i

THE CAVITY DUPLEXER

John E Portune W6NBC

[email protected]

rev 2019

NOTE FROM AUTHOR

This book was written several years ago and based on hardware-store copper
water pipe as the source of home-brew duplexer construction materials. Later I
began making from spun-aluminum commercial cake pans. Both require no
welding. Unfortunately the price of copper is today much higher.

Cake pans, however, are still reasonably priced, readily available and very
acceptable as the basis especially for VHF cavities. Because of maximum cavity
size limit, copper water pipe may still be indicated for UHF and above.

In any case, how a duplexer operates is basic physics. No matter what the
material, or whether the duplexer is commercial or home brew, the principles
herein are universal to duplexer construction, modification and tuning.

This book, however, is not finished. Repeater building is no longer my primary

interest in ham radio. Some subjects could be added. But as it contains the
essentials, I have placed on the internet incomplete. If you reproduce it, be so
kind as to give proper author’s credits.

W6NBC January 2019

.
 ii

CHAPTER OUTLINE
1. The Mysterious Duplexer
• The black box everybody uses but nobody understands
• Keys to understanding it
• This is not a cookbook
2. Let’s Make a Cavity
• Home-brew 2M aluminum cavity
• Example for the entire book
• The best way to learn
3. Cavities
• Mechanical and electrical properties of cavities
• Basic structure of a duplexer
• Why use cavities
• Getting energy in and out: loops, probes, taps and ports
• Three cavity types: Bp, Br, Bp/Br
• Creating the other types
• Helical resonators for 6M and 10M duplexers
4. Temperature Drift
• Commercial method – Invar rod
• Simple, elegant home-brew method
5. Performance
• Isolation
• Insertion loss
• Measuring hilltop noise
• The importance of hilltop noise
• Receiver sensitivity and selectivity, how to measure
• Transmitter purity, how to measure
• Pitfalls of preamps and power amplifiers
6. Tuning a Duplexer
• The simple equipment
• The basic process
7. Loops
• Position
• Placement
• Materials
• How critical?
8. Losses
• Skin effect
• Cavity size limitss
• Bandwidth vs. insertion loss
9. Lines
• Lines between cavities
• Rescaling a commercial duplexer’s lines
 1-1

Chapter 1 – The Mysterious Duplexer

The cavity duplexer is a very familiar part of both amateur and commercial two-
way radio. On a typical two-way hilltop, cavity duplexers are as common as the
forest of antennas that bristle from the towers. Most repeaters have one. They’re
well known to repeater owners and builders. Or are they? Don’t be so sure.

If you are like many hams and two-way radio professionals, you likely have more
questions about duplexers then answers. More than the other parts of a modern
repeater, an aura of “black magic” and many “old-wives tales” shroud the cavity
duplexer. It might be better to call it the familiar "black box" that everybody uses,
but nobody really understands.

What are the keys concepts? Why are there different type? What is the right way
to tune one? How can you minimize duplexer losses? And for the aspiring ham
repeater builder, “How do I modify a commercial duplexer for an amateur band,
and can I possibly build one for myself using hardware-store materials and
ordinary home workshop techniques?”

These same questions were running through my head when I first decided to
enter the world of repeater ownership. By then I'd already successfully installed
and maintained several duplexers. I’d even retuned commercial units to the ham
bands. But did I really understood them? Did this kind of ham experience really
qualify me as a knowledgeable duplexer user? Definitely not, and frankly I knew
it didn’t.

Like so many hams, especially those who want to put up their first repeater, my
experience had only given me a piece of the picture. It was also the dangerous
part – bits and pieces, acquired from other ill-informed hams and several old
wives tales. Cavity duplexer duplexers were truthfully then still a very big mystery
to me, as they currenly still are for many.

Purpose of This Book

My objective is to cut through the common “clouds of fog” that surround the
cavity duplexer, and to do so as simply and as non-technically as possible.
Duplexers aren’t "black magic" You really need only a handful of basic principles
to build, modify and tune one from a base of sound knowledge. So I have no
intention to write a definitive treatise on duplexers here. We won’t be able to
avoid some simple mathematics. I will leave out any that is of little value in the
practical world, however. Basic principles are the objective. That’s what most
repeater owner/builders need and normally lack.
 1-2

Also this book is not a cookbook. There are no cut-and-paste duplexer building
plans here. It it your responsibility to translate these concepts into actually
constructional designs. But don’t be discouraged, there’s precisely what you
need here to “roll your own,” to modify an existing commercial duplexer for an
adjacent ham band, or to knowledgably purchase a new unit. For perhaps more
than for any other aspect of the modern repeater, working with a duplexer from
sound basic principles and not blueprints is very important.

We will, though, build an 2M cavity. The idea is to illustrate the basic principles of
cavities in a practical way and to show that they can be built in a home workshop.
We’ll also look at a few examples from the 440 MHz band. 70 cm was my
specific interest when I began this book.

Lastly, there are many related topics of major importance to repeater builder and
owners other than duplexers – feed lines, antennas, isolators, to name just a few.
Some of these may get mentioned, but my objective is to limit this discussion, as
much as possible, to the common type of duplexer used in most modern
repeaters. The early chapters apply to all readers, the latter to home builders. All
are useful in concept, however.

Origin of The Book

In my original quest to understand duplexers I begin my search for an

understanding of the basic principles of duplexers in the highly-technical
engineering textbooks. My own engineering background suggested that for me
this was a resonable place to start. And this did yield a degree of useful principle.
I also did of course review the available popular amateur sources. There have
been a number of moderately-informative duplexer/cavity articles published in the
ham literature over the years. But to my general disappointment, from both
sources, I discovered quite quickly that the duplexer in the modern repeater
really remains somewhat of a “mystery” topic.

So I turned to a practical approach to close my understanding gap. I would jump

straight in and try to actually build a practical working duplexer for myself. Maybe
that was foolish. I’d probably make all sorts of mistakes, but I also sensed that I'd
probably learn a lot too. I’ve always been one who believes that trying to "roll
your own" is a good way to acquire practical understanding.

Immediately though, as you might guess, my ham friends loundly proclaimed that
I couldn't do it. “Only professionals can build duplexers, not amateurs. Duplexers
are beyond the weekend radio warrior. You don’t have the proper tools and test
equipment.” And yes, I will admit that at times it did seem that I’d bitten off more
than I could chew. Consequently, by the time I completed my first successful
duplexer, I’d gathered an impressive collection hardware-store leftovers.
 1-3

Today, thought, I can say that the struggle was worth the effort and I did save
“some” money. Though with the cost of metals these days, the savings in a
home-brew duplexer is less than one would like. The valuable part is what I
learned. That should be your primary objective in reading this book as well, not
saving money or finding cookbook-style plans for a duplexer. Learn the concepts
first. For despite how much mystery there seems to be surrounding duplexers in
the ham and commercial two-way radio world, they are not as difficult as they
seem.

Using the Ham Bands

For my first duplexer in the 1990’s, my interest was the 420-450 MHz amateur
UHF band, 70cm. At that time, common hardware store copper water pipe and
fittings were reasonably priced and were very suitable for building UHF cavities.
Today, aluminum is generally my preference, as we will see in the next chapter.
That why this book shows an aluminum 2M cavity built using home-workshop
tools and techniques. Again, the object is not to save money, but to give an
example fundamental duplexer principles and construction concepts.

If your interest lies on another band – no problem. Simply re-scale this 2M cavity
in direct proportion to wavelength. An important word of caution here. Do not
attempt to re-just simply re-scale the coaxial lines between the cavities. It is
much more complex than that. I will, though, show you how to easily deal with
the inter-cavity lines in a later chapter. It isn’t difficult either.

The home-brew 2M cavity in this book demonstrates one of the easiest way to
construct very-workable cavities in the home workshop. They are large in size,
but perform very well. I have actually used the technique for successful duplexers
on 2M, 70cm and 6M. I will even suggest a possible way to use the design on
10M.

Vital Opening Concepts

We really do need to begin with two very basic principles. Don’t mistake these for
“fluff,” however. In my view they are vital to one’s broad working knowledge of
duplexers. They have to do with, “Why do we need a duplexer in a repeater in
the first place?” Simple, you say? Yes it is, but many mistakes begin here.

At the most basic level, a repeater requires a duplexer for two reasons: (1) to
allow a very sensitive receiver, and a transmitter making power, to operate at the
very same time on the very same antenna. Also it must (2) isolate a repeater
from other radios I paid way too much attention to old wives tales in both these
areas in my early. So let’s very quickly get the essentials here. Every repeater
owner or builder must be “easy” with both.

DUPLEXER JOB ONE: The Same Antenna

 1-4

Consider, if you will, a typical repeater. How do we state its output? In Watts, of
course. Similarly, how do we specify its receiver sensitivity? This time in
microvolts. Simple, you say, but don’t pass too quickly over the difference.

Let’s take a real example by creating a theoretically repeater here in this opening
chapter. It will serve as an example throughout the entire book; we’ll return to it
often. So first let’s assume that the receiver can just hear a 0.22 microvolts from
a weak distant user station. I chose this specific value, as we’ll see in a moment,
to simplify the easy math tht follows. I think, though, that you will agree that 0.22
microvolts is a reasonable as a minimum receive sensitivity for a modern 2M
repeater.

The important issue now is this. As different as Watts and microvolts might seem,
they tare the same entity. Merely the size is different. We know this because we
can convert one into the other. It’s much like saying that a temperature of 23 C is
also 68 F. We can quite literally also state the Watts from a repeater’s transmitter
stated in microvolts, even though we normally don’t do so. So indulge me. This
conversion will emphasize a very important point about all repeaters.

To convert Watts in a repeater’s antenna system to microvolts we use Watt's law:

Watts = Volts2/Ohms

For our sample repeater, we’ll assume 100 Watts. But as you can see, we need
a resistance in Ohms to solve the equation. No problem. It is automatically
specified by the impedance of the antenna system. The 50 Ohm feedline
impedance gives us this value. Computing the equation we get:

100 Watts = 71 Volts2 / 50 Ohms

Work through the math yourself if you like, but notice that a 100 Watt transmitter
produces a 71 Volts in a 50 Ohm antenna system. That’s 71 million microvolts
compared to the receiver’s sensitivity of 0.22 microvolts Clearly, the transmitter is
making a signal very much too large for the receiver to handle. To illustrate this
point, which will become very fundamental to an understanding of duplexers, is
the reason we made this comparison.

Just to drive home the massive difference between the working signals
associated with both the transmitter and the receiver on the very same antenna,
consider Table 1-1. It lists the entire possible range, in power steps of 10, of the
signal levels possible between the two. It should make the equivalence of Watts,
microvolts and dBm as used in repeater discussions even clearer.
 1-5

Watts Volts dBm

-15
10 Watts 0.22 microvolts -120 dBm Our receiver
10-14 Watts 0.71 microvolts -110 dBm
10-13 Watts 2.2 microvolts -100 dBm
10-12 Watts 7.1 microvolts -90 dBm
10-11 Watts 22 microvolts -80 dBm
10-10 Watts 71 microvolts -70 dBm
10-9 Watts 220 microvolts -60 dBm
10-8 Watts 710 microvolts -50 dBm
10-7 Watts 2.2 millivolts -40 dBm
1 microwatt 7.1 millivolts -30 dBm
10 microwatts 22 millivolts -20 dBm
100 microwatts 71 millivolts -10 dBm
1 milliwatt 0.22 volts 0 dBm dBm reference
10 milliwatts 0.71 volts +10 dBm
100 milliwatts 2.2 volts +20 dBm
1 watt 7.1 volts +30 dBm
10 watts 22 volts +20 dBm
100 watts 71 volts +30 dBm Our transmitter
Table 1-1: Watts, Volts and dBm is a 50 Ohm antenna system

Remember from electrical theory that Volts are proportional to the square root of
the power. In any case, you should now be able to see why I chose 0.22
microvolts above.

Now notice the additional column on the chart, dBm. This is only another way to
specify signal strength. All three are exact equivalents. Which term we choose to
use in a repeater discussion depends mostly on which aspect of a repeater we
are discussing. For receiver sensitivity, microvolts is more convenient for output
power, Watts is best. dBm, being a logarithmic scale is good for both. Hams are
often intimidated by dBm so may avoid the term. But as Table 1-1 clearly shows,
one can state the sensitivity of a receiver, or specify how much power a
transmitter is making in dBm just as easily. I’ll therefore use all three terms
throughout this book. Just keep in mind that all are the same entity, though. Use
Table 1-1 for easy conversion.

A Little More on Signal Strength in dBm

“Plain” dB are not absolute. dB compare only the relative strengths of two
signals. dB say nothing about how many Watts or microvolts, for example, a
signal actually is. dBm do, however, specify the actual power or voltage of a
signal.
 1-6

To change dB into dBm we must give an actual value to one of the two signals
being compared (1 milliwatt is the usual practice for dBm). We must also specify
a system impedance, (typically 50 Ohms for radio antenna systems).

Therefore, in a typical repeater system,

E2 = W x R (Watt’s Law) You should now grasp the right column of Table 1-
E2 = 1 milliwatt x 50 Ohms 1. Our receiver has a sensitivity of -120 dBm and
our transmitter an output of +30 dBm. Again this is
saying exactly the same thing as 0.22 microvolts and 100 Watts.

Don’t, however, let all of this math confuse you. The issue here is not the
technical terms or the math. Though as a repeater owner/builder it is usually
worthwhile to master the math of dBm, if for no other reason than to not be
intimidated when the term comes up in repeater discussions.

What matters is that 100 Watts from our sample transmitter is equivalent to a
colossal 71 million microvolt signal (+30 dBm) on the very same antenna as our
receiver. Such an immense signal can never be feed directly into the input of our
sensitive receiver. So as you have already likely guessed, preventing the majority
of our transmitter’s +30 dBm signal getting directly to our -120 dBm receiver’s
input on the same antenna system is the duplexer’s number one job.

To make this point more evident, think about your car or home transceiver. Here
the transmitter is NEVER connected to the same antenna at the same time as
the receiver. That’s the reason for the T-R switch in a transceiver. Never do
receiver and transmitter operate at the same time. Therefore, the receiver never
has to look at the transmitter’s signal even though it is connected to the same
antenna, at least some of the time.

Again, this isn’t true for a repeater. Here it is necessary for the repeater’s
receiver and its transmitter to operate on the same antenna simultaneously, in
real time. Otherwise the repeater could not "repeat" the signal it hears from a
user, also in real time. Saying this one more time, but this time in actual
numbers, the receiver must efficiently be able to detect a tiny 0.22 microvolt
signal on the very same antenna that is simultaneously carrying a bone-crushing
71 million microvolt transmitter signal.

In relative dB this is a 150 dB difference. So to say one more time, solving much
of this difference IS the number one responsibility of a duplexer. In other words,
the duplexer provides a major part of a real-time isolation of roughly 150 dB
between the -120 dBm receiver and the +30 dBm transmitter. This amount of
total isolation is very typical of a modern repeater, and the duplexer provides a
major portion of it.

