Protecting Circuits from Transients: TVS Diodes, MOVs and Gas Discharge Tubes Explained

Your board worked perfectly on the bench for three weeks. Then it went into the field — or just got plugged into a different power supply — and the microcontroller is dead. No scorch marks, no smell, nothing visibly wrong. The replacement board dies the same way a month later. This is what an unprotected input looks like when it meets the real world: relays switching, motors spinning down, cables being hot-plugged, and electrostatic discharge from a hand on a connector.

Transient overvoltage — brief spikes far above the normal supply rail — is the single most common cause of unexplained field failures in student hardware. The fix is a small family of protection components: TVS diodes, metal-oxide varistors (MOVs) and gas discharge tubes (GDTs). They cost almost nothing, take up almost no space, and the difference between picking the right one and the wrong one is understanding five datasheet numbers and three layout rules. This guide gives you all of it, plus a worked sizing example.

What a transient actually is

A transient is a short, fast overvoltage event. Normal operation on your 12V rail might see 12.5V of ripple; a transient slams the line to 60V, 200V or more for nanoseconds to milliseconds. Semiconductors die when the electric field across their gate oxide gets too high — and that happens at voltages far lower than most students expect. A microcontroller input rated for 3.6V absolute maximum can be damaged by a spike you would never see on a multimeter.

The transients you'll actually meet:

• ESD (electrostatic discharge): a finger touching a connector. Human-body-model ESD hits 8kV contact / 15kV air in testing standards (IEC 61000-4-2). Extremely fast (sub-nanosecond rise), high voltage, low total energy.
• Inductive kickback: switching off a relay coil, solenoid or motor. The collapsing magnetic field forces current through whatever path it can find, generating L x di/dt spikes — easily tens to hundreds of volts for microseconds to milliseconds. This is the #1 transient in student projects with relays and motors.
• Hot-plugging and cable ringing: plugging in a cable creates ringing on the line inductance — repeated overshoot spikes at the connector.
• Load dump (automotive): the battery disconnecting while the alternator is charging. Hundreds of volts for hundreds of milliseconds — brutal, and why automotive electronics are heavily protected.
• Lightning-induced surges: not a direct strike (nothing on a student PCB survives that) but induced surges on long outdoor cables — sensor lines running across a roof, for example. Tested against the 8/20µs current waveform in IEC 61000-4-5.

Two distinctions matter. ESD vs EOS (electrical overstress): ESD is the fast, low-energy event above; EOS is longer-duration abuse like sustained overvoltage — different protection answers. And differential vs common-mode: a spike between the two wires of a pair vs between the wires and ground — your protector has to be wired to clamp the mode you actually have.

The three protectors: TVS diodes, MOVs and GDTs

All three do the same job — they look like an open circuit at normal voltage and turn into a near-short above a threshold, shunting the surge to ground. They differ in speed, energy handling and precision.

TVS diode (transient voltage suppressor): a silicon avalanche device, essentially a fast Zener built for surges. Fastest of the three (sub-nanosecond response), with the most precise clamping voltage — you know within a volt or two what your IC will see. Moderate surge current (tens of amps for a few hundred watts in SMA packages, more in SMC). This is the default choice for board-level protection of I/O lines, sensor inputs and data ports. Comes in unidirectional (for DC rails, cathode toward the rail) and bidirectional (for AC or bipolar signals) versions.

MOV (metal-oxide varistor): a ceramic disc whose resistance collapses above its varistor voltage. Handles far more energy than a TVS — hundreds to thousands of joules in the big discs — but responds slower (tens of nanoseconds) and clamps less precisely. Critical weakness: MOVs wear out. Every absorbed surge degrades the device slightly and its clamping voltage drifts upward over time; after enough hits it fails, sometimes shorted. Best at AC mains entry and power inputs, not on precision low-voltage I/O.

GDT (gas discharge tube): two electrodes in a gas-filled tube that arcs over above its sparkover voltage. Handles enormous surge currents (kA range) and is nearly invisible to the circuit when idle (picofarad capacitance). But it's the slowest (microsecond-scale) and has a dirty secret: follow-on current. Once the gas ionizes, it stays conducting even at normal line voltage — on a powered line it will sit there arcing until a fuse or series impedance clears it. GDTs belong at the building/cable entry of lightning-exposed lines, always with a fuse and usually as the first stage of a multi-stage design.

Property	TVS diode	MOV	GDT
Response time	Sub-nanosecond	Tens of nanoseconds	Microseconds
Clamping precision	Tight (± a few volts)	Loose	Loosest (sparkover varies)
Surge current	Tens of amps (SMB/SMC)	Hundreds of amps	Kiloamps
Energy per hit	Low–moderate	High	Very high
Wear-out	Minimal	Yes — clamping drifts up	Minimal
Idle capacitance	pF to nF (low-C arrays exist)	Hundreds of pF to nF	~1pF
Typical use	Board I/O, data lines, sensor inputs	AC mains / power entry	Lightning-exposed cable entry

Two supporting parts complete the picture: the polyfuse (PTC resettable fuse) for overcurrent — it goes high-resistance when current exceeds its trip point and resets when cool — and plain series resistance or inductance between protection stages, which lets a fast TVS and a slow GDT coordinate instead of the TVS taking the whole hit alone.

