K0LEE.com Sitemap

Other Hobbies

I have added this page with Chuck Duey's permission to help those folks who would like to learn a little bit about how duplexers work and maybe save some money by building their own.   This article first appeared in the Nov-Dec 1997 Amsat Journal.  The duplexer discussed here is designed specifically for the Arrow dual band satellite antenna.

KI0AG Micro-Duplexer

Chuck Duey, KI0AG, ki0ag@amsat.org

A good duplexer can make the difference between hearing AO-27 full quieting and hearing a bunch of muffled static. The main purpose of a duplexer is to separate the transmit and receive signals. In the case of an HT working AO27 the duplexer separates the signals for the 70 centimeter downlink and the 2 meter uplink. Depending on the type of feed, a 70 centimeter antenna can look like and open or short at 2 meters. Similarly at 70 centimeters the 2 meter antenna will not have a 50 ohm impedance. The problem is to combine the 70 centimeter antenna and the 2 meter antenna so that they do not interfere with each other’s signal or loading. One solution to the problem is using a duplexer which decouples the receive antenna from the transmit antenna.

Duplexers come in many models that can handle over 100 watts. This power handling capability intended for repeaters is more than what is needed for an HT. A large duplexer would be very inconvenient working an AO27 pass while aiming the antenna by hand.

A duplexer for 2 meters and 70 centimeters is quite simple in theory, but attaining low loss can be a challenge. A simple duplexer would employ a low pass filter from the radio to the 2 meter antenna and a high pass filter from the radio to the 70 centimeter antenna. If the filters have low insertion loss and the board layout is very clean, Wam! Bam!, instant duplexer. Unfortunately, it isn't usually that easy.

To start down the right path, look up a good filter in an analog filter book. My favorite is the venerable Handbook of Filter Synthesis by Anatol Zverev. A good filter should have a steep cutoff and not too much ripple across the passband. Many of the Chebyshev filters fit this requirement. The only problem is most books give numbers for a filter scaled to ‘Omega = 1’ or about 1/6 Hz. The values are in Farads and Henrys, a little big to fit in the boom of your antenna. Also the impedance is at 1 Ohm. At 1 Ohm the VSWR would be a bit large. So, the filter must be scaled. Here is an example:

Chebyshev n =5 Ripple = 0.5 dB, Rs = 1 Ohm, L1 = 1.8068, C1 = 1.3025, L = 2.6914, C2 = 1.3025, L3 = 1.8068 (From reference 1)

To scale Inductance: L new = L old* Rs new/(2*p *Hz)

To scale Capacitance: C new = C old /(Rs new*2*p *Hz)

Where Hz = new cutoff frequency (220 MHz)

Rs new = new filter impedance (50 W )

p = 3.14159

After scaling, the filter becomes L1 = 65nH, C1 = 18.8pF, L2 = 97.3mH C2 = 18.8pF, L3 = 65mH

Due to the cost of custom components, it is best to round the values to the closest standard value.

This should be done within reason. Adjusting the cutoff frequency a bit to get closer values can be done with a spreadsheet. After looking at the standard values, the filter becomes:

L1 = 68nH, C1 = 18pF, L2 = 100nH, C2 = 18pF, and L3 = 68nH

If a SPICE program is available, a careful look at the calculated values is always a good idea. Moving the filter cutoff around to give the lowest losses is also helpful. By choosing the ripples in the Chebyshev filter to correspond with the 2 meter and 70 centimeter frequencies, the losses and rejections can be optimized.

That takes care of the low pass side of the duplexer but what about the high pass side? Second verse same as the first, just flip it over. This means take the reciprocal of all the values in the table and change the capacitor to inductors and vice versa.

I.E. C_highpass = 1/L_lowpass and L_highpass = 1/C_lowpass

The un-scaled filter becomes C1 = 0.5535, L1 = 0.7678, C2 = 0.3716, L2 = 0.7678, C3 = 0.5535

Now apply the same formulas as above at 50 W and 380 Mhz.

C1 = 4.6pF, L1 = 16nH, C2 = 3.1pF, L2 = 16nH, C3 = 4.6pF

With Standard values C1 = 4.7pF, L1 = 15nH, C2 = 2.7pF, L2 = 15nH, C3 = 4.7pF

After calculating component values the most difficult part starts – construction. At 437 MHz the best rules are to keep the signal route as short as possible and keep the grounds as low impedance as possible. The other thing to keep in mind is the physical size of the components. Thanks to the Telecom Industry, there are many small RF components that can help keep everything nice and small. Because the circuit small, it was designed to fit inside the boom of a Yagi. After a few trial runs and quite a few AO27 passes a PC board was designed (Figure 1).

For this design C1, C2 = 18pF C3, C5 = 4.7pF, C4 = 2.7pF L1,L3 = 68nH, L2 = 100nH and L4, L5 = 15nH.

The voltage rating of the capacitors and the current rating of the inductors determines the power the duplexer can handle. If the VSWR = 1.0, the currents and voltages are easier to calculate:

V= SQRT(P*R), and I = SQRT(P/R)

At 50 Ohms and 10 Watts: V = 22 Volts and I = 0.45 Amps. This means the inductors in the Low Pass section must be able to handle over 450mA and the capacitors in the High Pass section must be rated at 22 Volts or better. Most surface mount capacitors in the low pF range are 50 Volts, so they present no problem. However, there are many inductors that are 250 mA or less. For a good margin add ~50% to the current rating so look for 650 mA or better. With the power ratings in the Parts list becomes:

Parts List:

2 4.7pF 5% 50V NPO 1206 or 0805
1 2.7pF 5% 50V NPO 1206 or 0805
2 18pF 5% 50V NPO 1206 or 0805
2 15nH 5% 650 mA 1008
2 68nH 5% 650 mA 1008 or 1210
1 100nH 5% 650 mA 1008 or 1210


Soldering the components can be tricky. Use a low wattage iron and metallic tweezers. Do not force the components around, or else the end caps will break off. For the novice with surface mount parts it is best to apply some solder to one pad on the board. Take the component with the tweezers and place it in the correct location while heating the pad with solder. After the component is in place solder the other side. Take a look through a magnifying glass to check the work. The solder between the component lead and the board should look smooth and shiny. For inductors a quick continuity check helps spot some bad joints.