DUPLEXER JOB TWO: The Neighbors

 1-7

Unfortunately, most mountain-top repeaters live on the “shady” side of town. The
RF occupants of a typical repeater site, the other radios in the same building, are
very often “bad” characters. The rogue’s gallery includes:

• Other amateur repeaters

• Commercial repeaters
• Remote base radios
• Broadcast radio and TV transmitters
• Paging and telemetry radios
• Link radios
• Data radios

Often overlooked by repeater builders, the secondary job of a duplexer is to keep

all these potentially bad customers from disturbing your repeater, and like a good
neighbor to keep your transmitter from bothering them. This isn’t frankly always a
cut-and-dried matter. Too often repeater builder/owners pay little attention to this.
They thoughtlessly assume that if the duplexer worked well on the test bench it
will work the same on the hilltop. In this they forget that the other hilltop
transmitters are also making millions of microvolts for your antenna to pick up.
The potential intermix problems can be severe and also be quite a surprise at a
new repeater installation. This is the main reason low-cost duplexers won’t work
on all hilltops.

There’s also the real possibility that hilltop repeaters, including yours, may be
“out of adjustment.” These two ever-present factors regularly cause every
repeater owner and its users, to be very familiar with “intermod.” Keeping
“grunge” minimized therefore is the second main responsibility or a duplexer. And
it is here that correctly determining how good a duplexer to use is vital.

Therefore, frankly, never can a repeater owner/builder ignore the other radios on
a hilltop. As you master the principles of this book, you will gain the tools to
correctly deal with these challenges. Many factors make up our arsenal of
defense. What’s more, you will also learn the practical lesson that sometimes
small compromises in transmitter output power and receiver sensitivity are much
better than living with grunge or being a bad neighbor on your hilltop.

So these are the two responsibilities of a duplexer:

(1) To keep your transmitter (and your neighbor’s transmitters) out of your
receiver.
(2) To keep your transmitter out of your neighbor’s receivers.

Neither is a simple, but is a challenge a repeater builder/owner MUST face. The

following chapters will give you the keys.
 2-1

Chapter 2 – Let’s Make a Cavity

We’ll step aside in this early chapter for a do-it-yourself project and build an
actual working 2M cavity. It’s practical for two reasons. It will make the remaining
chapters easier to illustrate. If you will build one and experiment with it, you will
gain the kind of practical knowledge most repeater owners' lack. This is how I
acquired mine. I highly recommend the process.

You can also, if you wish, use several of them to build an actual working 2M
duplexer. This cavity is a bandpass (Bp) cavity. We’ll learn about cavity types
and how to change this cavity into a notch (Br) cavity later. You can also scale
these cavities up or down for another band.

Construction Objectives

When I built my first duplexer I had two objectives. They’re still the same today.
First I wanted to use only hardware-store materials and home-workshop
techniques. Published home-brew designs often discourage would-be builders
with hard-to-get or expensive materials, and often the need for machine-shop
facilities. This design uses neither.

Second, the cavity(s) had to be easy to tune. Here you’ll only need a thru-line
wattmeter, a couple of dummy loads suitable for the frequency, and an ordinary
synthesized HT or mobile transceiver on the band. Any of the common ham
antenna analyzers is also useful handy. If you just happen to have access to a
spectrum analyzer or a vector network analyzer with a tracking generator, all the
better, but it isn’t necessary. These just makes things easier.

Materials and Special Tools

The first decision in building a cavity is the type of metal to use. As any home
craftsman knows that there are not many choices, steel, stainless steel, brass,
aluminum and copper. For low insertion losses, copper is by far the best. What’s
more, it solders easily. But copper is expensive. In recent years it has become
almost prohibitive. That’s why commercial duplexer manufacturers often use
aluminum. The losses are just a little higher and aluminum does present
fabrication problems for the home builder, but these are manageable in the home
workshop.

In case you are wondering, steel is unsatisfactory. Number one, it rusts. More
importantly, it has very poor conductivity compared to copper or aluminum. See
Table 2-1. A steel duplexer would have very high insertion loss. Stainless steel is
also of no interest either. It too has low conductivity.
 2-2

Some commercial manufacturers use silver-plated steel, but electroplating is

generally impractical for home builders. Common yellow brass, as Table 2-1
illustrates, is also not a good choice, again because of relatively low conductivity.
Red brass, which has much a higher percentage of copper, can be used.

A curious point here is that gold is a poorer

conductor than copper. Gold is used on electronic
Silver 1 connectors because of corrosion resistance, not
Copper 1.1 high conductivity.
Gold 1.4
Aluminum 1.6 Copper is the most workable choice for home-
Nickel 4.3 brew cavities, particularly common hard-drawn
Brass 5 household copper water pipe. It solders easily and
Iron 6.3 excellent ends caps are readily available in all
Tin 6.9 pipe sizes. These eliminate the need for
Lead 14 machining. That’s why I chose common copper
Table 2-1: Relative water pipe for my early duplexers. Today,
resistivity of common however, copper has become almost prohibitive
for the home duplexer builder at all but UHF.

Aluminum is to me the best choice for home brew 2M and 220 MHz cavities and
duplexers. True, aluminum cannot be soldered and there are no available end
caps as with common copper water pipe. But as you will see, there are easy
ways around these.

Also, is slightly lower conductivity than copper is not enough to make aluminum
unsatisfactory. One just needs to make the cavities slightly. Aluminum is,
therefore, the material we’ll use for an example cavity.

Connectors

The second decision is the connectors used to connect the cavities to a repeater
or between cavities in the duplexer. I universally recommend BNC connectors. N-
type are also excellent, but are more expensive and not really necessary for most
repeaters, which are generally under roughly 100 Watts. Common PL/SO-239
connectors are just barely okay for 2M and below but are quite poor at UHF.

Commercial Cake Baking Pans

Now for my “big trick.” This is how I easily tamed aluminum for home brew. I was
looking around a restaurant-supply store one day and noticed some heavy-duty
round commercial aluminum cake-baking pans. They have nearly vertical sides,
are about 2 inches high and come in diameters from seven to twelve inches.
And, they are reasonably priced. “Here are the end caps I have been looking for
to make aluminum cavities.”
 2-3

I then remembered that hardware stores sell flexible 10 mil (thin) hard-alloy
aluminum sheet in rolls. It was obvious that a sheet of this could easily be
wrapped around two commercial aluminum baking pans (bottoms facing) and
held in place with stainless steel hose clamps. The end result would be an
excellent aluminum cavity, easily made in a home workshop. If you can’t find
stainless hose clamps long enough, simply use several to form one longer clamp.

The example cavity

here is made from the
smallest commonly-
available cake pan size
(7 in.), Figure 2-1.

If you are
contemplating 6M or
10M cavities, larger
cake pans are ideal. I
have used 11 in. cake
Figure 2-1: Cross section of 2M sample cavity
pans on these lower-
frequency bands.
Frankly, even at 2M, larger diameter cavities perform better. Use them if you
have the room. There is, however, a size limit which we will see later.

The center conductor here

is a ¼ wavelength legth of
1 ½ inch common copper
water pipe with an end-
cap to make attachment to
the end of the cavity easy.
We’ll learn later how to
select the diameter of the
center conductor for other
bands or larger cavities.

Here the coupling loops

Figure 2-2: Example 2M cavity and loop detailare made from ordinary
3/16 in. soft-drawn copper
tubing. The connectors are chassis-mount BNC jacks, fitted with long ring-type
grouning lugs.

Tuning the Cavity

To tune the cavity and its loops, only a simple setup is needed.
 2-4

The transceiver needs

to be modified to
transmit outside of the
ham bands so that you
can cover more than
the ham band during
the initial design phase.

Tune-Up Procedure

Here’s the basic cavity-

Figure 2-3: Center conductor extension and tuning screw tuning process. It is
much the same as we’ll
also use to tune up a complete duplex later.

First connect one port of the cavity (either connector) to your transceiver and on
the other side a through-line Wattmeter and a dummy load. The dummy load
MUST be specified for at least the frequency on which you are working. If you
happen to have a spectrum analyzer with a tracking generator, connect the
tracking generator to one connector and the analyzer to the other. Figure 2-4
shows both setups.

For the basic setup, in small frequency steps, apply power briefly and record the
Wattmeter reading. Power throughput will peak at the cavity’s resonant
frequency. Plot the data on a graph. The graph wizard in Microsoft Excel is an
excellent tool for this, though a paper graph is completely satisfactory. It will
immediately show you the
frequency to which the
cavity is tuned, as well as
its overall frequency
response, and the
insertion loss (space
above the curve). See
Figure 2-5. This is the
basic tuning procedure
used for design and in the
field, for individual cavities
and complete duplexers.
No other tools are needed.
Figure 2-4: Cavity tune-up setups
If the graph shows that the
cavity is quite a bit off
frequency, you will now need to lengthen or shorten the center conductor. The
tuning screw shown in Figure 3 is only for fine tuning.
 2-5

To raise the resonant frequency, shorten the center conductor in this cavity by
roughly ¼ in. per Megahertz. To lower it, lengthen the center conductor. An easy
way to do this without installing a new center conductor each time is to fabricate
a slip-on extension from a short piece of the same-sized pipe, slit lengthwise
down one side. I cut the slit with a hand-held hobby grinding tool and a cutoff
disk. Pry the slit open with a large screwdriver until the extension fits snugly over
the end of the center conductor. You can solder the extension in place if you wish
after tuning, or replace the entire center conductor.

This configuration naturally

creates a bandpass (Bp)
cavity. Bp cavities have the
response shown in Figure
2-5. Our sample cavity here
has a power bandwidth of
roughly 2.5 MHz and an
acceptable small amount of
insertion loss. Later we’ll
modify it to form a band-
reject (Br) or notch cavity.
By building several cavities
and following the guidelines
Figure 2-5: Frequency response (Bp) of sample cavity given in the remainder of
this book, you will be able to
assemble a complete 2M duplexer.

Actually building and then experimenting with a cavity like this was the most
instructive part of the entire duplexer adventure. Hilltop experience does not even
come close. Try it. You will not regret it. If nothing else, you will at least find out
that duplexers are not black magic.
 3-1

Chapter 3 – Cavities
A duplexer is normally made up of four to six ¼ wavelength coaxial cavities.
Figure 3-1 shows the common configuration. We’ll have more to say about the
overall structure of a
complete duplexer later.
First, though, we need to
take a look at the
important mechanical and
electrical properties of the
individual cavities.

We’ll discuss three vital

issues: (1) why we use
Figure 3-1: Most common duplexer configuration resonant cavities (2) how
RF energy gets in and out
of a cavity, and (3) the three basic cavity types: bandpass (Bp), band reject (Br)
and bandpass-band reject (Bp-Br).

As a final footnote to this chapter, we’ll take a very quick look at a cousin of the ¼
wavelength coaxial cavity, the helical resonator. Especially to hams interested in
6M or 10M repeaters, helical resonators have much practical application.

(1) Why Resonant Cavities

Here is a very basic question. Why are quarter-wavelength coaxial cavities the
only real choice for duplexer filters? It’s because there actually isn’t much choice.
The still continues, even in today’s world of miniaturized solid-state electronics, to
be the only practical filter type for a duplexer.

The ¼ wavelength cavity has three essential properties for a duplexer all present
in one filter type. The other main types, namely discrete coil-capacitors filters and
active filters, lack one or more of the three essentials: (1) the ability to handle
power, (2) high Q, and (3) low loss. It is this unique combination of all three in
one filter type, that has long made the resonant cavity the only real choice as a
duplexer filter. The cavities are not about to disappear from repeater hilltops.

Of these three properties, perhaps the most significant is the first. For only a
passive filter, that is, one without active electronic components, can handle the
power of the repeater’s transmitter. Remember, our sample duplexer must deal
with 71 million microvolts (+30 dBm). Active electronic filters can’t.

Yes, it is true that filters made from discrete coils and capacitors also can handle
power, such as in an antenna tuner. But at higher HF and VHF frequencies, and
 3-2

especially at UHF, discrete coil-capacitor filters have poor Q and exhibit high
losses. So it is the combination of all the three properties, as found in the
resonant cavity, that has caused resonant cavities to be the filer of duplexers for
the last fifty years. The principles of this book were just as valid in the 1930’s as
today.

As an aside, but still on the basic concept of cavities, it’s conceptually useful to
point out that resonant cavities actually do behave as if they were made up of
discrete coils and capacitors even though they aren’t. That is, they exhibit real
inductance and real capacitance. Hence it is not surprising that the equivalent
circuit of the ¼ wavelength coaxial cavity is a parallel-tuned L-C “tank” circuit.
Notice Figure 3-2. Noting this simple fact might make the basic concept of the
resonant cavity a little easier to visualize.

Physically though, as
opposed to electrically, a
cavity resonator is just
an open volume of space
enclosed by highly-
conductive walls. It’s
metal container “rings” or
resonates very readily in
the presence of RF
energy, very much like a
soft drink bottle makes a
tone when air is blown
across its top. In the
Figure 3-2: Equivalent circuit of a cavity -- parallel L-C cavity, the vibrations
tank circuit aren’t in air in thee
electro-magnetic field.
You can liken a cavity resonator to an organ pipe, a penny whistle or a flute. The
math formulas describing both are almost identical.

A hollow metal
sphere is
technically the
best shape for a
cavity resonator,
at least in terms
of electrical
efficiency, but it
isn't a very
Figure 3-3: Components of a 1/4 wavelength coaxial cavity practical shape
physically. A
metal cylinder is much easier. We also add inside In a duplexer cavity there is
also a smaller metal cylinder about 1/3 the diameter of the outer cylinder. This is
 3-3

the center conductor. See Figure 3-3. The center conductor is connected to the
outer cylinder at one end of the cavity but not at the other.

You may recognize this configuration as a short length of large-diameter air-

insulated coaxial transmission line, shorted at one end. The common duplexer
cavity is a simply a ¼ wavelength shorted coaxial stub. And simply by making it
large we easily achieve the three basic desired filter characteristics. In a
duplexer, we also cover the open end with a conductive cap placed just a small
distance away from the open end of the center conductor. The cover has little
effect on the action of the filter. It would work perfectly well as a duplexer filter
without the cover, though undesired outside signals could enter.

Field Strength and Orientation

As we said above, our signal in the cavity is in the form of induce an electro-
magnetic (E-H) field. The electric lines of force (E) lie parallel to the length of the
cavity, as shown in Figure 3-4. The magnetic force lines lie at right angles, in
concentric circles around the center conductor.

The magnetic field

and the
associated
currents in the
cavity walls are
strongest at the
shorted end of the
cavity and weaker
Figure 3-4: Electric and magnetic field orientations
at the open end.
The electric field
and its associated voltages on the cavity walls are just the opposite. They are
strongest at the open end and weakest at the shorted end. Both, however, are
stronger near the center conductor and weaker near the outer conductor.

These orientation are important when we look at loops and probes for coupling
later. It is helpful, therefore, in a solid understanding of duplexers to have a
strong mental picture of the E and H fields inside the quarter-wavelength coaxial
filters.

