5: Operating Modes
The Class B amplifier is used where good efficiency with reasonable linearity is
required. It is most commonly used for audio amplifiers that need to deliver high power but where the distortion is not so important. Over the majority of the output power range it is fairly low distortion and acceptable for the intended purpose. A hi-fi amplifier it is not, but when the distortion does not have to be very low it suffices.
Class B is almost always used in push-pull audio amplifiers, as by operating two valves in anti-phase the distortion products generated in one half are cancelled by an equal and opposite distortion in the other half, at least in theory. When RF amplifiers are used, push-pull operation is not needed: a single ended circuit does everything we need. This is because the resonant tank circuit fills in the missing half cycle due to the ‘flywheel effect’ of the network. Hence, some, but not all, of the distortion is eliminated.
To obtain true Class B operation, the idle current of the valve is just brought to zero by adjusting the grid bias. In this condition the positive half cycle of the input waveform will drive the anode current over the full slope of the characteristic curve. This means that grid current flows over the upper portion of the anode current slope peak of the drive input waveform when the grid is driven positive. This can be seen in Fig 5.1.
Fig 5.1: Class B operating curve.
Although the ideal case is for the anode current to start at zero, very often this introduces gross distortion for very small input signals and so it is normally raised to about 2 to 5% of the peak anode current swing to obviate this problem. The anode current flows for the full 180º of the input signal on the positive half cycle. As can be seen the bottom of the transfer curve is bent over and this curvature causes second order distortion, commonly called harmonic distortion. (Not shown is the full top portion of the slope, this also has a similar curved portion when the anode current is very high. The total anode current slope is S-shaped, the top curved portion being a mirror image of the bottom portion.) By avoiding these operating regions the distortion is kept down to reasonable levels. However, very few, if any, valves have such an ideal slope as shown. Although they are substantially a straight line, some often show a second smaller distinct S-shaped curve over the middle portion when examined closely. As long as the deviation away from the perfectly straight line is not too pronounced the distortion will not be a serious problem. Another diagram of Class B (Fig 5.2) shows the full usable anode current slope with the non-linear portions more clearly.
Fig 5.2: Expanded graph of anode current slope.
The optimum biasing point is generally determined by extending the straight line portion downwards until it intersects the grid bias X axis. From this initial estimate small variations are made to optimise the linearity. It will be noted from the second chart that when the input signal drives the grid to 0V it then intersects the anode current slope at the 50% region of maximum usable anode current. This indicates that the optimum biasing point has been achieved and hence maximum undistorted power can be attained. Another method of finding the optimum biasing point is to adjust the grid voltage until it is zero. If the slope is as shown above the anode current will be at exactly 50% of the peak value.
The drive power required for a Class B amplifier is fairly high as the valve draws grid current over a large portion of the input cycle and hence the driving source must be able to supply sufficient grid voltage swing without a drop in signal amplitude. This implies a source with very low output impedance. To ensure that no input signal waveform distortion occurs this often means that the driving source needs to be able to supply a far greater power than is actually needed to just swing the grid voltage over the required range.
Derivation of Class B Efficiency
Terman derived the efficiency of an ideal valve operating in Class B when amplifying a sinusoidal waveform. The maximum possible efficiency is π/4, which is 0.785 (78.5%). This assumes that the valve is able to swing the anode voltage from the anode supply voltage down to zero volts. However, no valve made is able to do this. Therefore the full equation is modified by the term (Vamin / Vsupply). The full formula is hence:
η = π/4 (1 – Vamin / Vsupply)
For the case of a screen grid valve (tetrode, beam-tetrode, pentode etc) the limitation of how low Vamin can be is defined by the voltage on the screen grid (Vg2). Should the anode voltage swing down to, or below, Vg2 distortion will occur. Hence, if a valve operating with an anode supply of 750V and a screen supply of 250V were considered the maximum efficiency cannot be greater than:
η = π/4 (1 – 250 / 750)
0.785 x (1 – 0.333) = 0.523 (52.3%)
Hence, for maximum efficiency Vsupply needs to be much higher or Vg2 needs to be lower.
In practice, typical efficiency figures are more like 60% for the ideal valve case at low frequencies, such as audio, and when circuit losses are added somewhat less will be achieved in practice. If the waveform is distorted it no longer behaves like a true sinusoidal wave and the equation needs to be modified. This means that apparent better efficiency measurements are normally due to some distortion occurring in the output waveform.
