In this chapter we continue our overview of basics. Here we will ask the very important question, how efficient must a duplexer be? To what degree must it fulfill its responsibilities? This question is the subject of much misinformation in the mobile radio world. 

What you'l learn in this chapter added to what you learned in the second chapter we will enable to know exactly how much duplexer you need. You won't end up wasting money on too high-level a duplexer, and you won't end up with the miseries of an inadequate duplexer. 

But first, duplexer terms that we will need throughout this book.


As we saw in the previous chapter, one side of the duplexer does not have to be totally invisible to the other. The filters only have to adequately weaken the unwanted signals. In practice we usually only need a few tens of dB. We call this the isolation. One of the two important ways that we rate duplexer filters is in how many dB the filters reduce unwanted signals from other side.

Look again at Figure 1. Notice that the receiver must be able to successfully hear an incoming signal in the presence of a transmitter that is a thousand million, million times stronger. As you can see, 100 watts is 150 dB stronger than 0.22 microvolts. This is a total amount of receiver to transmitter isolation that we will need. But it does not all have to come from the duplexer.

At first glance you might therefore think that the transmit filters would have to isolate the receiver by 150 dB. Not so. The actual amount is much less.

The reasons for this are two. First is the fact a repeater's transmitter and receiver operate on d.ifferent frequencies. As you probably know this is called the repeater's frequency split. It is at least several hundred kHz. It varies from band to band. On the 420-470 bands it is 5 MHz, for example.

But because of its selectivity, the receiver does not hear the transmit frequency as well as it does its own frequency. The duplexer only has to provide part of the isolation. At the frequency split, the receiver will provide a significant portion of the total. This is a critical factor in repeater design. Don't forget it.

Similarly, on the opposite side of the duplexer, the transmit filters also have less of a job to perform than it might seem. Their job is only to remove what we call dirt. Dirt is unwanted energy made by the transmitter on the receive frequency. Notice Figure X.

Figure X Transmitter Output Power

Remember, again because of he frequency split, most of the 71 million microvolts made by our example transmitter is on the transmit frequency. If the transmitter is will designed, it will generate very little energy on the receive frequency. For example, for this transmitter, 5 MHz away from the transmitter's output frequency, the dirt is 60 dB less. We can therefore subtract this from the total isolation needed. This too is an important principle of repeater design. Don't forget it either. We'll put some real numbers to isolation in the next chapter.
Insertion Loss

The other major definition we need is called insertion loss. It is the amount of signal that we will unwillingly have to give up in the duplexer's filters? For even the very best duplexer has some insertion loss.

Insertion loss is usually stated in dB. For example if the 100 watts made by your transmitter became only 50 watts after passing through your duplexer, we would say that your duplexer has a transmit insertion loss of 3 dB. The same applies to the receive side. But now back to the story. 

How Much Duplexer

In most duplexers we skate on the thin edge of performance verses cost. For as much as we would like things to be different, there is no one design that is both inexpensive and highly efficient. Knowledgeable compromise is the name of the game. Let's understand. 

Suppose you could actually buy a totally perfect duplexer. It would have infinite isolation and zero insertion loss, wouldn't it? And surely you would agree, that it would work perfectly in all situations. Our fantasy duplexer would be a "fits-all." It would behave flawlessly no matter how terrible the hill top.

Coming down the hypothetical scale somewhat, how well does a really top-of-the-line duplexer perform? It would have say 120 dB of isolation for both receive and transmit and might have only one dB of insertion loss. Wouldn't such a duplexer also work in most situations too? With it you could make your repeater work on almost any hill top.

Instead, let's go to the other end of the scale. Imagine that you have a really low-priced duplexer with only 40 dB of isolation and over 3 dB of insertion loss. Could you use it? Yes, but not in as many circumstances. It might work okay, for example, in a vehicular installation, but it certainly wouldn't work as a repeater duplexer on a dirty RF hill top.

The vital point here is, the better the duplexer, the more places you can use it without understanding it. A really top-of-the-line duplexer is almost the perfect answer to your duplexer needs. Or is it? Remember, it will be very expensive. Can you always pay the price? Obviously, many can't. Certainly most ham repeater builders don't want to.

Wouldn't it be better if we could exchange knowledge for cost? Well, there's good news, you can. There are a number of well-informed compromises that you can make. And once you learn the rules, you will net big savings without any meaningful compromise in performance. For from real live experience I am keenly aware that only a few situations demand a high-priced duplexer. 