Resonant Frequency

Returning now to the soft drink bottle analogy, if we gently blow across the open
end, we produce the bottle's fundamental “note” or frequency. Like our cavity, the
soft drink bottle will now be oscillating in ¼ wavelength mode. If, however, we
blow harder, the bottle will break into an overtone mode. The note will now be
one or more octaves higher.
 3-4

These overtone or harmonic modes are why a trumpet, for example, can make
many notes with just three valves. For specific notes the musician excites an
overtone mode merely by blowing harder. Cavity resonators can also be driven
into overtone modes. But here it’s a “hazard” not an asset as it is in a trumpet. In
duplexer cavities we must avoid overtone modes. They exhibit high losses.

To establish the operating frequency of a coaxial cavity, we merely make the

length of the center conductor roughly ¼ wavelength. At 450 MHz, for example,
that’s about 6 inches. For other bands, the length is directly proportional to
wavelength. Hence a 2M cavity’s center conductor it is roughly 19-20 inches
long, again ¼ wavelength. On 6M and especially on 10M the cavities become
very large. This is where helical resonators, which we’ll look at very briefly at the
end of this chapter, can be useful.

Perhaps surprising to some, the outer shell of a cavity has virtually no effect on
the tuning of a cavity. So we simply make it a little longer than the center
conductor. Also, neither does the outer diameter of a cavity alter the frequency.
These two facts constitute a key concept. The resonant frequency is determined
almost exclusively by the length of the center conductor. This is not however true
for all possible resonant-cavity shapes where the resonant frequency is more
complex. This is another reason why duplexer cavities are generally ¼
wavelength coaxial cavities. For other electronic applications the harmonic
modes in other cavity shapes are useful.

Outer Cavity Diameter

Never the less, even though outer conductor diameter has little effect on the
resonant frequency, diameter is very important in a cavity. We’ll say more about
that later. For the moment, merely note that it should not be made larger than
roughly 1/3 wavelength. If we do, the cavity will break into a high-loss overtone
mode. This means that the diameter limit for 450 MHz cavities is roughly 8 in. A
2M cavity should not be larger than roughly 25 inches. But up to that limit there is
considerable benefit in a big diameter. Larger cavities filter better and with less
loss. For the home builder, a single large-diameter cavity could work better than
two smaller ones. With the construction techniques shown in the previous
chapter, large diameters are relatively easy to achieve. So this is an option worth
considering for home-brew duplexers.

(2) Coupling Energy in and out

The next main concept of this chapter has to do with how best to couple RF
energy in and out of a cavity. The most common method is a single-turn coupling
loop. This isn’t the only choice, it is just the most-frequently used method.
Actually, there are four practical ways to couple to a cavity: loops, probes, ports
and taps. In my investigations I examined each, and the results surprised me
somewhat. Let’s look quickly at each.
 3-5

Loop Coupling

A loop is a simple single-turn coil excited from a connector mounted through the
cavity wall. It is most often placed in the shorted end, but may also be put in the
side. The far end of the loop is grounded to the cavity. Notice Figure 3-3 again.
You may recognize that loops are analogous to the link windings shown in the
equivalent circuit, Figure 3-2.

A loop couples to the magnetic field and does this best when it is perpendicular
to the H field. Since the H field, as we learned, lies in concentric circles around
the center conductor, the loop is normally placed parallel to the length of the
cavity and on the cavity’s radius. It also couples best where the field is strongest.
This, as we also learned, is near the shorted end and near the center conductor.
Figure 3-3 shows loops in the maximum coupling position.

In my experimentation I was surprised to discover that the placement of the loops

isn’t actually critical. Despite what cavity design texts often imply, I have made
successful low-loss cavities with loops in the end, on the side and even near the
center of the cavity’s length. The loop size does change with these position
changes, in order to achieve equal performance, but the action of the cavity is
much the same no matter where one puts the loops. I will have much more to say
about loop placement in a later chapter.

A simple but meaningful analogy to the placement of a loop is pushing a child on

a swing. You may push anywhere you like up and down the ropes. If you push at
the bottom, you push gently with a long arc. If you push higher up the ropes, you
push harder but with a shorter arc. These are equivalent to larger and smaller
loops closer or farther from the shorted end of the cavity.

Probe Coupling

The second, though less frequently used coupling method, is a probe, Figure 3-5.
The probe is just one plate of a capacitor used to couple energy in or out. The
center conductor of the cavity acts as other half of the coupling capacitor. As
opposed to a grounded loop, a probe is open at the end. And as you may have
surmised, a probe couples to the electric field (E) instead of the magnetic field
(H). But like a loop, a probe couples best when it is perpendicular to the field, in
this case the E field, and is placed where the E field is strongest. This, as we've
learned, this is at the open end of the center conductor, near to it, as shown.

In my experiments, I
discovered that probes
function just as well as
loops, and this too
surprised me. For it

Figure 3-5: Maximum coupling position for probes

 3-6

made me wonder why commercial duplexer manufacturers don’t commonly use

probes. I soon discovered that there is a good reason. In that probes must utilize
the E field, they must be placed at the high voltage end of a cavity. A loop, on the
other hand, which utilizes the H field is placed at the low voltage end. With
probes, therefore, arcing is a potential problem, even at moderate power levels.

Recall as we saw, that a 100 watt transmitter places a 71 volt signal on a 50 ohm
transmission line. But in a good quality cavity Q can easily reach 1000. We
therefore multiply the 71 volts by 1000. This makes is clear that very high
voltages can exist at the probe end of a cavity.

Port Coupling

A third way to couple energy, this time normally between adjacent cavities, is to
cut a hole in outer walls of both to let some of the field leak through into the
adjacent cavity. This is called port coupling. A number of duplexer designs do
successfully implement this method. Also the helical resonators, often found in
receiver front-ends, commonly use port coupling. It is economical and a space
saver. Loops or probes usually need more room.

The main difficulty with port coupling for the home repeater builder is purely
mechanical. Varying position and coupling of loops and probes is easy. To
change the amount of port coupling one must physically change rhe size of the
port. This precludes easy experimentation. Also, since duplexer filters are often
made of cylindrical tubing, a port between cavities is also not easy to fabricate.
Small mobile duplexers of rectangular cross section often use port coupling.
Therefore, I only mention port coupling in passing. I did not extensively
investigate it, though I am confident that the end result would have been the
same as for loops or probes.

Tap Coupling

The final method, illustrated in Figure 3-6, is tap coupling. On the left is an actual
cavity with taps.

On the right is the equivalent LC

circuit again with taps. By adjusting
the position of the taps, one can
achieve a good impedance match as
well as efficient coupling. In small
cavities, where loops could be too
large for the space available, a tap is
an easy way to obtain tight coupling.

The disadvantage of tap coupling,

however, is isolation. If the type of
Figure 3-6: Tap coupling and equivalent circuit
 3-7

cavity you wish to use requires two ports, an input and an output tap, isolation
between ports is difficult to achieve. Tap coupling finds its best application in
single port cavities, such as notch cavities. It is seldom used in band-pass
cavities. We will discuss cavity types in a moment.

Which Coupling Type is Best?

Getting practical now, one might ask which coupling method is best? Does one
type make a better duplexer? By actual experimentation I found that the answer
is no. Surprising as it was for me, all four coupling methods ultimately perform
more or less the same after critical adjustment.

Experience, however, did lead me to the practical conclusion that loops are the
easiest choice for the home builder. They are also the most common choice for
the commercial manufacturers. If a loop will physically fit inside the cavity, you
can get it to perform just as well as any other method. It is much easier, however,
to mechanically implement and adjust. For this reason, I will stick to loop coupling
in the rest of this book.

(3) Cavity Types

Here is the final main concept of this chapter. It was another surprise to me to
discover that all one must do to change the same basic ¼ wavelength cavity into
any of the three basic types commonly found in duplexers: bandpass (Bp), band
reject (Br) and bandpass-band reject (Bp-Br) is to reconfigure the loops. Each
cavity type has a unique role to play in a duplexer, but the overall physical cavity
configuration is very much the same for all.

The best way to grasp

how the conversion from
one type to another is
achieved, is to examine
Figure 3-7. It is a classic
diagram found in many
electronic textbooks. We
won't labor over it here,
but only point out a
couple of important
points.

Bandpass Cavities

The behavior of a Bp
Figure 3-7: Impedance and reactance of an L-C circuit
cavity follows the total
impedance curve of a
parallel L-C circuit, Figure 3-7. The response of a notch cavity follows the
 3-8

reactance curve. To obtain a Bp response we must place the cavity in series with
the transmission line. Notice Figure 3-8.

First let’s look at a bandpass cavity. When a Bp cavity is off resonance, like a
parallel L-C circuit, its total impedance is low. At resonance, impedance reaches
a maximum. The absolute value depend on the Q of the cavity.

In a series configuration, all of the energy passes through the cavity. It is coupled
into the cavity by one loop and out by the other.

At the center frequency, the high Q cavity readily absorbs the energy supplied to
it by the input loop. Then at the output loop the H field couples back into the
transmission line. At resonance
very little signal is lost. To the
energy on the transmission line,
the cavity is invisible .

But transparency ONLY happens

at the cavity’s resonant
frequency. Off resonance, that is,
at a higher or lower frequency,
the cavity’s impedance rises very
rapidly. This greatly suppresses
off-frequency signals. Again, in a
Bp cavity, non-resonant energy is
Figure 3-8: Bp cavity, in series with transmission line
attenuated in response to the
impedance curve of Figure 3-7.

On the other hand, if we place the cavity in parallel with the line, we create a
band-reject (Br) or notch cavity or shunt configuration. Figure 3-9 shows the
common ways to do this. Parallel or shunt-connected cavities are sometimes
called or “suck out” cavities. The difference, therefore, between a Bp and a Br
cavity is merely the way the cavity is connected to the transmission line, series or
shunt. This alone
determines whether
it is a bandpass or
a notch cavity. Both
types still employ ¼
wavelength shorted
transmission line
stubs.
Figure 3-9: Br (notch) cavities, in parallel with transmission line
In (a) a single-loop
shunts the cavity
 3-9

across the line. In (b) and (c) the cavity shunts across a coil or capacitor. In (d) a
single-tap cavity shunts the transmission line. As we saw earlier, this is generally
how tap-coupling is implemented.

Here, however, is the key issue with Bp and Br cavities. A Bp cavity PASSES a
small band of frequencies. ALL others are rejected. A Br cavity REJECTS only a
small band of frequencies. ALL other pass on through. Said another way, a Br
cavity “sucks out" only a small band of frequencies.

In the Br or notch cavity,

behavior tracks the
reactance curve of Figure
3-7. As you can see, a long
way off frequency the
cavity reactance is
moderate.

As we approach
resonance, reactance rises
a little but then rapidly
becomes a deep notch. By
maximizing this response,
with either parallel
Figure 3-10: Br (notch) cavity response inductance or capacitance,
we achieve the familiar
response curve of Figure 3-10. Notice that the notch of a Br cavity is very sharp
compared to response of a Bp cavity..

Bp-Br Cavities

It is sometimes said that there is a third class of cavities, which supposedly has
both Bp and Br characteristics. How is it different? Actually it isn't. All Br cavities
have both a reject “notch” and a pass “bump.” When the frequency is significantly
off resonance the filtering action is only moderate, a few dB. Near resonance one
of the reactance excursions produces a small bandpass "bump" which is also
only moderate. But the
opposite reactance excursion, creates a deep notch. This is the action we are
looking for in a Br cavity, capable of many dB of filtering, far more than a Bp
cavity. The deep narrow notch is why Br cavities are the real work horses of
cavity duplexers.

In a few designs the bandpass bump is intentionally minimized and the cavity
called a pure notch cavity. In theory though, all Br cavities are Bp-Br reject
cavities. There is always both a notch and a bump. Hence there really are only
two basic cavity configurations, bandpass Bp and band-reject Br.
 3-10

Bump Up or Down?

Of great importance, however, to duplexer design is whether the bump is

configured to be above the notch or below it in frequency. Both are easy to
arrange. If the line bypassing the cavity has a shunt inductor, as in b, the notch
will be on the high frequency side of resonance. If the line has a shunt capacitor,
the notch will be on the low frequency side.

But as I just said, the relationship of the notch to the bump, as compared to the
cavity’s resonant frequency is very important. And again, that’s because the
transmit filters must always be one way and the receive filters the other. Which
one you will need for your repeater is dictated by the frequency split of your
repeater, that is, whether the transmitter is higher or lower in frequency than the
receiver. In Figure 3-10, a 2M cavity, the notch is high of the bump. This would
be for a repeater with a positive offset of 600 KHz. We'll get to the specifics in a
moment.

The Merits of Bp (Bandpass) vs. Br (Notch)

The final piece of basic theory of this chapter will be to compare the basic way
we use both types of cavities. For they fulfill different roles in a duplexer.

As far as the bandpass bump part of the response curve of either, both work the
same basic way. We want the bump to be centered on the frequency we want to
pass on either side of the duplexer. And at that frequency only a fraction of a dB
will be lost. The deep notch of the Br cavity(s) is another story as we shall see.

If bandpass cavities were good enough, that is, if its bandwidth were small
enough, then an all-Bp cavity duplexer would be ideal. Unfortunately Bp
bandwidth of a Bp cavity is not sufficiently narrow to be the only type of filter used
in a practical duplexer. Referring to our sample cavity, its bandwidth is only 3.6
MHz. At the normal frequency offset of a 2M repeater, 600 KHz, it can provide
only 3dB of filtering. This is why we must also use notch (Br) cavity cavities in a
duplexer. The notches provide many more dB of filtering than the off center
rejection of a bandpass cavity.

The notch cavity, however is incapable of rejecting anything but the small band of
frequencies on which its notch is centered. But in a duplexer that’s precisely what
we need. We primarily only need to keep the transmitter frequency out of the
receiver. Notch cavities are ideal for this.

On hill tops, however, and to a smaller degree in our own repeater, a

combination of both notch and bandpass cavities is necessary. The notch
cavities fulfill the duplexer’s primary job of isolating a repeater's receiver from its
transmitter. The bandpass filters fulfill the secondary responsibility of providing
general isolation from the outside world.
 3-11

Duplexers, therefore, that are used on radios where there are no neighbors, such
as in mobile installations if both transmitter need to operate at the same time, a
notch-only duplexer can be used. Unwittingly, however, many radio amateurs
make the mistake of trying to use this type of duplexer on a hill top. They are
attracted to the small size and low cost of mobile notch-only duplexers. But when
they do, they forget that their repeater is wide open to interference from the
mixes and intermod that is common at such a site. Small mobile duplexer by
themselves are not a good idea on mountain tops.

If economy is imperative, you can us a notch-only mobile duplexer on a hill top.

But you will need to add some outboard bandpass cavities to take care of the
neighbors. I have personally built good-performing hill-top duplexers from a low
cost mobile duplexer and outboard bandpass "bottles."

Fix in your mind then, that series-connected bandpass cavities are to keep the
neighbors out. Shunt connected band-reject cavities are to keep your transmitter
out of your receiver. This is a bit of a simplification, for both cavity types do also
provide other protections. But in tuning a duplexer to fulfill its main
responsibilities you will naturally cover all the bases at the same time.

Helical Resonators

As we promised, let’s take a very quick look at a close relative to the cavity
resonator, the helical resonator. You most commonly find them in the front end of
narrow-banded receivers. I have experimented with a hybrid form of the helical
resonator for ham 6M and 10M repeater application. I personally call a duplexer
built from helical resonators a heli-plexer. Here I will only very quickly overview
the concept of helical resonators in duplexer service. They do work, though, as
my limited experiments have verified.