It should be noted that if the amplifier is only driven to half its potential power the efficiency will be half of the peak efficiency calculated, so about 30% is to be expected. This is directly related to the term Vamin / Vsupply. If Vamin only swings down to half the full value (500V instead of 250V) the term (1 - Vamin / Vsupply) becomes smaller. In the above example the efficiency will be ~26%.
Another factor is how much the anode supply voltage droops under the peak anode current, if this is more than a few percent it slightly improves the efficiency, as the value of Vsupply is taken at the instance of peak anode current (I max) when Vamin is attained. Hence, there is not such an importance placed on the power supply regulation for Class B amplifiers.
The power output possible for ideal valves in a push-pull Class B amplifier is given by:
Po = (Imax Vsupply / 2) x (1 – Vamin / Vsupply) watts
Where Imax is the peak anode current in each valve.
Again, no circuit losses have been considered, which will dissipate some of the potential output power.
Class AB
To improve linearity, and lessen the drive power required, a change in the biasing method is often used. This now becomes a mixture of Class B and the more linear Class A and hence it is called Class AB. In this biasing method the valve behaves like a Class A type at small input drive levels and moves into the more efficient Class B as the input signal increases. There are two distinct types of Class AB, one in which no grid current flows and the other where there is some grid current.
The first type is known as Class AB1 and it behaves more like a true Class A amplifier over a wider range of anode current. The penalty we pay is that the anode dissipation is higher than that of Class AB2 and the idle current is much higher. Hence, the efficiency is lower than AB2.
In Class AB2 the same anode slope is utilised but the idle current is higher than the normal Class B. This means that the total anode dissipation is higher and as such we cannot obtain the same peak power without exceeding the anode dissipation rating. However, for an intermittent input signal type such as SSB this isn’t a serious limitation.
The deciding factor of whether it is AB1 or AB2 is whether any grid current flows. If grid current does not flow over any part of the input waveform, it is Class AB1. If any grid current flows over a substantial part of the input waveform it is AB2. Unfortunately, there are many permutations in this method and if a tiny amount of grid current were to occur just at the peak of the input waveform strictly speaking it has to be classed as AB2, when in reality for 99.9% of the input cycle no grid current flows. This led to some calling it Class AB1.5, which is nonsense.
Many of the popular tetrode valves for RF amplifiers show a small but significant improvement in IMD products if a little grid current flows, even though they are biased into Class AB1. The chart of the two classes is shown in Fig 5.3. As can be seen, the AB2 mode needs more driving signal but runs the anode current higher up the slope, and hence allows more output power. Although Class AB2 runs the anode current further up the anode current slope, it then begins to get closer to the top curvature region. The further we push the anode current up the slope, the greater the distortion products generated due to the slope bending over at the top.
Fig 5.3: Class AB types.
How far we can push the current will depend on whether the intermodulation distortion products (IMDs) are still acceptable or if we are exceeding the safe anode current and dissipation rating of the valve. It is tempting to keep on pushing the current upwards but there is a finite upper limit we must not exceed. Whether it is the IMDs or the dissipation that dictates the maximum will depend on the valve and the anode voltage used. Usually the IMD figure steps in first and to push the drive up any further will generate excessive splatter.
Tetrode Valve Limitations
The screen grid type of valves such as tetrodes, pentodes and beam tetrodes have a particular inherent problem when biasing for Class B, AB1 and AB2. If the chart in Fig 5.4 is examined this shows an expanded version of the bottom portion of the anode current transfer curve.
Fig 5.4: Lower transfer curve curvature diagram.
The point P on the grid voltage axis denotes the ideal biasing point. At this biasing point the anode idle, or standing, current is IaS on the anode current Y-axis. This corresponds with the point Q in the middle of the curvature at the lower end of the anode current transfer curve. The maximum linear anode current is denoted by point C. The upper curvature is not shown in the diagram.
The true straight-line portion of the anode transfer slope begins at point B. This is approximately where a Class AB1 amplifier would be biased; however, the biasing chosen is for Class AB2 as it gives better efficiency and output power. Between points A and B the curve is predominantly second order and will introduce harmonic distortion generation. Over the greater portion of the anode slope, from B it is a straight line, so it is substantially linear. For true Class B operation the correct biasing point would be point A. But this introduces more distortion and it often needs to be more aligned towards point Q. A bias point midway between A and Q is usually chosen as a compromise when screen grid valves are used.
