AVR Freaks

To implement a touch screen you use one of two technologies. The cheap and cheerful option is a resistive touch screen. This has a sandwich of glass panels placed on top of the LCD. The sandwich contains small plastic beads and the inner surfaces are coated in Indium Tin Oxide. This conducts electricity but has the same refractive index as glass so looks invisible (it's also used in the construction of LCD too). The beads keep two surfaces apart. When you prod with your finger or a platic stylus the two layers connect so a voltage can flow. The resistance (and hence voltage) varies in both X and Y directions so using two ADC channels to read voltage in X and Y tells you where the surface connection has occurred.

The modern kind of touch is capacitive rather than resistive. This is the kind of touch found in ipads, iphones and all decent "multi touch" devices. It's more complex than the simple two ADC resistive and usually there's a dedicated small CPU (often Atmel in fact) that is monitoring the induced charge created by the human operator in a field generated from behind the screen.

Most AVR projects would use resistive and you can buy "sandwich" panels with the same dimensions as most common small graphic LCD panels to go on top. Although the ITO is in theory "invisible" it does tend to attenuate the sharpness and clarity of the LCD.

Well you have an X ADC value and a Y ADC value. Presumably these can be scaled to a pixel x,y position. If your "buttons" are 20 pixels square in a 4 by 4 grid starting at offset 200,180 and your ADC converts to picel 234, 207 then its 14 pixels across the second button across and seven pixels into the 2nd one down. It's really just a question of intercepting rectangles. Picking an "easy" width/height (like a binary multiple) may simplify the maths.