Operating multiple bands in the 10 GHz and Up contest is difficult with separate antennas – after locating a station on 10 GHz and peaking the dish, we must start over on a higher band, usually with a narrower beamwidth. Using a dual-band feedhorn for 10 and 24 GHz would very attractive; the dish may first be pointed and peaked up on 10 GHz, then switched over to 24 GHz with no repositioning required.

At Microwave Update 2001, AD6FP and AA6IW described1 a dual-band 10 and 24 GHz feedhorn for shallow and offset dishes. The design was based on previous work of W5LUA2 and W5ZN3,4 to develop a dual-band feedhorn more suitable for conventional deep dishes. With the offset dish, we have a distinct advantage – the equipment may be located very near the feedhorn without being in the radiation pattern, minimizing the large feedline losses at the higher microwave frequencies without decreasing gain. Other advantages include higher efficiency feedhorns, less critical focusing, and the ready availability of modest-sized DSS dishes with good surface accuracy. Gary and Lars included computer simulated radiation pattern plots which look like potentially good feeds, but did not do dish efficiency calculations. However, they did include more important results – sun noise measurements and on the air performance! To calculate efficiencies, I took the published dimensions and resimulated. The results were good, so I wanted to make a feed.

The basis of the dual-band feedhorn design is the W2IMU dual-mode feedhorn5, dimensioned to feed an offset dish at 24 GHz and excited from the rear with a circular waveguide section. For 10 GHz, an excitation probe fed by an SMA connector is added on the side of the output section of the dual-mode horn. The tapered section of the dualmode horn acts as a closed end at 10 GHz, so that the output section behaves like a simple "coffee-can" feed at 10 GHz. Figure 1 is a sketch with the dimensions I used. While the 24 GHz dual-mode horn has a pattern (Figure 2) suitable for an offset dish, the simple 10 GHz horn has a much broader pattern (Figure 3), better suited to a deep dish, so it would have a lot of spillover feeding an offset dish. AD6FP improved the 10 GHz performance by adding a conical horn to narrow the beam, and AA6IW enhanced it further by using a corrugated horn. The dual-mode horn is intended to eliminate edge currents in the rim of the horn, so the addition of the conical horn outside the rim has a much smaller effect at 24 GHz. By varying the horn dimensions, it might be possible to make the patterns and efficiencies very close on the two bands. I had four different corrugated horns on hand, so I tried simulating with each of them. Results were promising, so I bored out the circular waveguide end of each horn on my lathe so that it could be slipped over the end of the dual-band horn.

Gary and Lars built their horns with copper plumbing and hobby brass, soft-soldered together. I tried this construction, but wasn’t happy with the dimensional accuracy, and it certainly didn’t feel robust enough for rover operation. Then I experimented with turning the tapered section out of solid brass, but found it difficult to get the taper right. Finally, I ran out of time before the 2004 contest and simply used a 25 dB horn on 24 GHz; at least it was easy to point.


Last fall, I was browsing through a tool catalog from MSC6 and found some 60º countersinks (normal is 82º or more). This would make a 30º flare angle for the tapered section, while Gary and Lars used my HDL_ANT program to calculate a 27.8º taper angle. The 60º countersink would be an easy way to machine a 30º taper, but is 30ºclose enough to 27.8º?"

I simulated the horn with 30º taper using Ansoft7 HFSS software. The results are shown in Figure 2 at 24 GHz: the rear lobe is a couple of dB worse than with the 27.8º taper, but the calculated efficiency of 76% is close, and the best f/D is about 0.7, just right for a DSS offset dish. At 10 GHz, calculated efficiency is still good, about 70%, but best f/D is about 0.38, as shown in Figure 3. At the 0.7 f/D needed for an offset dish, efficiency is down to about 47%, or nearly 2 dB worse. A few additional trials at 24 GHz suggested that a slightly longer output section might be a little better, if the countersink were long enough, but the improvement was not significant.

Another problem is that the nearest countersink size is ¾ inch, or 19.05 mm. I simulated with the inner diameter of the output section reduced from 20.4 mm. This change did not work well at all — the larger diameter is required.

The other mechanical problem is robust feedline attachment: WR-42 waveguide for 24 GHz and an SMA connector for 10 GHz. Using brass or copper for the feedhorn would allow soldering, but both are heavy and expensive. Turning the feed from aluminum rod was the best choice, but the size would have to be large enough for the WR-42 hole pattern: 7/8" square, or 1.25 inch diameter.

I found that 1" square aluminum was readily available in short lengths, so I ordered some, along with a ¾" 60º countersink. After a couple of hours with the lathe, my first attempt is shown in Figure 4. The machining was possible, but the ¾" countersink diameter is smaller than the 20.4 mm inner diameter that a small shoulder was left. I fiddled with a boring bar to minimize the shoulder so that I could at least measure the VSWR and make sure I was on the right track. Some improvement was necessary to make a proper taper. Matt, KB1VC, attempted to make a custom cutting tool, but the results were not encouraging – the countersink is clearly the right tool. I went back to the MSC catalog and found a 7/8" countersink, slightly oversized. The HSS tool steel is too hard to cut with ordinary tooling, so I used a toolpost grinder to grind the countersink to the exact 20.4 mm diameter. Now we are able to machine the correct taper cleanly.