6m Duplexer Project
Initally, when NZ5V said he was going to "Build" a set of 6m duplexers, the reaction was, to say the least, mixed.
Duplexers are perhaps one of the most difficult undertakings in ham radio. Both a mechanical and RF problem, they require skills at opposite ends of the spectrum - machine shop to aluminum welding to RF filter design and debugging.
The duplexersthat were under development by NZ5V, Gary, and Charles, KD5TKR using NZ5V's machine shop and some scrap irrigation pipe, have been completed
Finished Spec's on the Homebrew Duplexers
Yes. Mine do. 92db reject notch, 1.29-1.78db insertion loss from four cans.
So, how is it done? Read on... Or if you want the nitty-gritty up front click here for a PDF drawing of the 6 meter cans..
One IMPORTANT NOTE: Start with the heliax stubs being about 24" and trim them down from there to your frequency. Take too much off? Squeeze the heliax with a pair of pliers and it will move the notch back the other way.
Using 8" irrigation Pipe
The prototype cavity is an 8" can some 57" tall (the 8" can portion), using a ~3" plunger. The ratio of the conductors given by the formula Z=138*log(d2/d1) provides about a 59 ohm impeadence, not the ideal 77 ohms for a cavity of the optimum "Q" factor, but we are really limited by the materials that are available at an affordable cost. The design length of the cavity was determined by using the following formula:
Length in Feet = 234/Freq. in Mhz.
The velocity factor of an 8" coaxial conductor is about 98.5%. This figure was arrived at by using various manufactuer's spec's from commercial high power transmitting installations.
So, the length for a 52mhz. at 98.5% would be about 53 inches.
Add about 8-10% for a "safe" adjustment range below the lower target frequency and you have 57-58", the dimension for the length of the 8" irrigation pipe.
We used a 3" plunger that was about 12" long, providing about 11" of adjustment. The duplexer just nicely covers the entire 6m FM band from 52-54mhz..
Please note that the frequency adjustment range will depend upon the ratio of the inside and outside diameters, and the resonance point of your coupling loop. They all interact together.
Chosing the TYPE of Duplexer
Reviewing available duplexer types, at a 1mhz split, the BP-BR style of cavity offered the best solution. Not only did it have low insertion loss at the pass frequency, but the design offered an additional 5.5db per quarter wave connecting stub of additional isolation. This additional isolation was especially attractive because we did not expect our home made duplexers to perform as well as commercially made units.
We hoped to produce BPBR single loop cavities using a pass notch capacitor that will be made in the machine shop - UPDATE - a piece of 1/4" superflex heliax works great instead, many thanks to the Repeater Builder Web Site and WN3A, Jeff DePolo for the great idea!
A total of four (4) cavities will be made, two on transmit and two on receive. NZ5V's prior experience with plain aluminum cans on 2 meters was reasonable, though hardly as good as a commercial plated set. NZ5V expects to get in the range of 34db per can (we got 37db per can!) without some type of conductive surface finish, such as Alodine. If the cans are Alodined later, as much as an 3-6db increase may be obtained per can.
The plunger assembly is a ~3" tube with a 3/4" aluminum rod that was threaded on one end. Although it has poor temperature characteristics, the commercial material of choice, INVAR would cost about $30 per foot for 1/4" rod, and with some four (4) feet required for each can, the total cost for INVAR rods would be over $500.
Volume vs. "Q"
When you choose a center plunger diameter, there are two factors to consider that appear important; one is volume - the greater the volume of the cavity, the better it should work. Second, the "Q", and most will agree that 77 ohms yeilds the highest "Q".
An examination of most commercial cavities construction and you find small diameter center plungers. This has the advantage of requiring more plunger travel for a given change in frequency than a larger diameter plunger, and also a greater volume. In practice, the trade off seems about "even steven". My unplated scrap aluminum 8" cans with a 3" center plunger had 37db of notch with 0.5 db insertion loss. A commercial 10" DB Products can with a 2" center plunger had 41db of notch and 0.4 db of insertion loss.
In reality, I do believe that you can make your duplexers from whatever materials you have available and they will work well.
The secret to making duplexers
Once you have mechannically sound, basic cans, I really do think that the secret to making duplexers is having good test equipment. A spectrum analyzer with a tracking generator or a network analyzer that is in calibration is an absolute must. A return loss bridge, directional coupler, MFJ antenna analyzer, or a vector voltmeter all are very useful as well. To give you an idea of what a difference test equipment makes, I changed one cable between two of the duplexers cans for one that was 6" too long. The insertion loss on that side increased by almost double.
Make your duplexer interconnecting cables with test equipment, not a tape measure!
Be sure to make the cables between BPBR or Notch duplexers an electrical quarter wavelength, and the cables from the transmitter and receiver to the duplexer must be a half wavelength. Use of a quarter-wave multiple cable from the receiver to the duplexer will make your repeater seem deaf.
The coupling loops are very, very important. Some general suggestions are:
- Use copper tubing that is flattened out instead of narrow 0.63" flat stock for loops on 2m and 6m.
