Cavities, Diplexers, Duplexers, Theory, and More...   Compiled by Mike Morris WA6ILQ   (callsign) /at/ repeater-builder /dot/ com

Uninformed people sometimes get the two confused and call a diplexer a duplexer, and vice versa.

A cavity (and since duplexers are made from multiple cavities, therefore a duplexer) works because of simple physics based on frequency, wavelength, and the length of a 1/4 wave in coax and/or in a cavity. A cavity, depending on how it is made and used can be a peak filter or a notch filter. Different cavity designs have different peak/notch depth characteristics, and different insertion loss characteristics, and there is usually a very visible tradeoff between peak height/notch depth and insertion loss.

A Diplexer is a block filtering unit, usually L-C based, that allows multiple transmitters, or receivers on different bands to use the same feedline, antenna, or both. They come in 3 port, 4 port and 5 port designs. Diplexers are available in many different combinations of frequency bands. Three generic examples: a 2m radio and a 440 MHz radio both feed a single dual band antenna (a 3 port unit), or a 6m radio, a 2m radio and a 440 MHz radio feeding a triband antenna (a 4 port unit), or an HF radio and a 2m radio feeding a coaxial feedline, and on the other end another unit breaks it apart to separate HF and 2m antennas (two identical 3 port units). A specific example: an Icom 706Mk2G has two antenna connections, one is HF and 6m, the other is 2m and UHF. If you had a separate HF antenna, a 6m antenna, and a 2m/UHF dual band antenna you would use a diplexer to split the HF and 6m to two separate antennas and the dual band antenna would be connected directly. If you had separate 2m and UHF antennas you would need a second diplexer to split the single radio connector to the two antennas.

There are commercial grade diplexers (TX/RX is one manufacturer), and there are ham grade diplexers.

Two personal opinions from the author - WA6ILQ:
1) I won't use a ham-grade diplexer at mountain top radio site. The price difference isn't worth ruining my replationship with the site owner.
2) I do not like the diplexers with pigtail cables. Every one that I have owned and several that I have looked at or worked on for other hams have had problems with cable issues... one reason is that the manufacturers use the cheapest coaxial cable they can find - which means off-brand junk with thin jackets, thin conductors, and hence a fragile product. Then there were the intermittents, shorts and opens (I found the center conductor of one pigtail wasn't even soldered to the circuit board during manufacture... there wasn't any solder on the cable center conductor), or the cable pulls out of the metal body, or the connector falls off the radio end of the cable. On the other hand I have never had a problem with the ones that have real coaxial connectors mounted in the diplexer metal housing. Note that every manufacturer of ham-grade units uses nickel or chome plated connectors... and both nickel and chrome are known intermod generators in an RF environment.

On to Duplexers...

The normal purpose of a duplexer is to allow the simultaneous operation of a receiver and a transmitter in the same frequency band on the same antenna. Duplexers are made from multiple cavities arranged so that the peak (or notch) from each one is added to the previous one and the next one by means of critical length cables.

Unfortunately, saying "duplexer" is like saying "car". They come in various types, makes, models, and performance levels. Some are notch-only, others are notch-pass. There are 3-cavity, 4-cavity, 5-cavity, and there are 6-cavity models (more cavities give a higher level of receiver-to-transmitter isolation, but at a cost of increased insertion loss) and I have seen both a 7-cavity and an 8-cavity unit. There are also special purpose assemblies based on duplexers. In general, the higher the RF level at a site (i.e. how many other systems there are, how close they are (both physically and in frequency), and the power levels involved), the more cavities you need. I've also seen a 8-cavity 4-port duplexer (it was a custom order) that allowed a 144.39 MHz simplex APRS gateway radio to share one antenna with a 146 MHz range repeater. And you may not need a duplexer at all if you run a split site system at a low RF level site. Depending on circumstances even a single site system can be built with no duplexer at all (see the "Some thoughts on Repeater Receiver-to-Transmitter Isolation" article elsewhere on the Antenna page). And to continue the used car analogy above, used duplexers can have problems just like used cars. Unless you trust the used duplexer seller you will want to have your radio mechanic test it out before you buy it. As just one example - many duplexers have a variable capacitor as part of the tuned circuit (the "loop") on each port. Usually these are a precision glass vacuum variable capacitor. A new (replacement) one of those capacitors can cost you over US$100... and then you get to install it. And the effort involved in replacing one can be anything from simple to outrageous depending on how the cavity is built.

