Random Numbers

[locko]

This is untested, but was thinking of trying it someday....

The theory.
Leave an analog pin floating and take a voltage measurement to use as the seed for the random number.

Not sure how mach variation there will be in each of the measurements. The other thing would be to reduce the sample time for the conversion which may give to less accuracy and give better random results.
Also, if the pin is connected to a wire or long track, it should act as an antenna and increase the variation in the measurements.

Maybe some of the experts around here will have any ideas if this could work

Paul

[eggman]

I once did a project that used a DS18B20 temperature sensor. I used this sensor to also make the seed number for the random generator, worked quite well.