7: Measuring Amplifier Linearity
There are several different methods in common use to assess the linearity of RF
amplifiers. Although radio amateurs rarely consider anything other than the two-tone test, there are other means to assess the linearity. What needs to be kept firmly in mind is what we are trying to achieve. In most cases, the emphasis is on preventing, or at least minimising, the level of odd order inter-modulation products (IMDs). If allowed to get out of hand these cause the transmitted signal to spread over a wide portion of the band, commonly known as ‘splatter’, and causing stations listening on an adjacent frequency to hear these unwanted products.
In days gone by, before single sideband became popular, transmitters generally were amplitude modulated, full carrier, types. These transmitted a carrier as well as the two sidebands and, as such, they were wasteful of the power being used. By eliminating the carrier and one sideband, much better use was made of the available power as all the power was contained in a single sideband.
If one looks back at how things were in the days before SSB, the testing of a transmitter or amplifier was much simpler and took very little in the way of test equipment. Even though the amplifiers used for the final stage were not classed as linear amplifiers they in fact were pretty good in producing a clean signal with very little distortion. This was because the modulation was applied directly to the final amplifier stage.
The simplest AM transmitter would use a triode valve in the output stage. This is shown in Fig 7.1.
Fig 7.1: Amplitude modulated transmitter output stage.
In this method, the anode high voltage is fed to the secondary of an audio transformer to which the primary winding has an audio power amplifier connected. The modulating signal hence causes the anode voltage to vary. With no modulation applied, the carrier is unmodulated and the output power is constant. When a sinusoidal audio signal is fed into the microphone input the audio amplifier generates an additional voltage that both increases and decreases the applied anode voltage. On the positive half cycle of the audio signal the anode voltage increases to some maximum and on the other half cycle it decreases to some minimum. The depth of modulation is measured as to how it relates to the normal high voltage fed to the anode. If the anode voltage without modulation is 750V DC, when the audio signal swings the anode voltage upwards to full modulation the anode voltage rises to twice the normal level. Hence, it peaks at the top of the half cycle to 1500V. Similarly, on the other half cycle it swings down to zero volts. In practice it is very difficult to make a valve that can swing down to zero volts, so the maximum possible modulation can never attain exactly 100%, but 90% is generally possible.
If the RF output signal is viewed on an oscilloscope the carrier will look like the graph shown in Fig 7.2.
Fig 7.2: A 100% modulated RF carrier.
You will notice that the carrier rises to a peak but also falls to zero amplitude before rising to the next peak. An estimate of the modulation percentage is that if the carrier envelope does not fall to zero the modulation is less than 100%. For the condition where the anode voltage rises to twice the static condition, certain things are generally assumed.
Firstly, it is assumed that if the anode voltage is doubled the anode current should also double. Power input is a function of the product of the anode voltage multiplied by the anode current, assuming the stage is operated in a linear manner. For the case where the anode voltage and the anode current both increase by a factor of two, the instantaneous input power rises to a value that is four times that of the unmodulated condition.
The second assumption made is that the gain and the efficiency of the amplifier remains constant for varying levels of input power. That is, after all, what defines a linear amplifier: the output power versus input power is a straight line inclined at some angle on the graph. Hence, if the input power increases by a factor of four times, the output power must also increase by the same amount. Assuming the DC input power under unmodulated carrier is 150W and the stage efficiency is 66.6%, the carrier power is 100W. Under 100% modulation peaks the output power must be 400W PEP. This is where the UK amateur licence conditions are derived from.
Before the days of single sideband, transmitters were amplitude modulated types and the DC input power was limited to 150W DC into the final amplifier stage. It was assumed that the general efficiency of the Class C output stage would be an average of 66.6%, which was a reasonable assumption. However, with a very good Class C stage an efficiency of 75% was possible at low frequencies. Efficiency tends to fall off as the frequency is increased, but the 66.6% covered 99% of the amateur stations. When SSB became popular, the regulators simply applied the AM limits for a 100W carrier transmitter, so the SSB licence is the same at 400W PEP, which today is given as +26dBW, that is 26dB above 1W output.
Linearity of a Class C amplifier
An amplitude modulated Class C stage is in fact remarkably linear when correctly adjusted. Viewing the RF envelope on an oscilloscope shows virtually no distortion below about 90% modulation depth. To accurately measure the distortion specialised test equipment is required apart from an oscilloscope.
