Yet another bootloader stability problem

I have been using the Arduino Boards, standalone, in a number of large scale art projects. I previously was a PIC user, programming in C.

The Arudinos have been great - but I am consistently running into a stability issue: Code is programmed into the board (I’m currently using the mini - ver 3 with a custom PCB) and runs fine. However, after some period of time - sometimes days, sometimes minutes - the board doesn’t reboot the software correctly (this obviously happens when the power is turned off and then back on). I’ve also observed the same behavior in the outdoor piece happening spontaneously as well. In other words - coming back to the piece when I was sure there wasn’t a power outage, and having one of the boards having reset without rebooting. This is obviously harder to document, as one would have to sit around for hours/days waiting for this to happen.

The symptom is this: A steady pulsing of pin 13. When you press the reset button, there is no reset. The only way to reset is to power down/up. Then, the short wait followed by the pulsing pin 13 again.

If you reprogram the board with fresh code - it works instantly, until the next time it slips.

I’m assuming this pulsing on pin 13 is the bootloader cycle described elsewhere - and must mean that either my code has been erased and it is trying to read new code - or perhaps it received a spurious Serial signal and thinks a new program is being loaded?

This has been a very frustrating thing. I’ve followed the forum here, and added the 10K resistor from the Rx pin to ground - just soldering it the appropriate pins on top the mini board. I had hoped that this was the solution - but i am still getting the problem.

I wouldn’t mind spurious resets - or even the need to repower up to reset the work occasionally if it acts up. But the fact that the board is for all purposes dead until the code is reloaded is a huge problem.

Any ideas or suggestions anyone?