Vital Measurement Before Purchasing a Duplexer

To keep from spending unnecessarily for every application, you are going to have to do something. It will be necessary for you to make three simple measurements on every hill top. Most repeater owners don't, and most either spend too much money or end up with a poor-performing repeater. And often it is only because they don't understand these principles. So you really do need to equip yourself and learn how to make these measurements. For only after making them will you know what kind of a duplexer to buy, build or modify.

FACTOR ONE: The Noise Floor

The radio spectrum is full of noise. All radio people are aware of this. Hams in particular who work HF know noise well. Noise comes from a variety of natural and man-made sources. Solar flux, thermal activity and atmospheric noise account for most of the natural noise. Industrial machinery, power lines and other radio sources produce the bulk of the man-made noise. Together, they make up what we call the noise floor. Notice Figure 4.

Figure 4, Noise across the radio spectrum, freq. vs. dBm

As you can see, on average, noise in the RF spectrum is high at low frequencies and low at higher frequencies. This provides us with a basic duplexer principle,

Noise from the VLF to mid UHF decreases roughly in inverse proportion to frequency.

This is incredibly important in repeater design and in the selection of a duplexer.

We can state this more simply. Each time the frequency doubles, the noise drops to roughly one half. It is what we call a 6 dB per octave relationship. An octave is double and so is 6dB in voltage. And surprisingly, this relationship is remarkably constant over this wide frequency range. At about 1 GHz, noise reaches a minimum and then again begins to rise. Additional sources begin to make noise at these higher frequencies. In this book, however, we will not concern ourselves with higher frequencies, as I have said. 

If you want to determine the minimum duplexer required for a particular application, you must first measure the noise floor at your location. The reason is simple.

The noise floor sets a limit on useable receiver sensitivity. This is another basic duplexer principle.

The noise floor is just like a permanent "jammer." There is almost nothing that you can do about it. You can thoughtlessly increase the sensitivity of your receiver and make it more sensitive than the noise floor, but you will only amplify the jammer, not the signal. Your receiver simply can't hear a signal that is more that just a little weaker than the working noise floor. It is simply forever lost forever in the noise. 

At 440 MHz, for example, basic noise theory places the noise floor at very roughly 0.35 micro volts or minus 116 dBm. This is more or less the least amount of noise you can expect, on average for all locations, at the terminals of a 0dB gain antenna. If you want to know what it is on other bands, simply multiply this by the factor given in Figure 4, or use the simple rule above. As an example one can expect the noise to be roughly three times as high on the 2 meter band as it is on the 440 band. This would place it at roughly 1 microvolt. A receiver much more sensitive than this is more or less wasted except on rare occasion when the noise floor drops below average.

So despite what common wisdom may dictate about sensitivity, anything much below what the noise floor permits wouldn't improve your repeater. In actual practice, the noise floor on real hill tops can often be much worse than theory. It varies, of course, but I have seen it as high as minus 90 dBm (7.1 micro volts) at 450 MHz. In this situation, even a "numb" receiver would be totally adequate. It could actually be preferable.
Impact on the Duplexer

In our discussion here, another reason why you shouldn't use more sensitivity than the noise floor permits is that it places an unnecessary burden on the duplexer. Because, your duplexer must match every dB of sensitivity that you add to your receiver. The greater the gain, the harder it is for your duplexer to keep your transmitter out of your receiver. It's just basic physics. Greater gain also makes it equally more difficult to keep other transmitters and "grunge" out of your repeater. Measure the noise floor. It is vital information. 

With modern solid state RF devices it is easy to add gain that the situation does not require. Have you ever wondered why commercial manufacturers do not make their receivers more sensitive? Is it because they do not know about the fancy GASFET RF transistors we can buy today? Well hardly. It is because they know about the noise floor.

If you are familiar with the HF ham bands, you'l know that the noise floor is a permanent unwelcome companion. If it did not exist, "The world would be your oyster." When noise is low, you can talk around the world on the short wave bands. When it is high, all the receiver sensitivity in the world is worthless. My point is, that UHF and VHF repeater builders need to remember this too. Again, measure the noise floor before finalizing your repeater configuration.

Measuring the Noise Floor 

Precise measurements of the noise floor are difficult to make. But the type of measurement you will need are easy. All you will need is your repeater, an AC voltmeter, a signal generator and a hybrid combiner. 

The hybrid combiner is a passive mixing device that lets you connect your receiver to your antenna and to a calibrated signal generator at the same time. See Figure 6. The combiner isolates the two, so that one does not modulate the other. The combiner also maintains a constant 50 ohm line impedance on the line. This is important to the accuracy of the measurement.

