I've strapped a 6V zener diode across the 5V/GND rails of my circuit, and I want to know what will happen if my supply were to somehow fail pass 12V straight to the rest of my board.
The zener will definitely fry, but what happens afterwards? Can I say for sure whether the result will be a short or open circuit? Is there any way to ensure I get a short (thereby protecting the other components on the board)? I've learned all about diodes, but no one mentioned what happens after you fry one.
First, why will the zener fry? You may have a 12V supply powering your regulator, but your supply must have a fuse or circuit breaker somewhere to be safe. If this fuse or circuit breaker has a current rating which the zener can tolerate, then your PCB should be protected by the zener while the fuse shuts off, and no harm will come to the zener. If the current rating is higher than the zener can tolerate, well, you've made a mistake in choosing a current limiter.
What is the function of the 6V zener? If its function is to protect the 5V rail of your circuit, then I suggest that it's not going to do a very good job. Many 5V components have a maximum input voltage of around 5.5V or 6V, so a 6.2V zener won't help much. I'm not sure what your environment is like, but it's usually better to just shut everything down if your regulator fails than to try to let the zener run everything.
One common use of a diode for protection from error conditions is to use a reverse-biased rectifier diode across the input terminals. That way, if anyone connects the source backwards, the power will be dissipated in your designated diode. Make sure that this diode's forward voltage is less than the maximum reverse voltage on the protected components. In this configuration, when the source is plugged in backwards, nothing will work: The supply voltage will be at -0.7V. Presumably, you'd notice that the power LED was not on or that the circuit was not functioning, and correct your error.
Second, no, you can't say whether the result will be a short or open circuit. You've operated the device outside of the specifications, so anything could happen.
All zener diodes have a datasheet.
A datasheet like a promise the manufacturer gives to you -- if you stay within certain tight limits described in the datasheet, then the manufacturer guarantees that the device will work as described in the datasheet; if you stay within other somewhat looser limits described in the datasheet, then the manufacturer guarantees that no permanent damage will occur to the device, and when it returns to the tight range, the device will later work as described.
There are no guarantees as to what exactly the device will do outside the tight limits, and so good designers assume that the Worst Possible Thing will happen outside that range. Then when customers do things to their boards far beyond what they were intended to do, the designers are pleasantly surprised that there is anything left to salvage :-).
If you put a power resistor between the power supply and your board, you can choose a resistor such that it can limit the current such that even with 12 V on one end of the resistor, the zener can hold the other end of the resistor close to 6 V without exceeding the zener maximum power spec.
If the zener ever does exceed its maximum power spec, we should assume the Worst Possible Thing will happen -- in this case, that it fails open.
More sophisticated methods of protecting components from over-voltage include: various crowbar circuits (one example); using a similar resistor to limit power upstream before it goes into the voltage regulator; using spark gaps to limit the incoming voltage to 600 V; and various circuits that turn themselves off when the input "looks wrong" using a transistor that can easily hold off several hundred volts when turned off ( How do I protect against an automotive load dump? ).
You may consider a thyristor crowbar circuit.
simulate this circuit – Schematic created using CircuitLab
If the voltage rises enough to cause the zener to breakdown the thyristor is triggered and shorts out the supply. If the fuse doesn't blow (or doesn't exist) the thyristor will remain on until the current is switched off. C1 reduces susceptibility to noise.
The term 'crowbar' is a railway term from third-rail electrification systems. In the event of an emergency the lineman could trip out the electrical supply by throwing his crowbar between the live and one of the running rails on the track.
This circuit is not very precise due to variability in thyristor turn on voltages. See Axotron crowbar circuits for more ideas and improvements.
In general, you cannot count on undefined behaviour of components - because it can be, you know, undefined.
What I'd recommend is installing a fuse (either regular wire fuse, or auto-resetable polymer-fuse), that limit the amount of current the circuit can take without blowing the Zener.
Alternatively, you can install a resistor in series with your circuit - if you have a very low expected power consumption, this will not affect you much, and just protect the circuit as a secondary, less efficient supply.
Fuses in general are a good idea to have a controlled failure, I wonder why people forget them.
I conducted experiment with a zener. The zener normally fails SHORT circuited in reverse bias, but before it get short circuited, its terminal voltage goes up, that is it behaves like open circuit for few seconds and then goes to short circuit permanently. So using one zener will not protect the device always. So a better plan to connect 2 or more zeners in parallel. For e.g.: to protect a 5v line, connect 5.1v and 5.3v zener in parallel.
