Browsing through the forum, I have found several threads on the subject "I have fried my Arduino Board" because of some easy made mistake.
I suggest some boards could be completed with protective circuits.
I don't think a PTC is fast enough to protect semiconductors, and with a low-power device like an arduino, the protection diodes serve mainly to raise/lower the device power rails instead of just a single pin; just as deadly (maybe.) At about $0.40 per pin, it would be pretty expensive protection (and BIG, too.)
There are multi-line SMD ESD filters that might be a better option for some types of damage (National NUF8000MU), but frankly, An Arduino is at a price point where "replace the main CPU" is adequate for situations that would zap the board...
I have to second that notion, and need to vent a bit about it. I've been beat up for not having protection circuit on top of protection circuit (like $30 in parts worth on occasion), and had every worst case scenario presented as absolute certainty if I don't take so-and-so's advice. (like housefires, etc).
People can get really zealous about circuit protection I guess is the main point, even if it doesn't really make sense. One guys agenda, after describing Armageddon of course, turned out he just wanted to mount it in a hard to get location and wanted extra reliability "insurance". Never mind the fact that more components can lead to more failures. :
We have a diode to protect from polarity inversion, apolyfuse on the power supply and each arduino pin has internal diodes that protect the inputs.
I believe this is good enough for the basic Arduino.
Having taught a HUGE number of workshops I've never seen anybody completely fry their Arduino. It's pretty robust even for beginners.
I would never add more layers of protection on the standard board itself, it will make it look scary like all those microcontrolled development boards that engineers design that scare off beginners.
If somebody needs to interface to a "dangerous" environment they should make a shield with proper optocouplers etc etc.
You can heat the device without exceeding the maximum peak voltage (which is the V^2 / R in section A). You can also heat the device when you force it to breakdown which is the voltages listed in section B.
The size gives you a larger power rating and peak power rating. You still have
to deal with the voltage rating which is the same for 1206 - 2512. The peak voltage
rating is apx twice the operating voltage rating for each package.
I think that the 0805 offers a reasonable protection for the size.
It is not about the voltage. It is about the pulse power.
Um, my impression was that most of the failures of the sort the OP was talking about occurred when the owner did something really stupid like connected the other side of their LED/whatever to a 12V supply instead of 5V through a resistor... Most of the time, even the newbies realize they did something bad. In a way, this is part of the education one SHOULD receive from something like the arduino.
ESD is comparatively easy to protect against, since there's been a lot of interest in doing so. Protecting an explicitly experimental device against accidental mis-connections of all sorts is nearly impossible (assuming reasonable cost is a requirement.) We have a saying "It is very difficult to make something truly foolproof, because fools are so ingenious."
The other risk is that putting in protection for application A may affect the operation of the device in application B (especially with multi-function pins like the ATmega).
For example --- put too high a series resistance for current limiting and your A/D performance could drop.
Hopefully with a socketed ATmega the majority of minor mistakes will
be in the $3 range
