Been toying with a new(?) idea for ball joints for a delta-bot. No magnets, no PTFE, yet still shockingly awesome performance. The balls pop into the socket with a firm click, and then you can tighten the socket up with the little clamping screw. This eliminates lash, and lets you adjust the amount of grip. You can be tight enough to eliminate lash, while still being plenty friction-free. PLA on polished steel balls has an incredibly low friction coefficient. Thought @Shachar_Weis might like this one.
Cool. How do you connect the ball to the rod ? Weld a bolt ?
Epoxy works well. We do that at work with stainless calibration spheres
Using hardened 10mm steel balls (bearing grade) and 6mm OD aluminium tubing at present. Carbon fibre is a future possibility, but Al is cheaper and easier to work in the proto stage.
Tried CA glue, failed easily, tried cheap epoxy, failed also. Have on hand some good quality epoxy designed for joining metals that I will try next. Bonding the ball to the rod is the only unsolved problem thus far.
Because the balls are hardened steel they are essentially impossible to drill/tap, which was my go-to idea at first.
Yeah, attaching the balls is the hardest part. I tried roughing up the ball with a dremel and welding with solder, but that’s useless.
@Sanjay_Mortimer you may not be able to drill or tap the bearings but you could at least grind a flat surface to give a better adhesive fix against the end of the rod. You might also get away with putting a divot in the ground surface with a TC rotary cutting burr to increase surface area on the joint.
We scuff the glues site of the ball with abrasive. Bearing grade is an almost perfect surfaced sphere, glue will never stick, scuffing increases the surface area considerably. I believe we may use 3M DP-360, not positive. But our balls are .75 -1.00" diameter and once the epoxy cures, they’re practically indestructable.
maybe trying to source alu sphere instead of ss?
OR join the bearing to a short length of 4-5mm threaded rod, that way you can attach alu or carbon fibre or whatever to it with a simple adaptor.
Why not use RC car ball joints there’s many different sizes and you can use the thread to get accurate arm lengths.
@Nigel_Dickinson That’s where I’d seen them before! I spent 20 minutes Googling looking for that very thing. Good thinking.
Here we go. 9mm best I could find on my phone.
http://topshelf-rc.co.nz/catalog/product_info.php?cPath=23_83&products_id=2027
Lift support http://depot.com appears to have a big choice. Hope that helps. I’m sure there are other supplier’s but that’s the first I found.
These are the exact ones the RC truck i used to have uses: http://www.modellbau-seidel.de/index.php?firma=Tamiya&best=9404692&name=Aufhaengungsteile-Beutel-TGM-04-TNX-5.2R#bild
They are 10.5mm and, like others, can be depth adjusted with a hex key.
The T-Rex 450 also seems to use a vast amout of ballhead (sorry, another German link): http://www.ebay.de/itm/Kugelkopf-Set-Kugelkopfe-fur-Taumelscheibe-T-Rex-450-CopterX-HK-450-/150793625746?pt=RC_Modellbau&hash=item231c002092
This is like full circle to what was used on the 3DR design. Great minds think alike 
do you think you could place direct extruder on this head (without bowden). Although I am not owner of delta printer I see that bowden problem creates “hairy” prints.
@Sanjay_Mortimer - I hope this is for a triple head bowden E3D extruder - as you’ve mentioned in the past?
@Anthony_White The end-effector shown is just for a standard v5 - but the machine around it in general is purpose built as the testbed/prooving platform for the upcoming Kraken. Which actually has four bowden fed nozzles now.
It’s suffered some annoying setbacks due to frustratingly trivial reasons, but it is NOW in manufacture for alpha prototypes.
@Sanjay_Mortimer anyone in particular you had in mind to trial those? Pretty important to make sure they work correctly in the southern hemisphere I’d say 