The industry standard test for AM distortion is not made at 100% modulation but at 60% or 80%, as most amplifiers will have some non-linearity at very high modulation percentage. In commercial transmitters the modulation depth is limited by audio AGC or some other method so the transmitter cannot be over-modulated.
The method of testing is to feed into the microphone connector a pure sinusoidal audio tone, adjusting the input level to reach 60% modulation, and to measure the distortion generated in the modulated carrier. The standard test tone is 1kHz but 400Hz and sometimes 3kHz are also applied to see if the distortion is frequency related.
The RF signal is fed via a suitable attenuator to a modulation meter. This is a linear RF detector to demodulate the signal. The modulation meter can determine the exact percentage of modulation. The recovered audio is then fed to an audio distortion analyser, such as the Hewlett Packard HP-331A, that can measure the harmonic distortion present. Using an instrument such as this can measure the total harmonic distortion to very low levels. It is common to find that distortion levels as low as 0.1% can be achieved with a properly adjusted AM transmitter. This is despite the fact that a Class C amplifier is very non-linear for SSB signals.
It is understood that if the audio input signal contains harmonic products the output signal will always show some distortion products. The audio signal must be as perfect as possible and generally better than ~0.1% THD is necessary to make a valid measurement of the amplifier performance. The HP-331A can measure as low as 0.01% distortion.
The difference between AM and SSB transmitters
The reason a Class C amplifier can amplify an AM signal in a linear manner whereas the same amplifier, when SSB is used, generates a high level of distortion can be simply explained.
In the AM amplifier, the RF drive signal into the valve is a constant level signal; the modulation is applied to the anode. Hence, the drive signal is supplying the necessary RF voltage to keep the valve in the linear conducting region. By contrast, the same amplifier when fed with an SSB signal does not behave in the same manner. Here the carrier input is very low for some parts of the input cycle and at other times it is sufficient to bias the valve correctly. This effect is also noticeable with solid-state amplifiers biased for Class C. If the drive signal is above a certain level, the amplifier operates in a quasi-linear mode. When the input signal drops below a certain level the transistor becomes non-linear, as there is insufficient drive to turn it on. To reduce the non-linearity there has to be some minimum quiescent current flowing at all times and the DC bias fed into the device causes it to move from the non-linear mode to a semi-linear mode.
Some of the techniques to enhance linearity in a valve amplifier can also be applied to solid-state amplifiers, and was common in the days of AM transmitters using transistors. As the modulation depth increases, the anode current does not always follow the wanted curve of the valve characteristic. This is because at a high depth of modulation the drive level is insufficient.
In order to get around this problem the use of screen grid valves such as pentodes and tetrodes was adopted. In these, not only is the anode voltage modulated, but also the screen grid is supplied with a smaller portion of the modulated high voltage. This increase of the screen grid voltage increases the gain of the valve and offsets the lack of drive. Usually this was obtained from the anode modulated supply via a dropping resistor such that the anode and screen grid voltage went up and down in sympathy with the correct ratio of voltage. It is generally the case that when the correct value of resistor has been selected for the non-modulated condition it also fits well for the fully modulated condition. Sometimes the screen grid modulation is taken from a tap on the modulation transformer secondary to better suit the valve being used. The inclusion of the screen grid into the modulation system allows fuller modulation to be attained and with a well adjusted tetrode output stage 100% depth is achievable with low distortion.
Another fix is to partially upward modulate the driver stage so that on the upward swing of the anode current the driver anode voltage is also driven upwards. This increases the RF drive level during the peaks and reduces the distortion considerably. This method was nearly always used with solid-state AM transmitters to ensure adequate drive on speech peaks. It is simple to arrange with a diode to pass the required upward going voltage and blocking the downward going voltage. The diode fed the extra half sine wave DC pulse to the driver anode in parallel with the normal anode supply.
Carrier power vs 100% modulation peak power
This has already been discussed but it is useful to remind the reader that if a linear amplifier is used to amplify an AM transmission the correct setting is critical to ensure low distortion.
Under unmodulated conditions, the output power must not be more than 25% of the peak power rating of the amplifier. Assuming the amplifier is rated at 100W PEP the carrier power must not be more than 25W. If the amplifier is not perfectly linear – and very few linear amplifiers are in fact truly linear – a safer setting would be not more than 15 to 20W unmodulated carrier. This means that the maximum PEP possible is only 60 to 80W and will depend on the upper power linearity of the amplifier.