Figure 6 Setup for noise floor measurement

I built a hybrid combiner for myself. It is described in Figure 5. It's perfectly suitable for noise floor measurements over our range of interest. Figure 5 is the circuit. Commercial models are readily available.

Figure 5 Hybrid Combiner

Noise Floor Measurement Procedure

1. Connect the AC voltmeter, the hybrid combiner the signal generator and your antenna, as shown in figure 6.

2. Set the signal generator to your receiver's input frequency. Modulate it with an audio tone. 

3. Turn the signal generator's output all the way down. Disconnect the antenna and terminate its input to the hybrid combiner with a 50 ohm termination. 

4. Observe the reading on the AC voltmeter. With a signal from the signal generator and without the antenna, the AC voltmeter will be reading only the noise your receiver is making. Make note of this level.

5. Connect the antenna. This will introduce the noise signal into the system. The AC voltmeter should now rise. Make note of how many dB it increases on the AC voltmeter. Again remove the antenna and terminate its input.

6. Increase the output of the signal generator until the reading on the AC voltmeter increases by the same number of dB that it increased when you connected the antenna.

The generator is now producing the same audio power as the noise from the antenna. We call this the minimum discernible signal (MDS) that your receiver can detect.

The output level of the generator is now a very reasonable approximation of the effective noise floor at this site.

Antenna Gain

We'll need also to consider the gain of the antenna you are using at your repeater. To make an absolute measurement of the noise floor we'd need to use a different antenna. It would be the theoretical unity gain antenna that all manufacturers use as the reference to specify the gain of real antennas. It is called an isotropic radiator. Your antenna will usually have more gain by several dB. 

Remember, a gain antenna increases the noise. It is effectively adding passive sensitivity to the system. The amount of noise that it adds is not equal to its gain, however. It is usually less. Therefore, it is difficult to accurately compensate for the antenna's gain in a noise floor measurement. But this is not a serious problem.

What I usually do is assume that the error isn't large. Repeater antennas usually have less than 10 dB gain. And since the noise that they add is usually less than this, I merely take the noise coming from my real-world antenna as useable rough estimate of the noise floor.

I will, however, consider antenna gain if I thinking about increasing the gain of the antenna at an existing site. If I previously accuratelyspecified my duplexer only for a lower gain antenna, a higher gain-antenna may place too heavy a demand on the duplexer. Then I probably would elect not to change the antenna.

For our example repeater, we will assume that the signal generator after the noise floor measurement is showing an output of minus 110 dBm or .71 micro volts. This is the working noise floor. It is the starting point for how much duplexer to use. As a matter of interest, all the figures that I am using for the example come from an actual repeater. 
Variations in the Noise Floor

Before we leave our discussion of the noise floor, let's consider one of its additional characteristics. It is constantly changing. Variations in the earth's atmosphere, produced by changes in the radiation from the sun, cause the noise floor to change. At VHF and UHF, these variations are not as large as they are on the short wave bands, but they are still very real. That's why a mobile VHF or UHF station, that can access a repeater on some days from a distant location, won't be able to on another.

It is. therefore, best to take a series of noise floor measurements over a period of time to get a good idea of how low the noise floor can get at your particular location. I make a habit of taking a measurement on every early visit to a new hill top installation. Over a few trips I get a good idea of the lowest possible effective noise floor at that location. 

What you're trying to estimate is the minimum noise floor at your location. That determines the maximum receiver sensitivity that you can ever use at your site. You want to be able to take advantage of conditions at their best. At most sites I usually allow only roughly 10 dB of unneeded sensitivity to be sure. More than thisis a bad idea, as you've seen.

FACTOR TWO: Receiver Sensitivity and Selectivity

Next, you will need to make a measurement on your receiver. For you need to know not only how sensitive it is but how selective it is. As you learned, every dB of selectivity that the receiver does not provide, the duplexer has to provide. This is, therefroe, a critical measurement in determining the right duplexer.

Sensitivity, as we know, is easy to measure. Radio technicians do it usually when they visit a repeater. So I won't describe the procedure here For our example, our example receiver has a sensitivity of 0.22 micro volts or minus 120 dBm.

This turnes out to be a good match, without any additional gain, for the minus 110 dBm noise floor measurement that we made above. Remember, we want only a little more sensitivity than it takes to get down to the noise floor. The extra 10 dB in our example is a realistic amount of headroom.


While we are on the subject of sensitivity, let's pause briefly and talk briefly about preamps. In recent years the GASFET preamp has become the shining star of the repeater world. At least a lot of people look at it that way. Unfortunately, GASFET preamps can unwittingly cause serious problems.

