A differential pair is two conductors that carry equal and opposite versions of the same signal. Instead of measuring a voltage against a common ground, the receiving end reads the difference between the two lines. This simple idea is the foundation of nearly every high-speed digital connection you use today, from USB and HDMI cables to the Ethernet port on your router.
How a Differential Pair Works
In a single-ended signal, one wire carries the voltage and a ground wire serves as the reference. The receiver measures the voltage on that one wire relative to ground. A differential pair takes a different approach: it sends two complementary signals, one positive and one negative, on two separate conductors. When one line swings to +200 millivolts, the other swings to -200 millivolts. The receiver subtracts one from the other to reconstruct the original signal.
This subtraction step is the key. Any external electrical noise picked up along the cable or circuit board trace tends to affect both lines equally, since the two conductors run close together through the same environment. When the receiver subtracts the two signals, the noise appears on both sides of the equation and cancels out. Only the intentional, opposite-polarity signal remains. This property is called common-mode rejection: if a disturbance is applied equally to both inputs, the output is unaffected.
Why Differential Pairs Reduce Interference
Differential pairs don’t just resist incoming noise. They also radiate far less electromagnetic interference than single-ended signals. Each conductor generates a magnetic field as current flows through it, but because the two currents flow in opposite directions, their magnetic fields point in opposite directions as well. These fields bend toward each other and cancel out a significant portion of the emissions that either conductor would produce on its own. The tighter the two conductors are coupled (physically close and running in parallel), the more complete the cancellation.
This matters in dense electronics where dozens of signals run side by side on a circuit board, or inside a cable bundle where crosstalk between adjacent wires can corrupt data. Differential signaling keeps both radiated emissions and susceptibility to external fields low, which is one reason regulatory standards for electromagnetic compatibility are much easier to meet with differential interfaces.
Voltage Swings and Signaling Standards
Because the receiver only needs to detect the difference between the two lines, differential pairs can operate with very small voltage swings. Smaller swings mean faster switching and lower power consumption. Several standard interfaces define exactly how small those swings should be.
LVDS (Low-Voltage Differential Signaling) is one of the most widely used. It drives a constant current of about 4 milliamps through a 100-ohm termination resistor at the receiver end, producing a differential voltage swing of 250 to 400 millivolts. The common-mode voltage (the average voltage of both lines relative to ground) sits somewhere between 0.2 and 2.2 volts. That tiny swing allows LVDS links to run at hundreds of megabits per second while consuming very little power.
CML (Current Mode Logic) is common in multi-gigabit serial links. It uses a differential swing of about 800 millivolts and operates with the common-mode voltage close to the supply rail. PECL (Positive Emitter-Coupled Logic) is an older standard found in high-speed clock distribution, with output voltages referenced to the supply voltage and a common-mode level around 1.3 volts below the supply. Each standard trades off swing size, speed, and power differently, but all rely on the same differential principle.
Impedance and Trace Design
For high-speed signals, the two conductors of a differential pair must be designed to a specific characteristic impedance. If the impedance changes along the signal path, part of the signal energy reflects back toward the transmitter, distorting the waveform. Different protocols specify different target impedances:
- USB 2.0 and USB 4.0: 90 ohms differential (45 ohms per trace), with a tolerance of ±15%.
- HDMI: 100 ohms differential (50 ohms per trace), ±15%.
- DisplayPort: 100 ohms differential (50 ohms per trace), ±10%.
On a printed circuit board, impedance is controlled by the width and spacing of the copper traces, the thickness of the dielectric material between layers, and the distance to the nearest ground plane. PCB designers use impedance calculators to hit these targets, and board manufacturers typically verify the result with test coupons on every production panel. Keeping the two traces the same length is also critical. If one trace is even slightly longer than the other, the signals arrive at the receiver at slightly different times, reducing the effective voltage difference and degrading signal quality.
Termination at the Receiver
To prevent reflections, a termination resistor is placed across the two differential lines at the receiver end. The resistor value is chosen to match the characteristic impedance of the transmission line. For a 100-ohm differential pair, a 100-ohm resistor bridges the two inputs at the receiver. This absorbs the incoming signal energy cleanly, so none of it bounces back down the line. The TIA/EIA-422 standard, for example, specifically recommends a 100-ohm parallel termination resistor at the far end of the cable.
Placing the resistor at the receiver end (rather than at the transmitter) is deliberate. The signal travels the full length of the line and gets absorbed right where it’s being read. Any small impedance mismatches along the way produce reflections that travel back toward the transmitter and dissipate harmlessly, rather than bouncing between both ends and corrupting the data.
Differential Pairs in Analog Circuits
Differential pairs aren’t only a high-speed digital concept. In analog electronics, the differential pair (sometimes called a long-tailed pair) is the fundamental input stage of nearly every operational amplifier. Two matched transistors share a common connection at their emitters, which feeds into a high-value resistor or, more commonly in integrated circuits, an active current source. One transistor receives the positive input, the other receives the negative input, and the circuit amplifies only the difference between them while rejecting any signal common to both.
This is why op amps have two input pins labeled “+” and “−.” The entire purpose of that front-end transistor pair is to provide high common-mode rejection, so the amplifier responds only to the intended signal and ignores power supply noise, temperature drift, and other disturbances that affect both inputs equally. In modern IC designs, the basic long-tailed pair is enhanced with current mirrors in the output stage to improve gain and symmetry, but the core principle is unchanged.
Where You Encounter Differential Pairs
Virtually every cable you plug into a modern device carries differential pairs. A USB-C cable contains multiple differential pairs for data, each running at 90 ohms. An HDMI cable uses four differential pairs for video data plus one for the clock, all at 100 ohms. Ethernet cables contain four twisted pairs, each carrying differential signals. Inside your computer, the PCIe lanes connecting the processor to the graphics card, SSD, and other peripherals are all differential pairs routed on the motherboard.
The reason differential signaling dominates modern electronics comes down to a practical reality: as data rates climb into the gigabits-per-second range, the voltage swings must shrink and the noise margins get tighter. A single-ended signal referenced to a distant, noisy ground plane simply cannot maintain signal integrity at those speeds. Differential pairs solve this by making the reference local (the other conductor in the pair) rather than relying on a shared ground, while simultaneously rejecting noise and radiating less interference. The cost is doubling the number of conductors, but at high speeds that tradeoff is almost always worth it.