Skip to content
View in the app

A better way to browse. Learn more.

TEST SITE

A full-screen app on your home screen with push notifications, badges and more.

To install this app on iOS and iPadOS
  1. Tap the Share icon in Safari
  2. Scroll the menu and tap Add to Home Screen.
  3. Tap Add in the top-right corner.
To install this app on Android
  1. Tap the 3-dot menu (⋮) in the top-right corner of the browser.
  2. Tap Add to Home screen or Install app.
  3. Confirm by tapping Install.

Push-Pull Amplifiers

A single-transistor common-emitter amplifier has a fundamental limitation: one transistor handles both the positive and negative halves of the output signal cycle. For Class A operation, it must be biased to conduct a large quiescent current so that it can sink or source enough current for the full output swing. This wastes DC power as heat whenever no signal is being amplified. For Class B or AB operation, a single transistor can only amplify one polarity of signal — you need a second transistor to handle the other half of the cycle.

The push-pull amplifier solves both problems elegantly. Two transistors are connected so that one handles the positive half-cycle of the signal (pushing current through the load) while the other handles the negative half-cycle (pulling current through the load). The two halves are combined at the output. This configuration delivers twice the output power of a single stage from the same supply voltage, cancels even-order harmonic distortion by symmetry, and permits efficient Class AB operation that eliminates crossover distortion while maintaining acceptable linearity.

Push-pull design is the dominant topology for virtually every power amplifier stage — from the output stage of a hi-fi amplifier or an op-amp to broadband RF power amplifiers, motor drivers and the output stages of switching amplifiers. Understanding push-pull operation is essential for anyone who wants to understand or troubleshoot power amplifiers of any kind.

What you will learn: How push-pull operation works; transformer-coupled and transformerless (complementary-symmetry) topologies; even-order harmonic cancellation by symmetry; Class AB biasing for push-pull stages; output power and efficiency; impedance transformation with output transformers; balun transformers; and practical push-pull amplifier design.
Push-pull amplifier circuit diagrams: transformer-coupled BJT push-pull on the left, complementary-symmetry NPN/PNP push-pull on the right
Two push-pull topologies. Left: transformer-coupled using two matched NPN transistors — the standard for broadband RF power amplifiers and classic valve audio amplifiers. Right: complementary-symmetry using an NPN and a PNP pair — the standard for solid-state audio amplifiers.

The Push-Pull Principle

In a push-pull amplifier, the input signal is split into two equal and opposite signals — one drives transistor Q1, the other drives transistor Q2. Q1 amplifies the positive half of the cycle; Q2 amplifies the negative half. The outputs of Q1 and Q2 are combined in a way that reconstructs the full output waveform.

The key insight is symmetry. Because the two transistors handle equal and opposite signals:

  • Odd-order distortion products (fundamental, 3rd harmonic, 5th harmonic…) add constructively in the output — they are preserved and amplified.
  • Even-order distortion products (2nd harmonic, 4th harmonic…) cancel because they are in phase in both transistors and the output circuit subtracts one half from the other. This cancellation is automatic — it requires no tuning or adjustment.

The cancellation of even-order harmonics is a major advantage of push-pull design. A single-ended (one transistor) Class A stage produces significant 2nd harmonic distortion. The same transistors in push-pull produce no 2nd harmonic — their 2nd harmonic components cancel, leaving only 3rd, 5th, and higher odd-order distortion, which are typically much smaller in amplitude. This makes push-pull amplifiers substantially cleaner than their single-ended counterparts for the same transistors and bias conditions.

Transformer-Coupled Push-Pull

The classic implementation uses transformers at both input and output. The input transformer (driver transformer) has a centre-tapped secondary winding. When the signal goes positive, the top half of the secondary drives Q1's base positive and the bottom half drives Q2's base negative — giving Q1 and Q2 exactly opposite drive signals. The output transformer has a centre-tapped primary. Q1's collector current flows through one half of the primary winding; Q2's collector current flows through the other half. The two currents produce magnetic flux in the transformer core that adds to drive the secondary winding and the load.

