trackball3D

The trackball3D is a computer input device made for any axis rotation of camera view in pymol molecular visualizations. Unlike traditional 2D input devices like mouse, trackpad or ordinary trackball, this device senses rotations of a ball in any space axis and controls pymol camera angles accordingly. Ball surface accessible for hand manipulations is maximized to make rotations around any axis easy. This is not a replacement for normal mouse or trackpad, it is intended as additional device to be manipulated by another hand and fully devoted to space rotations only.

The related software works perfectly for pymol and Avogadro under linux only at the moment. It also works for FreeCAD, though some tuning is yet to be done.

I took electronics from two ASUS MU101C Blue Ray (or here) mice - for building one device, just buying an optical mouse is the cheapest way to get the motion sensor and USB electronics. For 3D trackball you want to maximize the ball area accessible to your hand so you want as small electronics board as possible. One trick to find suitable mice in an e-shop is to limit all the three dimensions to the same maximum (like 9.7cm) as you never know what they put as width or length or height :) and sort from cheapest and only then look for other parameters. The design is intentionally simple to make modification for other mouse easy.

You need to insert three 3.969 mm balls as bearings. Silicon nitride is ideal but steel ones are fine as well. The fit is very tight (balls do not move), use piece of wood to press balls in. The ceramic ones are very hard but could likely be broken when pressed by something hard as well.

The ball is a 61.5mm carom billiard ball, I got my blue Aramith one here. A cleaned new ball cannot be tracked by an optical mouse sensor - fingerprints are already enough to make it work but you do not want to rely on this. I scratched the whole surface with a very fine sandpaper and then slightly waxed it, this works fine (although the ball is not gliding as perfectly as before). Spraying droplets of some acetone containing paint might be an alternative. (You certainly need some aggressive paint, the original ball surface is quite able to avoid being painted.)

The ball can also be 3D printed in two halves and connected. It is quite possible to polish it using a series of sandpapers. You want to remove most 3D printed surface structure already by the first sandpaper. It is quite hard at the top of the ball, you really want the smallest layer height to make this easier. To make printing time acceptable, I guess variable layer height is a must. The 3D printed variant is interesting if you consider alternative construction with inertial sensors inside the ball (and some radio connection).

Parts are just glued together, this gives you some freedom in experiments. Especially gluing the light guide directly on the circuit board immensly simplifies things - use hot glue gun and just be careful to glue only on the sides. The two printed parts and the circuit boards are easily kept together by few drops of a super glue, as well as the pieces of silicon rubber tube used on the bottom.