As we can do with a full-sized antenna, we can also shorten the length of a

resonant cavity by inserting
a discrete coil into its center
conductor or by completely
replacing it The quarter
wavelength cavity now
becomes a semi-helical
resonator or full helical
resonator. See Figure 3-11.
To implement this concept
in a duplexer filter one only
needs to exert
knowledgeable caution.
The advantage of semi-
Figure 3-11: Helical resonator, semi-helical resonator, cavity
helical filters can easily
 3-12

outweigh their small disadvantages. But what are the advantages and
disadvantages?

The main advantage is the large reduction in the length, which on 6M or 10M is
highly desirable. However, for exactly the same reasons an antenna shortened
with a loading coil has disadvantages over its full-length cousin, so does a semi-
helical resonator. Still for 6M and 10M, a shorter filter may well be worth the
disadvantages.

Personally I have only just scratched the surface of substituting semi-helical

resonators for full-size cavities in repeater duplexers for one working 6M heli-
plexer. I have yet to try the technique on 10M. The 6M duplexer was, however, a
reasonably good performer. I used 10 in. baking pans for it. So I only offer the
following as grounds for experimentation. The 2M cavity in this book should be
easy to use as a test bed.

First of all I personally think that one should not reduce the length of the center
conductor of a semi-helical filter to more than roughly ½ the full-sized length. On
6M that would be just over two feet. Too much coil would be required. And as we
have seen, discrete coils exhibit too much loss at VHF and above to be
satisfactory as the center conductor. This is why a recommend the semi-helical
design, which retains a significant portion of the normal center conductor.

RF skin effect is the major enemy. So make the coil from copper tubing as large
as possible and use only a turn or two. This is also a good idea for rigidity.
Likewise keep the space between turns wide. To do both of these you will have
to retain as much of the normal center conductor as is required to bring the filter
to resonance.

I do not know if semi-helical cavities are more or less temperature sensitive than
full-sized cavities. I suspect more. So you will need to implement the brass tuning
screw technique we saw earlier. It will work much the same. Coupling loops and
shunts to create notch cavites should also be similar to full-sized cavities.

I have not seen the semi-helical resonator employed in the commercial two-way
radio world. There are two significant reason why. Low-band 30-50 MHz VHF
commercial repeaters have not ever been common in the two-radio world.
Simplex and remote base radios have rather generally been the rule. Duplexers
are not needed for either. Also the widespread use of commercial low-band for
land-mobile services is generally fading. Hams are generally the only current
uses of repeaters in this spectrum.

One final thought on the practicability of the heli-plexer on 6M and 10M is our old
friend band noise. It is three times higher on 6M, and five times higher on 10M,
than on 2M. We can live with the lower efficiency of a heliplexer. The 0.22
 3-13

microvolt sensitivity of our sample 2M is far more than the noise on the lower
bands will ever allow. Less duplexer is needed.
 4-1

Chapter 4 – Temperature Drift

This chapter is a side issue, but one of considerable importance. How does
temperature affect a duplexer? With a commercially-built unit, a repeater owner
does not need to pay much attention to this; it has been dealt with by the
manufacturer. For the home builder, it is of major concern, however. The do-it-
yourself builder must make provision during the design phase to keep home-
brew cavities “on tune” as the weather changes.

We likely learned in grade school science class that most substances in the
universe expand as they get warmer. The
Metal %/ Fo metal used in a cavity is no exception.
Aluminum 0.00131 Table 4-1 is the thermal expansion of
common metals, expressed in percent per
Cast Iron, gray 0.00058
degree F.
Chromium 0.00033
Copper 0.00098 Notice that copper expands or contracts
Invar 0.00007 0.00098% for every degree F. Aluminum
Iron, pure 0.00068 has a higher coefficient, 0.00131. How
Red Brass 0.00104 much then, you are probably asking, does
this affect the tuning of a cavity? Let’s take
Carbon Steel 0.00078
some real numbers and you’ll see.
Yellow Brass 0.00113
Table 4-1: Thermal expansion of metals For a range of 100 degrees F – reasonably
for a radio hilltop – the center frequency of
a copper 2M cavity drifts 143 KHz. Our aluminum 2M cavity drifts 191 KHz. It
would be 436 KHz for a copper 440 MHz cavity. These figures are obtained by
multiplying the percentage of change of the length of the center conductor per
degree, by the frequency and then by the total number of degrees. Thermal
expansion is essentially linear over a wide range.

Figure 4-1: Bp vs. Br bandwidth

 4-2

Now examine Figure 4-1. It is our sample cavity in both Bp and Br configuration.
You’ll easily see how significant temperature change is for the two. The
bandwidth in Bp configuration is 3.6 MHz, for Br it is 200 KHz. Therefore, it is
evident which configurations will experience difficulty with drift of 143 KHz, the Br
notch cavity, the configuration that fulfills the main filtering responsibility in the
duplexer.

If there were no temperature drift, uncompensated narrow-bandwidth notch

cavities would always keep a repeater at full efficiency. But if the notch cavities
drift down in frequency on a hot day, and no longer notch effectively, a noticeable
loss of repeater performance will be experienced. Therefore, home-brew notch
cavities clearly require temperature stabilization.

Stabilization Methods

The preferred method for commercially-built cavities is to employ a rod made of a

metal, called Invar, a nickel steel alloy that has a uniquely low coefficient of
thermal expansion (7% that of copper). See Table 4-1. The length of the center
conductor is controlled by the Invar rod. It changes length so little that the cavity
stays within acceptable limits as the temperature changes. Expansion of the
outer shell of the cavity, as we
learned earlier, does not affect
cavity tuning.
Invar, however, is not a
hardware-store material. Its use
by the home builder is
prohibitive. Also, sliding fingers
to permit the center conductor
to move up and down are
required. They are well beyond
home brew. So we will need to
find another method. The one I
Figure 4-2: Temperature stabilization prefer is simple, yet quite
elegant. As a bit of trivia, it is
based on a method devised over two centuries ago by famous British clock
maker, John Harrison of sea-faring longitude fame. It kept the length of clock
pendulums from changing length with temperature. Notice Figure 4-2 to see how
this works.

If we allow a short metal rod to extend a short distance into the open end of the
center conductor, we implement temperature stabilization. How? Quite simply.
The small rod is attached to the opposite end of the cavity as the center
conductor. As the cavity expands the short rod naturally tends to withdraw. It
does so in that the outer wall of the cavity is longer than the center conductor.
Further, if we make the outer wall of a metal that expands more rapidly, the effect
 4-3

will be greater. This is the case for our sample cavity. The inner conductor is
made of copper and the outer wall aluminum. Aluminum expands more rapidly
than copper.

How does this arrangement minimize frequency drift? The end of the center
conductor and the rod make up a small capacitor that is in parallel with natural
capacitance of the cavity. Recall that a resonant cavity act as if it were made up
of discrete coils and capacitors, So in other words, the little rod “tunes” the cavity
a small amount. The colder the cavity become the more the rod withdraws,
thereby lowering the frequency, just as we want. The opposite takes place at
warmer temperatures. The small rod, which I generally make from a threaded ¼
in brass bolt and a captive nut or a threaded hole, also provides, as a secondary
benefit, a convenient way to make small adjustment to the overall tuning of the
cavity. NOTE: The screw is designed to function correctly when it is screwed in
half way. If you discover that you must move it near its ends to tune the cavity,
change the length of the center conductor a small amount.

How long should you make the small rod, and how much shorter must the center
conductor be to compensate for the presence of the rod? Unfortunately these are
not simple to calculate. Like a good ham, I prefer a pragmatic approach. During
the design phase of a cavity I first shorten the center conductor a little and
introduce the bolt, screwed in half way. Then I retune the cavity to frequency by
shortening the center conductor a small amount.

Next I perform a rough measurement of how much the cavity drifts with changes
in temperature. A large cardboard box and a common hair dryer are quite
adequate tools to accomplish this. If the resonant frequency still drops at higher
temperatures, I install a longer bolt and shorten the center conductor a little bit
more. After a few such adjustments I arrive at a reasonable value. Remember
you don’t have to get it perfect. The temperature drift of commercial Invar rod
stabilized cavities isn’t perfect either. All you need do is to reduce the
temperature drift enough so that the notched frequency remains within the
bandwidth of the notch. The dimensions given for the bolt and the center
conductor of the cavity in this book satisfy these requirements for this cavity.

Duplexer cavities again are not black magic. The home builder can achieve very
acceptable temperature stability for home-brew cavities with this simple but
elegant little method. I have used it successfully on most ham VHF and UHF
bands.
 5-1

Chapter 5 – Performance
In this chapter well wade knee-deep into a sea of old wives tales often heard in
the repeater world. Much misinformation commonly exists in knowing what to
expect from a repeater’s duplexer. There are useful compromises here that we
can intentionally make that will benefit us? For all these performance factors
must be coordinated if top repeater performance is to be achieved.

To begin to get at this vital area of knowledge, let’s begin with the two most-
common terms used to describe how well a duplexer must perform, that is how
much loss it will introduce: (1) isolation and (2) insertion loss. Then, later in this
chapter we’ll also look at how well the (3) receiver and the (4) transmitter must
perform. These are the biggest area of old-wives tales in the repeater world.

Isolation

Ideally, in the “perfect” duplexer, the transmitter of the duplexer should be

“invisible” to the receiver on the opposite port, and vice versa. In practice,
however, this is never totally true. Actually the cavities on both sides of a
duplexer only ever “reduce” the unwanted signals, just sufficiently for the
repeater to function. Determining how much this reduction needs to be is the key
issue in knowledgably managing a repeater and its duplexer. We call this the
isolation of the duplexer. It is generally stated in dB, the relative ratio of signal
weakening. And again, it is never perfect.

Recall from Table 1-1, Chapter 1 that our sample receiver can hear a tiny 0.22
microvolt signal on an antenna also carrying a transmitter signal 150 dB (a
thousand million, million times) stronger. But does the duplexer have to provide
all these150 dB of isolation? Actually it doesn’t, and this is an important repeater
concept. The reason it doesn’t is simple; the repeater’s receiver provides some of
the dB, all by itself. That’s because its “front end” is frequency selective. This
partially rejects signals not on center frequency. That’s the meaning of selectivity.
The cavities of the duplexer, therefore, only have to provide part of the total
required 150 dB of isolation between receiver and transmitter.

This is also of course why a repeater has a frequency “split” or “offset.” The input
and output frequencies are intentionally placed apart. For example, on the 440
MHz band, the split is commonly 5 MHz and on 2M, 600 KHz. It’s different for
each band by convention, but the main function of a repeater’s offset is to
provide a large part the isolation needed in a repeater between receiver and
transmitter.

Insertion Loss
 5-2

The second main important performance characteristic of a duplexer is insertion

loss. It’s the amount of power or sensitivity that we give up for the duplexer to
function. There is always some, again normally stated in dB. For example, if the
output of our 100 Watt sample transmitter becomes 50 Watts after passing
through the duplexer cavities, the transmit insertion loss is 3 dB (half power) .
The same applies to the other side of the duplexer. Insertion loss is specified for
both sides and is usually a different value for each.

Jumping ahead briefly, insertion loss is mostly caused by RF skin-effect

resistance loss on the inner surfaces of the cavities. It is also affected by cavity
diameter and the coupling factor of the loops. For now, merely recognize one
basic fact about insertion loss. It is NOT always bad.

Insertion loss is the one performance characteristic of a duplexer that is most

open to knowledgeable compromise, even if most repeater owners consider it
their deadly enemy. For one can often derive increased repeater performance by
intentionally increasing the insertion losses in the duplexer. I have often seen a
repeater become better able to hear weak signals by doing just that. We’ll give
major attention to this later.

No Duplexer is Perfect

We of course would like to always have a “perfect” duplexer. Let’s suppose that
we could actually buy such a device – one with infinite isolation and zero
insertion loss. Our “fantasy” duplexer would be a universal "fits-all,” wouldn’t it? It
would behave flawlessly on any “RF dirty” hill top, no matter how much power our
repeater is outputting and how sensitive and selective our receiver is.

Coming down the scale just a bit, how well will a high-cost “top-of-the-line”
duplexer perform in the same situation? It would have perhaps 120 dB of
isolation and as little as one dB of insertion loss on both sides. Again, wouldn’t
such a duplexer work well in NEARLY all situations?

Now, consider the other end of the spectrum, a small low-priced mobile duplexer
with only 40 dB of isolation and as much as 3 dB of insertion loss. Can we use it?
Actually, we can. Recognize here, though, that such a duplexer would not
function well in nearly as many situations as the perfect duplexer or even the top-
of-the-line model.

For example, a low-end duplexer might be okay for a small repeater at a quiet
home QTH with a slightly “numb” receiver. But it certainly wouldn't be satisfactory
for a sensitive high-power repeater on a dirty RF hill top. Also a repeater with a
low-end duplexer might fail, had it been okay before, if its owner increases the
transmitter’s power output power or installs a receiver pre-amp. A high-end
duplexer, on the other hand, would likely tolerate such additions.
 5-3

The vital point here to realize is, the better the duplexer the more places one can
use it without understanding it. This might seem to suggest that you should
always use a top-of-the-line commercial duplexer. Not really. High-end duplexers
are expensive. Can you always afford the price?

Wouldn’t it be better to exchange knowledge for cost? Well, there’s good news.
Let’s now see some well-informed compromises we can make. You may net big
savings here without any significant compromises in performance.

FACTOR ONE: Band Noise

The biggest factor in knowing how good (or expensive) a duplexer needs be is
band noise on your hilltop. The radio spectrum is full of noise. Notice Figure 5-1.
Most repeater owners pay little attention to the average band noise at their site,
often to their detriment. On the other hand, hams who work the HF bands, know
well how important band noise is. On a noise-free day one can often “work the
world.” On a noisy day, a numb
receiver is as good as an
expensive one. To the repeater
owner, noise is just as important.
Working with it knowledgably is
vital to maximum repeater
performance.

RF noise comes from a variety of

sources. The sun, the earth’s
atmosphere and even our galaxy
all make radio noise. We call this
natural kind of noise QRN.
Figure 5-1: Generalized noise across the radio Industrial machinery, power lines
t and other transmitters create man-
made noise or QRM. Together all these sources create a background of noise at
your site. It is very real and always there.

Most repeater owners pay little attention to it, however. Yet in deciding how good
(how expensive) a duplexer needs to be, and how sensitive to make your
receiver, the average noise level at your site is vitally important. Here now are
the important basics.

As you can see from Figure 5-1, on average, noise in the radio spectrum is much
higher for low frequencies than for high. In fact, total band noise decreases
roughly in inverse proportion, all the way from VLF into UHF. As a simple “rule of
thumb” for noise on the radio spectrum, when the frequency doubles, the
noise drops to half (down by 6 voltage dB). Slightly above 1 GHz spectrum the
noise finally reaches a minimum and then begins to rise again. New noise
 5-4

sources become dominant, such as atmospheric ions. We needn’t, however,

concern ourselves in this book with frequencies above a Gigahertz.

