Common TVS Diode Failure Modes: Short, Leakage, and Open Circuit

A TVS diode often fails as a short circuit or low-resistance path after severe surge stress, but that is not the only failure mode. A device can develop higher leakage, a shifted breakdown window, or weaker clamping without becoming a hard short. An open circuit can also occur when a bond, solder joint, pad, via, or narrow PCB trace is interrupted. A single continuity check cannot identify all of these conditions.

Begin with the power removed and isolate the TVS from parallel circuitry before drawing a final conclusion. Measurements made in circuit include the input capacitor, power converter, interface IC, and other protection branches. They are useful for screening, not for confirming the health of the TVS itself.

What does a short-circuit failure look like?

When pulse energy exceeds what the junction and package can withstand, localized melting can create a permanent conductive path. The input supply may collapse, a fuse may open, a current-limited supply may repeatedly restart, or a communication line may remain clamped near one level.

In some systems, a shorted TVS is an intentional fail-safe outcome. That strategy works only when an upstream fuse, resistor, current limiter, or electronic disconnect removes sustained power. Without upstream isolation, the normal supply continues to heat the damaged diode. The package and PCB can then carbonize, changing a contained component failure into a board-level hazard.

A short circuit therefore answers only the last part of the event. The analysis still has to explain why the pulse exceeded the device capability or why sustained current remained after the TVS entered avalanche.

Why is increased leakage harder to find?

A surge can leave localized defects in the junction without forming a metallic short. At room temperature and low test voltage, the part may look normal on a multimeter. Leakage can rise sharply near VRWM or after the junction becomes hot.

Typical system symptoms include:

• higher standby current in a battery-powered product;
• a shifted bias voltage on a high-impedance input;
• local heating at the TVS during normal operation;
• intermittent failure during hot soak or burn-in;
• earlier failure during a later surge event;
• degraded communication when the damaged device loads a signal line.

Leakage must be measured at a documented voltage and temperature. An ohmmeter resistance value cannot be compared directly with a datasheet IR specification. Board contamination and parallel paths should also be removed from the result.

Can the TVS or its protection path fail open?

High energy can interrupt a bond wire or internal connection. It can also damage the external solder joint, pad, via, or narrow trace before the silicon becomes an obvious open circuit. A clean-looking package does not prove continuity, and a scorched package does not prove that the silicon itself is open.

Inspect the full current path:

• the trace and vias from the surge entry point to the TVS line terminal;
• the connection from the return terminal to chassis, protective earth, or local power return;
• solder joints for cracks, voiding, arc damage, or pad lift;
• upstream fuses, resistors, and current-limiting elements;
• the downstream IC, whose internal clamp may have carried part of the current.

An open protection branch can be deceptive because normal product function may return. The next surge then reaches the protected circuit with little or no TVS action.

What is a repeatable diagnostic sequence?

Separate “the product works again” from “the protection device is healthy.” A useful sequence is:

1. Preserve the event record: waveform, polarity, pulse count, product mode, supply current, and observed symptom.
2. With power removed, check resistance at the input and voltage drop across the suspected branch.
3. Lift at least one TVS terminal and measure the device separately from the remaining board.
4. Apply a controlled, current-limited DC voltage near the relevant working level and compare leakage with unused samples.
5. Check continuity through pads, vias, traces, and the return connection.
6. For parts without a hard short, compare VBR and a controlled clamp curve with a new sample.
7. Record post-event function and leakage instead of relying on visual inspection alone.

Do not keep increasing pulse severity on the original failed unit simply to make the damage easier to see. That destroys evidence about the initial failure. Use separate samples for destructive confirmation.

Why can devices in the same voltage class fail differently?

Consider three 24 V TVS models from ASIM. The SMA04J24V, SMB06J24V, and SMC15J24V all have a 24 V VRWM, a VBR range of 26.7 to 29.5 V, and a maximum VC of 38.9 V. Their associated IPP values are 10.3 A, 15.5 A, and 38.6 A respectively.

The equal voltage window does not give them equal pulse capability. Die size, package, peak pulse power class, pulse width, starting temperature, repetition interval, and PCB thermal conditions affect the failure boundary. Replacing a failed 24 V part with another 24 V model without checking those conditions can repeat the same problem.

The listed VC must also remain connected to its specified IPP and waveform. It is not a constant voltage source. If the actual surge current is higher, if the pulse lasts longer, or if layout inductance adds overshoot, the downstream pin may see a substantially different voltage.

How can the root cause be separated from the failure mode?

Replacing the TVS and seeing the product recover shows that the branch was involved. It does not prove that the original diode was undersized. The cause may be an incorrect VRWM, an unexpectedly long surge, a poor return path, or sustained power that continues after avalanche begins.

Ask and document these questions:

• Did the highest normal voltage approach VRWM, especially during charging or hot operation?
• What current actually flowed through the TVS branch during the event?
• Did the pulse width or repeated-pulse interval exceed the thermal assumption?
• How different were the voltage directly across the TVS and the peak at the protected IC?
• Did the event create a short pulse only, or did the power source continue supplying current?
• Was the device already damaged by an earlier handling or qualification event?

The answers determine whether the corrective action belongs in voltage selection, energy rating, upstream current limiting, PCB routing, thermal design, or process control.

What measurements are useful before and after surge testing?

A pretest baseline makes soft degradation easier to identify. Measure representative samples before testing and repeat the same checks afterward. Keep the fixture, temperature, and settling time consistent.

A compact record can include:

• leakage at one or more stated reverse voltages;
• breakdown voltage at the specified test current;
• clamp voltage at a controlled, nondestructive pulse level;
• capacitance for signal-line devices where loading matters;
• product standby current and functional behavior;
• microscope images of the package and solder joints.

Not every production test needs all of these measurements. Failure analysis does benefit from a baseline because a degraded part may still fall inside a loose continuity limit.

Common questions

Can a shorted TVS simply be replaced by a larger package?

Not without confirming the voltage window and event energy. If sustained overvoltage caused the failure, a larger package may only take longer to overheat.

Does an open continuity reading prove the TVS is healthy?

No. Increased leakage, shifted VBR, and degraded clamping can pass a simple continuity test.

Must a failed TVS look burned or cracked?

No. Electrical parameters can move without visible package damage. Visual evidence is helpful but not conclusive.

Why does the product power up after the TVS is removed?

The damaged TVS may have become shorted or excessively leaky. Removal is a diagnostic step, not a permanent repair, because the interface is then unprotected.

A complete failure report states the observed failure mode, the event that created it, and the design or process change that prevents recurrence. Replacing the damaged diode alone rarely supplies all three answers.