Circuit components:

  • Q1, Q2: matched NPN transistors (or matched MOSFETs in higher-power designs)
  • T1: input/driver transformer with centre-tapped secondary
  • T2: output transformer with centre-tapped primary, turns ratio chosen for impedance matching
  • Bias resistors/diodes: set Class AB idle current for both transistors
  • Emitter resistors (small value, 0.1–1 Ω): force current sharing between Q1 and Q2

Transformer-coupled push-pull is used in virtually all solid-state broadband RF power amplifiers producing more than a few watts, and in valve (tube) audio amplifiers. The transformers provide galvanic isolation, allow easy impedance transformation, and the center-tap arrangement ensures precise 180° phase splitting. Ferrite-core transformers (wound on toroidal or binocular ferrite cores) are the standard choice for RF work — they achieve low loss and wide bandwidth, for example from 2 MHz to 30 MHz, in a single winding.

Complementary-Symmetry Push-Pull

The complementary-symmetry topology replaces the output transformer with a matched NPN/PNP transistor pair. Q1 (NPN) handles positive half-cycles; Q2 (PNP) handles negative half-cycles. Because Q2 is a PNP transistor, it is connected so that a positive base signal turns it off and a negative base signal turns it on — the opposite of Q1. This provides the necessary push-pull action without a transformer.

Both transistors receive the same input signal — no phase-splitting transformer is required at the input. When the signal goes positive, Q1 (NPN) turns on and drives current through the load to the negative supply rail. When the signal goes negative, Q2 (PNP) turns on and drives current from the positive supply rail through the load. The load sees a full, bidirectional current waveform.

Complementary-symmetry is the standard topology for solid-state audio power amplifiers and for op-amp and IC output stages. Its advantages are fewer components, no transformer bandwidth limitation, and no core saturation risk. Its disadvantages are the difficulty of finding perfectly matched NPN/PNP pairs and the need for careful biasing to avoid crossover distortion.

Even-Order Harmonic Cancellation

The cancellation of even-order harmonics can be understood mathematically. Suppose each transistor has a non-linear transfer function and produces output with a second-harmonic component a₂:

Q1 output: IC1 = a₁·v + a₂·v² + a₃·v³ + …
Q2 receives −v (inverted signal): IC2 = a₁·(−v) + a₂·(−v)² + a₃·(−v)³ + …
= −a₁·v + a₂·v² − a₃·v³ + …

Output combines Q1 pushing and Q2 pulling:
Iout = IC1 − IC2
= (a₁·v + a₂·v² + a₃·v³) − (−a₁·v + a₂·v² − a₃·v³)
= 2a₁·v + 2a₃·v³ + …

The a₂·v² (second harmonic) terms cancel completely. Only odd-order terms remain.

This is why push-pull amplifiers have very little 2nd harmonic output (negligible in practice, if the circuit is well balanced). In an RF amplifier, the 2nd harmonic of a 7 MHz signal would fall at 14 MHz; a well-balanced push-pull stage can achieve 50–60 dB of 2nd harmonic suppression, dramatically simplifying the harmonic filter that follows. In an audio amplifier, cancelling the even harmonics removes a large part of the distortion before any feedback is applied.

Class AB Biasing for Push-Pull Stages

The most critical design element of a push-pull audio or linear RF amplifier is the idle (quiescent) current bias that places both transistors in Class AB operation. Without idle current, the stage operates in Class B and exhibits crossover distortion. With too much idle current, efficiency drops and the transistors overheat.

Diode Bias

The simplest approach for complementary-symmetry audio stages uses two silicon diodes (one for each transistor's VBE) in series between the bases. The diodes are thermally coupled to the transistors — mounted on the same heatsink. As the transistors heat up, VBE decreases; the diodes, also heating up, produce a lower voltage drop across the bias network, tracking the transistors' VBE reduction and keeping the idle current approximately constant. This is simple, cheap, and effective for audio power stages.

VBE Multiplier (Rubber Zener)

For more precise idle current control, a VBE multiplier circuit is used. This consists of a single transistor (Qbias) with a resistor divider from collector to base. It produces an output voltage that is a multiple of VBE, adjustable by the resistor ratio. The multiplied VBE voltage sets the idle current of the output pair. VBE multiplier circuits are thermally coupled to the output transistors for temperature tracking. They allow idle current to be set precisely by adjusting a single trimmer resistor.

Setting Idle Current

Idle current for push-pull RF power amplifiers is typically set at 50–200 mA total (25–100 mA per transistor). For audio amplifiers, 20–100 mA total is common. Too low gives crossover distortion and poor IMD. Too high wastes power and can cause thermal runaway in poorly heatsunk designs. The correct idle current is found experimentally by measuring distortion (THD, or IMD with a two-tone test) while adjusting bias, or by looking at the output waveform at low signal levels and adjusting until the crossover notch disappears. Many audio amplifier service manuals simply specify the voltage to set across the emitter resistors.