So what is the big concern about RF noise to the repeater owner? Simply this.
Both the RF noise spectrum noise and the noise in our own receiver, set a very
hard limit on how sensitive our receiver ever needs to be and how well a
repeater can perform. I used to hear ill-informed repeater owners say that their
repeater can hear a 0.1 microvolt signal. Well perhaps that is true when the
signal is coming from a quiet signal generator. But if the noise floor at their
repeater site never drops below say one microvolt, such a receiver is of no use.

Consequently, if you will knowledgeably match your repeater’s overall system

sensitivity to the hill top, you may well find that you do not need as much
duplexer as you think or a receiver with “killer” sensitivity. It is a simple fact of
repeater life, that a receiver can't hear signals lost in noise. If noise is high, all the
receiver sensitivity in the world is fundamentally worthless.

If you don’t believe this, just try telling an HF ham trying to work 40 M who can’t
hear anything because of a 10 over 9 noise level that he needs to raise the
sensitivity of his receiver. He knows that all that will do is bring in more noise.
Noise truly does set a very hard limit on how sensitive a repeater can practically
be, and therefore how much insertion loss a duplexer can have and still function
at peak performance at that site. I have several times seen that be intentionally
increasing the insertion loss in a duplexer, to narrow the overall system
bandwidth, enables a repeater to hear weaker signals in the noise.

For other bands, use the simple 6 dB per octave rule of thumb. That is, you can
expect the noise to be on average 5 dB (three times) higher on 2M, or -111 dBm.
This places the typical total noise on the average radio hill at roughly 1 microvolt.
A receiver much more sensitive than is largely unnecessary, most of the time. If
the noise drops below average, or if the user takes steps to reduce the noise
figure of the overall system, a more sensitivity receiver help. These are of course
“average” figures.

In actual practice, the noise on real hill tops does routinely rise and fall above
and below the average figure for that site. Haven’t you ever noticed that there are
days when you can’t get into your favorite repeater as easily from a particular
location? This is most likely due to higher noise on that day. I have for example,
seen the noise as high as -90 dBm (7.1 micro volts) at 450 MHz on an actual
hilltop. At such a time A VERY "numb" receiver or a duplexer with a lot of
insertion loss is perfectly okay. On a quiet day, it wouldn’t be.

Impact of Noise on the Duplexer

Many repeater builders wonder why commercial manufacturers don’t make their
receivers more sensitive. Is it because they don’t know how, or can’t? Hardly. It is
 5-5

because they are aware of the noise floor. They have already likely given the
receiver you are putting in your repeater as good a “front end” as it can
reasonably use. In many cases, in fact, it may be more sensitive than you
actually need. It is probably difficult for most hams to realize that a receiver can
be “too” sensitive. Just recall, your duplexer has to match every dB of sensitivity
your receiver has. Again, becoming knowledgeable of the average noise level at
your repeater site is vital.

Measuring the Noise

Fortunately there is a relatively easy way to make a satisfactory estimate of the

noise at your repeater site. just use your repeater’s receiver, a calibrated FM
signal generator and your repeater’s normal antenna. You will also need two low-
power 50 Ohm dummy loads or terminators and a hybrid combiner.

What a hybrid combiner? It is a passive mixing device that let’s you feed both
your antenna (through the duplexer) and a calibrated signal generator into your
receiver at the same time. The combiner keeps either signal from affecting the
other. A hybrid combiner is required for a proper estimate of total noise.

It is quite easy to build one as a coaxial hybrid ring combiner. See Figure 5-2.
Make it from 50 Ohm coaxial cable and BNC “T” connectors. The quarter
wavelength (in coax) phase
relationships between the
ports create the needed
isolation between the two
signals. This kind of
combiner is good only for
one band, however. It is a
“tuned” device. You can,
however, make alternate
cable sets for the same “T”
connectors.

Don’t forget that the coax

section must account for
velocity factor. On 2M, open
Figure 5-2: Ring hybrid combiner made from coax
air quarter wavelength in is
roughly 20 in. You need,
therefore, to make the distance from the center of one BNC “T” to the next
connector, 20 in. multiplied by the velocity factor of the coaxial cable – typically
0.6. Look up the velocity factor for the cable you are using in a radio handbook.
For a velocity factor of 0.6, the center-to-center distance on the righ would be 12
in. for a ¼ wavelength cable and 36 in. for a ¾ wavelength cable.

The Measurement Procedure

 5-6

1. Connect both the calibrated signal generator and your antenna to the
hybrid combiner. Figure 5-2.

2. Set the signal generator to the receiver’s input frequency.

3. FM modulate the generator with a continuous audio tone.

4. Now alter the RF output level of the generator until the tone coming from
the receiver’s speaker roughly matches the noise coming from the antenna.
A reasonable estimate here is completely adequate. You are not making a
laboratory measurement of the noise, only getting an estimate.

The noise level, and the output of the signal generator, are now roughly equal.
Read this level from the signal generator. We call it the minimum discernible
signal (MDS) your receiver can usefully detect. It incidentally also includes the
noise made by your receiver. By repeating this measurement on a number of
subsequent trips to your hill top, you can factor out your receiver’s own noise and
gather a perfectly reasonable estimate of total external hilltop noise. And from
this you can determine the required receiver sensitivity for your site as well as
begin to know how much duplexer you need. You may be surprised; I was.

FACTOR TWO: Receiver Sensitivity and Selectivity

By now it should also be evident that we need to know our receiver’s

performance if we ever expect to knowledgably match it to the duplexer and the
hilltop. Two factors matter here: (1) sensitivity and (2) selectivity. The second,
often totally overlooked by repeater builders, is most often much more important
than the first.

Sensitivity is easy to measure. Radio technicians often do so when they visit a

site. Merely connect a calibrated signal generator to the receiver and adjust the
generator’s output level until the receiver just barely can’t hear the signal. This is
the receiver’s sensitivity.

Lets use a 440 MHz repeater as an example. Again we’ll assume a receiver
sensitivity of 0.22 micro volts or minus 120 dBm. Without any additional gain, this
sensitivity is a naturally a good match to the average noise level at a typical UFH
repeater site, that is -110 dBm. Remember, we want only a little more sensitivity
than it takes to get down to the noise. An extra 10 dB is a realistic amount of
headroom. Now let’s measure our receiver’s selectivity, the other vital factor we
need to correctly match our receiver to its environment. Remember, a receiver
with the same sensitivity as another, but with poorer selectivity may work poorly
in a repeater where the other will perform flawlessly.

Selectivity Measuring Procedure

 5-7

1. This time, connect your receiver through the hybrid combiner to two calibrated
signal generators. You may substitute a hand-held transceiver for the second
generator with a small tone generator connected to its mic input. Remove its
antenna and wrap it in aluminum foil. Place it far enough from your test setup
away to create a weak signal. In this case, keep the repeater’s antenna
connected to the combiner.

2. Set the remote transceiver or the second signal generators to make a steady
weak FM modulated signal on the receiver's input frequency. Open the receiver’s
squelch so that you can hear the noise. The level from the second generator
should be just sufficient to begin to quiet the repeater’s receiver. Maintain this
level throughout the measurements.

3. Make a series of measurements with the signal generator set increasingly off
of the receiver’s center frequency. For each, increase the level of the signal
generator until you can just hear of “desense” caused by the off-channel signal. It
will be an increase in noise from the receiver’s loudspeaker. 500 KHz increments
should be sufficient, on both side of the receiver’s center frequency of the
receiver.

Plot the results on a graph. It should look similar to Figure 5-3, a graph I
compiled from an actual 440 MHz receiver. As you can see, at the center
frequency, a very tiny signal (-116 dBm) will desense the receiver. But the more
you move away from the center frequency the stronger the off-channel signal has
to be to cause receiver desense. This is due to the receiver’s selectivity.

This curve shows us that

for our example receiver
an off-channel signal five
MHz lower in frequency
must be 88 dB stronger to
cause desense than an on-
channel signal of -116
dBm. So at the normal 5
MHz frequency spilt of a
440 MHz repeater, the
receiver’s selectivity
already provides 88 dB of
Figure 5-3: 440 MHz receiver desense profile
the total 150 dB isolation
needed in our example.
The duplexer, therefore, only has to provide 62 dB of isolation. That's a lot less
than we pay for in a top-of-the-line duplexer.

We do though actually need a bit more than 62 dB for a couple of reasons.

Something more like 90 dB is in order. Much of this, as we’ll saw earlier, is
 5-8

because duplexer cavities drivt with temperature. So we add some additional dB

of isolation to compensate for this. Also, transmitters are never perfect. They
don’t put ALL their energy on only the output frequency. We often call a
transmitter’s off channel energy “dirt.” Hence the dirt from our own transmitter
causes additional desense. Outboard transmit power amplifiers are often
notoriously “dirty.” Adding one will frequently demand a better duplexer.

Preamps

While we are on the subject of sensitivity, let's pause briefly to consider

preamplifiers. In recent years preamps have become the shining star of the ham
repeater world. At least a lot of hams look at them that way. Unfortunately, they
can cause problems in respect to selectivity, hence how much duplexer is
needed.

The reason it, after-market preamps are broad-band devices. They have little or
no selectivity. They amplify the noise, the dirt and the grunge just as much as
they do the signals we want. Yes, of course, they do add sensitivity an can also
improve the noise figure of our receive system. But quite often they do this at the
expense of an undesirable increase in total receive bandwidth. As we have
painfully seen, reducing receive selective is deadly in a repeater. It instantly
places an increased burden on the duplexer, which may now no longer be able to
properly our isolate receiver and transmitter particularly if they are low-end
models. So don't be surprised if a preamp increases the “grunge.”

A possible benefit of using an after-marked preamp, as I just mentioned, is an

improvement in the overall receive system noise figure. Some high-end preamps
employ higher-quality and lower-noise front-end devices than some older or ham-
grade receivers. So if the repeater owner is also willing to add some additional
outboard bandpass cavities. to help control the increased receive system
bandwidth, a preamp can add useful overall noise figure reduction. This is often
much more helpful than simply increasing the gain.

So before installing a preamp, in hopes of creating a "killer" repeater, spend the

time to measure the sensitivity and selectivity of your receiver as well as the
working noise level at your site. You may be in for a surprise. For if you already
are slightly below the working noise floor, more gain from a preamp may not
help.

I go by a simple rule of thumb with preamps. On really high mountain tops, where
the desired signals are strong, and at remote sites that are difficult to get to, I
avoid preamps. Most modern barefoot commercial receivers have sufficient
sensitivity and noise figure for these situations. At lower altitude sites, such as at
a home QTH, or where the site is easy to get to, a preamp may help. Never,
though, look at preamps as a miracle cure.
 5-9

Selecting a Receiver

In light of what we have seen, you may now be able to guess that some receivers
are not suitable for repeater service. In commercial repeaters, the manufacturer
designs the receiver accordingly. Ham repeater can rarely do this. So let's briefly
talk about how to select a receiver for repeater use.

Commercial crystal-controlled mobile two-way transceiver receivers, the type

most amateur repeater builders have long used to build repeaters, are generally
a good choice. That’s because the more-recent models of crystal-controlled
commercial transceivers are narrow-band devices. They have good front-end
selectivity, often employing narrow-band helical resonators. They were designed
to operate on only a few adjacent frequencies in commercial mobile service. So
the manufacturer designed them with narrow-band filters directly in the front end.
This kind of receiver works very well in a home-brew repeater. Unfortunately they
are disappearing from the commercial two-way radio world.

Instead, the receiver used in most ham transceivers, and newer fully-synthesized
commercial mobile transceivers, are built to cover a wide range of frequencies.
This precludes narrow-band front ends and good front-end selectivity. Instead,
selectivity is created by IF filters. Compare the desense curve Figure xx, an older
crystal-controlled commercial UFH receiver to a more recent synthesized ham
440 MHz transceiver, Figure xx

This method
NOT COMPLETED
works okay in
Figure xx. Desense curve of a synthesized ham 440 MHz receiver.
mobile service,
but unfortunately the front end of such a receiver is wide open to off-channel
interference when used in a repeater. Repeater builders should try to avoid this
type. To state this as a simple rule of thumb, front-end receiver selectivity is
much less expensive than having to add more cavities to a duplexer.

Fortunately at the time of the writing of this book, there are still many late-model
crystal-controlled commercial mobile transceivers on the used commercial radio
market. Thousands are still in regular service. Look for one when you are
building a repeater. Avoid fully-synthesized receivers, especially those made for
ham mobile service. They are normally a poor choice. Again, isolation is much
less expensive in the front end of a receiver than as a higher-end duplexer or
more cavities.

The type of transmitter you select for repeater use is also important. We got a
clue of this above in mentioning the need for headroom in a duplexer. Two
transmitter characteristics are important (1) power and (2) purity.

Transmitter Power
 5-10

It is a simple law of physics, but the amount of duplexer you need is directly
related to the power of your transmitter. Take the time to look back at Table 1-1,
Chapter 1. It will be evident that a 10 watt transmitter needs 10 dB less duplexer
than a 100 watt transmitter. I'm amazed how many repeater owners do not grasp
this.

For example they assume that their duplexer will still automatically work well after
they have installed a power amplifier. More power ALWAYS theoretically requires
more duplexer. A duplexer that is entirely sufficient for a 10 watt repeater may
now badly desense if a 100 watt power amplifier is installed. We can’t escape
basic physics. Though if you have made the measurements above, you will know
if you have the headroom to tolerate a power amplifier.

Transmitter Purity

Transmitters, like receivers are not perfect devices. In Camelot, transmitters

ONLY produce power on their center frequency. Unfortunately, in the real world,
they don’t. Considerable off-center-frequency energy always exists. Notice figure
5-4. It is a spectrum analyzer display of a portion of the FM broadcast radio
spectrum containing several transmitters. Notice that each transmitter (the
spikes) creates energy over a range of frequencies, not just on the carrier
frequency.
Transmitter energy does not exist just on the carrier frequency. In other words,
transmitters also have bandwidth. Further, by its very nature, modulating the
carrier with audio, for example, creates sidebands – more off-carrier energy.

Returning briefly to the basic architecture of a duplexer for a moment, you should
now be able to grasp why the transmit side of a duplexer must also have notch
cavities. On the receive side. the reason for the notch cavities is simple. The
receive-side notches suppress the transmitter’s center frequency. But what do
the transmit-side notches do? They suppress off-carrier transmitter energy which
ends up on the receiver’s center frequency. This explains why, as the little tuning
chart in chapter three specifies, one tunes the transmit-side notches to the
receive frequency, just the analyzer
Figure 5-4: Spectrum opposite of the
display of receive-side notches.
part of the FM radio broadcast band

As we said earlier, we affectionately call the energy that exists off of the
transmitter’s carrier frequency "dirt." Much to our dismay, some of it always is
present on the center frequency of the receiver. Transmitters very a great deal in
how much dirt they generate, but getting rid of transmitter dirt is the primary job
of the notch filters on the transmit side of the duplexer.

