Chapter 18: The Transistor Amplifier

The power of a signal

Amplification means increasing a signal’s power, P = V × I, by raising the voltage, the current or both. A step-up transformer raises the voltage yet gives no amplification, because the current falls in step.

40

Power is V × I, so with the voltage unchanged and the current 40 times larger the power gain is 1 × 40 = 40. Unity voltage gain does not mean there is no amplification.

1

The input is 3 V × 40 mA = 120 mW, and an ideal transformer delivers the same 120 mW, now at 12 V and 10 mA: a power gain of 1. Its voltage ratio of 4 is not amplification.

10¹²

From −110 dBm to +10 dBm is 120 dB, and each 10 dB is a factor of 10 in power, so the factor is 10¹². Stopping at 0 dBm gives only 110 dB, 10¹¹.

To make up for losses in parts such as filters

Some stages, such as filters, lose part of the signal, so the receiver’s total gain must exceed the bare 10¹³ or so needed, to make up for them. Gain itself does not make a receiver more selective.

Its emitter is part of both the input and output circuits

The input goes to the base and the output comes from the collector, so the emitter belongs to both circuits. The stage has one transistor; common means shared by input and output, not by two transistors.

It passes DC but blocks the RF signal

The inductor passes DC, so the collector is biased positive and supply current can flow in, while its reactance blocks RF. Passing RF and blocking DC is the output capacitor’s job.

It passes the amplified AC on but blocks the DC

The output capacitor passes the AC signal but blocks the collector’s DC, so the bias does not upset the next stage. It is in series with the output, not part of a tuned circuit.

From the DC supply that feeds the stage

Power cannot be made from nothing: the small input only controls a larger current drawn from the DC supply, in the manual’s circuit the +12 V line via the inductor. The input itself supplies little power.

None: the transistor never conducts

A silicon transistor needs about 600 mV across its base-emitter junction before it conducts, and 0,4 V never gets there, so no collector current flows. Class C needs a large input signal.

27,6 V

As the transistor switches off, the inductor keeps its current flowing and drives the collector above the supply, by the manual up to about twice it: 2 × 13,8 = 27,6 V. It is not capped at 13,8 V.

It drops below the supply voltage

Collector current through the inductor makes it develop a voltage that pulls the collector below the supply. The collector rises above the supply only when the transistor switches off again.

Less than half of it, under 180°

Class C conducts for under half of each cycle, less than 180°: in the manual’s stage only while the input is above about 600 mV. Exactly half, 180°, is class B; the full cycle is class A.

The RF power stage of CW and FM transmitters

Class C distorts badly but is efficient, which suits CW and FM power stages, where a lowpass filter removes the harmonics. Amplifying an SSB signal needs a linear amplifier, which class C is not.

A linear amplifier

AM and SSB signals must keep their exact shape, so they need a linear amplifier that reproduces the waveform faithfully, such as class A or a good class AB stage. More gain cannot cure distortion.

A lowpass filter after the stage easily removes its harmonics

Distortion creates harmonics at multiples of the wanted frequency, above it, so a lowpass filter after the stage removes them easily. A highpass filter would pass the harmonics and block the signal.

60 to 70%

In class C most of the supply power reaches the output; the manual calls 60 to 70% common. About 25% is its figure for class A, which wastes power on a constant bias current.

80%

Efficiency is output ÷ DC input = 41 ÷ 50 = 0,82, or 82%, so 80% is nearest. The 9 W lost is heat in the transistor; 18% is that wasted share, not the efficiency.

Class A

A transistor that conducts for the whole cycle, 360°, is working in class A, as in the manual’s biased common-emitter stage. Class AB conducts for more than half the cycle but not all of it.

2,4 V

VB = 9 V × R2 ÷ (R1 + R2) = 9 × 12 ÷ 45 = 2,4 V. Working it over R1 gives 6,6 V, which is the drop across R1, not the voltage on the base.