Spectrum display showing a single-ended amplifier output with strong second harmonic and a push-pull amplifier output with the second harmonic cancelled by symmetry
Even-order harmonic cancellation in push-pull operation. The single-ended stage (left) shows a prominent 2nd harmonic. The balanced push-pull stage (right) has the 2nd harmonic suppressed by 40 dB or more through symmetrical cancellation.

Output Power and Efficiency

For a Class AB push-pull stage with supply voltage VCC and a load resistance RL (reflected to the primary through the transformer):

Maximum output power (Class AB, approximately Class B):
Pout(max) = VCC² / (2 × RL)

where RL is the load impedance as seen by one transistor (collector to collector / 4 for a centre-tapped primary, or the collector load of a complementary pair).

Maximum efficiency (Class B limit): ηmax = π/4 ≈ 78.5%

Typical Class AB efficiency: 60–70% at full output power
Worked Example: Push-Pull Broadband RF Power Amplifier

A 100 W broadband RF amplifier (of the kind used in industrial, broadcast and test equipment) uses two transistors in push-pull with VCC = 28 V. What collector load resistance must each transistor see?

Pout = VCC² / (2 × RL(each))
100 = 28² / (2 × RL)
100 = 784 / (2 × RL)
RL = 784 / 200 = 3.92 Ω

Each transistor should see approximately 4 Ω from collector to the center-tap (the output transformer provides this from a 50 Ω load).

DC input power: Assuming 65% efficiency:
PDC = Pout / η = 100 / 0.65 = 154 W

Heat dissipated: Pheat = PDC − Pout = 154 − 100 = 54 W
Each transistor dissipates approximately 27 W — hence the need for heatsinking in any push-pull power stage above a few watts.

Output Transformer Impedance Matching

The output transformer in a push-pull amplifier has two functions: it reconstructs the full output waveform from the two half-cycle contributions of Q1 and Q2, and it transforms the impedance from the transistors' optimum load to the load impedance — 50 Ω for an RF amplifier, or 4–8 Ω for a loudspeaker in a valve audio amplifier.

The transformer turns ratio required for impedance matching:

n = Nprimary / Nsecondary = √(Rprimary / Rsecondary)

For a centre-tapped primary, the effective primary impedance is Rcollector-to-collector = 4 × RL(each transistor)

Example: Rcc = 4 × 4 Ω = 16 Ω, Rsecondary = 50 Ω
n = √(16/50) = √0.32 = 0.566 (step-up transformer)

In practice, ferrite-core wideband transformers (wound on toroidal ferrite rings or binocular cores) are used instead of tuned transformers for broadband RF amplifiers. Wideband transformers can cover a range such as 2–30 MHz without retuning, so the amplifier works at any frequency in that range without adjustment. The transformer is wound with bifilar or trifilar wire so that each winding has equal coupling to the core.

Balun Transformers in Push-Pull RF Stages

A balun (balanced-to-unbalanced) transformer is used at the output of a push-pull amplifier when the load is unbalanced (e.g., a coaxial cable with one grounded conductor). The push-pull stage has a balanced output — equal and opposite voltages on the two collectors. The coaxial output is unbalanced — one conductor is at ground potential. The balun converts between these two forms without disturbing the symmetry of the push-pull stage or letting common-mode current flow on the outside of the cable.

Most broadband RF power amplifiers use a ferrite-core balun transformer at the output. A well-designed balun has insertion loss < 0.5 dB across its frequency range and provides 20–40 dB of common-mode rejection.

Push-Pull Output Power Calculator

Push-Pull Amplifier Output Power Calculator

Calculate maximum output power, DC input power, dissipation, efficiency, and transformer turns ratio for a push-pull RF or audio power amplifier stage.

Results will appear here.

Experiment: Observe Even-Order Harmonic Cancellation

Purpose: Demonstrate that a balanced push-pull amplifier produces no 2nd harmonic, while an equivalent single-ended stage does.

You will need: Audio function generator (produces sine waves at 1 kHz), two transistors (BC547 or 2N3904, matched beta if possible), a centre-tapped audio transformer (such as those used for 600 Ω audio matching), a multimeter or audio analyser/spectrum app on a smartphone, 12 V supply, resistors per circuit below.

