Mutual inductance is the property where a changing current in one coil induces a voltage in a nearby second coil through a shared magnetic field. In practical circuit design, this phenomenon is the fundamental mechanism that makes transformers and wireless chargers work, but on a high-speed PCB, it changes your clean switching nodes into noise-ridden voltage spikes that can destroy sensitive silicon if you ignore component placement.
V_2 = M × (di_1 / dt)Note: It does not matter how large the primary current is; only how fast it changes.
The Core Mechanism and the Math That Matters
When current flows through a conductor, it generates a magnetic field. If that current is steady (DC), the field is static and nothing happens to nearby conductors. But when the current changes, the magnetic field expands or collapses. If a second coil sits inside that moving magnetic field, Faraday’s law of induction dictates that a voltage will be forced across the second coil's terminals.
Let’s run a concrete numeric example to see how easily this generates destructive voltages on a workbench. Suppose you have a primary switching loop carrying a transient current, and a nearby secondary trace acting as an unintentional pickup coil.
Switching Time ($dt$): 100 nanoseconds ($100 × 10^{-9}$ s)
Rate of Change ($di_1/dt$): $3.0 × 10^7$ Amps per second
Mutual Inductance ($M$): 50 nH ($50 × 10^{-9}$ H) — typical for two loosely coupled loops 5mm apart
Induced Voltage ($V_2$): $50 × 10^{-9} × 3.0 × 10^7$ = 1.5 Volts
A 1.5V spike might not sound like much, but if your secondary loop is a 3.3V logic enable line or a sensitive analog feedback trace, that 1.5V glitch is enough to cause a microcontroller brownout or trigger a false over-current shutdown. If the mutual inductance increases to 200 nH because you routed the traces parallel for a longer distance, that spike jumps to 6V, easily exceeding the absolute maximum ratings of a 5V CMOS gate.
Where You Meet Mutual Inductance in Practice
You interact with mutual inductance constantly, both intentionally and accidentally. Here is where it shows up in real-world electrical and electronic systems:
- Power Transformers (Intentional): The 120V AC primary and 12V AC secondary in a linear power supply share a laminated iron core to maximize $M$, ensuring efficient energy transfer across the isolation barrier.
- Wireless Charging / Qi Pads (Intentional): The transmitter and receiver coils are separated by an air gap (and often plastic or glass). The system relies on precise tuning to maintain enough $M$ to transfer 15W+ of power over a few millimeters.
- Induction Cooktops (Intentional): The primary coil under the glass induces massive eddy currents in the ferrous cookware (the secondary), heating the pan directly via $I^2R$ losses.
- PCB Crosstalk (Unintentional): High-$di/dt$ switching regulator inductors coupling magnetic flux into adjacent high-impedance feedback traces, causing output voltage ripple or control loop instability.
- Ground Loops and EMI (Unintentional): AC mains wiring running parallel to low-voltage audio cables in a wall cavity, inducing a 50/60Hz hum into the audio signal path.
Bench War Story: When Unwanted Coupling Fries a Gate Driver
Theory is clean; the bench is unforgiving. Here is a real-world scenario demonstrating what happens when mutual inductance is ignored during PCB layout.
- The Setup: I was prototyping a 48V to 12V synchronous buck converter capable of 10A continuous output. To save board space, I placed a 4.7µH shielded power inductor (Coilcraft XEL4030) directly adjacent to the high-side bootstrap capacitor loop. The layout passed DRC, and the copper pours looked solid.
- The Numbers: At the switch node, the high-side MOSFET turned on, pulling 10A through the inductor in roughly 20ns. That yields a $di/dt$ of $5 × 10^8$ A/s. Even with a "shielded" inductor, fringe magnetic fields exist. I later measured the unintentional mutual inductance ($M$) between the power inductor and the bootstrap loop at roughly 40 nH.
- The Outcome: Applying $V_2 = M(di/dt)$, the induced voltage spike was $40 × 10^{-9} × 5 × 10^8$ = 20 Volts. This 20V spike was injected directly into the bootstrap rail, which was nominally sitting at 5V above the switch node.
- What Went Wrong: The gate driver IC had an absolute maximum bootstrap voltage rating of 7V. The 20V magnetic spike punched through the internal bootstrap diode, shorting the high-side gate driver to the switch node. The IC released its magic smoke on the third power cycle. The Fix: I rotated the inductor 90 degrees (aligning the air-gap axis perpendicular to the bootstrap loop) and moved it 8mm away. This dropped $M$ to less than 2 nH, reducing the spike to a harmless 1V, and the board has run flawlessly since.
Clearing Up the Confusion: Self vs. Mutual vs. Parasitic
People frequently confuse mutual inductance with other coupling mechanisms. If self-inductance is the water hammer effect in a single pipe when you slam a valve shut, mutual inductance is a pressure surge in that pipe bursting a valve in a completely separate pipe running parallel to it. Here is how to separate the three main culprits of circuit interference:
| Phenomenon | Driving Force | Coupling Medium | Primary Mitigation Strategy |
|---|---|---|---|
| Self-Inductance ($L$) | Current change ($di/dt$) | Component's own magnetic field | Add snubbers, use freewheeling diodes |
| Mutual Inductance ($M$) | Current change ($di/dt$) | Shared magnetic flux between two loops | Increase distance, orthogonal orientation, magnetic shielding |
| Parasitic Capacitance ($C$) | Voltage change ($dv/dt$) | Shared electric field between two conductors | Increase distance, add grounded copper shields (Faraday cages) |
For deeper reading on transformer coupling and magnetic fields, the Electronics Tutorials guide on Mutual Inductance provides excellent foundational math. For practical PCB layout rules to minimize these effects, refer to Texas Instruments' application notes on power supply layout.
Frequently Asked Questions
Can mutual inductance be negative?
Yes, in circuit analysis. The sign of $M$ depends on the physical winding direction of the coils relative to each other, denoted by the "dot convention" on schematics. If the current enters the dotted terminal of the primary, the induced voltage on the secondary will be positive at its dotted terminal. If you flip one coil physically, the mathematical sign of the mutual inductance in your mesh equations flips from positive to negative.
How do I measure mutual inductance on the bench?
You cannot measure it directly with a standard multimeter. You need an LCR meter. Connect the two coils in series. Measure the total inductance ($L_{aiding}$) with the magnetic fields aiding each other, then reverse the connections on one coil and measure again ($L_{opposing}$). The mutual inductance is calculated as: $M = (L_{aiding} - L_{opposing}) / 4$.
Does a grounded copper pour under an inductor block mutual inductance?
No. A solid copper pour will block electric fields (parasitic capacitance) and high-frequency eddy currents, but low-frequency or static magnetic fields pass right through non-magnetic materials like copper. To block mutual inductance, you need high-permeability magnetic shielding (like mu-metal or ferrite sheets) or physical distance and orthogonal orientation.
Why do switch-mode power supplies specify "shielded" inductors?
Shielded inductors (like those with a molded magnetic resin or a continuous ferrite core) contain the magnetic flux inside the component body. This drastically reduces the fringe field, lowering the mutual inductance ($M$) with nearby sensitive traces. Unshielded inductors (like drum-core chokes) are cheaper and offer higher saturation currents, but they spray magnetic flux everywhere, making PCB layout significantly more difficult.
