failure - What problems can arise from temporarily shorting power supply of a small microcontroller circuit?

A friend purchased a microcontroller-controlled lighting device and in the process of testing it, shorted the VCC and GND together for a second or two. Coincidental or not, the device stopped working thereafter, and he asked me if shorting the power could have been the cause.

It is my understanding that such a short could only really pose a threat to the battery or power supply (or the wire involved on the power supply side), given lack of current-limiting options or a fuse. The device should only experience power loss.

Can a power supply short cause a small circuit to permanently fail? If so, how?

I explained that the circuit might have failed because of the nature of the short: two wires making brief, intermittent contact. It might have created repeated instances of high inrush current which normally only happens once during power-on. Though, I doubt this theory because there shouldn't be much inrush current on subsequent power-on "events" as not enough time has elapsed for capacitors to discharge, etc.

Answer

A fast drop of DC input power can kill devices if they are not specifically designed to withstand this case. There are two main causes of damage:

1. Current was actually running backwards thru a path it was only ever meant to run forwards thru. Note that just disconnecting the power doesn't cause backwards current flow, but actively shorting it can and probably does. Many linear regulators, including the otherwise rock solid 7805, are damaged when the output is held at a higher voltage than the input. Since the output of a linear regulator will have some capacitance on it, it will momentarily stay the same voltage when the input is suddenly driven to 0 by the short.


2. Startup problems when some internal voltages exceed the power voltage as it comes back up. Think of any capacitance on the pin of a IC. If the power is suddenly brought to 0, that capacitance will discharge thru the IC pin, thru its positive protection diode, to the supply that is now at 0. If the supply comes back up while this current is still flowing, the whole chip can go into SCR latchup. That causes it to draw lots more current than designed for, and will often kill it or the linear regulator feeding it as a result.