I have a doubt about overflow.
When can we say there is an overflow?
I think there is one when a number dont fit in a 8bit register.
I mean:

lda #$FA
adc #$30

is there an overflow in that situation?

maybe this could be a fooly question but i dont know

This is indeed a type of "overflow" but you must be careful about your terminology. If you have an 8-bit data type to receive your result, it is considered an overflow. If you are adding two 16-bits and during the first 8 bits you get an "overflow" it's actually referred to as a carry. When adding, the 6502 always sets the C flag on an overflow.

Subtraction is a little different. A borrow is represented (in the 6502) by a cleared C flag. If you're subtracting two 16-bit numbers and the lower 8 bits requires a borrow, the C flag will be cleared and the value 1 will be subtracted from the upper 8 bits. However, if you're subtracting two 8-bit numbers and the result is incorrect because it can't be represented by an 8 bit, then that is considered an overflow. For example, -100 - 100 will end up being +72 because -200 cannot fit in 8-bits. The 6502 detects the sign discrepancy of this operation and flags it as an overflow in the V flag.

The overflow flag tells if the sign bit of the result is wrong. You'll see the earlier discussion on this if you down, under "General Discussions", down to "Carry and Overflow flags..." The URL seems to be:
http://www.6502.org/forum/viewtopic.php?t=62
Welcome to the forum.

Garth