Duplexers expect to see a 50 ohm load on every port, and many transmitters, many receivers and some antennas are not. A duplexer is 50 ohms only at resonance, anywhere else it is reactive. The attached feedline is 50 ohms, and the receiver or transmitter is SUPPOSED to be 50 ohms. If it is not, the most common way to "fix" the problem is to use a "magic" length cable - but the right way is to fix the radio so it is 50 ohms.

The proper way to tune a cavity or a duplexer is to use 6db or 10db 50 ohm pads between the test equipment and the cavity or endmost duplexer can, as that ensures that pad "masks" any non-50-ohm-impedance in your test equipment and ensures that the duplexer is really tuned to precisely 50 ohms.

Many folks that take a duplexer that is tuned on the bench with 6db pads on each port (to guarantee a 50 ohm environment) and then install it in a system discover that the duplexer is "mistuned" - that they have to make adjustments to get the same performance that they had on the bench. Well, that is telling you that your radios (or your antenna system) is/are NOT 50 Ohms. The transmitter can be fixed with a "Z-Matcher". Then retune the duplexer. Once the duplexer is tuned properly it should NOT need ANY adjustments from the bench to the site (exept for the bumpy 12-mile-long four wheel drive road - and that's why you take the pads and the service monitor with you to verify the tuning on site). And you drive that 4wd road "gently".

A posting on repeater-builder a while back made the comment that the GE MASTR II UHF transmitter output impedance was mostly 50 ohms resistive, but with a capacitive or inductive component whose value depended on what company made the output transistors (and this characteristic is not limited to GE equipment). In the early days of the Mastr II large fleet operators were having an unusually high number of station and repeater PA failures. One user that I am aware of had eight repeaters and was losing roughly one PA per month. After GE had several engineers investigate the problem they came up with a two-pronged fix. Solid state power amplifiers do not like reactive loads, and the engineers found that the best solution was a pi-network output matching stage followed by a circulator to protect the output device(s) from an unmatched load. GE had Decibel Products develop a "Z Matcher" device which functioned as the output tank circuit that the transmitter designer had left out. The second prong of the field fix was that GE ended up buying a circulator for each transmitter and this solved the rest of the problem (this situation is why the Motorola Micor line has a factory-installed circulator in every PA deck - even in the mobiles!). As mentioned above, every circulator must be followed by a low pass filter (some installations use a pass cavity, which itself might be separate or part of a pass-notch duplexer).

Unmatched loads did not start with solid state PA decks. Feedlines over the years have included 35 ohms, 50 ohms, 75 ohms, 93 ohms, 300 ohms and 450 ohms. As mentioned, one way to transform impedances is to use a "magic" length of the right impedance of coax as a matching transformer. The cable length between the end cavity and the transmitter is NOT supposed to be critical, and if it is then you have an impedance problem. While the most common situation is between the transmitter power amplifier and the duplexer, the same situation can happen between the receiving antenna and the receiver, or between the duplexer and the receiver... it is just much less visible. Matching the transmitter to the duplexer is covered in this article Cabling lengths between the Duplexer and Radio Set. Despite the article title, there is a lot of impedance matcher (usually called Z-matcher) info in there.

Tony King W4ZT has some information on the Z-Matcher that GE built into the later Mastr II station power amplifiers at the W4ZT Z-Matcher page. The closeup photos show how GE simply replaced the station antenna relay with the Z-matcher daughter board (prior to the Z-Matcher it had been replaced with a jumper).

The schematic to the GE Z-Matcher is here. It's part of LBI-30201G.The tuneup info for the DB Products Z-matchers can be found here.

Back to duplexers...