An indication of a typical amplifier is shown in Fig 7.3. The amplifier is officially rated as a 100W PEP type, but the claims are a little off the real case. To assess the linearity of an amplifier it is necessary to plot the output-input curve. In a perfect amplifier the gain would be constant, no matter what the output power happens to be. There will be some finite upper limit where the ideal straight line deviates away.
Fig 7.3: Input vs output curve for a typical amplifier.
Although this is claimed to be a 100W PEP amplifier it is not. The amplifier in fact delivers a bit more than 100W, it is plotted up to about 105W output. But the real linear portion stops at about 80W output. If the amplifier followed the ideal case, about 130W output would be available. At low drive levels the gain is 10dB, but when the output exceeds about 80W it starts to fall over. At 9W input it only supplies 85W and not 90W. At higher drive levels it becomes increasingly more non-linear. At 13W input it barely scrapes over 100W. It would be risky to claim this is any more than about a 70W PEP amplifier as the linearity deviates from an ideal condition. For FM or CW this doesn’t matter, but for AM and SSB it definitely does affect the purity of the transmitted signal. If the input drive is increased more the output power will eventually flat-line and the addition of any more drive will not give any further increase. The amplifier would be totally saturated. To ensure low IMD products it is necessary to operate the amplifier well below the 1dB compression point.
The generation of IMD products
Intermodulation distortion products are the result of a mixing process occurring in a non-linear amplifier. If an ideal amplifier were examined the output signal would be an exact replica of the input signal, only larger. In a typical amplifier some non-linearity occurs. This gives rise to the production of extra output signals in addition to the wanted one. The predominant type of extra signals are harmonics of the input and these can be considerable in a Class C amplifier. If the amplifier is only used for CW or FM this is no real concern as the harmonic power can be reduced by placing a suitable low pass filter after the amplifier to clean up the output.
In the case of an amplitude modulated Class C amplifier it is the mixing of the RF carrier with the audio modulating signal which produces the two sidebands either side of the carrier. This is a case where we need a very non-linear stage to produce the AM output signal.
If the amplifier were ideal, with no non-linearity, it would be impossible to generate an AM signal, as no mixing can occur. Trying to generate an AM signal using a linear amplifier has poor results as there isn’t enough non-linearity to perform the mixing process.
In the case where the input signal is not a single carrier frequency, but two carriers, the mixing products have two possible extra generated signals. In the AM transmitter we know that if the audio signal is 1kHz the two sidebands will occur –1kHz and +1kHz either side of the carrier. In odd order IMDs the second harmonic of one of the carriers mixes with the fundamental frequency of the other carrier input to generate the extra product. These fold back to occupy a place adjacent to the carriers, above and below, defined by the frequency difference of the two input signals. This first IMD product is known as the third order product, because it is the result of the mixing of the second harmonic (a second order signal) plus the first harmonic – known as the fundamental frequency, hence 2 + 1 = 3. For the case of the fifth order product it is the result of the third harmonic and the second harmonic of the two carriers which also fold back to sit above and below the third order product. If the amplifier is very non-linear these IMD products extend to infinity, gradually falling off in amplitude. The greatest products are normally the third order, then the fifth order is lower and so on all the way up the IMD series.
In a typical AM transmitter these extra products are also visible as extra side bands above and below of the normal ones, but are generally so low they are not a problem in practice. These are due to distortion products (harmonics) contained in the audio input signal, often generated in the AF power amplifier output stage, and are exactly the same mechanism as SSB IMD generation. Hence, with a 1kHz audio signal the second harmonic of the audio occurs at 2kHz and gives rise to the extra sidebands observed. If the audio signal is pure these cannot be generated.
Generating a test signal
In RF design laboratories the classic method of generating a test signal is to utilise two very high quality RF signal generators and to combine the two signals in a 0º power combiner. The amplitudes are adjusted so each is identical and the frequencies are set to produce a small difference in frequency, say 2kHz. This signal is then applied to the amplifier under test and the output viewed on a spectrum analyser. Any non-linearity will be evident by the IMD products seen. In a perfectly linear amplifier only the two discrete carriers should be visible. In a practical amplifier some distortion will be generated in the form of intermodulation distortion products.
