Can Two TVS Diodes Be Used in Parallel?

Two TVS diodes can be connected in parallel, but their IPP or peak pulse power ratings should not simply be added and treated as a guaranteed new rating. Breakdown tolerance, dynamic resistance, temperature, and PCB branch inductance can make one device carry much more current than the other, especially during the leading edge of a surge.

When the goal is higher surge capability, a single device with enough margin at the required waveform and temperature is usually easier to specify, lay out, test, and maintain. Parallel TVS devices should be used only when there is a clear design reason and when current sharing is validated on the actual board.

Why do parallel devices not divide current equally?

TVS diodes are nonlinear avalanche devices. A part with a slightly lower VBR, lower dynamic resistance, or warmer junction may begin carrying more current first. That current raises its junction temperature and can change its electrical behavior further. On a fast pulse, the electrical path with lower inductance can dominate before thermal effects become important.

Several sources of imbalance appear in a real assembly:

• VBR, VC, and dynamic-resistance tolerance between parts;

• unequal package temperatures, copper areas, and cooling conditions;

• unequal trace length, trace width, via count, or return route;

• mixed polarity structures, packages, or land patterns;

• fast di/dt, where a small difference in branch inductance changes the initial current path.

Using the same part number reduces some variation. It does not guarantee that each diode carries half of a surge current across temperature, production spread, and pulse shape.

When is parallel TVS a poor choice?

Avoid relying on parallel ordinary TVS devices when the downstream voltage margin is narrow, the energy is high, reliability requirements are demanding, or the compliance boundary must be clear. Long pulses, repeated pulses, and sustained overvoltage increase the importance of thermal distribution and fault management.

These conditions often call for a different architecture. A larger single TVS may be appropriate. The system may also need an upstream fuse, current limiter, series impedance, inductor, electronic disconnect, or a staged clamp. Parallel TVS diodes are not a substitute for sustained-overvoltage protection.

What does a careful parallel layout require?

If the design must use parallel devices, make the two branches as electrically and thermally similar as practical:

1. Use the same part number, polarity, package, and preferably the same production lot.

2. Match the connector-to-device trace length, width, via pattern, and copper area.

3. Match each device’s route to the intended return reference; both paths should be short and wide.

4. Keep thermal conditions similar. Do not place one device on a large copper pour and the other at the end of a narrow trace.

5. Measure current sharing at the required pulse condition with a suitable current probe or low-inductance shunt method.

6. Repeat the check at relevant temperature extremes and after representative repeated stress.

Symmetry reduces known sources of imbalance. It does not prove perfect sharing, so the design still needs margin for the more heavily loaded branch.

Why should a single higher-rated part be considered first?

For a 24 V working window, the ASIM SMA04J24V, SMB06J24V, and SMC15J24V all have a 24 V VRWM and a maximum VC of 38.9 V under their stated conditions. Their associated IPP values are 10.3 A, 15.5 A, and 38.6 A respectively.

If a system needs a higher specified surge current, first review whether a single SMC-class candidate covers the real waveform and temperature. This approach often gives a clearer electrical path than paralleling smaller packages. The comparison still needs temperature, pulse width, copper area, and downstream voltage checks. Equal published VC values do not guarantee equal protected-pin voltage in different layouts.

Can series resistance solve current sharing?

Small balancing resistance can improve sharing in some slower or controlled-current circuits, but it adds residual voltage and dissipates energy. During a fast ESD or surge edge, the resistor package and its own parasitic inductance matter as well. It is not a universal correction.

Before adding balancing parts, verify whether the system actually needs parallel TVS devices. If a single qualified part and a better return path solve the requirement, that is normally the more robust option.

What should be measured during validation?

Measuring only total clamp voltage cannot show whether one branch is overloaded. The validation plan should include current in each TVS branch, voltage at the TVS network, voltage at the protected load, and device temperature where relevant. Keep the pulse source, cable, probe arrangement, polarity, and sample identity in the record.

Post-test checks matter too. A device can develop higher leakage or a shifted breakdown characteristic without becoming a hard short. If one part has been repeatedly overloaded, the parallel network no longer represents the original design.

How do faults change the network?

One TVS can fail short, leaky, or open. A shorted part may pull the supply down and force upstream protection to operate. An open part leaves more stress to the remaining branch. Higher leakage can alter normal bias and make later surge behavior unpredictable. After a failure, inspect and replace the relevant devices as a set, then check pads, vias, and the complete return path.

Do not assume the remaining TVS provides the original safety margin. The network has changed, and the cause of the first failure may still be present.

Common questions

Can two identical TVS diodes double peak pulse power?

No guaranteed doubling should be claimed without current-sharing evidence at the required waveform, temperature, and layout condition.

Will parallel connection lower VC?

Not automatically. Total current, current split, branch inductance, and measurement location determine the observed voltage.

Can one TVS remain in service after its partner fails?

The network should be inspected and normally restored as a matched pair. A fault in one part can alter stress and normal operation in the other branch.

Is a larger package always the better alternative?

It must still meet VRWM, VC, waveform, thermal, and PCB constraints. Larger package size alone is not a selection rule.

Parallel TVS devices are a controlled engineering choice, not a simple arithmetic shortcut. Approve them only after the reason for not using one adequately rated device and the evidence for acceptable sharing are both documented.