Hi Daniel,

what you have observed is typical for all in system programmable flash micros unfortunately. Power up and power down cycles without proper reset, over- and undervoltage conditions on Vcc and any other pin, even of very short duration can accidentally invoke the automatic in-system-programming micro routines internally of micro resulting in a destroyed or corrupted flash code memory.

You should control the rise and fall time of power up and power down. You should allow Vcc to reach 0V before a new power up, especially if the internal power-on-reset and brown-out-reset circuitry is used. And you should shield and filter the whole micro controller circuit so that no harmful over- and undervoltage events or radiation or static fields can reach the pins of micro. Put a Faraday cage arround the micro to even eliminate the bad effects of electrostatic influence and filter all lines running to and from the pins.

Electronic industry is making us believe that microcontrolles become more and more powerful and less and less susceptible to EMI. But the opposite is true. The more the geometries on the die shrink the more the micro becomes susceptible to EMI. And finally the simple touch of a finger is enough to make the house of cards collapse.

To be honest, it's kind of shock to me, that it's a Silab's micro that shows such a behaviour...

Kai Klaas