I have a custom PCB designed to interface an Arduino Nano's ADC pin that had 12v from a bench power supply inadvertently connected to a header pin meant for 5v input. The ATMEGA328P now gets extremely hot even removing the Nano module from the PCB. It seems to program, though it will occasionally spit out an error during verification that can be overcome by programming it again. The ADC can still be calibrated and matches within reason (<1% error) the displayed bench DMM. The Nano module normally reads approximately 4.5V, which now reads 3.5V on the 5v rail.
I thought I had designed robust protection into the PCB after doing research into diode clamps, crowbar circuits, etc. I settled on the following circuit that takes the incoming raw signal, passes it "through" an ESD array (DRTR5V0U2SR-7) and then into the circuit below.
simulate this circuit – Schematic created using CircuitLab
My calculations say that 1.2mA should get through the 10k resistor and simulations in LTSpice (with modeled internal clamp diodes of the microcontroller) show at most 726uA (assuming a 4.5V rail.) Most of it is going through the bat54 diode and approximately 400nA through the internal diodes.
Removing the Nano module and powering the bare board draws very little current, so I do not believe anything shorted on the adapter/support PCB. Testing the BAT54S array shows a voltage drop of 0.25V from input to 5V rail and 1.25V from rail to ground. Reversing the leads reverses the readings. Comparing the values to the other BAT54S in circuit show the same values.
I thought originally that maybe I backfeed the regulator on the Nano due to the clamping diodes, but at 1mA and various power hungry modules, like a connected 20x4 line LCD, I think the rail would absorb that amount of current. My next thought was the TVS conducted a large amount of current that overwhelmed the rail, but due to the low current when the bare PCB is powered I do not believe the TVS is damaged.
I'd like to improve my understanding and hope to learn and prevent this in the future. What am I missing?
Edit: I've updated the schematic with the DRTR5V0U2SR-7, to more clearly show the internal diodes from the DRTR5V0U2SR-7 datasheet. The 12V was applied to the node labeled RAW INPUT which is the pin header on the support board. The 12V from the supply was connected while the USB was powering both the Nano, and the adapter board (the Nano was in the socket on the PCB.) The voltage was applied for approximately 5+ seconds before I realized I had moved the leads to a pin expecting 5v not 12v. Later, I measured the short circuit current to be current limited at 1 amp on the bench supply.
The BAT54S diode array was measured while still mounted to the PCB. However, there are about 7 of the BAT54S spread around the PCB and they all measure identical while measured in circuit.
When I removed the Nano, the fault still exists, and it is definitely the microcontroller getting scorching hot as that part heats up to finger burning levels while the heat slowly transfers through the board to the regulator on the opposite side.