Bartosz Wucke wrote:

Oh, well!
If there's say, 5 ohm between the pin and the ground left, if you connect enough power to blow up, erm, power up the pin to get more than 2.5V across the resistance, you have it on "logic 1" level. It means though, that there will be 0.5A of current going through the shrapnel, err, microcontroller, which gets dissipated (or cumulated) as heat. The 1.2 watt over 30 seconds will produce 8.5 calories, within volume of thin transistors and tracks inside the microcontroller, I'd say 0.01mm^3, raising the temperature to over 500 centigrades and possibly causing rather impressive-looking explosion.
If the resistance is high enough though, and you manage to upload the program within, say, 5 seconds, there is a small chance it will survive.

Hallo Bartosz,

your highly accurate calculations demonstrate, that you have a deep theoretical insight to physics of mircocontrollers. Your recommendation to solve problems by applying an extra portion of calories is amazing!

Unfortunately, with the 5sec duration there's also a certain probability to finally damage the chip.
But since scientific world has gained some remarkable understanding about 'quantum fluido mechanics' in the last years, there's another, a more scientific way to fix the chip, which much less probability of ending up with total damage. It works like this:

If an ensemble of quantum mechanical particles is exposed to a pressure which undergoes a very high change rate, quantum mechanical particles' states become coherent and a superfluid current of string states is invoked. The effect is, that if these quantum mechanical particles form a transistor the logic state of the whole transistor can be changed! So, if a transistor was in logic low state, then it can be changed to high state, and vice versa. Most amazing fact is now, that this so called Hammer-effect will distinguish between two certain states: AL-state and NON-AL state (read: alive-state, non alive-state). So, if the high pressure change rate is applied, then only the NON-AL components are involved. Undamaged components remain unchanged!

There's only one disadvantage: Hammer-effect only takes place in the moment of applying high pressure change rate. So, only when the hammer hits the microcontroller, damaged transistor can change its logic state. So, you need to apply the hammer many times. I think, at least complete code bytes can be programmed with each hammer hit. So, it seems to be possible, that you can program your damaged microcontroller by bytewise hammering the code into the Flash.

I think, in future Hammer-effect will play an enormous role in fixing damaged electronic components.

Very best regards,
Kai