The original version of this project was happy when it was pointed ±2° of the sun. To be used as the basis of a sun clock that's accurate to within a minute (which is where I'd like to go) I need it to get that down to about ±0.25°. But I want to make this thing as simple as I can.

After kicking around possibilities, I decided ti add a sun-blocking disc to the end that points toward the sun. The disc is the same diameter as the spacing of the four LED light sensors. Now when it points directly toward the sun, the disc casts a shadow that passes through the center of each of the sensors.

As you can see from the photo, it really does point at the sun now, keeping the disc's shadow right where it needs to be. Is it good enough? I don't know yet. It's a lot easier to detect a 2-degree error than an 0.2 degree one. I'll need to think about how to measure it (without getting all fancy). But, yeah. A big gain.