Magnetic ball joint design

[Bob Bilbrey]

I’m of the opinion that a “well designed” magnetic ball joint systems should be superior to a Traxxis style pivot for Delta use.  But I’m not totally sold on the mag-ball approaches I’ve seen. 

Initial assumptions for the M3W MagBall design. 

1 – Seating in a 45 degree or similar cone provides only line contact which can overload polymer seats and will squeegee lubricants from metal and/or polymer seats.  Line contact can create very high contact pressures and friction.

2 – Ball end mills will NOT reliably and accurately produce the closely conforming seat for first class spherical socket mating with a precision bearing ball ( +/- 0.00005 for a run of the mill bearing ball – better for a really good one ).  You machinists out there know this to be true.  A top-end CNC can possibly single-point an adequate ball socket but that is beyond the usual capacity ( and cost ) of this gang.

3 – A ball mag socket must have a tightly coupled non-contact consistently gapped magnet face.

4 – The magnetic element should NOT be the ball, but should be a strong discrete magnet that always pulls the ball along the centerline of the socket ( is this obvious or not ? ).  Note this arrangement possibly does not yield the lowest mass; mag balls might be more mass effective ( -1)

5 – For a metal-metal ball joint generally a brass socket might be considered superior to aluminum except for the mass.  I think aluminum wins because it is lighter and , if the socket is of high  nesting precision and the lubricant system is robust, there will be NO actual metal-to-metal contact ( very light loading ).

6 - Mag joints have a much greater range of angular freedom than do mechanical sockets.

7 – W2S lubricants are phenomenally slippery and diamond based stuff might be even better (don’t really know as yet – maybe some hype).   I think at least the equal of PTFE, possibly superior but I haven’t got the data.  Sounds like I am shooting in the dark here which is probably true. So, given all the above assumptions, I’ll undress my M3W MagBall system.  Not finished and tested yet but I feel pretty good on the concept.  If you don’t buy the above, thanks for reading.

10 - The shot below is the exterior bits at the effector - very convention looking

11 - Cross section thru the diagonal rod - note the ball seat ( spherical ) is the yellow band.  It will be shown that the magnet is first loctited (or other) and carefully positioned in the socket bore

12 - The mag-ball sub assembly

13A - The key to this design idea is the formation of the ball seat as well as the method of creating the magnet gap ( 0.004in).  The socket is a simple lathe part with a 45 degree conical "seat primitive".  The cone should be turned to a slightly rough finish - I estimate this to be optimally 63-120 RMS.  A bright shiny very smooth surface may not work depending on how smooth.  The best way to create this cone is to "single point bore" with a sharp tool tip.  This will create slightly rough spiral tool marks.

13B - The spherical seat is produced by "peening" or "indenting" with a high precision steel bearing ball.  With aluminum a force of 1000 to 1500 pounds should form an indented seat about 1.0 to 1.5mm width ( yellow band ).  The actual force is dependent upon the degree of roughness and the alloy used.  It is easily determined by trial and error.  I used a dropped weight to form the seat by impact which works equally well - very repeatable.  The ideal seating band will be reflective and clearly curved.  It will also still have remnants of the spiral lathe tool marks.  These micro-scratches are desirable as lubricant retainers. 

13C - A robust mag-ball design must have a very small gap (but with a positive separation) between magnet face and the ball surface.  Note in my case the joint uses a 0.5000 inch bearing ball.  So to create the magnet gap I employ a 12mm precision bearing ball firmly seated in the socket by magical means.  Since this ball is slightly smaller than the .500 inch ball it seats in a little deeper - in this combination the ball surface difference is about 0.004 inches and is VERY repeatable.  I lightly coat the small ball with release agent and then loctite the magnet in place.

14 - The socket/magnet sub-assembly is then epoxied into a suitable carbon fibre tube per standard fixturing processes.  The .004 gap yields quite powerful socketing forces when combined with an N52 Neo magnet  ( .250 dia x nnn ).  Dump some W2S lube into the socket , wipe off the excess and you have a very accurate, powerful, low friction, no-slop floppy joint.   It's the bees knees.

[Ryan Carlyle]

I like the peening idea, that's slick. (Pun intended.)

At work, we have some crazy ball valves that have to see about a half million pounds of force on a spherical ring bearing surface. We have a few different materials for that. My favorite is graphite-filled PEEK. We also use brass and glass-filled PTFE. I know they're all precision-machined out of solid blanks, but I'm not sure what tools are used for that. The mating ball surfaces are either precipitation-hardened 718 inconel or tungsten-carbide hard-coated inconel. I'm often surprised by how much force those surfaces can take without any appreciable lubricant and not wear away any faster than they do. You get enough contact area and a good enough fit, and the longevity can be extremely good.