After all the surface mount components are on, it’s time to add the cables. Strip back the center conductor ½ inch, and the shield ¼ inch. Bend the center conductor 90 degrees down about ¼ inch out. Put the center conductor through the hole from the side without components and solder. Then using ¾ inch of wire, loop it over the shield and pull it tight and solder to the board and the shield. After the cables are on, check things out with a dummy load and SWR meter.

Once everything is assembled and tested, it is time to put on the finishing touches. First, check the fit of the duplexer. For the square boom of an Arrow antenna, make sure the components and cables to not bind up. Sometimes quick adjustment to the board with the file makes it a good fit. Be careful no to file the components. Some heat shrink tubing or electrical tape over the duplexer board and cables should prevent any shorts and protect from damage.

This duplexer has reduced the weight and the number of cables for my AO27 setup. I have used it to work many passes from a wide range of locations and conditions. I hope it works as well for all who try building this project as it has worked for me.

73 and Catch Ya’ on the Birds!

Chuck Duey, KI0AG


  1. Handbook of Filter Synthesis, Anatol I. Zverev, John Wiley & Sons, 1967

Digi-Key 1-800-DIGIKEY Delevan Inductors and Panasonic Caps
Far Circuits 1-847-426-2431 PC board for the Micro-Duplexer
Arrow Antenna 1-307-638-2369 Fully assembled and tested Micro-Duplexers and Dual band Yagis

--------  end of original article -----------

If you would like to purchase parts to build the duplexer, you can get the capacitors and inductors using the following Digikey part numbers:

Panasonic SMT Capacitors:
PCC2R7CCT-ND 2.7pF cap (2) $1.60/10
PCC4R7CCT-ND 4.7pF cap  (1) $1.60/10
PCC180CCT-ND 18pF cap (2)  $1.60/10
Delevan SMT Inductors:
DN08150CT-ND  15nH inductor (2)  $1.93 ea
DN08680CT-ND   68nH inductor (2) $1.93 ea
DN08101CT-ND  100nH inductor (1) $1.93 ea

I've recently become aware that since writing this page, some of the components have gone to RoHS compliant (i.e., no-lead) versions and thus the capacitors  mentioned above are obsolete.   You can find capacitor substitutes in the Digikey catalog with similar properties in RoHS versions, which will work fine with lead solder.  For example, a good substitute for the Panasonic capacitors would be as follows:

CC1206CRNP09BN2R7 Yageo 2.7pF cap (2 req'd) $.92/10 Digikey P/N 311-1216-1-ND
CC1206CRNP09BN4R7 Yageo 4.7pF cap (1 req'd) $.92/10 Digikey P/N 311-1218-1-ND
CC1206JRNP09BN180 Yageo 18 pF cap (2 req'd) $1.70/10 Digikey P/N 311-1152-1-ND

The board can be found at Far circuits (http://www.cl.ais.net/farcir/control1.htm) for a cost of $3.50.  However, you may want to order a few of them and split up the order with a friend, since the shipping is $5.50 for up to 4 boards.  You probably won't need to buy the article reprint (which is another $1.50) if you just print out this one.  Just be aware, that single $3.50 board will end up costing you $9.00.

You can see a pictorial of the microduplexer and how the coax is attached at the Arrow Antenna website.

The RG58A/U cable lengths are:  24 ¼" for the 2m coax, 14" for the 70cm coax, and the radio coax is 45".  A good source for this cable and BNC connectors is at http://www.cablexperts.com.

I tested my duplexer after I built it to check the performance.  Here is a Word Document file with the frequency response curves for the Microduplexer.

I also have a .dxf file of the board artwork drawn to scale.  If you are handy with a CAD program, you can download the .dxf file and fill in the copper and print out the artwork.  For some reason, the CAD programs I used created problems on the copper fill when I did the .dxf conversion.  I tried to fill it with two other CAD packages, but the result would not convert and restore the filled areas properly so I just decided to call it quits and figured anyone who could use the .dxf file would be able to fill the traces and print off the artwork themselves.  Shown below is a .gif image of the artwork with dimensions in inches.

Richard Holbert, KB8DNR, has made up a nice pdf file of the artwork which should print out full size.

If you don't already have a general purpose CAD program, I would recommend you give DeltaCad a try.  A fully functional version is free for 45 days and then it only costs $39.95.  It is an easy-to-use general purpose CAD program.  Here is a file that contains the filled copper pattern in the Deltacad .dc format.  It prints out exactly to scale when request 1:1 printing.

I would have used a PCB layout program for doing layouts called Circad but this small single-sided board was already laid out by Chuck and easier for me to copy using a general purpose CAD tool.  If you're a ham and would like to get a very powerful yet inexpensive CAD tool for schematic capture and PCB layout, you can get a substantial discount on Circad by visiting his website.  It normally sells for $995, but with your ham callsign, you qualify for a $600 discount!  Hat's off to Jay Craswell, W0VNE for making this happen.  Jay also hosts a YahooGroups discussion on Circad. Return to Lee Devlin's Ham Page