2,2 V

VB = 12 V × 18 ÷ (82 + 18) = 2,16 V, so 2,2 V is nearest. Putting R1 in the top of the fraction gives 9,84 V, the drop across R1, nearest the 10 V option.

2 mA

The emitter sits about 0,6 V below the base, at 2,6 − 0,6 = 2,0 V, so IE = 2,0 V ÷ 1 kΩ = 2 mA. Forgetting the base-emitter drop gives 2,6 mA.

10 µA

The emitter carries IB + IC = IB + 149 IB = 150 IB, so IB = 1,5 mA ÷ 150 = 10 µA. The 1,49 mA is the collector’s share, the other 149 parts in 150.

It rises by under 1%, so barely changes

IC = IE × β ÷ (β + 1): 1 mA × 100 ÷ 101 = 990 µA, and with β = 500 about 998 µA, under 1% more. The bias, not β, sets the current, so IC ≈ IE.

6 kΩ

The supply is at AC ground, so R1 and R2 appear in parallel: 15 × 10 ÷ (15 + 10) = 6 kΩ. Adding them as if in series gives 25 kΩ, the DC path through the divider.

7 kΩ

R1 and R2 act in parallel for the signal: 56 × 8,2 ÷ (56 + 8,2) = 7,15 kΩ, so 7 kΩ is nearest. Adding them in series gives 64,2 kΩ, nearest the 65 kΩ option.

The supply line is held at AC ground by the decoupling capacitor

Decoupling capacitor C3 gives the supply line an AC path to the chassis, so for the signal the top of R1 is grounded, like the bottom of R2. C1 only couples the signal to the base.

17

Voltage gain = RL ÷ RE = 8 200 ÷ 470 = 17,4, so 17 is nearest. Squaring it gives about 300, which is the power gain (with equal input and load resistances), not the voltage gain.

26 dB

With equal resistances the power gain is the voltage gain squared, 20² = 400, and 10 log 400 = 26 dB. Taking 10 log of the voltage gain alone gives only 13 dB.

Both voltage gain and current gain

A common-emitter stage can raise both the voltage and the current of a signal, so its power gain is large. Current gain without voltage gain describes the emitter follower, whose voltage gain is about 1.

2,6 V

The emitter sits at 3,2 − 0,6 = 2,6 V, so the input can swing 2,6 V negative before the emitter reaches 0 V and conduction stops. Forgetting the base-emitter drop gives 3,2 V.

20%

Efficiency = power to the load ÷ power from the supply = 2 ÷ 10 = 20%. The other 8 W, 80% of the input, ends up as heat in the transistor.

About 25%

The constant bias current wastes power, so typically only about a quarter of the supply power reaches the load and the rest heats the transistor. 60 to 70% is the manual’s class C figure.

So little power is involved that the waste is unimportant

Class A wastes much of its supply power, but a small-signal stage handles so little power that the waste is unimportant, and class A copies the waveform faithfully. It draws the most idle current, not the least.

Quite high, like a current source

Collector current is set by the base and hardly depends on collector voltage, so the collector behaves as a current source of fairly high impedance; L1 also has high reactance. Low output impedance is the emitter follower’s.

Its emitter voltage follows its base voltage

Its output comes from the emitter, which stays about 0,6 V below the base and so rises and falls with it, giving a voltage gain of about 1. The collector is the common terminal, fixed at the supply.

1,8 V

A change at the base gives an equal change at the emitter: the voltage gain is about 1, so the 1,8 V swing comes through whole. The 0,6 V drop is a fixed DC offset, not taken off the swing.

Low, so its output voltage barely changes with load

An emitter follower’s output impedance is fairly low, so it acts as a good voltage source whose output hardly depends on the load. High output impedance, a current source, belongs to the common-emitter and common-base stages.

High, so it draws little from that stage

An emitter follower’s input impedance is high, so it draws little from the stage before it; that, with its low output impedance, makes it a good buffer. Low input impedance is the common-base stage’s.