As said above a duplexer is made up of a number of cavities and a number of critical length cables beteeen them. Each cavity adds to the total transmit-to-recieve isolation provided by the assembly. Note that duplexers are initially tuned and the ports labeled at the factory. Most are labeled "REC", "ANT" and TRANS", or something similar (see this photo). A duplexer is built with a "higher" side and a "lower" side, and the coupling loops and inter-cavity cabling are selected for those frequencies. The application an amateur has may conflict with the original port labels. Most commercial repeaters (definitely on UHF, usually on VHF) receive on the higher frequency and transmit on the lower frequency, whereas amateur repeaters can swing either way. You need to look at the original label and note if the receive side is the higher of the two frequencies, or if the transmit side is... and keep the high/low relationship the same as you retune it. You may end up using the original receive side as your transmit side. Some manufacturers, including TX/RX call the sides "channels", and their documentation refers to the high channel and low channel and "isolation between channels". Personally, I remove the factory "Receiver" and "Transmitter" labels and replace them with my own permanent labels reading "High Side Pass/Low Side Reject" and "Low Side Pass/High Side Reject", then I add temporary "Receiver" and "Transmitter" labels for each application. This is expecially true on UHF GMRS (+ offset) and on UHF amateur (- offset in Southern California, Denver, and many other areas). Some duplexer manufacturers have adopted better connector markings that read "Pass Low, Reject High" on one side and "Pass High, Reject Low" on the other.

Interestingly, if you order a untuned duplexer it arrives with labels that are almost appropriate (it doesn't say "pass" or "reject")... see this photo.

The cable length between the cavities in a duplexer is very critical, to the eighth of an inch (about 3 mm) at UHF, and the actual length of any one cable is dependent on three factors: (1) the cable velocity factor, (2) the length of the coupling loops, and (3) the frequency involved. On a 600 KHz offset 2 meter duplexer the frequencies are so close together that the receive side and transmit side lengths are usually the same, however on a VHF commercial 5 MHz (or more) offset or on a UHF duplexer the optimum length is different between the receive and transmit side. The lengths are also different depending on the length of the loops inside the cavity, and if you are at 425 MHz, 435 MHz, 445 MHz, 455 MHz, 465 MHz, 475 MHz, or 485 MHz. I once measured a difference of 7/8 of an inch between the two sides of a UHF duplexer on a 5 MHz offset with one type of cable and 1-1/8 inches with a different type.

From an email to repeater-builder:

While it is fairly straighforward to calculate the wavelength in a certain piece of coaxial cable, the length of an intercavity jumper cable on a duplexer is only a portion of the whole length that must be considered. The coupling loop inside the cavity has length that must be added - which depends on the frequency and on UHF may amount to 5 to 6 inches, and also varies among the many manufacturers of duplexers. The cable lengths shown in the chart for any particular duplexer have been corrected to include the lengths of the connectors plus the coupling loops or probes inside the cans, and as a result the cable lengths by themselves will not directly correlate to a portion of a wavelength.

From another email to repeater-builder: (EMR, REMEC, TX-RX, Sinclair and Cellwave are all manufacturers of cavities and duplexers):

A technician at EMR once told me that he had racks of pre-made jumper cables in 1/4" increments, and although he had tables giving him the typical starting points, he would try jumpers above and below that length to achieve the optimum match between each pair of cavities. I have heard pretty much the same story from the factory-floor techs at REMEC, TX-RX, Sinclair, and Celwave.

Likewise the lengths of the cables between the antenna "T" fitting and the end cavities (some folks would call them the "middle" cavities) are frequency dependent and are a critical length, and usually a different length than the intercavity cables. The "T" fitting in the center of the duplexer is critical, and can take any one of several forms. Sometimes it's a special made-for-the-job device that combines a female connector and a "T" fitting (making up a "T" cable), sometimes it is made up of connectorized cables, a simple coax "T" fitting and maybe a barrel connector. Sometimes you find it miscabled, as shown in the article here. Look at the fifth photo, and the last photo.