Most good-quality after-market preamps are broad-band devices. They have little or no selectivity. They amplify the interfering signals just as well as they do the signals that we want. They do add sensitivity, but also add bandwidth. This is a problem. This extra bandwidth automatically places an increased burden on the duplexer. 

Before adding a preamp, your duplexer may have been able to properly isolate your receiver from your transmitter and shield you from your neighbors. But if you add a preamp, not only will the signals that you want be stronger, but so will be the ones that your duplexer is fighting to eliminate. It may not be able. So don't be surprised if a preamp only makes the "grunge" worse. 

Before installing a preamp, in hopes of creating a "killer" repeater, spend the time to measure the amount of sensitivity that you actually can use at this site. It may surprise you. If you are already slightly below t the noise floor, a preamp probably won't help. More than once, by making a noise floor measurement, I have saved the expense of a preamp and I have avoided a lot of grief by not inviting the interfering signals. 

I have a rule of thumb about preamps. On really high mountain tops, or at remote sites, where maintenance is difficult, avoid them. Most barefoot commercial UHF receivers have more than enough sensitivity for these applications. At a lower altitude site, they can help a lot, however. 

Selectivity Measuring Procedure

1. Connect the AC voltmeter, the hybrid combiner and this time, two signal generator, as shown in figure 6.

2. Set one of the signal generators to make a weak signal on the receiver's input frequency. The level should just sufficient to quiet the receiver. You should see roughly a 1-2 dB decrease on the AC voltmeter. Maintain this level throughout the measurements.

3. Set the other signal generator to a similar level. It will represent the off-channel interfering signal, such as your transmitter or your neighbor's transmitter.

4. Increase the level of the off-channel generator until you begin to notice desense. You'll hear it in the loudspeaker and will be able to see it as a slight increase on the AC voltmeter.

Perform this step at a number of small uniform frequency increments across the center frequency of the receiver. Record the desense level and the off-channel frequency for each measurement. Plot the measurements on a graph. It should look similar to the example shown in Figure 5. It's a graph I compiled from an actual receiver.

Figure 5 Receiver Desense Profile Graph

As you can see, at the center frequency, a very tiny signal will desense the receiver. Away from the center frequency, either up or down, you'l need a stronger signal to produce desense. 

Returning to our example, the selectivity graph shows us that our transmitter has to be 88 dB stronger than minus 116 dBm to desense the receiver. This is isolation that our duplexer does not have to provide. Our receiver's selectivity provides it.

Then, by subtracting 88 dB from 150 dB (the signal difference between our 100 watt transmitter and our .22 micro volt receiver), we discover that our duplexer only needs to provide a 62 dB of isolation at the frequency of the receiver. That's a lot less than the 120 dB that we pay for in a top-of-the-line duplexer.

We need, of course, a little more than exactly 62 dB. Something more like 80 to 90 dB would be ideal. We'll discover later that duplexers drift due to temperature and humidity. The extra headroom provides for this.

Selecting a Receiver

In light of what we have been discussing, it should be obvious to you that some receivers may not be suitable for repeater service. In a commercial repeater, the manufacturer will make the right choice. Hams, however, may not. So let's briefly talk about selecting a receiver for repeater use. 

Most commercial radio receivers, the type found in mobile radios, are frequently a good choice for a ham repeater. They are narrow-band devices. They have good front-end selectivity. They were originally designed to normally operate on only one or two nearby frequencies This allows the manufacturer to use narrow-band filters in the front end. Sharp helical resonators are common.

Most ham receivers, and an increasing number of commercial receivers must cover a wide range of frequencies. They can not be narrow-band devices. They lack front-end selectivity. Their working selectivity comes mostly from the IF filters. The front end of this type of receiver is wide open to off channel interference. Repeater builders should avoid this type of receiver when building a repeater. Let me prove the point by reviewing the issue of selectivity above.

Figure x compares two receivers. One is a typical wide bandwidth receiver, such as many ham receivers. The other is the narrow-band commercial receiver we've been using as our example.

Figure x A Wide and a Narrow Receiver's Selectivity Curve

How much duplexer would we need to use the wide-band receiver in a repeater? Our narrow-band receiver needed 62 dB of isolation from the duplexer. As you can see, the wide-band receiver would need a duplexer with xx dB of isolation. Saving money on a receiver is very costly in terms of the duplexer. To state a simple rule of thumb, selectivity is less expensive in a receiver than it is in a duplexer.

FACTOR THREE: Transmitter Power and Purity

The type of transmitter that you use is also very important in selecting a duplexer. Two issues are at work here.