A buffer, so later stages do not upset earlier ones

With high input and low output impedance, an emitter follower makes a good buffer: changes in the stages after it do not reach the stage before it. Its voltage gain is only about 1.

4,4 V

A conducting silicon transistor keeps its emitter about 0,6 V below its base: 5,0 − 0,6 = 4,4 V. The emitter follows changes at the base, but always about 0,6 V lower.

When it delivers more current than it takes from its source

Its voltage gain is only about 1, but power is V × I, so it gives power gain whenever its output current exceeds its input current, which it easily does. Drawing more current than it delivers would be a loss.

Very low, a few ohms or tens of ohms

The common-base input is very low: the manual says typically only a few ohms, and in practice it is about 25 Ω at 1 mA of emitter current. High input impedance is the emitter follower’s.

High, many kilohms or more

Like the common emitter, the common-base stage has a high output impedance, making it a good current source. A low output impedance and good voltage source describe the emitter follower.

One: its output current equals its input current

The common-base stage has a current gain of one, which is why the manual calls it a current follower; in practice it is just under 1. A current gain of about β belongs to the other two configurations.

When its load resistance is greater than its input impedance

With unity current gain the same current flows in and out, so by P = I²R the power rises in proportion to resistance: a load larger than the input impedance gives power gain. The reverse gives a loss.

200

With equal currents, P = I²R makes the power gain the ratio of the resistances: 5 000 ÷ 25 = 200, about 23 dB. Unity current gain does not rule out power gain.

Less widely used than the other two

The manual calls common base the third and less common configuration, behind common emitter and common collector, though it has its uses as a current follower. It is an amplifier, not a rectifier.

Class AB

Between 180° and 360° is class AB: each device is biased slightly on, so it conducts for a little over half the cycle (the manual says about 180°). Exactly 180° is class B, the full 360° class A.

Exactly half of it, 180°

Class B conducts for exactly half the cycle, 180°, and the manual notes it is not commonly used. A little more than half, between 180° and 360°, is class AB.

It keeps a small current flowing in both transistors

VBias is just enough to keep both transistors conducting a little with no input, so neither starts from fully off; this reduces crossover distortion. Holding them fully off would be class B behaviour.

Its device never cuts off at any point in the cycle

A conduction angle of 360° means conduction for the whole cycle, class A: bias current keeps the device from ever cutting off. That constant current makes class A the least efficient class, not the most.

Class A

Class A conducts for the whole cycle, so it copies the waveform faithfully with little distortion. Class AB comes next but adds some crossover distortion; class C distorts heavily.

Crossover distortion

Crossover distortion appears where one device stops conducting and the other takes over; good design keeps it within reasonable limits. Clipping is the flattening of peaks when a stage is overdriven.

It feeds the bases in opposite phase

The input transformer is a phase splitter: its two halves drive the bases 180° out of phase, so one transistor conducts harder as the other turns off. Adding the collector signals is the output transformer’s job.

It adds the two collector signals together

Each transistor handles about half of each cycle, and the output transformer adds the two collector signals back together, so the output has both half cycles. Feeding the bases in opposite phase is the input transformer’s job.

It suits AM and SSB signals as a power stage

Class AB is much more efficient than class A and much more linear than class C, so it suits AM and SSB power amplifiers. Being too nonlinear to amplify AM signals is class C’s failing.

Class C

Class C conducts for only part of each cycle, so little power is wasted, and the manual calls 60 to 70% efficiency common. Class AB beats class A but cannot match class C.

It is controlled by a voltage rather than a current

A FET is controlled by the voltage on its gate, which draws almost no current, while a bipolar transistor needs base current. The close option describes the bipolar transistor instead.

FETs need voltage bias rather than current bias

A FET’s gate takes a bias voltage and draws almost no current, so networks designed to feed a base current may need changing. It is the bipolar transistor that needs bias current.