Screen grid valves, such as a beam tetrode, are sensitive to the screen grid voltage. If the anode constant current curves are examined in detail for different screen grid voltages the relationship between the anode current and screen grid voltage can be plotted on a separate chart. It shows that the anode current is heavily dependent on the screen grid voltage and has a 3/2 law. This means that if the screen grid voltage is raised from, say, 300V to 500V the anode current doubles for the same control grid voltage and anode voltage. The reason this occurs is that the screen grid is acting as a better accelerator of the cathode electrons flowing towards the anode.
To correct for this, the anode current has to be lowered to keep the anode current and dissipation within the safe limits for the valve. Increasing the grid bias voltage can do this, so that it is more negative. This restores the idle current back to the required value and sits it firmly in the bottom of the lower anode transfer curve at point Q. As the grid voltage is now much higher, a greater grid driving voltage is required and, as grid current flows under AB2 operation, this means a higher drive level is required.
This goes against all the apparent reasoning for raising the screen grid voltage to obtain more gain and output power. The output power to a first order is the anode voltage swing and the ability of the valve to draw sufficient current. If the anode voltage is low, the anode voltage swing and current will also be low for the same drive signal. For a typical valve, if the anode voltage is raised from 1kV to 2kV the potential output power increases by a factor of four.
The other major problem of raising the screen grid voltage is that the anode voltage cannot swing below the screen grid voltage by a great amount. If the anode voltage swings below the screen grid voltage a virtual cathode exists, the screen grid tries to assume the role of the anode and it draws excessive screen current. This diverted anode current causes the anode voltage waveform to flatten at the top, as the anode current cannot swing low enough to follow the correct sinusoidal shape. The result is that distortion products are generated when this occurs.
By raising the screen grid voltage it brings the point at which you can swing the anode voltage downwards to a higher voltage and hence limits the anode current it is possible to draw without excessive screen grid current flowing. Lowering the screen grid voltage has the opposite effect. The anode can now swing the voltage lower before reaching the screen grid voltage. This effect was exploited in the G2DAF linear amplifier, which we will cover later.
Class C
Although the Class C amplifier is unsuitable for modes requiring a linear transfer curve, it has its uses for AM anode modulated transmitters as well as FM and CW modes. The basis of operation is that the anode current is cut-off by a large negative grid voltage and the anode current only flows in narrow pulses for a portion of the positive grid signal waveform. The anode current transfer curve is shown in Fig 5.5.
Fig 5.5: Class C anode current curve.
Normally the grid bias is set to be about twice the value of that required to just cut-off the anode current. This means that on the positive half of the input-driving signal no anode current flows until the grid voltage approaches the 0V grid-cathode point. Once the 0V point has been exceeded the anode current is driven right to the top of the slope and over the top by a small amount. This places the anode current into saturation and any small variation in input signal level has little effect on the anode current and hence output power.
The grid driving power is the highest of all the operating modes, but as the grid resistance is fairly high the power required is not excessive. The gain that a typical Class C power amplifier can produce is from around 20 times (13dB) to about 40 times (16dB), depending on the frequency and power output level. The grid current is fairly high and hence a valve designed to cater for this high current is essential. Fortunately, at the power levels that amateurs use this means that many normal valves can be operated in Class C without serious problems. However, there are some valves that must not be used with high grid current as the grid structure can be damaged.
Of all the operating classes, Class C has the highest efficiency and a figure of 75% or a little higher is common. In a really efficient anode circuit it is possible to see figures a little above 80% at low frequencies.
Although the anode current pulses are much less than the 180º of the input signal, and hence are rich in harmonics, the resonant anode tank circuit restores the input signal waveform to a near perfect sinusoidal waveform due to the flywheel effect. Only harmonics of the fundamental frequency exist at the anode. These are largely suppressed by the tank circuit loaded Q, and any further suppression can be made with a low pass filter connected in series with the output if they are deemed to be excessive. In a correctly proportioned tank circuit, when the anode circuit is brought to resonance the harmonics normally dip as the tank circuit is adjusted to maximum output power.
