AVR Freaks

I'd suggest that there is probably no regulator or diode in the cheapy boards. Applying 12V in this instance would equal smoke. Some investigation of the cheapy board is required.

You haven't looked too hard - there's datasheets on the web. The usb->serial is probably the least of your worries. I'd be sketching out what happens to vin.

I'd sketch out the circuit first. Use a multimeter to 'beep' out where VIN goes to and compare this against the 'official' schematic. I'd guess there is no diode to isolate Vusb from VIN.

Ok, so the regulator is completely burned. I have nanos exactly like yours, and others that are a little bit different. You can see in the picture attached, on the right, the one like yours, on the left, a different model.

I notice that my other model has more copper near the regulator and maybe that helps to spread heat and prevent the burning out, maybe your other nanos are also like this, so they survive. However, the AMS1117 is supposed to have thermal shutdown circuitry and this shouldn't happen... fake AMS1117, perhaps? check this: https://plus.google.com/+JamesSleeman/posts/Q5s4VmbQwj6

I got brave and decided to test my nanos with 12V, I observed no problems, the 5V rail was perfect, but they were not connected to anything, so they were "just" drawing about 20 mA current. In this situation, the regulator is dissipating (12-5)*20 = 140 mW power.

Now, the datasheet says:

Thermal Resistance:

 

SOT-223 package  ϕ  JA = 90°C/W*

* With package soldering to copper area over backside
ground plane or internal power plane ϕ  JA  can vary from
46°C/W to >90°C/W depending on mounting technique and
the size of the copper area.

So today is pretty hot were I am, lets say 30 ºC, if the thermal resistance was 90 ºC per watt my regulators should be at 30 + 90 *.140 = 42.6 ºC

However, if I was drawing lets say 200mA from the 5V rail of the nano the regulator would dissipate 1.4W and would heat to 156 ºC, maybe blow up.

But if the nano was well designed with more copper near the regulator, maybe the thermal resistance would be less, if it was for example 65 ºC/W it would only heat to 121 ºC and survive (from the datasheet, max is 125 ºC).

Maybe your application draws a lot of current from the 5V rail, so some nanos can handle it and others can't, depending on their thermal design.

Who knows? Maybe these are fake AMS that don't even have any overcurrent protection, so they just blow up? I searched and there are several people complaining about fake regulators that blow up in various conditions...

There are three voltage levels on the Nano clone.  There is Vin (pin #1) which is +7_+12VDC.  There is Vusb, which is about +4.9V, and there is Vcc, which powers the AVR and the CH340G.  This is about +4.3V with USB power and +5V with Vin power.   There is also a 3.3V source that comes out of the CH340G IC, but the amount of current available from this pin is undocumented.

The Schottky diode between the USB +5 (Vusb) and the AVR Vcc prevents the voltage from Vin from being applied to the Vusb when both the USB cable and the Vin power source are applied at the same time.  This is a common occurrence when the AVR is only using the USB for a COM channel and bootloading the program, and the main power for the Arduino comes from Vin.  Vin can be from +5.5V to about +15V, although at higher voltages the Arduino board will get noticeably warm.

When powered by the Vusb, the Arduino runs at about +4.3 to 4.5 volts DC, not +5V.  The Shottky diode drops about 0.3V, and Vusb is often less than exactly +5V.   It seems to be a characteristic of the AMS1117 regulator that if an input voltage is applied to the Vin that is less than the rated output voltage, then that input voltage will be passed to the AMS1117 output pin without damage to the AMS1117 IC.   This is why all the Nano clones have a +5 AMS1117 regulator instead of the +3.3V regulator in the original Nano designs.  If you put in +7V -- +9V into Vin, then the AVR Vcc is exactly 5 volts.

To make a 3.3V system from a Nano clone,  use solder wick to clear the solder from the cathode (the side towards the center of the Nano board) of the Schottky diode and lift this end of the component off the pad.   Then exchange the AMS1117 5.0V regulator for an AMS1117 3.3V regulator: (use the solder wick on the four IC pads of the AMS1117).  Lastly, connect a thin (AWG30 wire-wrap works best) wire from the cathode of the Schottky diode to the Vin of the AMS1117 3.3V regulator.   Now voltage from the Vin pin (@ 9V) goes to the input pin of the AMS1117 and 3.3V comes out onto the Arduino Vcc.  But the Schottky diode blocks this high voltage from going into the Vusb +5.   With no external voltage on Vin, the Vusb goes from +4.9 to about +4.5 through the diode, and then to 3.3V through the AMS1117 regulator.

I believe that your Nanos were damaged by accidentally applying +12 volts to the AVR/Arduino's Vcc pin (+5V) instead of the Vin pin.

I'd suggest that something in your circuit precipitated the failure.

After several days of troubleshooting, measuring voltages and currents etc.. and rebuilding just about everything in my circuit, I think I figured out what at least one of the causes of trouble was, I had a BME280 module connected to the device, which I had actually repaired a few weeks ago - by reflowing a new sensor onto the module. My guess is that this attempt to reflow was intermittently causing a short. I was going to write up a schematic and post that, as I thought I may have designed something wrong. I have spend several days in a maddening process rewiring everything - except for the sensor itself. I have replaced the sensor, although the replacement is evidently only 3v3 capable.. realising this after rewiring the sensor cable several times.. Going to switch-out an ESP32 so that should be fine.. - The most frustrating part was thinking I had solved this issue, go to reinstall my system at differently location and each and every time have this issue magically reappear...