Arduino (Pro Mini) locks up and "loses" bootloader

Hi!

Today, for the second time, my Arduino Pro Mini 328 was running a sketch, then suddenly froze. a reset did not help. The LED on pin 13 no longer lights up after a reset, indicating a lost bootloader!!

my questions:

  • how could this happen? I was running a simple sketch, controlling some LED drivers and reading analog inputs, nothing fancy. I was powering the whole thing using a regulated 5V supply. The only thing I can think of, is that the high power LEDs might have drained too much power from the supply at some point. But that would just have triggered a reset, i don't think the ARduino's memory should be damaged by something like that.

  • how can i load up a new bootloader? I have a second Arduino using the AVR ISP emulator, and I hooked them up, but the Arduino IDE freezes when I select "Burn Bootloader". I've disabled the auto-reset using a resistor.

Has anyone ever experienced something similar?
Looking forward to answers! Thanks a lot!

I just thought of something else:
I am reading the voltage on an audio cable going from my PC to my stereo to control an Arduino-powered VU-Meter. Sometimes, when turning off the light in my room, my VU-Meter spiked shortly, as if noticing some interference from turning off the light.
While i didn't to anything similar at the time the Arduino froze, do you think this could have been the cause of the problem? Some sort of overvolting via an analog input? Is that possible?

EDIT:
another note: When I use AVRdude via terminal, I receive a different device signature each time, usually with blocks of 0x00 or 0xff, e.g. 0x00ffff, 0xffffff, 0x000000. They change everytime I execute AVRdude. Weird, huh?

The LED on pin 13 no longer lights up after a reset, indicating a lost bootloader!!

How does it behave when it's first powered?

No reaction whatsoever. The power LED lights up, but nothing else.
I must say I find the fluctuating device signature the most intriguing symptom.

I must say I find the fluctuating device signature the most intriguing symptom.

Is the Arduino connected to anything while the device signature is read?

If "yes", how does it behave when it isn't connected?

No, the arduino is not connected to anything besides the ISP (another arduino, which has successfully programmed an ATtiny13, for example).

I am glad it is not just me! I have got two pro minis that run for about 20 minutes, stop completely and will not restart unless they are powered down for a few hours. On reconnecting the power they will spring to life and run for 20 minutes again. I also have another two that have stopped working completely after working perfectly for more than 5000 hours.
I also had a number of Duemilanoves working at the same site, clocking up in excess of 10000 hours without any failures at all.

No, the arduino is not connected to anything besides the ISP

Very carefully check the ISP connections (including ground). If the device signature is still unreadable, you very likely have a damaged processor.

:cry:

@jabber: Voltage regulator getting hot?

@Coding Badly: That is what it seemed like but the 5V pin checked out ok with a meter.

This has piqued my interest enough again to see if my middle age eyesight will allow me to poke around with a scope probe to see what is going on without shorting everything on the board.

I have got two pro minis that run for about 20 minutes, stop completely and will not restart unless they are powered down for a few hours. On reconnecting the power they will spring to life and run for 20 minutes again.

OMG! Now that you mention it, I had the same problem with the Pro Mini before! However, back then I blamed it on poor coding on my side, some other hardware trouble or on too weak a power supply. But I guess it must have been what you're experiencing.
Do you have any Pro Minis running flawlessly as well?

I have contacted my retailer and am hoping for a replacement now. Still, it's nice to know I'm not completely alone :slight_smile:

I have still got about ten of them running and a very uneasy feeling about how long they will last. The pieces of kit they are installed in are still under warranty for another 3 months.

I just put a scope probe on one at power up, the voltage regulator output is 4.9V and the oscillator is running at 16Mhz. Twenty minutes later the regulator output is 4.9V and the oscillator has stopped. Very odd.

Is it possible that the ATmega got damaged by a negative voltage on its analog input? I was reading the voltage on a line-level audio signal, so relative to ground, it was sometimes positive and sometimes negative. Any possibility that might have been the cause?
(I attached the audio signal using a 1:1 audio transformer, audio signal on one side, the analog in and Arduino GND on the other)

...aaaaand, it's alive again! Just like that! After sitting around overnight, it greeted me with a little green "blink" and accepted my uploaded sketch without troubles. So, I am definitely experiencing the same problem as you, jabber!
Does your ProMini freeze while running a specific program? Could you try and reproduce the crystal stopping while running the stock "Blink" sketch? That would be most interesting.
Thank you!

For my own case there are about 20 ~ 25 microprocessors running at the site where I have been working and there have been 5 identical failures of 4 pro minis and one full size pro. None of the Diecimilas, or my own design Mega644 boards have failed and none of the other boards (a mix of Atmel & Microchip) which are not my responsibilty have failed.

A quick Google turns up a few more links like this one Nerdipedia | Arduino Pro Mini

scroll down to the bottom.

You posted while I was still editing my last post.
Yes I test the boards by programming them with the blink sketch and they definitely stop after a period of time :slight_smile:

My supplier suggested it might be a bad fuse bit that is causing some problems with the oscillator.
Would you mind reading the fuse bits on one of your "bad" ProMinis and one of your "good" ones?
The fuses of mine (it started working again today... :-?): Low 0xFF; High 0xDA, Ext 0x5 (which I guess is 0x05). Is there anything wrong with that? I used the AVR Fuse Calculator but don't really know what's right for me :-/

The only board that will cooperate at the moment is a pro mini with the intermittent fault, the others are refusing to run at all today.

The fuse settings are Low 0xFF; High 0xDA; Ext 0xFD.

Here is another link about the problem. http://forum.sparkfun.com/viewtopic.php?f=7&t=22955&hilit=pro+mini

I have emailed them today about my experiences with the boards.

My supplier recommended changing the low fuse to 0xF7 (to Full-Swing Crystal), arguing that the Oscillator would run with a higher voltage and thus more stably (is that a word?).
After setting the fuse and verifying it successfully, the chip died again.
Coincidence? Is it bricked? Or just playing dead as usual? I guess only time will tell.

Thanks for taking the time to check, by the way! :wink:

Which software did you use to read the fuses?

For the first reading above, I used Avrstudio because it was quick & easy.

I have just tried again with command line avrdude and got a reading of 0xFF; 0xDA; 0x5.

I do not know if it applies here but there are some differences about how unprogrammed bits are regarded in the fuse settings. I have only had a very brief read about it.

I used AVRdude. Hence the 0x5 reading.