TVS Diodes - Do they limit current or do you need a current limiting resistor?

[aiq25]

Hi. I was wondering do I need to have a current limiting resistor for TVS diodes? I always thought TVS diodes by design limit the current but recently I was told that was not the case.

What I mean is... say I have a power supply capable of supplying 100's of Amps for few milliseconds at 25V. Should I include a current limiting resistor in my design to limit the current to say 10A max? Or less....

[T3sl4co1l]

Why is your power supply higher than the TVS diode rating in the first place?

A TVS is just a zener diode.  It has a low (incremental) resistance (at the clamping voltage).  If you have a very low impedance source, like a power supply that's accidentally twice the voltage rating being hot-plugged, yeah, current limiting is required if you want to avoid damage.

For transients, it's not usually required, but any impedance you can add of course helps.

IEC surge has a source impedance of 2 or 20 ohms; fast transient and ESD have impedance 50-300 ohms (ballpark).  This impedance, combined with the short duration of these transients, is how the bare diodes withstand such treatment.

When the source voltage is high, things get more difficult.  Say, mains transients: you have to deal with ~kV surges, but you can only clamp to maybe 400V because that's normal peak line voltage.

Now, within a given family of TVS diodes, they all handle about the same power, whatever the voltage rating.  So, high voltage diodes can't handle high currents.

If you're clamping a 1.5kV mains surge to 400V, that's 1100V / 2 ohm = 550A peak.  This may well be more current than hot-plugging a 48VDC supply into a short circuit!

So the power and energy are much higher (~MW, and joules at a time) -- beyond what a mere TVS can handle (or an affordable one, at least).

This is why MOVs are used for mains protection: way more energy capacity.  Downside is, higher internal resistance.  Which, yes, does reduce the surge current a bit -- but that makes it more difficult for your equipment to deal with what's left, because now the surge looks like a Norton source with a peak (clamping) voltage of, say, 800V and a source resistance of fractional ohms.

If your risk scenario is different supply voltages, not induced lightning surge -- consider an overvoltage protection circuit instead.  This is also relevant to automotive load dump (a very slow transient (~100ms) that therefore delivers tons of energy, maybe 10-100J), where shunting the surge is difficult.

Tim