A Darlington amplifier (or Darlington pair) is a compound transistor circuit where two bipolar junction transistors (BJTs) are cascaded so the emitter current of the first feeds directly into the base of the second, multiplying their current gains to achieve massive amplification from a tiny input signal. What this changes in a real circuit is the effective input impedance and current-driving capability: it allows a high-impedance, microamp-level signal from a sensor or microcontroller GPIO to directly switch amp-level loads like solenoids, relays, and motors without needing complex driver ICs.

The Core Trade-off: You gain massive current multiplication ($\beta_{total} \approx \beta_1 \times \beta_2$), but you pay for it with a higher base-emitter voltage drop ($V_{BE} \approx 1.4V$) and a higher collector-emitter saturation voltage ($V_{CE(sat)} \approx 1V$ to $1.5V$), which generates significant heat at high currents.

The Math: A Worked Numeric Example

Let's look at a common bench scenario: driving a 12V, 2A automotive relay using an ESP32 GPIO pin. The ESP32 outputs 3.3V logic, and while absolute maximum GPIO current is 40mA, the recommended continuous source current for reliability is under 5mA. We need to switch 2A on the load side.

Scenario A: Standard Single BJT (e.g., 2N2222)

A standard small-signal BJT has a DC current gain ($\beta$ or $h_{FE}$) of roughly 100. To drive a 2A collector current ($I_C$), the required base current ($I_B$) is:

$I_B = I_C / \beta = 2A / 100 = 20mA$

This exceeds our 5mA safe target for the ESP32 GPIO. Furthermore, a 2N2222 will physically melt at 2A (its max rating is ~800mA). We need a power transistor, but even a power BJT like the 2N3055 ($\beta \approx 20$ at high currents) would demand a staggering 100mA of base current.

Scenario B: Darlington Amplifier (e.g., TIP120)

The TIP120 Darlington pair has a guaranteed minimum $\beta$ of 1000 at 3A. Let's recalculate the base current:

$I_B = 2A / 1000 = 2mA$

This is well within the ESP32's safe 5mA limit. Now, we calculate the base resistor ($R_B$). Because the Darlington contains two base-emitter junctions in series, the voltage drop ($V_{BE}$) is roughly $0.7V \times 2 = 1.4V$.

$R_B = (V_{GPIO} - V_{BE}) / I_B = (3.3V - 1.4V) / 0.002A = 950\Omega$

We select the nearest standard E12 resistor value: $1k\Omega$. This yields a base current of 1.9mA, safely saturating the TIP120 to switch our 2A relay.

Bench Reality Check (Thermal Dissipation): When a Darlington is fully ON (saturated), it doesn't drop to 0V like a mechanical switch. The TIP120 datasheet specifies a $V_{CE(sat)}$ of up to 1.5V at 3A. At our 2A load, the transistor dissipates $P = V_{CE(sat)} \times I_C = 1.5V \times 2A = 3W$. The TO-220 package can only dissipate ~2W in free air before overheating. You must bolt the TIP120 to a small heatsink for this 2A load.

Where You Meet This in Practice

You will rarely build a discrete Darlington pair from two raw transistors on a modern PCB unless you are designing custom silicon or ultra-specific analog front-ends. Instead, you meet them integrated into standard, cheap, and ubiquitous driver packages:

  • ULN2003A Driver ICs: The backbone of hobbyist stepper motor control. If you are driving a 28BYJ-48 stepper motor with an Arduino or ESP32, the ULN2003A board you plug it into contains seven discrete Darlington pairs with built-in flyback diodes.
  • Power Switching Modules: Pre-wired "TIP120 modules" sold on Amazon or AliExpress for PWM dimming of 12V LED strips or solenoid control.
  • Touch Switches and High-Impedance Sensors: Because a Darlington can amplify the tiny leakage current of human skin, they are the classic component inside touch-sensitive desk lamps and moisture sensors.
  • Audio Output Stages: Historically used in the output stage of analog audio amplifiers to provide the massive current gain needed to drive low-impedance (4-ohm) speakers from a low-current voltage amplifier stage.

Decision Tree: Darlington vs. MOSFET vs. Sziklai

The biggest mistake beginners make is defaulting to a Darlington amplifier for every switching task because "it has high gain." Modern electronics offers better alternatives depending on your exact constraints. Use this decision matrix to select the right component.

