Common-Mode vs Differential-Mode#

Every voltage on a pair of conductors can be decomposed into two components: the part that’s the same on both (common-mode) and the part that’s opposite (differential-mode). This is not a special property of balanced cables or differential signaling — it’s a mathematical identity that applies to any two-conductor system. Understanding it is the key to noise analysis, filter design, and debugging interference problems.

The Decomposition#

For any two voltages V(A) and V(B) on two conductors:

Differential-mode voltage: V_dm = V(A) - V(B) This is the difference. It’s the intended signal in a differential system.

Common-mode voltage: V_cm = (V(A) + V(B)) / 2 This is the average. It’s the voltage that both conductors share relative to some external reference (usually ground).

Any combination of V(A) and V(B) can be fully described by V_dm and V_cm. The original voltages reconstruct from these components:

• V(A) = V_cm + V_dm/2
• V(B) = V_cm - V_dm/2

What Common-Mode Noise Really Is#

Common-mode noise is any unwanted voltage that appears equally on both conductors of a pair. It doesn’t change the difference between them — it shifts both up or down together.

Sources of common-mode noise:

• Ground potential differences. Two devices on different outlets have slightly different ground potentials. This offset appears as a common-mode voltage on any cable connecting them
• Electromagnetic coupling. An external field (from a motor, a power line, a switching regulator) induces voltage in both conductors. If the conductors run close together (twisted pair, parallel traces), the induced voltage is nearly identical on both — common-mode
• Power supply coupling. Noise on a power rail that feeds both sides of a circuit appears as common-mode noise on any signal referenced to that supply
• Capacitive coupling to nearby structures. Both conductors of a pair capacitively couple to a nearby noisy trace or cable, picking up common-mode noise

A differential receiver rejects common-mode noise because it takes the difference: the common-mode component cancels. A single-ended receiver cannot distinguish common-mode noise from signal — it sees both.

Why CMRR Is Finite#

No real receiver has infinite Common-Mode Rejection Ratio. CMRR is limited by asymmetries in the receiver:

Component mismatch. A differential amplifier uses matched resistors. If the resistors differ by 0.1%, the CMRR is limited to about 66 dB. Precision instrumentation amps use laser-trimmed resistors to achieve better matching.

Input impedance imbalance. If the two inputs have slightly different impedances (from parasitic capacitance, PCB trace mismatch, or connector asymmetry), common-mode current flows unequally, creating a differential voltage from a common-mode source.

Frequency dependence. At DC, CMRR can be very high (120+ dB for precision instrumentation amps). As frequency increases, parasitic capacitances create asymmetric paths and CMRR degrades — often at 20 dB per decade. An amp with 100 dB CMRR at DC might have only 60 dB at 1 kHz and 40 dB at 100 kHz.

Practical consequence: CMRR is excellent at 60 Hz, making it reliable for rejecting ground loop hum. At 1 MHz switching-noise frequencies, rejection may be 40–60 dB lower — the CMRR vs frequency curve in the datasheet is the relevant specification.

How Imbalance Converts Common-Mode to Differential-Mode#

This is the mechanism by which common-mode noise becomes a real problem — even in differential systems. When the two signal paths are not perfectly symmetric, common-mode noise partially converts to differential-mode noise, which the receiver cannot reject.

Sources of asymmetry that cause conversion:

• Different conductor lengths (in a cable, or as PCB trace length mismatch)
• Different impedances to ground on each conductor (connector pins, PCB pad sizes, cable routing)
• Asymmetric coupling to nearby structures (one conductor closer to a noise source than the other)
• Connector contact resistance differing between pins
• Mismatched termination on the two lines

The conversion is described by the Longitudinal Conversion Loss (LCL) in telecommunications or the balance ratio in audio. A well-constructed twisted pair cable might achieve 40-60 dB of balance, meaning common-mode noise is converted to differential-mode noise 100-1000× weaker than the original common-mode level.

Differential-Mode Noise#

Not all noise is common-mode. Differential-mode noise appears as a voltage difference between the two conductors — it looks exactly like signal to a differential receiver.

Sources of differential-mode noise:

• Common-mode to differential-mode conversion (the mechanism above — the most common source)
• Crosstalk from adjacent differential pairs (capacitive or inductive coupling that’s inherently asymmetric)
• Noise generated within the signal source itself (thermal noise, shot noise, oscillator phase noise)
• Impedance discontinuities that cause reflections on one conductor but not the other

Differential-mode noise cannot be rejected by CMRR. It requires filtering, shielding, or reducing it at the source.

