A push-pull amplifier is an electronic circuit configuration that uses two active devices (like BJTs, MOSFETs, or vacuum tubes) operating in opposite phases to alternately source and sink current into a load, effectively canceling even-order harmonic distortion and increasing power efficiency. By splitting the waveform into positive and negative halves, this topology fundamentally changes how a circuit handles power delivery: it eliminates the massive, wasteful continuous bias current required by single-ended Class A designs, allowing for significantly higher output power from smaller, cooler-running components. Beginners frequently confuse the push-pull audio output stage with an H-bridge motor driver; while both use a four-device bridge or a two-device half-bridge, an H-bridge switches DC polarity to reverse motor direction, whereas a push-pull amp linearly amplifies an AC signal waveform.
The Core Mechanics: Sourcing vs. Sinking Current
In a standard complementary symmetry push-pull stage, an NPN transistor (or N-channel MOSFET) handles the positive half of the AC waveform, while a PNP transistor (or P-channel MOSFET) handles the negative half. Think of it like a two-person crosscut saw: one person pulls while the other pushes, but they never pull at the exact same time. When the input signal swings positive, the NPN device turns on, sourcing current from the positive supply rail through the load to ground. When the signal swings negative, the PNP device turns on, sinking current from ground through the load to the negative supply rail.
This alternating action is what cancels even-order harmonics (2nd, 4th, 6th). Because the two halves of the circuit are symmetrical but operate in opposite directions, any even-order distortion generated by the non-linear transfer curve of one transistor is mathematically mirrored and canceled out by the other when the halves recombine at the load. This leaves primarily odd-order harmonics (3rd, 5th), which are generally less fatiguing to the human ear in audio applications and easier to filter in RF applications.
Worked Numeric Example: Power, Efficiency, and Heatsink Sizing
To understand the real-world thermal and power implications, let us calculate the output power, efficiency, and transistor dissipation for a Class B push-pull amplifier driving an 8Ω loudspeaker from a dual ±20V DC power supply.
1. Calculate Peak and RMS Output Power:
With a ±20V supply, the peak voltage ($V_p$) across the load is 20V (ignoring minor transistor saturation voltage drops for this theoretical model).
$V_{rms} = 20V / \sqrt{2} \approx 14.14V$
Output Power ($P_{out}$) = $V_{rms}^2 / R_L = (14.14)^2 / 8\Omega = 200 / 8 = 25 Watts$.
2. Calculate DC Input Power:
Peak current ($I_p$) = $V_p / R_L = 20V / 8\Omega = 2.5A$.
The average DC current drawn from both rails in a Class B stage is $(2 \times I_p) / \pi$.
$P_{dc} = V_{cc} \times I_{avg(total)} = 20V \times [(2 \times 2.5A) / 3.14159] = 20V \times 1.59A = 31.83 Watts$.
3. Calculate Efficiency and Dissipation:
Efficiency ($\eta$) = $P_{out} / P_{dc} = 25W / 31.83W = 78.5\%$.
Total power dissipated as heat ($P_{diss}$) = $P_{dc} - P_{out} = 31.83W - 25W = 6.83 Watts$ (split roughly 3.4W per transistor).
Do not size your heatsinks based on maximum power output! Maximum transistor dissipation in a push-pull amp does not occur at full volume. It occurs when the output voltage is roughly 63.6% of the supply rail ($2/\pi$). For our ±20V supply, max dissipation happens at an output of 12.73V, where the transistors must burn off 10.13W total (5.06W each). If you size your heatsink only for the 3.4W dissipation at full power, your transistors will overheat and fail at half-volume.
The Crossover Distortion Problem (And the Class AB Fix)
The pure Class B push-pull topology has a fatal flaw for audio: crossover distortion. Because a silicon BJT requires roughly 0.7V at its base-emitter junction to begin conducting, there is a 1.4V dead zone (from +0.7V to -0.7V) in the center of the waveform where both transistors are completely off. As the signal crosses zero, the output flatlines momentarily, creating severe high-frequency harmonic distortion that sounds like harsh fizzing on transient peaks.
The industry solution is the Class AB push-pull amplifier. By inserting a small bias network—typically a $V_{BE}$ multiplier transistor or a pair of forward-biased diodes—between the bases of the output transistors, we hold them slightly turned on even when the input signal is zero. This quiescent bias current (usually set between 20mA and 100mA for discrete power amps) bridges the dead zone, eliminating crossover distortion at the cost of a tiny drop in overall efficiency. According to Electronics Tutorials, maintaining thermal stability in this bias network is critical; the biasing diodes or $V_{BE}$ multiplier must be thermally coupled to the output transistor heatsink to prevent thermal runaway as the silicon heats up and its forward voltage drop decreases.
Where You Meet Push-Pull Topologies in Practice
You will encounter push-pull architectures across multiple domains of electrical and electronic design:
- Audio Power Amplifiers: Almost all modern analog audio amplifiers, from the LM3886 integrated circuit to massive discrete Class AB/HD designs like the classic Bob Cordell designs, rely on complementary push-pull output stages to deliver high current to low-impedance (4Ω to 8Ω) speakers.
- Microcontroller GPIO: The digital output pins on an Arduino, ESP32, or Raspberry Pi use a miniature push-pull topology called a "totem pole" output. A small PMOS pulls the pin to VCC (push), and a small NMOS pulls it to GND (pull), allowing the pin to actively drive both high and low states rather than relying on a passive pull-up resistor.
- RF Power Amplifiers: In radio frequency transmitters, push-pull stages are used to cancel even-order harmonics before the signal reaches the antenna. This is crucial for meeting strict FCC or CE spectral emission masks, as even harmonics can fall directly into adjacent licensed communication bands.
- Line Drivers and RS-485: Differential line drivers use a form of push-pull logic to drive signals across long twisted-pair cables, rejecting common-mode noise in industrial environments.
Frequently Asked Questions
What is the difference between a push pull amp and a single-ended amplifier?
A single-ended amplifier (like a Class A design) uses one active device that conducts continuously for the entire 360 degrees of the signal cycle, requiring a massive bias current that generates constant heat regardless of volume. A push-pull amp uses two devices that split the workload, each conducting for roughly 180 degrees (or slightly more in Class AB), which drastically reduces idle power consumption and allows for much higher peak power output without melting the components.
Why does my push pull amp output have crossover distortion?
Crossover distortion occurs when the amplifier is operating in pure Class B, meaning there is a dead zone at the zero-crossing point where neither the N-type nor P-type transistor is conducting. To fix this, you must add a bias voltage (typically 1.2V to 1.4V for BJT pairs) between the bases of the output transistors to push them into Class AB operation, ensuring a small quiescent current flows through both devices even when no audio signal is present.
Can I use a push pull amplifier topology to drive a DC motor?
While a basic half-bridge push-pull stage can drive a motor in one direction with variable speed via PWM, it cannot reverse the motor. To achieve bidirectional control, you need an H-bridge (a full bridge of four switches). Furthermore, audio push-pull amps operate in the linear region to preserve waveform fidelity, whereas motor drivers operate the MOSFETs as hard switches (fully on or fully off) to minimize $I^2R$ heating losses.
How do I bias the transistors in a Class AB push pull amp safely?
The safest and most thermally stable method is to use a $V_{BE}$ multiplier circuit (a single transistor with a voltage divider across its base and collector) mounted directly onto the main output heatsink. As the output transistors heat up, their required turn-on voltage drops; the heatsink-mounted $V_{BE}$ multiplier senses this heat and proportionally reduces the bias voltage, preventing thermal runaway. Never use fixed resistors for biasing high-power output stages.