The topic of anode tank circuits, and the need to select the correct value of loaded Q, is covered later in another chapter.
Controlling the RF output power
In the Class C amplifier the normal method of driving the grid of a beam tetrode is to supply a fixed carrier level from the driving stage. The grid bias can be either by a fixed bias supply, which has some variation to set the correct cut-off voltage, or to utilise the grid current flowing in a resistor connected from the grid to ground. Of the two this is the simpler method. Since the anode current is dependant on the grid bias voltage if the driving signal is removed the grid voltage rises to 0V and the anode current will rise to a maximum, as the grid bias voltage developed across the grid resistor has now disappeared.
To prevent this condition, a common method with CW transmitters is to arrange some method so that if the grid bias fails the screen voltage is brought to a low voltage, so reducing the anode current to a low level. This technique is called a clamp stage. A simple clamp stage using a triode to control the screen grid voltage of the RF amplifier is shown in Fig 5.6.
Fig 5.6: Simple clamp circuit.
The triode valve connected to the screen grid of the RF amplifier pulls the voltage down when it conducts fully. When the driving signal is applied to the grid of the output stage the negative grid bias developed biases off the triode and the screen grid voltage rises to the maximum set by the resistor R from the high voltage supply. When the key is up, no grid bias is available and the clamp valve pulls the screen grid down to close to ground, reducing the PA valve anode current to a safe value. Where a variable RF output is desired, a minor modification allows some variation of the screen voltage, and hence the anode current, and this is shown in Fig 5.7.
Fig 5.7: Clamp circuit with variable output control.
The clamp valve is a small beam pentode and the grid bias developed in V1 is fed via a level setting pot to the grid of V2. By adjusting VR1 the screen grid voltage applied to V1 can be set to suit the output power required. Using an arrangement such as this circuit the RF output can be varied between about 10% and the maximum the amplifier can deliver.
Power output capability of a valve
A question which is often asked is: “How much RF power can I get from a particular valve?” The simple answer is: “It depends on what sort of amplifier you are intending to use and the frequency of operation.” The question is a bit like “How long is a piece of string?” We need more data to give an accurate estimate of the power likely to be achieved.
It is important to understand how valves are rated for power. In items such as internal combustion engines and electric motors the general way of rating them is to quote the power output in either horsepower or kilowatts. Valves are generally not rated this way, although the manufacturer will often give an example to show its capability. Valves are rated in what they can not deliver as power. Although this seems a strange way to rate them, it is for a very logical reason. The reason is because of the efficiency at which the valve can be safely operated in a practical circuit.
Efficiency
Efficiency is the measure of how much power output we can achieve for a certain amount of power input. It is given in percentage terms. If all the DC power fed into a valve were available in some other form, for example audio or RF power, then it would have an efficiency of 100%.
The Conservation of Energy Law states that: “Energy may be neither created or destroyed. It may only be converted from one form to another.” The crux of this law is that although the power output is different to the power input, the difference between the two can always be accounted for.
In a valve, or any other device which converts energy in to energy out, the difference is usually found to be heat generated within the device during the conversion process. In valves this is termed the anode dissipation, and it is expressed in watts. The DC power fed into the device that could not be converted into output power of the type required appears as heat. It is only ‘lost’ insofar as it isn’t in the form that we would like: it is fully accounted for as thermal energy. The efficiency (η) is generally considered to be:
Power out
η = x 100%
Power in
If a DC input power of 150W is fed into an amplifier and a power of 100W appears at the output, the efficiency is 66.66%.
Anode Dissipation
The vast majority of the lost power is attributed to be consumed by the anode. This isn’t strictly true: some may be used to heat up other parts of the valve, but the concept is a viable way of expressing the mechanism. If no power were lost in the anode then the conversion process would be perfect and the efficiency would be 100%.
How much power a particular valve can convert from the DC input power to useful output power is solely determined by the safe anode dissipation of the valve. If the power being dissipated is greatly in excess of the manufacturer’s maximum rating it will overheat and damage the valve. The valve can only get rid of the heat energy up to a certain rate, by transferring the heat energy to another object. The cooling of the anode, and other parts of the valve, is therefore an important factor in determining how much anode dissipation is possible.
