[MCahoon]

Apologies for not having a diagram at work, and having to go from memory.

We used an IR Emitter/Detector pair (Radio Shack p/n 276-142). This is powered using a 7805 voltage regulator, connected as a custom circuit to the Power Distribution Board. The Emitter is driven from 5 V, with a 51 ohm series resistor (gives about 90 mA drive, max current on the emitter is 150 mA as I recall), and the sense voltage of the detector photo-transistor is developed across a 10 K (as I recall, possibly 15K) resistor.

The emitter and detector were each mounted in a 1-inch oak block (one of our mentor is a wood-shop artiste), with a hole drilled through just the size of the emitter or detector body. The emitter and detector were each super-glued into one of the blocks. The blocks were mounted inside frame rails on each end of our ball control roller with the hole in the blocks aligned with holes CNC'd into the frame rails so they were closely aligned. The holes are located at the height of the center of the ball. The whole assembly is under the frame and bumpers, so is additionally shaded from ambient light.

The distance between the rails is about 19 inches. This emitter/detector pair work great at that distance. With no ball in the roller, approximately 4.1 V is read (on the Analog Breakout), and with a ball captured, approximately 10 mV. The difference in ball position between not being detected, and being detected is about 1/2 inch.

Actually works much better than I originally thought it would. I was afraid it would not be able to transmit across that distance, or that the detection cone would be so wide that it would be hard to distinguish when the ball was captured.
We used these at the Oregon regional, and once we got the alignment and the kicker power dialed in, we score 1 ball from the far zone about half the time, and clear either one or two of the other balls.