Reading the datasheet: the five numbers that matter

Protection datasheets look intimidating but you only need five parameters:

1. Standoff voltage (Vrwm) — the maximum voltage the part tolerates while doing nothing. Your circuit's highest normal operating voltage, including tolerances and ripple, must sit below this with margin. Violate it and the protector conducts during normal operation, overheats and dies.
2. Breakdown voltage (Vbr) — where the part starts conducting significantly. Between standoff and breakdown is the grey zone; design on standoff and clamping, not breakdown.
3. Clamping voltage (Vc) — the voltage the protected line actually sees while the part absorbs its rated surge current. This is the number your IC experiences. It must be below your IC's absolute maximum rating, with margin. A TVS with Vc = 30V protects nothing on a 5V-tolerant input.
4. Peak pulse power/current with its waveform — TVS parts are rated in watts against a 10/1000µs waveform; MOVs in amps against 8/20µs. The waveform matters enormously: the same part survives far more current on a short 8/20µs pulse than on a long 10/1000µs one. Never compare ratings across different waveforms.
5. Capacitance — the protector hangs on your signal line. A 1nF TVS on a USB data line will round off the signal and kill the link. For high-speed interfaces (USB, Ethernet, HDMI) use low-capacitance TVS arrays (single-digit pF), often built as a steering-diode bridge plus one TVS.

How to pick one: the selection recipe

Work through these steps in order:

1. Find your maximum normal voltage. Not the nominal — the worst case: supply tolerance, regulator overshoot, ripple. A "12V" rail with a ±10% supply and startup overshoot can touch 14V. Your standoff voltage must clear that with 10–20% margin.
2. Find your IC's absolute maximum. Read the "Absolute Maximum Ratings" table, not the operating conditions. Clamping voltage must sit below it with margin — this is the constraint most students miss.
3. Estimate the surge. For relay/motor kickback on a 12V board, tens of amps for microseconds is a reasonable design assumption; for lightning-exposed lines, design to the IEC 61000-4-5 levels your product claims. Pick the peak power/current rating (on the right waveform) to cover it.
4. Check the signal speed. DC power and slow sensor lines tolerate nanofarads; anything above ~1MHz needs you to check capacitance, and USB/Ethernet need dedicated low-C arrays.
5. Choose unidirectional vs bidirectional. DC rail or a signal that never goes negative: unidirectional (and get the polarity right — cathode band toward the positive rail, or it becomes a short). AC, RS-485, audio or any bipolar signal: bidirectional.
6. Check the package can survive repetition. A 400W TVS handles one big hit or many small ones; if your circuit switches an inductive load thousands of times a day, verify the repetitive rating and consider an RC snubber or freewheeling diode at the source too — protection at the victim plus suppression at the source.

Placement and layout: where protection actually works

A perfectly chosen TVS placed badly protects nothing. Three rules:

• Put it at the connector. The protector must be the first thing a surge meets. Every millimetre of trace between the connector and the TVS is inductance the surge will develop voltage across — 1nH/mm adds roughly 1V per amp per nanosecond of rise time, and ESD rises in under a nanosecond.
• Give it a short, fat path to ground. The surge current has to go somewhere: a via straight into the ground plane right at the TVS pad, not a thin trace wandering to a ground pin three centimetres away. The ground return is half the protection circuit.
• Don't route the protected signal past the protector. The trace from TVS to IC should leave directly; never run the sensitive trace alongside the incoming unprotected trace where the surge can couple across.

For serious threats, use two stages: a GDT or MOV at the cable/building entry, then a series impedance (resistor, ferrite bead or inductor), then a TVS right at the IC. The impedance drops the residual voltage between stages so the fast, precise TVS finishes the job the coarse first stage started. Without the impedance, both stages fire at once and the TVS takes current it was never rated for.

Worked example: sizing a TVS for a 12V DC sensor input

Scenario: a 12V-powered sensor feeds an analog input through a wire that also runs near a relay. The sensor output can reach 12V; the 12V rail worst case is 14V (tolerance plus overshoot); the ADC buffer op-amp has an absolute maximum input of 20V.

Step 1 — standoff: needs to clear 14V with margin. A 15V-standoff part fits (a 12V-standoff part would conduct during normal 14V peaks and cook itself).

Step 2 — check clamping against the IC: consider a 400W-class SMA part with 15V standoff — its breakdown is around 16.7–18.5V and its clamping voltage at rated pulse current is about 24V. That's above the op-amp's 20V absolute maximum, so a lone TVS does not protect this input. This is exactly the trap: standoff fits, clamping doesn't.