Calculating output power from anode dissipation
The formula to convert the anode dissipation rating in watts to input power uses the formula:
1
Pin = x Pdiss
(1 - η)
Where:
Pin is the DC power input in watts
Pdiss is the rated anode dissipation
η is the numerical efficiency (e.g. 75% = 0.75).
For a valve such as the RCA 811, the maximum CCS anode dissipation is 45W. Assuming an anode efficiency of 75% for a Class B audio amplifier, it can safely handle 180W DC input.
The output power is simply the input power multiplied by the efficiency:
180 x 0.75 = 135W
The anode dissipation is (power input – power output) = Pdiss
(180 – 135) = 45W.
This assumes that the power output is continuous. When the output is not at 100% rating intelligent derating can be applied to obtain a safe working condition. We call this the duty cycle.
Duty Cycle
In the case of Morse code the dot-space ratio is 1:1, so if a string of dots is transmitted the transmitter is on for half the time and off for the remainder. Hence, the duty cycle is 50%. Under this condition, using the example above, it could be safely doubled for the DC input and RF output power without exceeding the 45W anode dissipation. The dissipation quoted is the average value, so although during the transmission of the dot it is dissipating twice its rated dissipation, during the space period the dissipation is zero. It has a resting period where the heat can be transported away.
In radar and similar pulsed transmitters the power level is very high during the transmit pulse, but the time between pulses is long. Therefore, the valve has ample time to move the heat away and the average anode dissipation is low.
When assessing how far the envelope can be pushed before the valve is unduly stressed, we need to know the duty cycle of the various modes in common use. For Morse code if a string of dashes is transmitted the on time is 75% of the total time, because the dash to space ratio is 3:1. Hence, in this mode we could only increase the DC input power by 1.33. For FM, where the carrier is at 100% during transmissions, clearly the 45W anode dissipation is the ruling factor and running at more than the safe DC input power would be unacceptable.
For single side band operation the average voice has a duty cycle of between 30 to 40% of the peaks and so the rating could be increased quite a bit before the average anode dissipation is exceeded. If, however, a speech processor is used this raises the average voice amplitude to about 60% and dictates how much extra power we could safely run.
Amateur transmitters, however, are not normally run continuously: they have short periods of transmitting and about the same period receiving, so the peak power could be increased a little. If the equipment is used in a contest things are very different, especially if heavy speech processing is used. These factors determine how much ‘intelligent abuse’ the valve is able to tolerate.
Potential Anode efficiency
It has been calculated from first principles by several sources what the maximum anode efficiency can be for the various conduction modes in common use. The popular operating modes are Class A, Class B and Class C. Added to these are two other variations of Class AB1 and Class AB2, which are combinations of Class A and B. Which operating mode is used will depend on the application and other factors.
Terman first calculated these maximum values and several others have confirmed the values as being the upper limit for an ideal valve. However, nobody makes an ideal valve, only practical valves are available. Hence, although useful, it is simply a guide as to the absolute maximum efficiency it would be possible to achieve if the ideal valve were available.
Generally, the anode efficiency falls as the frequency rises, so a value for an audio amplifier will always be better than an HF amplifier and a VHF / UHF amplifier. To quote from manufacturers’ valve data sheets for the three types of operation is not usually possible, as a particular valve is not usually made to cater for all types. However, sufficient data does exist for operation at areas of frequency, for example audio and low RF or HF and UHF.
A typical high power valve at audio frequency will approach the theoretical value but fall short by a small amount. Similarly, a microwave valve at low RF frequency is notably better than at the upper range in the microwave bands. The 2C39A triode at 3GHz has an efficiency of about 25 to 30% at best for Class C operation, at 144MHz its efficiency is up around 70% and at 432MHz it is typically 60%. This trend is common for all valves. Typical theoretical values for the various operating conditions can be stated:
Class C has the highest at 84%;
Class B is next best at 78.5%;
Class A is the worst at about 25 to 30%.
The in-between modes of Class AB1 and AB2 are similar, Class AB2 being slightly better at 65% and Class AB1 at 55%. However, it must be repeated that these are the maximum theoretical values when operated at low frequencies. At higher frequency these values are lower, sometimes much lower! Generally, the higher the linearity of the amplifier the lower the potential efficiency.
