Protecting buck converter from shorted input?

[Dago]

I'm currently working on a project which has a TPS562200 step-down converter generating 5V from the 12V input: http://www.ti.com/lit/ds/symlink/tps563200.pdf

The circuit has a PTC->TVS->100n PP cap -> common-mode choke-> 100µF electrolytic cap plus two 10µF ceramics and then the chip itself and after the chip 2x22µF ceramics plus more ceramics littered around the board.

The problem is that if I accidentally short the input (like when testing with alligator clips) it blows the chip! I was fairly certain that it was due to reverse bias and putting a (regular, slow) diode from the output rail to the input rail would solve it... nope. The chip still gets shorted from the SW pin to the ground. I've even had one chip blow without a short, just a bad DC-plug connection to the PSU.

How should I protect the chip? All I can think of right now is TVS from the switching node to the ground (never seen this done) or a series diode preventing any current from flowing back (increased losses).

[Mad ID]

What do you mean by shorting the input? If you short the input the only thing that can go malfunction is the supply supplying your buck...

Also, can you post a schematic of the input stage?

[Dago]

What do you mean by shorting the input? If you short the input the only thing that can go malfunction is the supply supplying your buck...

Also, can you post a schematic of the input stage?

Shorting the input as in shorting the input. Like I often power the device with alligator clips and sometimes they happen to touch eachother... That is what I would assume so too but this does not seem to be the case. I measured my lab PSU when shorting it and it does not peak one bit. Everything looks overdamped like it should. Reverse bias would explain it well because then power would flow from the output rail through the inductor (increasing the voltage) to the short, but not sure why adding a diode from the output rail to the input rail did not solve that.

Also never had this problem before and I've probably used tens of different step-down ICs and I'm confident I have accidentally shorted the input on pretty much all of them without any issues

When measuring it looks like the SW pin gets shorted to the ground so looks like the voltage still manages to peak even with the reverse bias diode.

Attached is a schematic of the PSU section.

[Richard Head]

I suspect the output tantalum caps are discharging through the inductor, upper MOSFET parasitic diode and short circuit. I would expect the upper MOSFET to short out. Is that the failure mode?
A series diode in the IC supply should solve it.

[Dago]

I suspect the output tantalum caps are discharging through the inductor, upper MOSFET parasitic diode and short circuit. I would expect the upper MOSFET to short out. Is that the failure mode?
A series diode in the IC supply should solve it.

The output caps are ceramic. And it seems like the lower MOSFET gets shorted to ground since SW pin is shorted to ground.

[Niklas]

When shorting the supply input, the possibility to do a proper shutdown with de-energizing is very limited. I suspect that the maximum or minimum potential at the SW pin is exceeded, with breakdown of the lower MOSFET as a result. The body diode of a MOSFET is seldom very fast as it is only parasitic and not the primary element of the component. The output voltage (5V) is fixed, so the SW end of the energized inductor will act as a whip.

[T3sl4co1l]

Possibly, what's happening is, without power, the synchronous transistor isn't holding, and the voltage gets unclamped and bad things happen.  Probably a schottky diode from GND to VCC, and another from OUT to VCC, will help, but also a pair clamping the switching node, GND to SW, SW to VCC.

Short circuits in general are startlingly fast and violent transients, and getting them anywhere near a semiconducting device is sure to cause problems.  It could be something as unexpected as sheer ground bounce, for all we know.

Tim

[Mad ID]

Shorting the input as in shorting the input. Like I often power the device with alligator clips and sometimes they happen to touch eachother...

Did you scope the output voltage during shorting event? Set the trigger to little above normal output voltage and see what you get. One thing you can try is to reduce the output capacitors, you don't need 50uH for a buck (most of the ripple is at the input)

Let us know