Some folks run a "split" duplexer without the "T" fitting and cables (or "T" cable), with one antenna connected to the transmit side of the duplexer, and a second antenna connected to the receive side (or the receive antenna may be a one port from the multiport splitter on a community receive antenna). If you find a "split" duplexer and want to go back to the normal configuration then don't be surprised if you have to acquire a replacement center "T" cable. More than once I've seen the situation where it was obvious that the installer simply removed the "T" cable, and ran receiver and transmitter cables right to the connectors on the end cavities. Other times he removed it and tie-wrapped it to the intercavity cables. More than once I've had to acquire or manufacture a T-cable for a duplexer that arrived with it missing. Please folks, if you aren't using the T-cable just tie-wrap it to the duplexer harness !

Note that many duplexer assemblies have cavity adjustments that are wider in frequency than their cabling harnesses - for example if you move a VHF duplexer from 160 MHz or 170 MHz to 2 meters you will find that the cavities will generally go there (as long as the adjustment rods have not been cut off) but you will usually have to aquire a new harness, or lengthen the old one. One trick to see if lengthening the harness is the correct option is to add an elbow adapter (silver plated!) to one or both ends of each cable and test. Depending on the internal construction of the particular elbows just adding one can add 1/2 to 1 inch (13 to 25.4 mm). Hams that are moving a 482 MHz or 470 MHz duplexer to 440 MHz will have the same type of harness cable length problem.

Double-shielded silver-plated coax is the best thing to use to connect a cavitiy to an adjacent cavity in a duplexer, and to connect a radio to a cavity or to a duplexer. For example, RG-214 has two braids made up of individually silver-plated strands and a silver-plated inner conductor for maximum noise rejection. Avoid any coax that has disimilar metals rubbing against each other such as LMR-(any 3-digit or 4-digit number) or Belden 9913, both of which use an aluminum foil shield rubbing against a raw copper braid. Many amateurs have purchased LMR-400 or similar cable, installed it, and it works fine (for a while). They just rave about how good it is. Then anything from 9 months to 5 years later they find themselves taking it down as the cable itself causes "duplex noise" (see the article on this page titled "Help!! I have a crackling noise in my repeater" by Kevin Custer W3KKC). The expensive LMR cable gets reused somewhere else (in any simplex situation). RG-142 and RG-400 are both a smaller version of RG-214 (but sharing the same construction) and can also be used (but see the article on this page titled "Double Shielded Coax Cable, the differences between RG-142 and RG-400, and why you DON'T want to use RG-142..." by Eric Lemmon WB6FLY)"). My personal preference is to use RG-214 as the top-of-the-tower antenna-to-Heliax jumper, nothing but Heliax and Superflex on the antenna side of the duplexer and nothing but Superflex and RG-400 on the radio side. Yes, it's expensive, but like my late father used to say about hand and power tools, "Buying quality only hurts once". Good quality cable with the right connectors is going to be low loss and last a long, long, long time, and it's one less thing to worry about. If you have to use coax in a duplex situation then look at RG-214 or RG-393 for the large diameter (i.e RG-8/213/214 size) and RG-142 or RG-400 for the small diameter (RG-58 size). All are double shielded with silver plated shields and will not cause duplex noise.

The type of cable that is used between the duplexer and the receiver and between the duplexer and the transmitter is important. Many manufacturers scrimp and use cheap single-shielded cable, and in many cases get away with it.   The Motorola GR series repeaters and many of the Yaesu/Vertex VXR series repeaters were (still are?) supplied with single-shield jumpers inside the cabinet, and this allows random low level desense to occur. Replacing the factory jumpers with RG-400/U, RG-214 or RG-393/U double-shielded coax eliminates that. And when you do make cables, make them with the correct silver-plated connectors on each end, so that you do not have to use any adapters. Use a right-angle connector where you need it rather than a normal connector and a right angle adapter. All it takes is one chrome or nickel plated connector or adapter to ruin your day.