Circuit efficiency
The overall efficiency of a power amplifier is not entirely determined by the anode efficiency of the valve. There are other factors that add extra loss to the transformation process of DC input to RF output. The main factor is the inherent loss in the anode network. At audio and low RF frequencies these losses are fairly small and can often be ignored. Such is not the case when the frequency is very high. At UHF and above the anode network losses can be significant and these can eat into the available power to feed the antenna.
As has already been mentioned, the valve manufacturers often give two different RF power output measurements in their data sheets. These are the Available anode power and the Useful output power. The first of these is the calculated power exiting the anode, the other is the actual power measured at the amplifier output connector. The difference between the two is the power lost in the anode transformation network, which can be measured without the valve in circuit.
Cooling
Cooling is one of the most critical areas in any high power amplifier design. If the heat generated in the valve is not removed quickly enough it will exceed the safe operating temperature.
When a valve is designed the manufacturer decides on the optimum cooling method to suit the application likely to be used. For the smaller glass envelope type of valves often simple convection cooling is sufficient to keep the valve below a certain maximum temperature. For large, higher power valves the envelope has a larger surface area and sometimes convection cooling is adequate. Glass envelope valves will dissipate the heat more readily if a gentle stream of cool air is blown on to them. Often the volume of air required is not large and this is sufficient to cool the valve. In others, special glass chimneys are made that fit over the valve and a greater volume of air is blown in from beneath the valve to form an efficient heat transfer method.
For the physically small valves, where the available area is small, often a finned cooling radiator is fitted to the anode and air is forced through this to keep the anode at a safe temperature. In other high power valves air cooling is simply not enough and liquid cooling jackets are fitted to transport the heat away.
If the rate at which heat can be transported away and dissipated is much greater than the valve generates it is then feasible to increase the anode dissipation to a much higher level. The 2C39 series of microwave triodes are fitted with a finned anode cooler that allows a maximum anode dissipation of 100W. However, it is simple to replace this finned cooler with a liquid cooling jacket. With a liquid cooling system the anode dissipation can be raised by as much as four times over the normal air cooling.
Experiments carried out by amateurs with liquid cooling showed that using this technique allows more latitude in the intelligent abuse factor. The classic N6CA 23cm amplifier using the 2C39BA ceramic microwave triode has been widely copied and it raises the potential power from a puny 45W to 250W, when used intelligently for CW operation. Valve life is shortened a little, but 2 to 3 years from a valve is easily achievable, as we don’t transmit 24 hours a day.
Actual efficiency
Although the calculation of the efficiency is an easy factor to determine, we may be surprised that what we achieve in practice is much lower than expected. It is easy to measure the DC input power: simply measure the anode current and the voltage and multiply the two together. It is also simple to measure the RF power output using a calibrated wattmeter. But there are cases where we get unexpected results.
Suppose the amplifier in use has a known efficiency of 60% when driven to its full rated output. The operating mode is Morse code. When the amplifier is switched into the transmit mode, but before the key is pressed, it is showing some idle current on the anode current meter, and of course the RF wattmeter shows zero output. What is the efficiency under this condition? The answer is 0%, because although the amplifier is drawing power there is no output power.
Now assume we send a long string of dots that are correctly spaced with the dot length being the same time period as the space period. What is the efficiency now? When the key is down the amplifier supplies full power at the measured efficiency, so it is operating at 60% efficiency. But in the key up time it has an efficiency of 0%. Therefore the average efficiency is the maximum divided by two or 30%. The same holds for the average power output. When the key is down it transmits, say, 100W but with the key up it is of course zero power. Hence, the average power is 50W.
If we now send a long string of dashes, and again we assume the correct dash to space period is observed, what is the efficiency figure under this condition? The key down dash period is three times longer than the space period, so the duty cycle is 75%. Hence the average efficiency is 60% x 0.75 = 45%. The average power is 75W.
When using SSB as the input the same argument applies. SSB with a human voice has a duty cycle of about 30% average to peak ratio. So the average efficiency we attain is 60% x 0.3 = 18%. Although this sounds terrible it needs to be viewed in the correct context. Under this operating condition the anode dissipation also averages only 30% of the full key down condition. For a mode such as FM, where the carrier is at 100% of the power, the anode dissipation is the maximum.