For cases where the input power to the amplifier is a large signal the output of the combined signal generators needs to be amplified. RF signal generators generally can only develop about 100mW output of unmodulated carrier. These when combined produces a composite signal of 400mW PEP, assuming there is no loss in the combiner. The linearity of the amplifier following the signal generators has to be quite exceptional if the test signal is to have very low IMD products. This often means expensive amplifiers with extra filtering and obtaining a signal sufficiently clean is not a trivial matter, especially if the drive level required is high.
For amateurs this is not a viable system, so an alternative is to generate the test signal at audio frequencies and to feed this composite two-tone signal into the transmitter microphone connector. The audio tones selected must not be harmonically related, so a 1kHz and a 2kHz tone cannot be used. Generally, the audio tones must be able to pass through the SSB filter so they have to be below about 2.5kHz. A good choice would be 1kHz and 1.8kHz, as these are not related and are both in the full pass band of the filter.
Of the two different methods the phase coherence of the composite signal is not important. In the case of the two signal generator method the phase is not a fixed value, it varies as the two frequencies are different. In the alternative audio two-tone method the composite output signal from the double balanced modulator is phase coherent as they are generated from a common carrier frequency. In the double balanced modulator the carrier is nulled out with the balancing adjustments so that just the two sidebands are visible. In practice a carrier null of –50dB or lower is possible. If the carrier cannot be nulled low enough this equates to a third carrier signal, similar to a true AM transmitter. In an SSB transmitter the IF crystal filter helps to remove one of the sidebands and the residual carrier so the two-tone carrier signal is generally clean enough, if the transmitter is correctly adjusted.
Either of these two methods will generate a composite signal which when viewed on an oscilloscope will look identical to that of Fig 7.2: they are both the same as a 100% modulated AM signal. When viewed on a spectrum analyser, however, they are not the same. The AM signal has three major signals, the large central carrier and the two lower amplitude sidebands, spaced either side of the carrier by a distance corresponding to the modulating frequency. If the carrier frequency is 1MHz and the modulating signal is 1kHz, the sidebands will occur at 999kHz and 1001kHz, hence they are separated by 2kHz. In a 100% modulated AM signal the sidebands are exactly 6dB below the carrier level. In an SSB transmitter, the audio two-tone signal will show just the two sidebands, assuming that the carrier balance is correctly adjusted to null it into the noise.
Of the two different signals, the AM modulated carrier is more searching, as it is a three-tone stimulus. Measuring the IMD products with just the two-tone stimulus often shows good IMD performance, but the three-tone signal shows IMDs not present under the two-tone test.
The AM modulation is closer to what happens when a complex voice signal is applied to the transmitter. The human voice contains not just two discrete fixed frequency signals but a whole spectrum of varying amplitudes and frequency separation. If the spectrum analyser has a peak hold function and a person speaks into the microphone for a few seconds, the picture seen shows the passband response of the SSB filter as it gradually fills up.
The test method mentioned above is widely used in the cable television and other industries where a three tone stimulus signal is used to better explore the linearity of the amplifiers used. In cable TV (CATV) systems as many as 10 channels pass through the amplifiers and each one is a varying signal level. The IMDs can get interesting in these types of systems. This is the reason that the amplifiers used are capable of very high power but run at a fraction of their peak capabilities.
Purity of two-tone signal requirement
The purity of the two-tone audio signal has to be very good. If one considers the conversion of THD in percentage terms to dB, a 1% audio distortion equates to
–40dB. If the IMD of the amplifier is likely to be lower than this the audio distortion must be considerably lower for this method to work. A block diagram of a two-tone oscillator is shown in Fig 7.4.
Fig 7.4: Two-tone audio oscillator block diagram.
The two-tone audio signal is used to generate the SSB signal and then it passes through various mixers and gain stages in the transmitter to generate the final frequency at the required output level to drive the amplifier under test. If the THD of the combined tones is only 1% at the microphone the chances are that at the output of the transmitter it will be considerably worse. It is common to find that there is a degradation of 10 to 15dB in the driving transmitter. So the –40dB input signal degrades to as little as –25dB, which is far too high to be used for accurate testing.
It is the norm in IMD testing to require a drive signal which is at least 6dB better than what is being measured. Hence, the two-tone composite signal needs to be at least –60dB when applied to the microphone input and this will generate an output signal of about –45dB IMD at the output of the transmitter. This means that the lowest it is possible to measure is about –39dB IMD. An input signal of –60dB equates to an audio distortion of 0.1% and that is becoming difficult to generate with simple audio oscillators.