Criteria Darlington BJT (e.g., TIP120) Logic-Level MOSFET (e.g., IRLZ44N) Sziklai Pair (Complementary)
Drive Type Current-driven (needs continuous base current) Voltage-driven (draws almost zero steady-state gate current) Current-driven (similar to Darlington)
ON-State Voltage Drop High (~1.0V to 1.5V) Extremely Low ($I \times R_{DS(on)}$, often < 0.1V) Moderate (~0.7V, one $V_{BE}$ drop)
Thermal Loss at 5A ~5W to 7.5W (Massive heatsink required) ~0.1W (Runs cool, no heatsink needed) ~3.5W (Medium heatsink required)
Switching Speed Slow (especially turn-off time without pull-down) Extremely Fast (ideal for high-frequency PWM) Moderate (faster than Darlington)
Best Use Case Amplifying ultra-low current sensors; legacy replacements Microcontroller PWM, high-current DC motors, LED strips High-fidelity audio output stages
The Concrete Default Pick: If you are switching any DC load greater than 1A from a microcontroller (Arduino, ESP32, Pi Pico) using PWM or simple ON/OFF logic, do not use a Darlington. Buy a logic-level N-Channel MOSFET like the IRLZ44N or STP16NF06L. They cost roughly the same ($0.50 - $1.00), require no continuous base current, and will not require a heatsink for loads under 5A. Reserve the Darlington (TIP120) strictly for situations where you are driving from an ultra-high impedance source (like a piezo element) or replacing a legacy part on an existing PCB.

Common Confusions and Pitfalls

Confusing Current-Controlled vs. Voltage-Controlled Devices

People frequently confuse Darlingtons with MOSFETs. A MOSFET is a voltage-controlled device; once you charge the gate capacitance, it draws virtually zero current to stay ON. A Darlington is a current-controlled device. The base current must flow continuously to keep the load energized. If your microcontroller goes to sleep and the GPIO floats, the Darlington might partially turn on due to noise, overheating the transistor and the load.

The Missing Turn-Off Resistor

Darlington pairs suffer from slow turn-off times because the minority carriers in the base of the second transistor have nowhere to go when the input signal drops to zero. If you look at the internal schematic of a ULN2003 or a discrete TIP120, you will see built-in base-emitter resistors (usually a few thousand ohms) across the junctions to bleed off this charge. If you build a discrete Darlington from raw 2N3904s on a breadboard and omit these bleed resistors, your circuit will suffer from severe switching delays and thermal instability.

Forgetting the Flyback Diode

Because Darlingtons are frequently used to drive inductive loads (relays, steppers, solenoids), the collapsing magnetic field when the transistor turns off will generate a massive reverse voltage spike. This spike will instantly punch through the collector-emitter junction of the BJT, destroying it. Always place a 1N4007 or 1N4148 flyback diode in reverse bias across the inductive load. (Note: The ULN2003 and TIP120 have these diodes integrated internally, but external discrete Darlingtons do not).

FAQ: Darlington Amplifier Edge Cases

Can I use a Darlington amplifier for high-frequency PWM dimming?

No. Darlingtons have high storage times and slow turn-off characteristics. If you attempt to run a TIP120 at 20kHz PWM for motor control or LED dimming, the transistor will spend a significant portion of its time in the linear (active) region rather than fully saturated or fully cut off. This causes massive switching losses, leading to rapid thermal runaway and component failure. Use a MOSFET for anything above 1kHz PWM.

Why does my Darlington get hot even when switching a small 500mA load?

Because of the inherent $V_{CE(sat)}$ penalty. Even at 500mA, a Darlington will drop roughly 1V across its collector and emitter. $1V \times 0.5A = 0.5W$ of heat. While 0.5W won't instantly destroy a TO-220 package, it will make the tab too hot to touch (reaching roughly 60°C to 80°C in free air). This is normal physics for the topology, not a sign of a defective part.

What is a Sziklai pair and why is it sometimes better?

A Sziklai pair (or complementary feedback pair) uses one NPN and one PNP transistor. It achieves the same massive current gain as a Darlington ($\beta_1 \times \beta_2$), but it only suffers from a single $V_{BE}$ drop (approx 0.7V) instead of two. This makes it more thermally stable and slightly more efficient, which is why modern high-end audio amplifier output stages heavily favor the Sziklai topology over the traditional Darlington.