Peak efficiency
In an RF amplifier the anode-matching network is normally adjusted to deliver maximum power when the valve is fully driven. This means that when the valve is not driven up to its maximum input power the efficiency suffers. This is because the matching has been adjusted to suit the maximum power, at lower power outputs the valve is now not correctly matched to deliver the full potential power. This scales almost linearly in practical amplifiers. As with the Morse code dots example above the same effect applies.
If the valve is only driven to give half the peak power the efficiency falls to 50% of what it would be at full peak power. If the peak efficiency is 60% the valve only delivers 30% efficiency for that particular DC power input. This is not generally a problem as the anode safe dissipation rating is far higher at full power output. We can plot the efficiency curve against power input and it is essentially a straight line that starts at 0% and meets the maximum efficiency point at the top of the line.
Anode dissipation under idle conditions
The allowable anode dissipation when a valve is biased into a linear mode is of concern. Normally, the idle current is set to quite a high value so that the valve under low drive conditions is running in Class A. As the drive is gradually increased it moves into the Class B operating region. The operating mode with the highest idle current is Class AB1. Finding the optimum idle current is often a matter of experimentation. On the one hand we wish to have good linearity at small drive levels but the anode dissipation is a concern. If we lower the idle current then we reduce the anode dissipation but we may reduce the linearity for small drive levels. Fortunately, Eimac came up with the definitive answer to this problem.
Eimac recommends that when a valve is biased for Class AB1 the idle current should be set so that the anode dissipation level is at only 80% of the valve’s maximum anode dissipation. This is the best compromise setting. This means that for a valve with a maximum anode dissipation of 250W, such as the 4CX250 series, the anode dissipation will be 200W. For the recommended DC anode supply voltage of 2kV the 4CX250 valve should draw 100mA of idle current.
If we examine the Eimac data sheet for the 3-500Z triode we see their recommendation for operation in Class AB2. The 3-500Z is most linear with an anode supply of around 3kV and when idling at this voltage it draws 62mA. This is an anode dissipation of 186W. However, to provide this state the cathode voltage needs to be raised to +10V instead of the normal zero bias condition. If the anode supply is lowered to 2.5kV the +10V extra bias can be dispensed with. It now draws an idle current of 130mA when run in true zero bias condition. Under this condition the idle anode dissipation is 325W.
If the anode supply is raised to the maximum of 3.5kV the cathode bias needs to be raised to +15V to keep the anode current in check. This idles at 53mA, a dissipation of 185W.
Under full drive at the 3kV supply it draws a single tone DC input power of 1200W and typically outputs 740W. The anode dissipation is therefore (1200 – 740) = 460W. At the lower anode voltage of 2.5kV it draws 1kW DC input and outputs 600W, a dissipation of 400W.
If we choose the maximum anode supply of 3.5kV, the DC input is 1400W, the output is 890W, and the anode dissipation is 510W. The maximum anode dissipation is stated to be 500W.
In all of those modes it needs about 45W of drive power when in grounded grid configuration.
Grounded Grid Amplifiers
These are covered in more detail later but one interesting factor about them is the apparent efficiency possible. When measuring the power output and power input it gives an efficiency figure greater than seems realistic. This was eventually traced to the fact that some of the input drive power is feeding through the valve from the cathode to the anode and adding to the anode power. If an excessive amount of drive were used an increase in output power of as much as 11% is possible.
As a result of this the FCC in America changed the wording of the amateur licence to cater for this effect. They stated that if the percentage of feed through power was above a certain value the DC input to the anode needed to be reduced to keep within the allowable 1kW DC input level.
In a grounded grid amplifier the cathode and the anode signal are in-phase and add. In the grounded cathode amplifier they are 180º out of phase and hence the feed through power is subtracted from the anode power.
This was in the days before RF output could be measured accurately and transmitters were of the AM and CW types, i.e. prior to SSB, where only the output power was measured. Today DC input power isn’t used except, in some countries, for FM and CW modes. If you have to use a DC input power of 1200W to attain 400W PEP that is of no concern, it only affects the operator who has to pay the electricity bill. Many of the earlier pieces of commercial amateur equipment were specified as PEP input and not output. This made the equipment look better, but when the efficiency factor was considered they often gave less than half the power of the quoted input figure. The early Yaesu FTdx500 was an HF transceiver rated at 500W PEP input: it typically gave about 180W PEP output.