Conversion of THD to deciBels
The THD figure in percent is a voltage measurement. To convert to the corresponding power measurement in decibels the following formula is used:
Distortion in dB = 20 log (% / 100)
10% distortion is equal to 1 / 10 = 0.1 and equates to –20dB
1% distortion is equal to 1 / 100 = 0.01 and equates to –40dB.
0.1% distortion is equal to 1 / 1000 = 0.001 and equates to –60dB.
Two Tone Generation and precautions required
Audio oscillators to generate the two discrete tones can be simple feedback oscillators. One of the simplest types is the phase shift oscillator which uses common components and is simple to make. A schematic for a phase shift oscillator is shown in Fig 7.5. Each oscillator has a switch so either or both may be operated at any time.
Fig 7.5: Audio phase shift oscillator.
The frequency of oscillation is defined by the values of C2 and the resistors R6, R7 and R8. To increase the frequency these resistors are all decreased in value. With the values shown the frequency is about 1100Hz. To modify for a higher tone R6, R7 and R8 are changed to 8.2kΩ, which increases the frequency to ~1700Hz. The exact frequencies used are not critical as long as they are not harmonically related. The composite signal when the two oscillator outputs are added together creates a beat note of about 600Hz.
The oscillator consists of TR1 which is an inverting amplifier. The output at the collector drives TR2 which operates as another inverting amplifier for the signal at the collector. The emitter acts a voltage follower and has zero phase shift between the base and emitter. This emitter signal is at a low impedance to drive the RC network consisting of R6, R7 and R8. The capacitors, which are all the same value, and resistors form a three-stage phase shift network of 180º, with 60º per section. This signal is then applied to the base of TR1. The total phase shift around the network is (180 + 180 = 360º) or (180 – 180 = 0º), both of which equate to 0º. Hence, the oscillator will run at the frequency where the phase shift in the RC network equals 180º. The RC network behaves as a high-pass filter and so the harmonics of the amplifier transistors are contained in the output signal.
To ensure that the oscillator output is clean, a low pass filter is required between each oscillator and the combiner network. Any oscillator contains harmonics and if these are allowed to pass to the transmitter microphone audio stages they will generate their own IMD products. Consequently a sharp cut-off low pass filter is inserted between each oscillator and the combiner. The combiner is a simple resistor network which allows adjustment of the individual oscillator levels and the combined output to the transmitter. This is shown in Fig 7.6.
Fig 7.6: Oscillator combing network.
Resistors RS set the maximum output voltage into the combiner and may be omitted in some applications. Variable resistor VR1 is a balance control to obtain the same amplitude for each tone. The output level is set by VR2. This should be adjusted to suit the transmitter under test, often as little as 10mV RMS is adequate. The higher the value of RS the better the isolation between the two oscillators and the lower the total output signal amplitude. The phase shift oscillator shown will deliver at least 1V peak-to-peak output, so RS will be quite a high value. Resistors R9, R10 can be used to set the attenuation level by varying RS. If desired one of the RS resistors can be replaced by a variable series resistor for finer level control of the output. Only one needs to be variable.
The connection between the generator and the transmitter must be made with a well screened cable to ensure no RF can get into the microphone amplifier stages. The entire generator also needs to be in a well shielded enclosure for the same reason: building this two-tone generator in a plastic box will lead to problems! The best option is to power the unit with an internal 9V battery rather than trying to supply it with an external power supply. The current drawn is low and a small 9V battery transistor radio battery should last a long time.
The audio low pass filter schematic is shown in Fig 7.7. The values of the resistors and capacitors define the cut-off frequency. The values shown suit a cut-off of about 1000Hz. (The component values for the 1700Hz filter are R1, R2 = 27kΩ and C1 = 3.3nF, C2 = 1.5nF.) The attenuation of the second harmonic is –20dB for both filters.
When setting up the transmitter to perform a two-tone test it is important to have the correct levels. If an oscilloscope is available the signal levels can be checked and adjusted more accurately. The composite output signal when viewed on an oscilloscope should look like a 100% modulated AM signal: the two peaks should be exactly the same amplitude and the signal should drop to zero with no discontinuity at the zero amplitude point. However, there are a few undisclosed problems we may encounter!
Fig 7.7: Audio low pass filter.