So again we ask the question, how well must the transmit notch cavities work? It
probably won't surprise you, but the answer is similar to the notches on the
receive side. It depends on how strong the dirt is, and how selective and
sensitive the receiver is. Sound familiar?
 5-11

Let's expand this just a little for emphasis. Dirt is definitely weaker than the signal
the transmitter is generating on its carrier frequency. But it is always still far too
strong not to need to be notched out on the transmit side. Remember, the
receiver is very sensitive. The dirt only has to be 0.22 micro volts for the receiver
to hear it quite well. Again by using Table 1-1, Chapter One, we discover that
0.22 micro volts is a mere 0.000,000,000,000,001 Watts. The cleanest
transmitter in the world makes a lot more dirt than this on the receiver’s
frequency, even with the repeater’s frequency split. That's again why we need
notch filters on the transmit side of the duplexer. They remove transmitter dirt.

Therefore, just like on the other side, we need to know the strength of the dirt
before we can specify how much isolation we need from the transmit-side
notches. To measure dirt, a spectrum analyzer IS required.

Transmitter Dirt Measurement Procedure

1. Connect your transmitter directly to the input of the spectrum analyzer,

through an adequate attenuator pad. This is required to protect the spectrum
analyzer’s front end from the transmitter’s power and to keep the load
impedance on the transmitter a constant 50 Ohms.

2. Set the spectrum analyzer to display just a bit more than the frequency
split of the repeater.

3. Cause the transmitter to make power, but without modulation. Adjust the
spectrum analyzer to display the center frequency at roughly full scale. Note
the reading in dBm.

4. Now observe the strength of the dirt at the receive frequency. Note how
many dB lower it is at the receive frequency than at the transmit.

Let’s say its 85 dB weaker than the transmitter, with an absolute level of -35
dBm. Our duplexer, therefore, only need to have enough isolation in the notch
filters on the transmit side to reduce the dirt to below the noise of -110 dBm in
our example. This means that the filters on the transmit side of our duplexer
would have to provide at least 75 dB of isolation. I’ve added another 10 dB for
headroom. Even so, 85 dB is much less than provided by a top-of-the line
duplexer.

As a final point on dirt, add-on power amplifiers normally generate more dirt than
barefoot transmitters. Therefore, if you add a power amplifier to your repeater, be
forewarned. You may need to add quite a bit of additional transmit-side notch
cavity isolation.

A Simple Desense Test

 5-12

If you don’t have access to a spectrum analyzer (most hams don’t) there is a
simple useful way to evaluate everything we have covered in this chapter. It is a
simple test all repeater owners should perform on their repeater(s). It won’t show
you where the problems lie, if they do exist, but it will tell you if your transmitter is
desensing your receiver.

In other words, if your repeater passes this test, it is working fine despite how
much power you are running, how expensive your duplexer is, what kind of
antenna your repeater has, if it has a receiver preamp or after-market power
amplifier. This test also won’t however tell you anything about the noise level at
your site. But it is a good starting point.

1. Make provision so that you can manually key the repeater’s transmitter,
instead of the receiver doing so via the carrier/PL operated relay.

2. With the repeater otherwise in normal operating mode, with the duplexer
well tuned, generate a weak signal on the input frequency. It should be at
a level that just begins to quiet the receiver. Again, a foil-wrapped hand-
held transceiver works well for this or a signal generator connected to a
small antenna.

3. Then, will receiving the weak signal, manually switch on the repeater’s
transmitter.

If the duplexer is properly isolating the receiver from the transmitter, you will hear
no desense in the weak input signal. If you do hear a change, you are
experiencing desense. If you have installed either a preamp or an accessory
power amplifier, try disabling it. If the desense goes away, you now have some
indication of what the desense is being caused by. I never take a new repeater to
a hill top until it can pass this test.

In review, isolation, insertion loss, receiver and transmitter performance are all
critical to knowledgeably work with. Mainly by not asking too much of any of
these we arrive that the most efficient and economical configuration for a
repeater. Many hams, and perhaps a few commercial operators, need to re-
evaluate the many old-wives tales that exist in this arena.
 6-1

Chapter 6 – Duplexer Tuning

We’ll now assume that you have built a complete duplexer or that you have a
commercial unit to tune. How do you do it correctly? It’s quite easy, even in the
field. Again, a spectrum analyzer with a tracking signal generator makes things
easier, but you can definitely also do it with the basic setup.

First, review the basic architecture of a duplexer, Figure 6-1. It is essential to

keep it in mind as you are tuning. I’ve seen many an ignorant novice badly
misadjust a duplexer by not paying careful attention to this basic structures.

Again, there are two sides,

receive and transmit. Both
typically have two notch
cavities and a single
bandpass cavity,
configured as shown.
Figure 1. The Bp cavities
are adjacent to the
antenna input. Despite the
Figure 6-1: Layout of the average duplexer disposition of the cavities,
the tune-up procedure will
be similar.

Tune-up Procedure

Table 6-1 lists the basic rules for tuning a duplexer. Notice that there are four
combination. One pair applies to the transmit side, the other to the receive side.
At a casual glance both may seem the same. Look closer. The frequencies are
reversed. Hence there are four rules, one for transmit Bp, one for transmit Br,
one for receive Bp and one for receive Br. Note the distinct differences and the
pattern. Memorize these rules; they are absolutely basic and essential..

For the actual tune-up procedure, there’s

TX Bp Max. at the transmit freq. an overall rule that I consider mandatory
TX Br Min. at the receive freq. – tune each cavity separately, first .
Completely disconnect one cavity at a
RX Bp Max. at the receive freq. time from the duplexer by removing its
normal inter-cavity cables. Then tune it on
RX Br Min. at the transmit freq.
its own. The need for individual cavity
Table 6-1: Tune-up rules tuning is this: a mistuned cavity down the
line can very easily induce incorrect
 6-2

tuning in the one you’re working on. What’s more, individual cavity tuning will
almost always get you very close to an ideal over-all tune up. You’ll rarely need
to readjust cavities assembled as a complete duplexer.

When tuning a duplexer, it is not necessary to make a graphs as we did during

the design phase of the cavities. All you are doing now is adjusting for either a
peak or a dip. Here an MFJ-259, or similar, is easier to use than the simple
transceiver method. With it you’ll find the Br notch at maximum SWR, the Bp
peak at minimum SWR. Notice that this is the reverse of power.

Tuning is also a little easier, if you realize something else. Duplexer cavities are
symmetrical. In other words, they have no specific input or output. The two inputs
are the same and can generally be reversed. You may therefore connect the
Wattmeter and a 50 ohm load to either connector, and your transceiver or MFJ-
259 to the other.

FIRST, as Table 6-1 shows. tune the bandpass cavities for maximum through-put
power (or minimum SWR). Remember, always tune Bp first, it’s always vital. Also
don’t forget to select the correct frequency when you move from one side of the
duplexer to the other, again according to Table 6-1.

Then, only after you have tuned all the Bp cavities, tune the notch cavities on
BOTH sides. Again, as Table 6-1 shows, tune the notch cavities for minimum
power (or maximum SWR). Once again remember to change frequency as you
change sides. Notch cavities are more critical to tune. You could use a receiver
with an S meter for a finer adjustment. If so, place a 50 Ohm pad (20 dB or more)
in front of the receiver to prevent damage to its front end. The pad also insures
that the receiver presents a 50 Ohm load to the duplexer. Without proper loads
you WILL always misadjust.

Finally, after tuning all the cavities individually, reconnect the entire duplexer.
Now you may, if you still feel it is necessary, perform a fine adjustment of the
whole unit. DO NOT go straight to this without first tuning the cavities individually.
The importance of this can not be overemphasized.

For an overall fine tune, connect the Wattmeter and dummy load to either the
transmit or the receive port of the duplexer. Also terminate the other port with a
50 Ohm load. This is very important. Then connect your transceiver (or MFJ-259)
to the antenna port . Again following Table 6-1, first touch up the Bp cavity(s) on
both sides. Finally, touch up the Br cavity(s). Never “tweak” just the notch cavities
without going through the full procedure. Experience has shown me that notch
cavity adjustment in a fully-connected duplexer can be the pathway to disaster,
especially for the newcomer.

For the Home Builder

 6-3

In the following chapters we’ll now complete the picture, especially for those who
want to roll their own or modify a commercial unit for ham-band use. We’ll dive
into loops a little deeper and then into lines and losses. Even for those who do
not choose to build their own, these final chapters involve important
comprehension principles.
 7-4

Chapter 7 – Loops
Before I began my study, the coupling loops in cavities were especially
mysterious. I could not find much about how to design or modify them in the
published literature. Why were there so many variations? I’d seen fat loops, thin
loops, wire loops, strap loops, side loops and top loops. Why had the designers
made these choices? Is one better than another for the home builder, I
wondered? I couldn't see a pattern. That's why I devoted a great deal of time to
coupling loops in my early experiments.

First, is the shape critical, and what about the placement? Do either of these
require great precision? How about loop material? How important is that? My
intention in answering these questions in this chapter will be to take you through
the experiments that gave me the answers. The principles here present a
practical picture of how to design your own loops for peak performance.

Loop Shape

My first question was, is there a magic shape for cavity coupling loops? As I
mentioned, I had inspected many duplexers, and the loops came in a baffling
variety. I wanted to know what effect loop shape would have on duplexer
performance. So I built a cavity, something like the one illustrater in an earlier
chapter and I began to experiment.

When I first began experimenting with loop shape, I kept confusing several
factors. For examples, two loops of different geometry have a different
inductance even if you use the same amount of wire. They also have a different
geometric center. Since the magnetic field in the cavity is not uniform, two
different loops will couple differently to the field.

Fortunately, though, I found a way around this. Just keep changing two loops
until they exhibit the same bandwidth and losses and you eliminate everything
but loop shape. So using this technique I then tested circular loops, rectangular
loops and loops of irregular shape. What I discovered was that the shape of the
loop makes little difference, provided it is made to perform the same as a loop of
another shape.

This led me to realize that the only characteristic of a coupling loop that matter
very much is the total area of the. I later found in an engineering book that
coupling is proportional to the square root of that area. Therefore, if two loops
have the same area, they will perform much the same place in a cavity even if
shape and the amount of wire is quite different. This simple generalization has
limits of course, but for practical purposes, loop shape and size is not significant
 7-5

factors in cavity design. Only the area of the loop and how it is oriented in the
cavity determines how much it will couple to the magnetic field.

Where Do You Put the Connectors?

Another factor I wanted to know about, was where to put the connectors attached
to the loops? I had seen a lot of variations in commercial and amateur-built
duplexers. Two locations seemed to be common. The connectors were typically
installed either in the shorted end of the cavity or a short distance down the side
wall. Is one better than the other, or will it change how well the cavity performs, I
wondered?

Once again, the answer to these two question is no. For any given loop, it does
not make any significant difference whether its connector is in the side of the
cavity or in the end. As long as it ends up in the same place in the cavity, it
performs the same.

So why the difference in connector location in cavities? Why should you select
one over another? It is my opinion it is only a matter of convenience. If it is
handier for the connectors to be in the side, then put them there. Where you put
the loops in a cavity also does not turn out to matter very much. We’ll see this in
a moment. But as far as connector position goes, if a designing a single cavity
application I often employ the side position. For a group of cavities, such as for a
duplexer, the end is generally easier. Though take your choise.

Loop Construction Materials

Next I wanted to know if the piece of metal or wire used to form the loop matters.
I knew for example that conductors of different dimensions have different
characteristic impedances when used in transmission lines. Does this affect the
loops in a cavity? For, example, in that the loop in a cavity is fed with a 50
transmission line, does it also perhaps have to look like a 50 ohm line section?

Again I began experimenting. I tried round wires of widely differing diameters. As

before, I adjusted all factors until the performance of each loop was equal to
those of another loop under comparison. Then I tried flat metal straps bent into
loops. I did this in that I had seen loops made of flat strap in commercial cavities.
But again, after trying all these variations, while always keeping performance
equal, I came to the conclusion that loop conductor size, shape or material has
little significant affect of loop performance. Ordinary wire is perfectly acceptable.
In fact, it is probably the best choice.

There is, however, one factor that does matter in the material used for loop
construction – current handling capacity. Notice Figure 7-1. It shows RF current
in the transmission lines (and loops) of a repeater at different power levels.
 7-6

These values may not seem high if you think in DC terms, but RF conductors
need to be much larger conductors due to
1 watt .14 Amps skin effect. We will go into the problems
3 watts .25 Amps caused by skin effect in a later chapter, but as
10 watts .44 Amps a general rule, above about 100 watts, loops
30 watts .77 Amps should be made from heavy wire. Below that
100 watts 1.4 Amps power level, 16 AWG wire is completely
300 watts 2.5 Amps adequate.
1000 watts 4.4 Amps
Table 7-1: Loop current vs. Flat strap actually turns out not to be a good
transmitter power at 50 Ohms choice. It has skin effect problems compared
to round wire. But it’s easier to bend into loops
for cavities used at higher power levels, and if made a bit larger, it works fine. It is
a common choice in commercial cavities.

Loop Placement

In an earlier chapter we saw how a loop must be placed in a cavity in order to

couple to the magnetic field. The H or magnetic field, as you will recall, lies
concentrically around the center conductor. The loop couples best when it is
perpendicular to the H field. This is along the cavity’s length and radial along its
diameter. We also learned that the magnetic field is strongest near the shorted
end of the cavity and close to the center conductor.

If, however the loop is moved within the cavity or it is rotated compared to the
magnetic field, coupling will be less. But does that matter? Must loops always be
perpendicular to and placed where the magnetic field is maximum in order to
work well? I spent a lot of time researching this too, and once again concluded
that the answer is no. I experimented with loops at several locations, near the
shorted end, away from the shorted end, near the center conductor and away
from the center conductor.

I also experimented with rotating a loop so that it was not perpendicular to the
field. In fact, making a loop so that it can conveniently be rotated from outside a
cavity is a useful feature found in some commercial cavities, especially large
bandpass cavities. To accomplish this, the connector and loop are installed on a
small circular plate. The plate can be rotated in the field and locked down with a
set screw. This adjustment permits the user to determine how much insertion
loss and bandwidth the cavity will exhibit. As we saw in an earlier chapter, the
decrease in bandwidth caused by looser coupling can be a big asset.

To study the effect of loop location, I would merely keep changing the area of the
loop in a different location until it would again couple equally to the field present
at that location in the cavity, And when I did I would obtain performance equal to
any other location. This includes both insertion loss and bandwidth. So my final
conclusion was, that as long as the total amount of coupling between the loops
 7-7

and the cavity is made to be the same by adjusting loop area, shape, material
and orientation, the position of the loop within the cavity, loop position does not
matter, at least within wide practical limits.

Loop Grounding

Armed then with knowing how tolerant loops are, it then came to me as no
surprise that it also does not matter much how you ground a loop. That's
probably why I again saw many variations in commercial and amateur designs.
The three common configurations are shown in Figure 7-1.

As we have learned, only the area of the loop matters. For example, in the left
cavity of Figure 7-1, a section of the loop is actually a part of the cavity's outer
wall. The area contained by the wall and the remainder of the loop performs the
coupling. My personal favorite for loop grounding is the right configuration. If you
ground the loop to the body of the connector that feeds the loop, the loop and the
connector can be made easily removable and rotatable. The convenience of this
method makes it a very common grounding configuration.

All this leads to a Golden Rule.

