Transmitter IMD and the Two-Tone Test for SSB Signals

The most useful receiver specification is its dynamic range. The most useful transmitter specification is almost the mirror image: how clean is the signal when it is loud? On SSB the answer is given by the transmitter IMD, short for intermodulation distortion, measured with the two-tone test. The number tells you how far below the peak power the unwanted products lie, and so how much splatter your signal throws into the neighbouring channels. A transceiver or an amplifier with poor IMD can sound fine on the air to its owner and still be a nuisance for everyone within 5 kilohertz. This guide explains the test, what the products are, which figures are normal, what causes poor IMD and how to improve your own signal without a laboratory.

Why a Linear Transmitter Can Still Distort

An SSB transmitter has to amplify a signal whose amplitude changes constantly, with peaks that may be many times higher than the average. The amplifier stages must reproduce that waveform faithfully, which is why they are called linear. No real stage is perfectly linear. As the signal approaches the maximum output, the gain starts to fall, the peaks are flattened and the waveform is distorted. Distortion in a signal made of many frequencies produces new frequencies that were not there before, sums and differences of multiples of the original ones. This is intermodulation, the same effect that limits a receiver in the presence of strong signals, as the article on receiver dynamic range explains from the receiving side.

A voice contains many frequencies at once, so the products of the speech distortion are a smear of noise on both sides of the signal. This smear is what operators call splatter or buckshot, and the part that lands outside your own channel is heard by others as a rough, wide signal, often with a clicking sound on the peaks. The effect is much worse on strong, nearby signals, and in a contest the splatter of a single loud station can block a neighbouring frequency. The related effects at the keying edges are discussed in the article on key clicks and splatter.

The Two-Tone Test

Because voice is difficult to measure, engineers use two pure tones of equal amplitude. The American amateur laboratories, among others, use frequencies of 700 and 1,900 hertz, which fit within the audio passband of an SSB transmitter. The two tones are fed to the microphone input, the transmitter is set to a chosen power, and the output is examined with a spectrum analyser. The ideal output has two lines only, at the carrier frequency plus 700 and plus 1,900 hertz. In practice there are extra lines on both sides, and the heights of these lines compared with the wanted tones give the IMD figure.

The tones have a useful property: the peak envelope power of two equal tones is four times the power of one tone, so each tone is 6 decibels below the PEP. The IMD figure is usually quoted in decibels below one of the two tones, or in decibels below the PEP, which is a different number by 6 decibels, and it is important to see which reference is used. A product 30 dB below one tone is 36 dB below PEP. Our article on peak envelope power explains how the power is defined and measured.

Where the Products Fall

Take two tones f1 and f2. The third-order products appear at 2f1 minus f2 and at 2f2 minus f1, close to the wanted tones. With 700 and 1,900 hertz, the product 2 times 1,900 minus 700 is at 3,100 hertz, and 2 times 700 minus 1,900 is at minus 500 hertz, on the other side of the carrier. The first lies outside a normal 2.7 kilohertz channel, which is why it hurts neighbours. Fifth-order products appear at 3f1 minus 2f2 and 3f2 minus 2f1, at minus 1,700 and 4,300 hertz, further out and usually weaker. Seventh-order products continue the pattern further out, and for a good transmitter they are small.

What Numbers to Expect

Independent measurements of commercial HF transceivers at full output typically show third-order products between 25 and 35 dB below PEP, with the best radios reaching 40 dB or more at a slightly reduced power. A third-order intercept point specification expresses the same property in another way. Linear amplifiers add their own distortion, and a tube amplifier that is well driven and correctly tuned can reach 30 to 35 dB, while an amplifier that is overdriven may show 20 dB or even worse. A product 25 dB below PEP is audible for several kilohertz either side of the signal when the station is strong; at 35 dB it is far less noticeable; and at 40 dB it is hard to hear unless the listener is very close.

Notice that the IMD is not a fixed property of a radio but a function of the drive and the power level. Backing the output down by 10 to 20 percent often improves the IMD by 5 to 10 decibels, which is why a radio set to 80 watts frequently sounds cleaner than the same radio at 100.

Causes of Poor IMD

The most common cause is too much drive. A microphone gain that is set too high pushes the audio into the processing stages and then into the final amplifier at levels beyond their linear range. The ALC meter is the guide: it should barely move on voice peaks, and any steady, deep reading means that the stages are being compressed. The article on SSB audio settings describes how to set the gain and the compression, and the similar rules for digital modes in the guide to FT8 operating etiquette apply, because any ALC action in a digital mode is the same distortion.

The second cause is the power supply. A supply that sags under load, or a car battery with poor connections, reduces the voltage at the output transistors on the peaks, which flattens them. The 12 volt wiring tips in the article on 12 volt wiring show the effect of voltage drop. The third cause is the amplifier itself: tubes that are past their life, a wrong bias, a tune-up at the wrong loading, or a transistor amplifier with poor filtering. The fourth cause is a poor antenna match, since a high SWR makes the radio fold back its power and distort. Finally, an excessive speech processor setting increases the average power at the cost of linearity.