Correct level adjustment
If the two-tone oscillator is set up correctly, when the output is viewed on an oscilloscope it should appear exactly the same as that shown in Fig 7.2. If one tone has a greater amplitude the waveform resembles an under-modulated AM signal. When the two oscillator levels are identical the waveform will fall to zero at the crossing point. It is simple to achieve this but we often find that when we connect this signal to the transmitter and look at the spectrum it produces a poor result. The RF signal at the transmitter output when viewed on an oscilloscope appears as an under-modulated AM signal. Lots of head scratching!
The reason the signal no longer looks like the correct waveform is that, unlike a wide band audio hi-fi amplifier, the SSB transmitter has some stages which are very narrow band. The two portions of the SSB transmitter generally responsible are the microphone amplifier and the crystal filter. In many cases neither of these stages are truly flat in frequency response.
The microphone amplifier stages may have a pronounced roll-up or roll-off for certain frequencies. Similarly, SSB crystal filters are rarely totally flat across the entire passband. Depending on the type, they can have an upwards sloping response or a downwards sloping response. Both of these sloped responses colour the voice signal and either enhance or degrade the audio quality. Also, if the carrier insertion oscillator is not correctly positioned on the slope of the filter response it also upsets the bandwidth response for high or low frequency voice components.
The only certain way to ensure we have two equal carriers is to adjust the tone levels using the transmitter.
To set up the transmitter we firstly need some test equipment. At a minimum we need an average reading RF wattmeter with a dummy load and an oscilloscope suitable for the frequency the transmitter is designed for. If it is an HF transmitter a 10MHz ’scope will work fine up to about 30MHz. The amplitude response of a ’scope falls off as the frequency gets higher, but the relative amplitude of the signal normally does not. The best method is to start on a lower frequency band such as 80m where the ’scope will give perfect results. At 10m, for the same power output, the trace’s vertical height will be a bit less, that’s all.
Set the transmitter to either LSB or USB and ensure that the speech compressor (or whatever it is called on the transmitter) is switched off. We need just the normal microphone stages operational. Connect the two-tone test box to the microphone connector.
To start with, we turn off one of the audio oscillators so we have a pure sine wave tone going into the microphone stages. We load up the transmitter in the normal manner to produce a CW signal (continuous wave) – simply a pure carrier. Observing the RF power meter we now adjust the microphone gain control to set the carrier to exactly 25% of the rated peak output power. (You need to set the two-tone output level pot to suit the transmitter microphone sensitivity. Most HF transmitters need about 10mV RMS to attain full output with the microphone gain control about midway.) If the transmitter is a 100W PEP type we need to adjust the output to exactly 25W indicated on the wattmeter.
Having set this level correctly we must not adjust the microphone gain control in the later setting up.
The first oscillator is switched off and the second oscillator is switched on. If the second oscillator has exactly the same amplitude the transmitter will show exactly 25% on the power meter. Often it doesn’t. This is because the transmitter microphone amplifier stages or the crystal filter have a gain sloped response. So we carefully adjust the output balance control of the oscillators to obtain exactly 25% carrier level. If we switch between the two audio oscillators when correctly set they should both give exactly 25% carrier level as indicated on the wattmeter. Now the audio settings are correct and we can now, if needed, adjust the microphone gain control for later measurements.
Key the transmitter and switch on both tones and observe the wattmeter. It should now indicate 50% of the transmitter rated power. The wattmeter is reading the average of the power that the two-tone signal is generating. Observe the RF envelope on the ’scope and look for any non-linearity on the peaks of the waveform. If it looks exactly like the AM waveform of Fig 7.2 it means that the transmitter is working within its linear region.
What is more likely is that some minor flat topping will be observed. If so, carefully adjust the transmitter microphone gain control so these just disappear. Read off the average power on the wattmeter. The true PEP is double the average reading, so don’t be surprised if your supposedly 100W PEP transmitter only appears to give 80W PEP under this condition; that is fairly common with most transmitters. If it is a solid-state transmitter expect to see as little as 60W PEP, because transistors are not very linear at the top end of their power range.
The RF envelope test using a humble ’scope is not as accurate as a spectrum analyser, which can measure the exact level of IMD products, but it is far better than relying on meters alone. When the transmitter is set to the optimum linearity, note the current reading for future reference. Also note the average RF power, this is the most you should use to ensure linearity.