Nothing about loops is critical.
All you need do is to alter the
area and the orientation of the
loop until it correctly couples to
the magnetic field and you will
obtain the same results for a wide
range of cavity locations. I like a
little analogy here. Imagine you
are pushing a child on a swing.
Where on the ropes should you
push? Actually anywhere is fine. If
you push at the bottom, you only
need to push gently, over a long
Figure 7-1: Loop grounding, end, side, to connector distance. If you push nearer to the
top, you’ll have to push harder but
over a shorter distance. This is a perfect analogy to the size, placement and
orientation of the loops in a cavity.

Feel free to experiment with the loops in the sample cavity, it is excellent
instruction. They are only mounted on the ends, on the baking pans, for simplicity
and to permit easy rotation, should that be desired. A side-mount position would
work well too. It would, however, not likely be as convenient for use in a complete
duplexer.
 8-1

Chapter 8 – Losses
This is undoubtedly the most important chapter. It contains the hidden secrets.
Wrapped up here are the major compromises that create optimum performance
with maximum economy. We’ll talk about energy transfer, skin effect, cavity
proportions, and most important of all, bandwidth vs. insertion loss. These are
vital concepts.

Energy Transfer

How does energy, our receiver or transmitter signal, get into and out of a cavity?
It works the same way for all cavity types, Bp or Br. Energy enters from an input
transmission line most often via an input loop. The loop, for all practical purposes
is just a small inductor. It is more complicated then that, but this is an adequate
approximation.

The magnetic field around the loop excites the cavity into oscillation. As we said
earlier, it is like blowing air across the top of a soft drink bottle. In so doing, the
cavity absorbs the energy delivered to the loop by the input transmission line. At
the output loop, provided it has been configured to have the same performance
as the input loop, the complement takes place. The energy returns to the
transmission basically unchanged, little is lost except for that which we don’t
want.

Said in simpler terms, what goes in at the frequency we want comes out. Said
another way, a cavity should essentially be invisible to energy passing by at the
resonant frequency. At other frequencies, however, the losses are made as high
as we can. That is the basic idea of a cavity filter – keep the good stuff, get rid of
the junk.

But unfortunately cavities are not perfect. They always loose a little of the good
stuff even at the resonant frequency. It is unavoidable and is called insertion
losses. Not all the energy that goes into the cavity at the resonant frequency
comes out. Understanding what causes the loss and how to minimize it, though,
is the main subject of this chapter. Knowing how is vital to home-brew design in
particular.

Skin Effect

The biggest cause of cavity loss is conductor loss on the inner surfaces of the
cavity, particularly as it is affected by RF skin effect. In most RF devices, skin
effect is not a problem, but in a cavity it is a very major factor. Energy on the
transmission line coming to the cavity becomes RF current on the surface of the
inside of the outer wall of the cavity and also on the center conductor. If these
 8-2

surfaces could be resistance free, that is perfect conductors, there would be

virtually no insertion loss. But the walls of all real cavities do have real resistance,
more than we realize. It is what causes insertion loss in a duplexer primarily.

As many hams know, the resistance of a electrical conductor increases as the

frequency increases. A conductor carrying RF exhibits more Ohms of resistance
than if it were carrying DC current. The reason is, as the frequency increases, AC
current in a conductor moves to the surface of the conductor. This is called skin
effect. As a result, deeper in the conductor there is less current for RF. At low RF
frequencies the effect is small, but at VHF and UHF skin effect is quite
significant.

Current in the wire creates a magnetic field

concentrically around the wire. But because the wire
can never be 100% conductive, there also exists a
weak internal magnetic field. This internal field links to
the current flowing in the wire causing a progressive
increase of inductance toward the center. The net result
is that the current is forced outward in the conductor.

The net result is that the effective cross section of the

conductor is less electrically than it is physically. Notice
the current density graph in Figure 8-1. It illustrates that
no current at all exists at the center of the wire shown.
Another way of looking at this is to say that the current
tries to redistribute itself to be encircled by as few
magnetic field lines as possible.
Figure 8-1: Current
distribution in round
wire (black). No current In conductors that do not have a round cross section,
at center/ skin effect is even worse. Round conductor are the
least affected by skin effect. Notice Figure 8-2

What significance does this have? Simply this, round conductors are best for RF.
It is counterproductive to make the coupling loops of a cavity from flat metal
strap. Commercial manufacturers often do use strap loops. This is only because
they are mechanically easier to make for high current cavities. I prefer round
conductor loops. Similarly,
square cavities are less
efficient than cylindrical
cavities. The effect is not
great, however, and
successful duplexers can be
made by soldering flat sheets
Figure 8-2: RF current and magnetic field in flat together.
conductor. Current mostly at ends.
Another case in point, not
 8-3

related to duplexers, where flat metal straps are definitely not a good idea, is for
compact transmitting loop antennas, sometimes called magnetic loops. These
perform better made from round tubing, precisely for the same reason.

How bad is skin effect?

Skin effect is definitely a major problem in duplexers. It is significant at VHF and

higher at UHF frequencies. Above that it becomes extreme. Let’s look at some
real numbers. The complete equation for calculating skin effect resistance is
complex, but it may be simplified to:

One skin depth (inches) = 0.0026 / square root of Frequency (MHz)

To avoid calculation, Table 8-1 is this equation solved for RF frequencies 1 MHz
through 1 GHz.

We consider a skin depth to be the depth at which

Frequency Skin Depth the current has decreased to 36.8%. Obviously
1 MHz 0.0026 in. some current is still flowing at one skin depth, but it
3 MHz 0.0015 in. will again decrease by another 36.8% for every
10 MHz 0.00082 in. additional skin depth. In practice we consider the
30 MHz 0.00047 in. current below three skin depths insignificant. Quite
100 MHz 0.00026 in. close to the surface is where most off the current is
300 MHz 0.00015 in. flowing.
1 GHz 0.000082 in.
At 450 MHz, skin depth is 0.00012 inches.
Table 8-1: One skin depth
vs. frequency Therefore, the effective thickness of the
conducting surface is only .00036 in. Copper and
aluminum may be good conductors, but for a conductor this thin there is now
appreciable resistance. Multiply these figures roughly by three at 2M

Surface Treatment

A common way to reduce skin effect losses is to electroplate the inside of the
cavity with a more conductive metal. Since most of the current flows on the
surface this technique can be effective, but generally only at or above UHF
frequencies.

Silver, and sometime copper are the only practical choices for plating, and
neither is possible on aluminum. Surprisingly, gold is not a worthwhile option
here. As we saw earlier, gold is only used on connectors for corrosion resistance
not conductivity. Copper is a better conductor than gold, and silver is only
marginally better than copper.
 8-4

For ham use, the cost and the marginal benefit of silver plating is impractical.
Below roughly 1 GHz the thickness of the plate required is too great. At 450 MHz
and below, especially for home construction, bare copper is perfectly adequate.

However, at lower frequencies there is one case, where electroplating can be

effective. That is copper on steel. Unlike silver, as a plating material, copper is
relatively inexpensive even if heavy plating is required,. Several commercial
manufacturers make excellent cavities out of copper-plated steel. Steel is
otherwise totally unacceptable as we saw due to poor conductivity. I have
personally not attempted copper plating at home, though I suspect it could be
successfully done by some home builders.

Corrosion of Copper and Aluminum

What about the corrosion that forms on copper or aluminum? Won’t that reduce
the effectiveness of a cavity? I had the same question when I first started making
cavities. So I did an experiment. I took one of my copper UHF cavities, and
measured its performance while it was still very dirty. I hadn't cleaned it up after
assembly. It was black with copper patina.

Then I polished its inner surface to a mirror shine. I was dumbfounded to

discover that the cavity’s performance remained the same. Apparently the patina,
as it is called, that collects on the surface of copper is not a problem in cavities,
most likely because it is very thin and is also a conductor. Skin effect works for
us here. As a result, and since that time, I only give my cavities modest cleaning
merely for aesthetic reasons.

Aluminum is more or less the same, though the nature of the surface corrosion
on aluminum is quite different. Copper patina is conductive, aluminum corrosion
is an insulator. It is aluminum oxide, the very same substance rubies and
sapphires are made of. Aluminum corrosion also has no noticable effect on the
RF currents running on the inside surfaces of a cavity.

Cavity Impedance

The second major loss issue in duplexer filters is induced by the characteristic
impedance of the cavity itself. Remember, a quarter wavelength coaxial cavity is
just a section of large-diameter open-air transmission line. Like any transmission
line it has a characteristic impedance. For air-insulated circular coaxial
transmission line, the characteristic impedance is determined simply by the inner
to outer conductor diameter ratio.
 8-5

An early question for me was, does the characteristic impedance of a cavity have
anything to do with its performance in a duplexer? That is, is the inner to outer
diameter ratio
important? And the
answer is yes, quite a
lot. I was surprised to
find out that there
actually is a "magic”
characteristic
impedance for a
cavity. Figure 8-3,
though surprisingly it
isn’t 50 Ohms, but 77
Ohms. At the specific
inner-to-outer ratio for
Figure 8-3: Inner to outer conductor diameter radio vs. relative loss 77 Ohms, cavity loss
is lowest.

Notice that insertion loss is lowest when the inside of the outer conductor is 3.6
times as large as outside of the inner conductor. Again this translates to a
characteristic impedance of roughly 77 ohms.

Table 8-2 is the 3.6 to 1 ratio calculated for commercial aluminum cake pans with
the copper pipe size to use for the center conductor. As you can see from the
graph, small variations do not introduce significantly greater losses. Also, the
price of copper pipe takes a sharp
Pan Size Center Conductor Copper Pipe jump above 2 in. I stick to 10 in. or
12 in. 3.3 in. 3 in. smaller cake pans, even for 6M or
10 in. 2.8 in. 2 ½ in. 10M cavities. For example, the
9 in. 2.5 in. 2 ½ in. inner conductor of our sample
8 in 2.2 in. 2 in. cavity is 1¼ in. copper pipe. This is
7 in. 1.9 in. 1 ½ in. a little small, but is close enough. If
you intend to build an actual
Table 8-2: Cake pan vs. center conductor pipe size duplexer from this design, 2 in. pipe
will yield slightly better insertion
loss.

As a sidelight, the graph also illustrates why 75 ohm coax is normally used for
receiving purposes, such as TV cable. At this characteristic impedance, coax has
the lowest losses? Why then, you may ask, do we use 50 ohm coax for
transmitters? Shouldn’t they also use the lowest loss coax? Yes, they should if
one only considers loss. 75 Ohm coax can't handle nearly as much power as 50
Ohm coax. The best power handling capacity in coaxial line is achieved at
roughly a 30 Ohm characteristic impedance. 30 Ohm cable is difficult to
manufacture, so 50 Ohm cable has become the accepted compromise for coax
carrying RF power. In a cavity, however, losses are extremely important. So we
 8-6

want to use the optimum impedance of 77 Ohms, again an outer-to-inner

diameter ratio of 3.6 to 1.

When I learned about this magic ratio, I wondered, "Doesn't that create a
mismatch between the 50 ohm coax and the cavity?" Yes it does, but it does not
matter. I'll have more to say on this later when we talk about lines, but for the
moment, let me state another basic cavity principle. Any mismatch that takes
place at the input of a cavity is reversed at the output, that is provided both loops
are equally configured.

Bandwidth Verses Insertion Loss

Next we come to the most important duplexer concept of all, bandwidth vs.
insertion loss. If everything were perfect, a duplexer would pass only the
frequencies we wanted and totally reject all others. What's more, there would be
no insertion. In other words the bandwidth would be extremely narrow and the
insertion loss would be zero. As we know, this does not happen in practice.

It is, though, possible to have reasonably narrow bandwidth. Bandwidth is directly

proportional to the Q of the cavity. The Q is determined mostly by skin effect
losses. If we assume that we have the ideal conductor diameter ratio of 3.6 to 1,
a reasonable approximation for the Q of a copper cavities without any load is,

Q = 107 x Diameter (in.) x [root-2 Frequency(MHz)]

From this we can determine the bandwidth by,

Bandwidth = Frequency / Q

Table 8-3 lists value calculated from these equations vs. outer conductor
diameter, at 450 MHz.

Notice the bandwidth figures. You may be surprised that they are so narrow. If
you've had practical experience with duplexers, you probably were expecting
bandwidths of Megahertz not Kilohertz, and you’d be right.

The reason for the difference is quite simple. The values are for unloaded
cavities. In use, duplexer cavities exhibit far poorer bandwidth because they are
loaded by the external equipment connected to them. In a real duplexer, there is
a 50 ohm load on both the input and the output. One is the 50 ohm load of the
antenna at the tee junction. At the other ports, the duplexer sees the 50 ohm load
of either the receiver or the transmitter. Each individual cavity also sees the same
loads, since in a well designed duplexer the cavities are more or less
transparent, as we learned above.
 8-7

Therefore, every cavity is

doubly loaded by 50 ohm
Diameter Q (unloaded) Bandwidth(MHz) loads. What this does is to
1 in. 2300 195 KHz establish a new effective
2 in. 4500 98 KHz working Q for each cavity
3 in. 6800 65 KHz called the loaded Q. It is
4 in. 9100 49 KHz far, far less than the
5 in. 11300 39 KHz unloaded Q given in Table
6 in. 13600 33 KHz 8-3. That's why the
Table 8-3: Q vs. bandwidth for 450 MHz copper cavities with working bandwidth of
a 3.6:1 ratio duplexer cavities is much
greater.

The Effect of Coupling

Before we can get an idea of how much the external loads reduce the unloaded
Q of the cavity, however, we must look at another process that happens inside a
cavity, that is, coupling. The 50 ohm load impedance created by an external
device is fixed, but the amount it actually loads the cavities is not.

The amount of loading depends on the arrangement of the loops in the cavities.
You will recall from an earlier chapter that the size and orientation of a loop
determines how much it couples to the cavity. If a loop is made large enough and
correctly positioned, it will place the entire external load on the cavity. If,
however, it is made smaller than this, or is rotated away from being perpendicular
to the field, it will not place the entire external load on the cavity.

This leads us to a most important

cavity concepts: the tighter the
coupling, the wider the bandwidth.
Notice Figure 8-4

Well if that’s true, you may be

thinking, then let's never use tight
coupling, and the rejection in our
duplexer of unwanted signals will
always be good. That's a valid idea,
except for one thing – and it leads us
to another important principle: the
less the coupling the more the
Figure 8-4: Bandwidth and insertion loss insertion loss. Notice Figure 8-4.
vs. coupling. for over coupled, critically Here I show the same cavity with the
coupled and under coupled. loops progressively adjusted for
loose (under) coupling, optimum
 8-8

(critical) coupling and tight (over) coupling. Notice that bandwidth decreases as
the insertion loss increases. We can't have both – it's simple physics.

If we did not care how much insertion loss our duplexer has than under coupled
cavities would provide all the Q we would ever need for good isolation between
receiver and transmitter and protection against the neighbors. But if we don't
want to lose receiver sensitivity and transmitter power, we need to use critical
coupling and the largest cavities we can afford. In review, remember this. The
loaded Q of a large critically-coupled cavity can be the same as that of a small
under-coupled cavity, but the insertion loss will be less.