Improving Your Signal Without a Laboratory

You do not need a spectrum analyser to improve the IMD. First, set the audio so that the ALC shows nothing, or only a flicker, on loud syllables, and then use the compression at a modest level. Second, run the radio at a power somewhat below the maximum, which gives a better figure for little loss of signal: reducing power by 20 percent reduces the signal by a mere one decibel. Third, drive a linear amplifier only to the power the manufacturer states for the full output, and use the amplifier’s ALC feedback if it has one. Fourth, check the supply voltage under transmit.

Fifth, listen to yourself. A second receiver, or a software defined radio, tuned a few kilohertz away from your signal while you speak, shows how wide you are. A panadapter or a waterfall makes it visible, as described in the guide to the panadapter and spectrum scope. Ask a friend a few kilometres away to look at your signal on a monitor, and use a station with a good receiver for a comparison between settings. Many clubs have a member with a spectrum analyser who will run a two-tone test for you, and the result is a revelation. If the IMD is poor at a particular setting, change one thing at a time and test again.

A Home Two-Tone Test With an SDR

An inexpensive software defined receiver can serve as a rough spectrum analyser for the test. Connect the transmitter to a dummy load, and take a small sample of the signal from the load through a heavy attenuator or a loosely coupled pickup, so that the receiver is not overloaded; the guide to the dummy load explains why the test must not radiate. Generate two audio tones of equal level with a free program on a computer, feed them to the microphone input through an attenuator that brings them to the normal speech level, and set the power to the value you normally use. The spectrum display shows the two tones and the products around them.

The reading is only as good as the receiver. A SDR with a poor front end produces its own intermodulation when the input is too strong, so add attenuation until the products on the screen stop changing. If an extra 6 dB of attenuation makes the displayed products fall by 18 dB, they were made in the receiver, and you have gone too far in the wrong direction. When the products fall by exactly the same amount as the tones, they are real. This simple check, plus a little care with levels, gives a figure that is within a few decibels of a laboratory result and is more than enough to compare settings.

Digital Modes and IMD

The same physics governs the digital modes that are sent as audio tones through an SSB transmitter. FT8 uses a constant-amplitude signal, which is easy on a linear amplifier, but if the audio drive is too high the ALC acts and the stages distort the tones, creating products on either side of the signal. The classic sign is a wide base to your signal on other stations’ waterfalls. Digital operators who run at full power often discover that a power of 30 or 40 percent of the radio’s maximum gives a cleaner trace and no loss of contacts. The thermal side is described in the article on the FT8 duty cycle, since reduced power also helps the radio stay cool.

For voice, a modest speech compression can raise the average power without increasing the peaks, and so improve the readability of your signal at a given PEP. Too much compression raises the average to the point where the transmitter works like a continuous-carrier mode and the IMD worsens, so keep the compression within a few decibels and use the test to decide.

Amplifiers and the Two-Tone Test

Linear amplifiers deserve special mention. A test with two tones at the amplifier’s rated power, followed by a test at a lower power, shows how the IMD changes with drive. A good amplifier shows a small change; a poorly adjusted one deteriorates sharply when it is pushed to the rated level. Operators sometimes see an apparent improvement in their signal reports after they reduce the drive of an amplifier by a few watts. The article on HF linear amplifiers mentions the connections between power, linearity and cost, and it is better to buy an amplifier with a good record of two-tone test results than one with a higher nominal power and unknown behaviour.

The Bottom Line

Transmitter IMD measures how much unwanted distortion a signal contains, and the two-tone test with 700 and 1,900 hertz shows it as extra lines around the two wanted tones. Third-order products at 2f1 minus f2 and 2f2 minus f1 fall just outside the channel and appear as splatter. Good transceivers show 30 to 40 dB below PEP, and each tone is 6 dB below PEP, so check the reference. Overdriving, a weak power supply, a mistuned amplifier and a poor antenna worsen IMD. Keep the ALC quiet, back off the power by a little, and observe your signal on a second receiver or a panadapter.

Frequently Asked Questions

What is transmitter IMD?

Intermodulation distortion in a transmitter: unwanted products created when a nonlinear stage amplifies a signal with several frequencies, such as voice or the two tones of a test.

What is a good IMD figure for an SSB transmitter?

Third-order products 30 dB or more below PEP are good, and 35 to 40 dB is excellent. Poorly adjusted amplifiers may show 20 to 25 dB.

Why are two tones used for the test?

Two pure tones produce products at predictable frequencies, which are easy to measure with a spectrum analyser. Voice would give a smear that cannot be measured simply.

Does reducing power improve IMD?

Usually yes. A small reduction in power often improves IMD by several decibels, at a cost in signal strength of only a fraction of a decibel.

Can ALC action cause splatter?

Yes. A deep ALC reading means the stages are compressed or overdriven, which creates distortion. Reduce the audio or drive until the ALC barely moves.

How can I check my signal without test equipment?

Listen on a second receiver or watch an SDR waterfall a few kilohertz off your frequency while you talk, or ask a nearby station to check.