Part A — Single-ended stage:

  • Build a simple common-emitter amplifier with VCC = 12 V, RC = 2.7 kΩ, RE = 470 Ω (bypassed), voltage-divider bias for IC ≈ 2 mA.
  • Apply a 1 kHz, 200 mVpp sine wave to the input.
  • Connect a spectrum analyser app to the output and observe the fundamental (1 kHz) and the 2nd harmonic (2 kHz). Record the 2nd harmonic level relative to the fundamental.

Part B — Transformer-coupled push-pull:

  • Connect two identical transistors in push-pull using a centre-tapped input transformer (or a phase-splitter circuit) and a centre-tapped output transformer.
  • Apply the same 1 kHz, 200 mVpp input.
  • Connect the spectrum analyser app to the output and observe the 1 kHz fundamental and the 2 kHz second harmonic.

Expected results:

  • Single-ended: visible 2nd harmonic, typically −30 to −40 dBc depending on signal level and transistor characteristics.
  • Push-pull: 2nd harmonic reduced by 20–40 dB compared to the single-ended stage. The circuit symmetry should give at least −50 dBc for the 2nd harmonic.

To explore further: Deliberately unbalance the push-pull stage by changing one transistor's bias slightly — observe the 2nd harmonic reappearing. This demonstrates that perfect cancellation requires good balance, which is why power amplifiers use matched transistors and care is taken with transformer winding symmetry.

Frequently Asked Questions

Why do push-pull amplifiers cancel even-order harmonics but not odd-order?

Even-order harmonics (2nd, 4th…) are produced by the squared term (a₂·vin²) in the non-linear transfer function. When one transistor sees +vin and the other sees −vin, squaring gives (+vin)² = (−vin)² = vin² — the same positive value for both transistors. These equal components cancel when the output transformer subtracts one transistor's output from the other. Odd-order harmonics (1st, 3rd, 5th…) come from terms like a₁·vin and a₃·vin³: cubing a negative gives a negative (−vin)³ = −vin³, so these terms are opposite in the two transistors and add constructively at the output. Hence the fundamental and all odd harmonics are amplified; even harmonics cancel.

Why must transistors be matched in a push-pull amplifier?

Even-order harmonic cancellation depends on the two halves of the push-pull stage being identical — same gain, same VBE, same non-linearity. If one transistor has significantly different characteristics than the other, the cancellation is incomplete and even-order harmonics appear in the output. For audio amplifiers, transistors from the same production batch are usually matched closely enough, and audio manufacturers sell output transistors in matched pairs. For high-power RF amplifiers, transistors are sometimes characterized individually and paired by matching their gain and saturation characteristics. Even a 5–10% gain mismatch between Q1 and Q2 degrades 2nd harmonic cancellation from 60 dB to perhaps 30–40 dB.

What happens if the idle current in a Class AB push-pull stage is too low?

Insufficient idle current places the stage closer to Class B operation. Near the zero-crossing of the output signal, neither transistor is fully conducting and the output passes through the non-linear crossover region, producing crossover distortion. This appears on a spectrum analyzer as increased odd-order harmonics (3rd, 5th) and on a two-tone test as increased IMD. In an audio amplifier, crossover distortion sounds harsh and grating, and is most obvious at low volume, where a large part of the signal sits in the crossover region. In an RF linear amplifier, it causes excessive IMD and splatter. The fix is to increase idle current by adjusting the VBE multiplier or diode bias network — usually via a bias trimmer.

What is a ferrite toroidal transformer and why is it used in RF push-pull amplifiers?

A toroidal transformer is wound on a doughnut-shaped ferrite core. The closed toroidal path contains the magnetic flux efficiently, resulting in low leakage inductance (important for wideband frequency response) and minimal stray magnetic field. Ferrite cores are chosen for their permeability and loss characteristics at the frequencies involved — a Type 43 ferrite works well from about 1 to 10 MHz; Type 61 ferrite is better from 10 to 200 MHz. By winding bifilar (two wires twisted together) on a ferrite toroid, wideband 4:1 or 9:1 impedance transformers can be made that cover 2–30 MHz with less than 0.5 dB insertion loss — essential for broadband push-pull RF amplifiers that must work at any frequency in their range without retuning.

Check Your Understanding

Answer each question and click Check to see the explanation.

Account

Navigation

Search

Search

Configure browser push notifications

Chrome (Android)
  1. Tap the lock icon next to the address bar.
  2. Tap Permissions → Notifications.
  3. Adjust your preference.
Chrome (Desktop)
  1. Click the padlock icon in the address bar.
  2. Select Site settings.
  3. Find Notifications and adjust your preference.