- Choose a capacitor for your BPBR loops that has the highest possible "Q". A length of 1/4" Andrews Superflex or brass tubing and a dialectric a la Wacom are good alternatives to high quality, expensive, piston trimmer capacitors for lower frequencies. The higher the "Q" the lower your insertion loss and the greater the notch depth, all else being equal.
- Make sure your loops have a "dip" (a resonance point outside the can) just above and below the frequencies you are using. For 6m, 55-56mhz and 49-50mhz are a good starting point.
Making the Homebrew 6m Duplexer Cans
Picture 1. Above you can see one of the ~3" diameter center stubs for the duplexer being finished off square on the end using the steady rest on the Scharer lathe. This lathe just accomodates the nearly 60" length of the 8" irrigation pipe.
Picture 2. This is a closeup of the end of the center tuning stub being turned off clean and square.
Picture 3. This is the 8" outer can being turned off clean and square. As you can see, it barely fits in the lathe.
Picture 4. This is the coupling loop that makes the can a BpBr or Band Pass - Band Reject cavity. The small 1/4" superflex "heliax" was selected for this project as the tuning capacitor after reading on the repeater builder web site how high a "Q" the superflex gave the DB Products cavity, and how low an insertion loss it presented by WN3A, Jeff DePolo. It's another version of the legenday WACOM design. It really works quite well and is easy to implement mechannically. Please note, this loop didn't work as well as using 3/8" copper tubing that was flattened in a vice.
Picture 5. Another close-up view of the BPBR coupling loop.
Picture 5-2. The final loop assembly was made from some refrigeration copper tubing, crushed flat.
Picture 5a. Plunger top. This is the piece that attaches the plunger tube to the top plate of each duplexer cavity.
Picture 5b. Here are 3 of the 4 partially assembled center tube and duplexer can tops, and one of the 8" sections of irrigation pipe that have been cut to length.
Picture 6. Charles, KD5TKR threading one of the four (4) duplexer tuning rods. Each is a 3/4" aluminum rod about six (6) feet long. Steve, KG5QH did one of these too, but I didn't get a picture of you Steve. Sorry about that!
We used a fine thread standard 3/4"-16 die. No special ultra-fine thread was needed. The duplexers tuned very nicely.
Picture 7. KB0JOS, Wayne, milling (using a rotary table) one of the six (6) square 9"x9" 0.375" aluminum plates that were then put in the lathe and made into duplexer tops and bottoms.
Picture 8. NZ5V, Gary, at the Schaerer lathe making the plunger top piece. It's the aluminum part you see the threaded rod going into at the top of the long 3" tube below.
Picture 9 A view of the plunger tube attached to the duplexer top. Note the hole in the circular top - that will hold the coupling loop after assembly.
Picture 10. The duplexer top completed and attached to the plunger tube - except for the three 120 degree threaded holes that will hold the coupling loop.
Picture 11. This shot shows the duplexer bottom (notice the 3 holes used to hold it in the lathe), a partially completed plunger top (threaded, but needing 3 tapped holes to hold it to the duplexer top), and some just started punch-out looking parts that will become coupling loops.
Some Tested Performance Specifications
The prototype can has been tested by W5MUH, using a Cushman CE-50-A1/TG service monitor, and although somewhat dirty inside and not plated, it was measured at 1.2 db insertion loss, and 40db rejection. While somewhat short of .25db insertion and 45db rejection of a commercial cavity (as measured by Jeff DePolo), it has shown us a homemade set of cans is workable. While a Total of 2.4db insertion loss and only 80db notch would not be very good, it is enough we feel to go forward with the construction of the other three cans.
An update on the 6m duplexer cans - the center plunger and interior of the cans were cleaned throughly and re-tested using my newly aquired HP Vector Voltmeter, and now have an insertion loss of 0.6-0.7db and a rejection of 42db (note: the cans measure 37db, plus about 5db for the quarter-wave connecting cable). This won't provide adequate isolation for the tube type 330w PA unit that may be used at a later date - we think more like 97db notch will be needed. As a note, we think the insertion loss is overstated and rejection understated because our signal source is not as clean as it could be - we think the vector voltmeter is seeing "off-frequency" RF from the old Singer CSM-1 Service Monitor.
With each can providing 0.6db insertion loss, the total insertion loss is projected to be 1.4db, or 2x 0.6db plus 0.2db for the 1/4 wave connecting stubs. Isolation is projected to be 2x 37db, plus 15db for the three 1/4 wave connecting stubs for a total of 92db isolation - not quite enough to duplex with a 330w tube type power amplifer.
HP Network Analyzer
October 2004. As time passes, and the test equipment is improved, our measurements and work with the duplexers does too.
We now have the prototype 6m can operating with 42db isolation and about 0.60-0.45 db insertion loss (it varies from can to can). With three cans connected, we are measuring about 1.2db insertion loss total.
The network analyzer makes a service monitor with a tracking generator and spectrum analyzer look like a signal generator and a radio when it comes to working with duplexers.
Now with two cans and a 1/4 wave coupling, we have a trace that is off the screen at 10db per division vertically. That is over 80db isolation. It measures out to about 92-93db for two cans. Some tweaking of the 1/4wave connecting coax may improve that figure. Overall, we are very happy with the results of our home-made duplexers. By adding a third can to the receiver side we should duplex the 330w Micor power amp without a problem.