A Basic Compromise

It should be obvious then, that the correct balance between bandwidth and
insertion loss needs to be established in every duplexer. If we try to achieve for
too little insertion loss, bandwidth and isolation will suffer. Anything tighter than
critical coupling is counterproductive. The small improvement in insertion loss is
not worth the price of the increase in bandwidth.

Therefore, when one is building a duplex for general purpose applications, you
should set the loops for critical coupling. Most commercial duplexers are this
way. I also begin with In my home-brew designs I make the loops a little larger
than necessary to achieve critical coupling when they are perpendicular to the
magnetic field. Then all I need do is to rotate them slightly to achieve critical
coupling. As we learned earlier, rotating the loops does not degrade the basic
performance of a cavity, it merely reduces the amount of coupling.

What is a “Reasonable” Insertion Loss?

A very significant point here is to ask how much loss should we expect to have at
critical coupling? This is where the uninformed duplexer may hear
misinformation. Hams often strive for too little
Diameter Insertion loss insertion loss. By now you should realize that
3in. 2.2 dB learned, insertion loss is determined almost
6 in. 1.1 dB entirely by skin effect, which is a function of
9 in. 0.75 dB cavity diameter and cavity construction
12 in. 0.6 dB material. The informed user realizes that
Table 8-4: Insertion loss for 2M insertion loss is a fact of life. It is unrealistic
aluminum cavities. to expect otherwise. Table 8-4 gives my
personal rough guidelines, based on
experience, for aluminum cavities for 2M. For copper cavities at 450 Mhz, divide
the diameters by three and the insertion loss by the conductivity difference
between aluminum and copper of 1.5.
 8-9

The figures I have given are for loaded cavities installed in a completed duplexer.
Therefore, a small amount of the loss included here is due to other factors. With
care you may be able to achieve slightly better results.

The Next Step

Now that we have a starting point, critical coupling, we are going to take another
step that most duplexer users will complain loudly if told to do so. We are going
to increase the amount of insertion loss, to knowledgeable decrease the
bandwidth of our duplexer. The object is achieve the best balance between
economy and loss. For as we’ve seen, an oversized, overpriced critically-coupled
duplexer will always work, but it is often not the best economy. This of course will
evoke an attitude problem for some repeater owners. They do not like ANY
increase in insertion loss, even if it will bring benefits.

Let’s begin on the transmitter side. Again hams probably won't like this, but the
two-way industry is generally willing to accept a 3dB loss of transmitter power to
achieve the correct bandwidth. Yes, that's giving up half of the transmitter’s
power in the duplexer. Admittedly, that's not what we’d like, but it truly is an
acceptable compromise. Amateurs want “absolutely every microwatt” of power
they can get. Commercial operators usually know better.

So unless you have all the money in the world and have all the space you need
in a repeater cabinet, transmitter insertion loss of 3dB is acceptable in practical
repeater installations. Often we can do better, but don’t feel “put upon” if you
can’t.

The reason this is so is because losing half the power in a worse case situation is
really not that bad. Especially at UHF, but even at VHF, the service area of a
repeater with only half the power is actually very little different than it would be at
greater power. Terrain is a far bigger factor than power. That's why most
commercial repeaters are less than a hundred watts. Modest power is really quite
adequate. A kilowatt isn’t necessary. So neither is there any need to cry over
losing half of your power in a duplexer if it will net significant gains in economy
with performance. A couple extra watts isn’t worth the price.

The same is true for the receiver side of a duplexer, except that acceptable
insertion loss can surprisingly sometimes by much greater. Again we would like
as little loss as possible, but once again it is not always that important. I've seen
several cases where intentionally increasing the insertion losses on the receive
side beyond of 3dB has improved the repeater's actual performance quite
noticeably. It’s ALWAYS an option worth investigating.

As we saw in an earlier chapter, the average level of band noise at your repeater
site is the biggest determining factor of how much sensitivity you can practically
use. For example, in chapter 1, we saw that if the average noise level at your
 8-10

location is -110 dBm, then all you need is an overall receive sensitivity of roughly
0.71 microvolts. But if your receiver has a natural sensitivity of 0.22 microvolts,
like our sample receiver, simple math says that you can afford to take a big 10
dB loss in insertion loss on the receive side of your duplexer without any loss in
system performance whatsoever. Again, don’t get upset at having to give up 3dB
on the receive side of your duplexer to achieve good filter action.

Never forget that if you intentionally reduce overall system sensitivity by admitting
greater insertion loss on the receive side, you will always instantly improve
system bandwidth. So don’t strive for the last microvolt of gain and the absolute
minimum of insertion loss in a duplexer. It is not always the best choice. You will
gain much more by using only enough gain and as much insertion loss to let your
receiver hear efficiently, and no more.

I can recall the first repeater site on which I saw this principle applied. I had
adjusted my home-brew duplexer for the absolute minimum insertion loss on the
bench. When I put it on the hill top it worked okay, but I was experiencing a little
interference from a nearby paging transmitter. So reluctantly, I twisted the loops
in the receive bandpass cavities for an increase of insertion loss to eliminate the
paging transmitter. The interference immediately went away, and to my great
surprise the repeater could now actually hear weaker signals. Case in point.

Increasing receive-side insertion loss in a duplexer in order to narrow the

bandwidth is always a compromise worth considering. It is much like adding
another cavity. And if you are using a preamp or an outboard power amplifier, it
can be worth its weight in RF gold. Duplexer insertion loss is NOT always your
enemy.
 9-11

Chapter 9 – Inter-Cavity Lines

This chapter will probably be the most mysterious if you’ve ever tried to modify a
duplexer for a different band. Of all the parts of a duplexer, the lines that connect
the cavities and those that connect the duplexer to other devices, generate many
questions. They also spawn many old wives tales. Fortunately, we’ll learn that
determining the length of the lines is actually quite easy, especially if we employ
a practical approach.

In this chapter we will look at the types of cable to use and the basic theory what
the line lengths between the cavities must be. We’ll then explore a practical way
to determine those lengths easily. Finally we’ll see what to do with the lines that
connect the duplexer to the repeater.

Line Type

Like almost everything else in a duplexers, my experiments demonstrated to me

that the type of coax that you use to couple the cavities together is not critical.
Only three things matter. The first is shielding. All the coax that you use in a
repeater must be 100% shielded. It is basically a necessity. Otherwise all of the
effort to isolate your receiver from your transmitter or from the neighbors can
easily be lost by direct pickup through the shield of the coax.

Ordinary coax, with a single braided shield, the kind we most often use, is
inadequate. It is too leaky. Only two types of coax are satisfactory for repeater
use, double-shielded flexible coax or so-called hard line. Most of the common
varieties of flexible coax are available in double-shielded versions. The only
difference is that they has two layers of either braided wire loom as the outside
conductor instead, or one of braid and one of foil. Double-shielded coax
essentially 00% shielded.

Hard line, either rigid or semi-flexible is also totally shielded. Its outer conductor
is solid copper or aluminum tubing not braid. Hard line, though, is impractical to
use inside a repeater or between the cavities of its duplexer. We normally employ
is only for external feed lines. Its main benefit is its ability to withstand weather.
Braid-shielded coax with a single or a double shields over a period of time allows
moisture to pass through. This eventually corrupts the dielectric of the coax.
Except for a few high-quality coax types, losses increase rapidly in braid-shielded
coax exposed to the weather at VHF and UHF. Inside a repeater, though, braid-
shielded coax is no problem as long as it is double shielded.

The second factor to consider in the type of cable that you select, is the
connectors. Especially for the home builder, this can increase costs considerably.
You can find connectors to fit almost any type of cable, but the cost varies
 9-12

dramatically. You won’t need many feel of cable to build a duplexer, but you will
need quite a few connectors. Shop for connectors carefully.

The exact type that you use is not critical, as long as it is correct for the cable
type. I personally prefer crimp-on connectors for convenience and durability, but
the tool to attach them does add some expense. The screw on type are totally
satisfactory, however, even if obtained from a local bargain retail electronics
outlet.

I recommend two types: N and BNC. If you are running roughly 50 watts or less,
BNC is the best choice. They are moderately priced and have good RF
characteristics. N connectors are better for higher power and similarly have good
RF characteristics. Do not, however, be tempted to use SO/PL 239 connectors
above 2M. Even at 2M they are marginal. Such are common on RF devices, but
can have poor RF characteristics. One or two in a repeater is not usually a
problem, but try to avoid them if you can. Many commercial duplexer
manufacturers do use them, but generally they select SO/PL 239 connectors
made from special materials. The local radio store variety can cause grief.

The third consideration in selecting cable is power handling capacity. As with

connectors, for power levels under roughly 50 watts, double-shielded cable
similar in size to common RG-58 is quite satisfactory. I use an economical variety
of foil double-shielded RG-58 that is similar to the cable developed for cable TV
applications. Most cable manufacturers now offer it. My point is that almost any
type will work. If you wish to use the expensive silver plated double braid
varieties, that's fine, but less-expensive types work just as well.

Determining Line Length

We now come to a very important issue of comprehension for the duplex builder
or the home repeater builder wishing to retune a duplexer to another band. It is
how we determine how long the lines should be between cavities. It is not
common knowledge. And the best part is that it will lead us to a simple practical
way to quickly configure the lines between cavities. It will also explain why one
cannot simply proportionately rescale the lines for a change of bands.

Here is the concept in a simplified nutshell. We need to make the electrical

pathway from one cavity to the next “disappear.” In other words, we must “tune”
the length of the line until the two cavities think there is no line between them.

To accomplish this perhaps-seemingly difficult feat, we employ a very basic fact

about transmission lines, including coax. That is, ½ wavelength of transmission
line does not change the impedance of what it is connected to it on one end, at
the other end. This is of course ½ wavelength electrically, not physically. The
velocity factor of the line must be factored in.
 9-13

But because whatever we connect to one end of an electrical ½ wavelength of

transmission line also appears unchanged at the other end, effectively, as we
require, the line does not exist. This simplification ignores losses in the ½
wavelength line section, but for the small involved, the losses may be ignored.

So theoretically we need to make the total length of the pathway between the
cavities such that the cavities think they are directly connected to each other,
essentially with no line in between them. For when we do this, the cavities
behave as they should and only then.

But here is something very important. The ½ wavelength pathway we need is not
made up just of the coax between the cavities. It also partially includes of the
coupling loops at both ends to some degree. Notice Figure 9-1. The ½
wavelength pathway is roughly the total length shown as the dotted line in the
figure, not just the electrical length of the inter-cavity coax.

The full theory is more complex than this, but for the technique we will use to
actually constructing the lines, this simplification is quite adequate. The practical
procedure we will use, bypasses the theory. I give it merely to help the reader
visualize what is going on with inter-cavity lines.

In properly inter-connected
adjacent cavities, the total
length of the dotted line, Figure
9-1, should appear to be ½
wavelength. Notice that the
coax is only a part of the total
length. The loops must also be
included.

Realize then, that if you need to

reconfigure a duplexer for
Figure 9-1: An inter-cavity line. Total electrical length
another band, the length of the
= 1/2 wavelength inter-cavity lines is almost
impossible to compute. It isn’t
just a matter of simply rescaling the coax lengths proportional to frequency. This
will severely upset the filtering properties of the duplexer. Insertion loss and
isolation will suffer.

Instead, the best way I have found to easily determine the correct length for inter-
cavity lines is a pragmatic approach. I never attempt a mathematical solution.
Actually, the method I use is also used by some commercial duplexer
manufacturers.

Make up a set of a dozen cables of the same type of coax and connectors that
you’ll be using in the final configuration. Make each cable just a little longer in
 9-14

small equal increments. The total set needs to span a full ½ electrical
wavelength. For example, ½ wavelength on 440 MHz is roughly 12 inches.
Multiplying this by the velocity factor of the cable I used (0,6), twelve cables in 1
inch increments is a suitable. For 2M, 3 inches increments are fine. Label each
cable for easy identification. Make the shortest one roughly 18 in. long for 2M, 6
in. for 440 MHz. Any beginning length is fine as long as it will fit between the
cavities.

To use the cable set, begin with either the longest or the shortest cable and
connect together two adjacent cavities. Each cavity must have already been
individually tuned to frequency. Do not alter cavity tuning during this procedure

On one side of the pair of cavities connect a spectrum analyzer or your

transceiver. On the other side of the pair, connect a though-line wattmeter with a
50 Ohm dummy load or the input of the spectrum analyzer. The lengths of the
outboard cables are not critical as long as the source and load impedances of the
outboard equipment are truly 50 Ohms.

Now evaluate the cavity pair as described earlier in this book. For the simple
setup, collect data points and draw a graph for the cavities in combination. If you
have a tracking spectrum analyzer available, it will display the response curve for
you. Then simply change to the next cable increment and re- evaluate the cavity
pair.

Once you go through the complete set of cables, the correct cable length will be
very evident. If two cables seem similar, use either. This is now the cable length
that best satisfies a ½ wavelength electrical pathway between cavities. Make up
a cable of the same length for permanent use.

In essence what you have done pragmatically is to essentially eliminate the

inter-cavity line, making the loops think they are in both cavities. At this line
length your inter-connected cavities will perform in the complete duplexer without
difficulty. They will perform very close to basic cavity theory. Incorrect inter-cavity
line lengths cause serious problems.

Then, after you have interconnect all the cavity pairs individually, assemble the
compete duplexer and then re-evaluate it as a complete unit in much the same
way. Do the input side and output side separately. Just remember to keep any
unconnected ports terminated with 50 Ohm dummy loads.

At this point you will be able to estimate the total isolation and the total insertion
losses of your complete duplexer from the graphs or the spectrum analyzer
display.

External Lines
 9-15

You may now be wondering about the lines that connect the duplexer to the
receiver, the transmitter and the antenna. Do they also have to be a special
length? The simple answer is no, not if you have tuned and configured your
duplexer correctly. All external devices should present a 50 Ohm load to the
ports of the duplexer, and if as such the lengths of these external lines is not
important. Do not tune external lines; it is unnecessary.

There is one exception to tuning an external line – an add-on cavity. Let’s say
you want to add a pre-amplifier to your repeater. Furthermore you decide that an
additional outboard bandpass cavity in front of the preamp would be a good idea.

In this case you also do not need to tune the lines. The preamp isolates the
outboard cavity from your duplexer, and presents a 50 Ohm load to each. But if
you decide to connect the outboard cavity directly to your duplexer, then you do
need to tune the line. For now the external cavity is a working part of your
duplexer. Do so in exactly the same way you tuned the inter-cavity lines in the
duplexer. Remember, if a line is terminated in 50 Ohm devices, the line does not
need to be tuned. A cavity is not a 50 Ohm load, so any line to it needs to be
tuned.

Always keep the basic principle in mind for inter-connecting cavities. The total
electrical length of the line, including the coupling loops should be ½ wavelength.
This essentially causes the lines vanish from the picture hence not to change the
filtering action of the cavities in any way. That’s the objective for inter-cavity lines.

Author’s Note

Well, that’s as far as I got. I think, though, that most of it has been said, and in
pretty simple terms. So enjoy! Roll your own. You will be pleased

I will be pleased to correspond: John Portune W6NBC. [email protected],

w6nbc.com. 519 W Taylor St., Unit 111, Santa Maria, CA 93458.
January 2019