Pass-notch cavities (and the duplexers made from them) have a quirk that can cause you some unexpected problems. Picture this situation: you have a transmitter connected to a carefully tuned cavity with the pass adjustment peaked on the transmit frequency and the notch is centered on the receive frequency. Then a new somebody on the tech committe gets the smart idea to "touch up" the duplexer...

He thinks that it was out of tune because he's now seeing a few watts more going to the antenna - but the system performance drops. And he doesn't know why. Or maybe he doesn't see a difference, so he thinks that his tweak didn't do anything or maybe the previous guy didn't do a very good job, or maybe it really needed tweaking. What he doesn't realize is that before he tweaked he used to have have 12 to 18 dB of headroom but now he has maybe 3 or 4 dB (but he doesn't know that because there was no visible or audible change). But when winter comes and the ice builds up on the antenna things change, and the system now has a case of desense. But he doesn't realize that turning the knob 4 or even 6 months ago and having desense today are cause and effect... (and it takes snowmobiles, a SnoCat, snowshoes or even a helicopter to get to the site to fix it... and you never go to a site alone, so it inconviences two or more people. And the last time I priced a helicopter for a hill trip it was over $250 per hour, flying or sitting...)

What happens is that as soon as the person turns any one of the big knobs on the cavity the peak moves and the notch in that cavity also moves... since the notch tracks the peak... and the total rejection notch depth of the three cavities on that side of the duplexer is now shallower. If the total notch depth is not not deep enough, the receiver is desensed.

Once a duplexer is properly set to your frequency there is NEVER any reason to adjust it!

I've seen duplexers that haven't been touched since installation (and some were in the late 1960s and early 70s). A lot of folks that I know actually tie-wrap a note to the tuning rods - and the note says (in very strong, coarse and blunt language) not to touch the tuning for any reason, and if necesary to contact (someone's name - usually the head of the tech committe, or the system owner). Some folks remove the knobs in addition to posting the note. There's a reason Sinclair offered a steel cabinet (with either bolt-on front and rear panels or optional locking doors) for their high band duplexers.

Most "mobile duplexers" are reject-only and are sometimes called flatpacks, notch-only, or notchplexers. Some are 3 transmit and 3 receive (6 cavity), some are 2+2, and I've seen a 3+2 and a 3+1).

A note on small mobile duplexers from Eric Lemmon WB6FLY:

Most UHF mobile notch-only "flatpack" duplexers will work fine below 50 watts at a 5 MHz split- but their performance rapidly deteriorates when operated outside of their stated band and/power limits. For example, a commercial-band mobile duplexer that was manufactured to operate in the 450-470 MHz band will likely perform poorly in the 440-450 MHz Amateur band. Although such a duplexer may SEEM to be working in the Ham 70 cm band, it may display shallower notches or have excessive insertion loss that the owner may not realize. That's because the internal coupling loops are set at the factory during manufacture and not readily adjustable by the owner.

I once tried to use a commercial-band mobile ("flat pack") duplexer on 70 cm in a portable repeater, and was disappointed in its performance- especially in the receive sensitivity at 441 MHz (editors note: Southern California uses low-in, out-high on the UHF repeaters). I then ordered a new Celwave duplexer of the same model, but factory-tuned for my Amateur frequency pair. What a difference! Once the new duplexer was installed, the receive range of the portable repeater was significantly improved, with no other changes.

Above all remember that reject-only / notch-only duplexers are not appropriate for high level RF sites as they provide zero protection for the receiver from any other transmitter - they are designed to protect a single repeater from itself, and nothing else (and remember the "higher side" and "lower side" rule above). If you are going to have a quality system at a busy site you generally need to use a pass-notch duplexer with at least six cavities.

If you're just starting out, I strongly suggest that you read the first nine articles in this section on the Antenna page, especially those by Jacques Audet VE2AZX, John Portune W6NBC and by William F. Lieske. Yes, there is some repeated material between the articles, but if you're new to duplexers, it won't hurt. When you have multiple people writing on the same topic (any topic) some repetition is inevitable. The 9 articles collectively will give you a basic and solid education on duplexers.

This web page split from the main Antenna Systems page 12-Nov-2011.