I have a grid PCB setup out in corridor that has run sensor, soil sensor, DS18B20 and LDR hooked up, few days back due to some mistake high voltage went to grid and all sensors got burned , so now I am thinking to put a voltage protection before voltage is passed to sensors coming from Arduino Mega. My luck was I am using Arduino Mega, which tolerated high voltage of 18 volts in VIN and it was fine but sensors that were taking supply from 5volt of mega couldn't survive.
I was under the impression that voltage outpput from 5volt pin is regulated no matter what volage is sent by VIN it output 5 volt at 5 volt or 3.3 volt.
Please help.
The first attack should be prevention - you don't say what the "mistake" was, but the system should be configured to prevent those sorts of things - well labeled connections etc. One brute force method you can use for protection is called a "crowbar overvoltage protection" - see Crowbar (circuit) - Wikipedia for an example. If the voltage goes too high, the SCR triggers on shorting out the supply and blowing the fuse. Again, prevention is a better approach though.
What is a "grid PCB", and what circuit are you using with both Arduino circuitry and high voltage?
As gpsmikey said, this sounds like an overall design problem. The voltage regulation circuitry on an Arduino has limits to the input voltage. Look at the documentation under the menu item "products" above.
i had the same problem, happend twice to me, broke 2 Nanos and 2 port expanders 
:D... i had no school for electronics, my only school was those incidents(and more) which made me read again and again about it and then i realised the best protection method is to check the damn circuit for 5 times before plug it in
If you really want to protect them the simplest way is to use a external regulator(they also have different protection about the temp, the voltage), some caps, i also used zener diode protection but i consider it to be a difficult idea
Good external regulator and opto-couple all IO. There is also the "hardened" UNO board. Long cables can pick up big spikes so of your sensors are not in the same box as the Arduino, isolate them. Use quality twisted pair wire. RF or IR also does a good isolation. I like optical because not only do transients not pass, neither do ground loops.
In my many years in failure analysis, I can tell you 99% of our component failures were connected to the outside world. I am surprised more shields have not been made with protection circuits. I bet one could use cheap TOSLINK drivers and receivers for short range optical connections. Going multimode glass is a bit more expensive. There are also a lot of line drivers and receivers with some amount of protection built in.
I don't know how to program yet, but I do remember my lab days!
Hi All,
I'd agree this may be poor design but the problem didnt happened at the grid side, (not wire short circuit) but instead the
the tiny screw which we rotate to adjust voltage on the blue box like thing on LM2596S got changed by itself, or I accidently but I rememeber I never touched that tiny screw ever, all i was doing was covering with the box that came along with it
I hope it didnt got readjusted due to heat dissipated from LM2596S , so for now I readjusted it. I now need a way that the voltage output from LM2596S should be cut if voltage is not what was expected.
