useful for high multiplication order LO’s, and a typical LNB shottky mixer (back to back pair in an SOT 23 package) driven with 100mW  at 280 MHz will give useful output at 10 080 MHz (ie x 21) with spurious harmonics all at least 40 dB down.

Written as a filler for 'Scatterpoint', mid 2006