* Transmitter Power

* Transmitter Purity.

Transmitter Power

It is a simple law of physics, that the quality of your duplexer is directly related to how powerful your transmitter is. You might think this goes without saying. I'm amazed by repeater owners who forget this when installing a power amplifier, for example. 

More power simply means more isolation in the duplexer. Refer back to figure xx. If you increase the power by 10 dB, you will need an additional 10 dB of isolation on both sides you your duplexer. A duplexer that is just sufficient for a 10 watt repeater will permit harmful desense if a 100 watt power amplifier is added.

Transmitter Purity

What we have been saying so far has applied mostly to the filters on the receive side of the duplexer. What about the filters on the transmitter side? How well must they perform?

A transmitter, like a receiver is not a perfect device. In an ideal world, the transmitter would produce power only on its center frequency. Unfortunately, it doesn't. Notice figure 7. It is a spectrum analyzer display of a correctly functioning unmodulated transmitter.

Figure 7. Spectrum Analyzer Display of a Transmitter

Notice that most of the energy is where we want it to be, on the center frequency. Unfortunately the transmitter also generates lesser amounts of energy on all nearby frequencies. We affectionately call this energy on nearby frequencies "dirt." Some of it is will fall, much to our dismay, right on the center frequency of the receiver. Getting rid of it is the primary job of the filters on the transmit side of the duplexer.

So again we ask the question, how much? How well must the filters work? It probably won't surprise you, but the answer is similar to the other side. It depends on how strong the dirt is, and how selective and sensitive the receiver is. Sound familiar? Let's talk about signal strength first. 

This time, however, we are only talking about the small amount of energy that the transmitter is making on the receive frequency, the dirt. It's much less than the energy the transmitter is generating on its center frequency. It is still, however, a lot more than the receiver can handle. Remember, the receiver is very sensitive. Notice figure x again in chapter two.

The dirt only has to be 0.22 micro volts for the receiver to hear it well. Again using Watt's Law, 0.22 micro volts is only .000,000,000,000,001 watts. The cleanest transmitter in the world makes a lot more dirt than that, 5 MHz away on the receiver's frequency. That's why we need filters on the transmit side of the duplexer. They remove the dirt.

So just like on the other side, we will need to measure the strength of the dirt before we can specify how much duplexer we need. To do this, a spectrum analyzer is the best instrument to use.

Transmitter Dirt Measurement Procedure

1. Connect your transmitter directly to the input of the spectrum analyzer. Use an 20 dB attenuator pad to protect the spectrum analyzer and to keep the load impedance on the transmitter constant at 50 Ohms.

2. Set the spectrum analyzer to display only a little more the transmit and receive frequency split of your repeater.

3. Place the transmitter into transmit mode, without modulation. Adjust the spectrum analyzer to display the center frequency of the transmitter at roughly full scale. Note the reading.

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

Notice figure xx again. We may now determine the strength of the dirt on the receiver's input frequency? We do this by comparing the difference, in dB, between the dirt and the transmitter. In our example, the transmitter is making 100 watts. Most of the 100 watts are on the center frequency. 100 watts as we can see from figure xx has an absolute value of +50 dBm.

The dirt, we note, is 85 dB weaker than the transmitter. Its absolute level is, therefore, minus 35 dBm.

Our duplexer only has to provide enough isolation in the filters on the transmit side of the duplexer to reduce this signal below the noise floor. It's minus 110 dBm in our example. This means that the filters on the transmit side of our duplexer must provide at least 75 dB of isolation. We'll add another 10 dB for headroom. Even so, at 85 dB, we have again arrived at far less than a top-of-the line duplexer.

One final word on power amplifiers. The main problem is that they tend to increase dirt more than they do center-channel power. Therefore, if you add a power amplifier to your repeater, be forewarned. You may need more isolation in your duplexer than you would expect from the power increase alone.

Chapters 2 and 3 Summary

The central concepts of the last two chapters have been quite simple. To specify a duplexer you must first understand its responsibilities. Then you must measure the characteristics of both your repeater and proposed site to know how much duplexer you actually need. These measurements include the noise floor, receiver sensitivity, receiver selectivity, transmitter power and transmitter purity.

In the next chapter, I will take you through an actual example from a real hill top. I'll also give you a simple worksheet for you to use on your own repeater. This will be the concepts from chapters one and two put into practical action.

But by not treating the duplexer as a black box, you will net a big saving in cost. You'll be able to specify a lower cost duplexer without sacrificing optimum performance. It's an exchange of knowledge for cost.