The drain

In general the gate replaces the base, the drain the collector and the source the emitter. Emitter and base are bipolar transistor terminals, which a FET does not have.

Gate, source and drain

A FET’s electrodes are the gate, the source and the drain; a voltage on the gate controls the drain-source channel. Base, emitter and collector are the bipolar transistor’s names.

The source follower

The source takes the emitter’s place, so the emitter follower becomes the source follower, also called common drain. The drain stands in for the collector, so a drain follower is no emitter-follower counterpart.

Large-scale integration makes them cheaper to produce

Making MOS devices costs less and goes more smoothly in large-scale integration, so FETs dominate logic and computer chips. They are still made of doped semiconductor, like junction transistors.

Switching at high power levels

MOSFETs are now everywhere in high-power switching, where they have almost completely displaced junction transistors; the manual’s summary also names logic and RF power amplifiers. It makes no such claim for small-signal audio.

The doping of its channel: N-type or P-type

FETs come in two basic types, named after the semiconductor type of the drain-source channel: N-type or P-type. The supply polarity follows from the type but does not name it.

An extremely thin layer of oxide

MOS means metal-oxide semiconductor: an ultra-thin oxide insulates the structures (the manual names aluminium oxide; silicon dioxide is the classic choice). A forward-biased PN junction conducts, so it cannot insulate.

RF power amplifiers of 1 kW and above

Semiconductors have largely replaced tubes, but tubes survive in high-power RF work such as linear amplifiers of 1 kW and above, and are robust enough for home-built amplifiers. Receiver front ends use semiconductors.

Their harmonic distortion pleases some listeners

Tubes distort differently from solid-state devices, and some listeners find their kind of harmonic distortion more pleasing. They do distort; the difference is in the character of the distortion.

The cathode

The heated cathode emits electrons, which form a cloud around it; the anode (plate in American usage) collects them when it is positive. The anode receives electrons rather than giving them off.

When its anode is positive relative to its cathode

A positive anode attracts the electrons given off by the heated cathode, so current flows. With the anode negative relative to the cathode the electrons are repelled, and no current flows.

The amount of anode current

The grid sits between cathode and anode, and a small change in its voltage controls how much current reaches the anode, much as a gate voltage controls a FET. It does not heat the cathode.

Thermionic emission

Heating gives the cathode’s electrons enough energy to escape, forming a cloud around it: thermionic emission, which gives the tubes their name. Photoelectric emission is driven by light, not heat.

ICs do the same job in less space, at lower cost

One IC can hold thousands, millions or more components on a tiny chip of silicon, far smaller and cheaper than the same circuit built from discrete parts. High power is where discrete parts still win.

Twice as many transistors fitting on a chip about every two years

Moore’s law, named after Gordon Moore, is that twice as many transistors fit on a chip about every two years; the manual puts it as ICs shrinking by half. Chips get denser, not bigger.

By writing new software for it

Its microprocessor simply follows its software, so new software turns the same board to a new task, which makes such boards so handy for amateur projects. No hardware needs changing.

Arduino and Raspberry Pi

Arduino and Raspberry Pi boards carry a reconfigurable processor with its ports brought out for easy connection. The LM741 is an op-amp and SOIC a package style, not project platforms.

High input impedance, low output impedance and high gain

An op-amp offers high gain and linearity with little power drawn, taking almost nothing from its source (high input impedance) and driving loads easily (low output impedance). Swapped impedances suit neither job.

The difference between the voltages at V+ and V−

An op-amp’s output is an amplified copy of the difference between its V+ and V− inputs, and either input may be grounded. Adding the two input voltages is not what it does.

VS+ and VS−

The supply goes to VS+ and VS−, for example +5 V and −5 V, or one of them may be grounded. V+ and V− are the signal inputs.

Amplifiers, active filters and oscillators

With simple networks of outside passive parts, an op-amp can make almost any signal amplifier, buffer, filter or signal generator (oscillator). Oscillators are well within its reach.