Having now learnt how to set up for a distortion test you can try adjusting the microphone gain control to see the effect it has on the waveform. As the microphone gain is increased the envelope will begin to flat-top and the amplitude will rise accordingly. If the gain is advanced too far the waveform becomes very distorted. Note the average power indicated and mentally double the figure to get an approximation of the PEP reading. The PEP reading will not be correct now because the signal is distorted and the average to PEP ratio of 2:1 no longer applies.
Let us assume that when you advance the microphone gain to get a noticeable amount of flat-topping the average power indicated is 70W. Theoretically, if the transmitter were perfectly linear, it would indicate a PEP value of 140W. But this is never going to happen with a nominal 100W PEP transmitter, as you have run out of available power!
Testing an external Linear amplifier
To test an external linear amplifier, we set up the transmitter as previously described in order to obtain a perfect waveform and then use this to drive the amplifier. We go through the same setting up procedure with each tone individually to find the correct 25% power level setting. Switch between the two different audio tones to ensure that the average power of each is identical. Then switch on both tones and observe the RF envelope waveform. If it is showing signs of flat topping back off the RF drive from the transmitter by turning down the microphone gain control. When you have established the optimum setting, read off the average power and double the value. This is the absolute maximum linear power in PEP that the amplifier can produce with low distortion. Again you may be surprised how low it actually is!
Whilst you have the test equipment hooked up, now is the time to connect and set the external ALC adjustments so you cannot exceed this power level. All the other users of the band will thank you for doing so!
Do Not Fiddle with the Settings!
Amateurs, being inquisitive people, often make adjustments to the internal bits of their equipment. That is fine if you are doing an alignment to correct a problem after a component has been changed, but there are the inveterate tweakers who get itchy fingers when they see a pot or some other control. I try to discourage people from tweaking pots or other controls inside the equipment if they do not understand what it does and how it can negatively affect the operation.
In this country there is a famous Japanese make of HF mobile transceiver which is popular with the long distance trucking industry. Some of these pensioned-off mobile transceivers have passed into the hands of newly licensed amateurs (and some not so newly licensed). The official power output of this mobile radio is 100W PEP ±1dB, which means it meets specifications if it outputs a minimum of 80W or a maximum of 120W, to comply with the commercial licence conditions. When the agents set one up for a customer they normally set them to about 80W, so they comply with the law and also they have less reliability problems and hence lower warranty issues at the lower power, and I fully agree with this approach. The transceiver is very clean when correctly set up, as it should be. 1dB is one-sixth of an S-point and you are never going to be able to see that on a receiver.
The problem is that the RF output transistors can deliver up to 180W when fully saturated. The manufacturer chose this particular device for several reasons. Firstly, by using a larger device and under-running it by about 50% they get better linearity and, secondly, the reliability is much higher.
The problem is that amateurs have latched on to this ‘free’ extra power and many transceivers have been tweaked up to over 150W PEP. By turning the ALC control all the way up the transistor runs open loop, so it is very simple to turn up the power on this transceiver. The signal not only sounds grossly distorted by doing this but the tweakers also have a go at the mic gain pot. The splatter these transmitters generate has to be heard to be believed! You can spot one a mile away: when you hear splatter on the band you just know that ‘Walter’ is on the air and using one of them. They also appear to need new PA transistors fitting quite regularly: I wonder why that is? Hopefully, they will all blow up and get consigned to the trash bin!
How Eimac tests linearity
When a manufacturer such as Eimac needs to test the IMD on experimental valves they have a special test fixture which operates at 2MHz. Hence, all the common Eimac valve data sheets are shown for 2MHz IMD measurements. In an Eimac Engineering Bulletin Bill Orr, W6SAI, divulged the details of their in-house test system. The scale of this is mind boggling!
They have two high stability crystal oscillators running at 2MHz with a 2kHz difference in the frequencies. These oscillators are of course valves (what else?!) Each oscillator then has a severe low pass filter to reject any harmonics so each signal is perfectly sinusoidal. They then combine these two high power signals in a special hybrid combiner to get a composite two-tone signal. Finally, this signal is fed to an amplifier valve, but not just any valve. It is – wait for it – a 4CX5000 tetrode, with a variable high voltage supply, running in Class A. The anode load is a 50Ω non-inductive resistor and the output is coupled out via a capacitor. This presents a perfect 50Ω source to the amplifier under test. The spectral purity of this set up is such that it is virtually impossible to see the IMDs – they are very well suppressed. This can, if need be, generate a near perfect signal of several hundred watts to apply to any amplifier needing test data.