Step 3 — fix it with a second stage: add a 100Ω series resistor between the TVS and the op-amp input, plus a small low-voltage clamp (a second TVS or a Schottky diode pair to the 5V/ground rails) at the op-amp pin. Now the 24V residual appears across the 100Ω resistor: the current into the op-amp's internal protection is (24V − 20V) / 100Ω = 40mA, which you check against the op-amp's input current rating — and the local clamp holds the pin near its safe range. Alternatively, a resistor divider that scales the 12V sensor output down also scales the residual transient by the same ratio, solving both problems at once.

Step 4 — sanity checks: the resistor must survive the pulse power (I²R for the pulse duration — usually fine for microsecond events), the TVS unidirectional with cathode to the 12V line, placed at the connector with a ground-plane via at its pad.

(Part numbers and values above are typical of the common 400W SMA TVS family — always confirm the exact Vrwm, Vbr and Vc on your specific part's datasheet before ordering.)

Common mistakes students make

• Clamping voltage above the IC's absolute maximum. The #1 mistake. Standoff fitting is not enough — Vc is what the IC sees.
• No standoff margin. A 12V-standoff TVS on a rail that touches 13.5V conducts continuously, overheats and fails shorted — sometimes taking the board with it.
• TVS placed centimetres from the connector, with the sensitive trace running alongside the incoming line.
• Thin, long ground trace from the TVS instead of a ground-plane via at the pad.
• Unidirectional TVS installed backwards on a DC rail — it forward-conducts like a normal diode and shorts the supply. Check the cathode band twice.
• High-capacitance TVS on USB/Ethernet — protection that kills the signal it was meant to save. Use low-C arrays.
• MOV on a DC line using its AC rating, and ignoring MOV wear-out in circuits that see daily switching transients.
• GDT on a powered low-voltage line without a fuse — follow-on current keeps the arc alive until something melts.
• Expecting any board-level part to survive a direct lightning strike. Board protection handles induced surges and ESD; direct strikes are a building-protection problem.

FAQ

Can I just use a Zener diode instead of a TVS?
For tiny, slow overvoltages, a Zener plus series resistor can work. But Zeners are rated for continuous power dissipation (fractions of a watt), not surge pulses — a real ESD or kickback event will destroy the Zener before it clamps meaningfully. TVS diodes are Zener-process devices rated for hundreds of watts of pulse power. Use the right part.

Do I need protection on every input?
Protect every conductor that leaves the board or comes from outside it: power in, sensor lines, communication buses, anything a user can touch. Purely internal signals between two ICs on the same board generally don't need TVS — but do keep the decoupling and layout rules from the PCB layout guide.

TVS or MOV for a 230V AC mains input on my power supply?
MOV, sized for the AC line (a 275V AC-rated varistor for 230V mains), placed across line-neutral at the mains entry with a fuse upstream. TVS diodes top out at modest voltages and can't absorb mains-level energy. This is one place to be extra careful: mains work has safety implications — fuse it, keep clearances, and get the design reviewed.

Why did my TVS fail shorted?
Usually one of three causes: standoff voltage too low so it conducted during normal operation; a surge far beyond its pulse rating; or repeated smaller surges that eventually punched through. A shorted TVS across a power rail will then blow the fuse or overheat — which is actually the fail-safe direction, since it forces you to notice.

Can protection components affect normal circuit operation?
Yes, three ways: leakage current (nanoamps to microamps — matters for battery-powered and high-impedance sensor nodes), capacitance (matters for fast signals), and the standoff margin (too tight and it clips your normal signal peaks). Check all three against your circuit's needs.

Is a freewheeling diode across a relay coil enough, making TVS unnecessary?
The freewheeling diode handles the coil's own kickback at the source — always fit it. But it does nothing for ESD on the connector, hot-plug ringing, or surges coming down the cable from elsewhere. Source suppression and victim protection are complementary, not substitutes.

Limitations

• Transient protectors clamp brief overvoltage. A sustained 24V on a 12V rail is a regulator/power-supply problem — a TVS will conduct continuously and destroy itself trying to fix it.
• They don't protect against reversed polarity (use a series diode, ideal-diode controller or bridge), overcurrent (use fuses/polyfuses), or bad grounding.
• MOVs degrade with every hit — in high-transient environments they are a service item, not fit-and-forget.
• GDTs need follow-on current management (fuse + series impedance) on any powered line.
• Protection adds cost, board area and (slightly) capacitance and leakage — on ultra-low-power or RF-sensitive designs, these trade-offs need explicit checking.
• No board-level protection replaces proper mains safety practice: fusing, creepage/clearance distances and enclosure on the AC side.

Keep reading

• Buck vs Boost Converters Explained — know your power stage before protecting it.
• Power Supply Design: Linear vs Switching — where transients come from inside your own supply.
• EMI/EMC Basics for Student PCBs — the other half of real-world robustness: emissions and immunity.
• 10 PCB Layout Rules That Prevent Failures — grounding and trace discipline that make protection actually work.