Note: This was not test-printed yet and the SCAD code is still alpha.This design is copied from:"ThinkGeek :: Nanodots GYRO DUO: Magnetically Gyroscopic Balls"http://www.thinkgeek.com/product/18ae/?pfm=homepage_BestDeals_10_18aeand brought into a well FDM printable form.Default parameters are for magnetswith: h = 5mm - d1=25mm - d2=15mmpossible source:https://www.magnet-shop.net/neodym-magnete/scheibenmagnete/scheibenmagnet-250-x-50-mm-n50-nickel-haelt-86-kg::250.htmlhttps://www.magnet-shop.net/neodym-magnete/scheibenmagnete/scheibenmagnet-150-x-50-mm-n40-nickel-haelt-48-kg::48.html86kg & 45kg holding force on full contact ! Take care of your fingers !### How to find the optimal magnet(s):You may need to recompile the stl files for your specific magnet size.It is assumed that all magnets have the same thickness.Try to choose the two magnet radii such that they fill a common sphere as good as possible. For feedback the OpenSCAD code echoes two "corner radius factors" back. If one of the magnets has a "magnet corner radius" factor below 1 you may want to increase this magnets radius or alternatively decrease the other magnets radius until both factors are near one.### Tipps for making the gyro-ball:*) Thinner & harder bearing pins should significantly reduce the probability of mechanism clogging since the magnets then create higher lever forces.one could use pieces of:iron nails / copper wire / brass rod / printing filament / toothpicks? / ...*) Tip for making pins from filament:Use a retractable utility knive to make a notch into a piece of ~3mm diameter PLA filament then use pliers to break the piece of (ABS may not be brittle enough to break) then cut eventually occuring pieces that deviate from a flat fracture off.*) use thinn oil to lubricate the joints*) keep mass of the gyro parts small (low infill) but note that this is not that importatnt since the magets are heavy anyway.### Further notes:*) through all the mechanics it's unavoidasble that the distance between the magnets becomes quite significant => big magnets that create big balls with relatively thinn shells are preferrable.*) when it comes to gimbals you should know about the gimbal lockhttp://en.wikipedia.org/wiki/Gimbal_lock*) the OpenSCAD design is animatable!*) it might be interesting to try adding excess mass to one magnet side*) are there sources for fully transparent plastic balls? ### Infos about some of the design decision:*) All sizes are dynamically generated from magnet size so that the maximum amount of magnet power can penetrate through the shell*) For simplicity of design and assembly an irreversible assembly design was choosen that includes glueing and or plastic welding.*) the gap between the gimal ring and the gimbal sphere is determined indirectly via the "bearflatadd"-patrameter (no specific reason for that) <<< change?*) in the code:radii with underline postfix correspond to the radii of the bearing flatsradii without underline postfix correspond to the majority of the gimbal surfaces*) the parameter relationships became so complex because all the stuff has to be cramped together to the absolute minimal space *) GPL was choosen as license because lack of original idea but quite a lot of code