Filtering: Common-Mode and Differential-Mode Are Filtered Differently#

Because the two noise modes have different sources and different behaviors, they require different filtering approaches:

A common-mode choke is an inductor wound on a single core with two windings carrying the two signal conductors. Differential-mode current creates opposing magnetic fields that cancel — the choke is invisible to the signal. Common-mode current creates reinforcing fields — the choke presents high impedance and blocks it. This makes common-mode chokes an elegant filter: they suppress common-mode noise without affecting the differential signal.

Why “Fixes” Often Just Move the Problem#

A common pattern in noise debugging:

1. Common-mode noise appears on a cable
2. A common-mode choke is added — the common-mode noise drops
3. But differential-mode noise at the receiver increases

What happened? The common-mode choke reduced the common-mode noise on the cable, but the asymmetry that was converting common-mode to differential-mode noise is still there. With less common-mode noise, less converts — but other noise sources (now unmasked) may become visible. Or the choke itself introduced a small asymmetry.

The real fix for this situation is to address the root cause: improve the balance of the cable or connection (fix the asymmetry), or reduce the noise at its source. Filtering is a symptom-level treatment. It works — but understanding the conversion mechanism reveals where the deeper fix is.

Similarly: adding shielding to a cable might reduce noise pickup, but if the shield creates a ground loop (shield connected to ground at both ends with a potential difference), it can introduce new common-mode noise that converts to differential at the receiver. The fix introduces a new problem. Understanding the modes of noise and their conversion mechanisms prevents this whack-a-mole pattern.

Tips#

• If adding a ferrite clamp to a cable has no effect, the noise is likely differential-mode. Ferrites only attenuate common-mode current; differential currents cancel in the core and pass through unimpeded. A ferrite that does nothing is diagnostic information, not a failed fix
• Separating conducted emissions into common-mode and differential-mode components identifies which filter topology is needed. A LISN measures total conducted emissions; separating them into CM and DM prevents wasting effort with the wrong filter type — CM chokes for common-mode, X-capacitors and series inductors for differential-mode
• A current probe clamped around both conductors of a pair reads only common-mode current. Differential currents flow in opposite directions and cancel in the probe. Any current the probe registers is common-mode — a fast way to quantify CM noise without breaking into the circuit
• When selecting a common-mode choke, check its impedance at the noise frequency, not just its DC or low-frequency rating. A choke rated at 1 kΩ at 100 MHz may present only 10 Ω at 1 MHz. The impedance-vs-frequency curve determines whether the choke actually attenuates the noise that matters
• A two-channel oscilloscope can approximate the CM/DM decomposition using math mode. (Ch1 + Ch2)/2 displays the common-mode component; Ch1 − Ch2 displays the differential-mode component — separating noise into its two modes without specialized equipment

Caveats#

• Common-mode noise is invisible to a standard oscilloscope measurement. Connecting a scope probe between the two conductors of a differential pair measures differential-mode voltage only. To see common-mode voltage, measure each conductor individually against earth ground, or use a current probe clamped around both conductors (any current the probe reads is common-mode)
• CMRR specifications at DC are misleading. Many amplifiers and receivers are specified at DC or low frequency where CMRR is highest. Always check the CMRR vs frequency curve for the frequencies where the noise exists
• A “balanced” cable with poor balance is worse than advertised. The cable’s balance (symmetry) determines how much common-mode noise converts to differential-mode. A damaged cable, a cable with a broken twist lay, or a cable with mismatched connectors may have 20 dB worse balance than specification — converting 10× more common-mode noise into signal
• Common-mode range limits are hard limits. A differential receiver with a ±10 V common-mode range will clip, distort, or be damaged if the common-mode voltage exceeds that range. CMRR only applies within the specified common-mode range. Ground offsets, noise peaks, and transients must all stay within bounds

In Practice#

Noise that disappears when switching from a single-ended to a differential measurement is common-mode — present on both conductors equally but rejected by the differential measurement. The single-ended measurement was showing the common-mode noise as if it were signal; the differential measurement reveals the actual signal underneath.

Noise that persists regardless of measurement method is differential-mode — it appears as a voltage difference between the two conductors and looks identical to the intended signal. Differential-mode noise cannot be rejected by the receiver and requires filtering or reduction at the source.

Noise on a balanced cable that exceeds what the common-mode level alone would predict often traces to CM-to-DM conversion from asymmetry in the cable or connectors. The common-mode noise itself may be modest, but even small imbalances in conductor impedance, length, or termination convert a fraction of it into differential-mode noise that the receiver passes through.