magnetic joints for delta machines

I am now testing my self designed and recently built delta printer. I am using chinese magnetic joints, 13 mm balls and same diameter sockets, which are stated as having 50 N holding force.

from the beginning I had a vague suspicion that the high friction in the joints can cause:

- wear and tear
- somehow, because of the "stick-slip" effect, it will affect accuracy

I searched for information about wear and could not find a single word on it. anyone?

now for accuracy. I set up a test for positioning repeatabily using a dial indicator securely mounted on the frame. I had macros move the effector in three dimensions and finally return the effector to the same position, with the last move always in line with the dial indicator.
I was alarmed to see variance in positions in the range of anything from 5 hundredths of mm up to 0.4 -0.5 mm. I checked the assembly, the diagonals, the balls mounting, the rigidness of the effector as held by the diagonals, and everything was fine.
at that point I was convinced, still without proof, that the stick-slip action in the magnetic joints interfere with the last micro movements.
I then performed another test: repeatability in moving only in Z - up and down - direction. why? in such moves there is no change at all in the joints - no movement of the ball relative to the socket. my expectation was that the repeatability will be good. and indeed the variance was very small - within 1 hundredth. this gave me an initial indication that the joints are at fault.

then I took the critical step: I added teflon pieces (cut from 0.25 mm sheets), between the balls and the sockets, to act as friction reducers. each piece was pressed in place by forcing the ball into the socket (in a vice, with a piece of wood) so the teflon piece achieved the exact shape of the cavity. the feeling of the joints was now very smooth. there was some reduction in the holding force but not significant. (I am now waiting for 0.1 mm thick teflon sheets for further testing).

the results: perfect repeatability! under the same tests, the position variance were within one hundredth of a mm.

[1drv.ms]

[1drv.ms]

note:JCB posted a year ago few words about doing something similar. I was not aware of it until today, but credit for the idea goes to him.

Edited 1 time(s). Last edit at 11/11/2019 10:48AM by ooly.

I don't know how you're keeping the teflon sheets in place; but if it helps, CS Hyde mouse tape is 0.1mm or a little less and adhesive backed. It's 1/2" wide which will probably work for you. I've been using that for about 5 years. BTW I don't think I came up with the idea independently but I don't remember.
Even 0.1mm will reduce the magnetic force, perhaps more than you would expect; but I'm sure it won't be a problem.

Sorry, I can't see your video and the still image isn't quite clear... are these sockets spherical, or the typical countersunk profile? If spherical, I'd really like to give this a try... one of the weaknesses of countersunk is that there is very little surface area in contact, so the attraction force isn't great. Can you please give a link to (I assume) AliExpress for the sockets?

Incidentally, N50 is a grade of neodymium magnet, and doesn't mean 50 N of force. Attraction force is a complex 1/(d*d) relationship very dependent on the shapes of the magnet and its position and orientation relative to the object.

JCS: I apologize ! I wrote your name as JCB instead of JCS,but you easily realized it is you...

frankvdh and jcs: I realize the still image is not clear and my explanations need more details. it's night here but tomorrow I will add videos and still images plus more detailed explanations. the preparation of the teflon will hopefully be made clear. by the way it is very easy to make and install.

jcs: I did test some adhesive backed teflon tapes but at the end I preferred the sheets. among the small problems with the tapes: since they stick in the socket once you push them in, it is more difficult to ensure they are absolutely spherically flat and have no kinks.

frankvdh: what video file type you can use? you can always convert my video to that. the sockets are of course spherical, otherwise the whole joint idea will not work. and whatever the contact area, and whichever formula you wish to use, the holding force is well within the stated number. I did run holding force tests and 50 N is, in my experience, safe and sound. even if the holding force is reduced to 30-40 N it is probably sufficient. here is the supplier:

[www.aliexpress.com]

I use KD413, 50 N, but you can see in the chart they have 70 N and even more. like I said, the force numbers are reliable. notice the higher forces use larger balls and sockets.

Edited 1 time(s). Last edit at 11/11/2019 03:33PM by ooly.

a question to jcs: in the five years you are using your delta printer, have you noticed any change/degradation in performance? what speed do you use in printing?
I have tested today positional repeatability with a 0.08 mm adhesive teflon tape and the results are close to the 0.25 sheet teflon. the holding force is of course higher, which is good. I don't know if it is the same tape as yours.

continuing tests with the 0.1 mm adhesive teflon tape, I noticed there is a small creep of 2-3 hundredths after the movement was seemingly stopped.

here are some photos showing the 0.25 mm teflon "hat" preparation:

a hat after shape forming: (squeezing a ball into the socket) [1drv.ms]

magnet barrel assembly, with hat in place: [1drv.ms]

there is very low friction in these magnet joints, but I am waiting for the 0.1 mm teflon sheet to conduct more testing. my feeling is the friction in the joints using the 0.1 mm adhesive tape is higher. this, and the possibility that the adhesive "flows", might explain the creep.

a video of positional repeatability test without teflon: please mute the video! there is background noise and music [1drv.ms] note the positions in the 5 repetitions changed as follows: 0, -3, -2, -4, +22, +21 (hundredths of mm)

positional repeatability with 0.25 sheet teflon "hats" : [1drv.ms] in the 3 repetitions, the positions were: 0,0,0

[reprap.org]

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

I've changed a number of things on my printer during the last 5 years, so printing quality has only improved slightly. During that entire time I haven't replaced the teflon tape.
My printing speed is mostly 60mm/s but if time is an issue I print at 90mm/s. Travel speed has been set to 200 - 300 mm/s as the mood strikes me when setting up a slicer.
I use "CS Hyde PTFE Mouse Tape" (see amazon) trimmed exactly the socket. That's easy for me because the edges are fairly sharp since I machine them with a 3/8" ball end mill. I use 3/8" diameter ball magnets and an arm using a short cylinder magnet with a machined poll piece.

jcs - have you ever measured the holding force of your joints?

No, I've never measured the force. My gut feel, based on my professional experience with magnetic devices, is that it is probably a bit less than the 13mm magnets without (I presume) the additional magnets in the arm.
A while back I rebuilt my original rostock and I used the same design which should tell you what I think of it. I've never had an arm come loose except in the case of a major head crash. Under those circumstances I consider an arm or two coming loose a good thing. I also have had the head snag on a warped print (or whatever) that resulted in a belt slipping or skipped steps without the arms coming loose.
The weakest point on my design is the attachment of the spherical magnets to their support. That failed a few times until I got it right.

jcs, thanks. by the way, your presumption is wrong - in the chinese joints I use the balls are not magnets. the brass barrels are ( a magnet is set inside) . the balls come with m3 short screws and are very convenient for assembly. I have just noticed that the magnetic force is strong enough to slowly unscrew the balls from their seats because of the back and forth joints travel. so I now use strong loctite to secure the balls in place.

I've been using some magnets now for my delta, but I noticed their holding force isn't as good as I expected it would be. Now I had to tune down the speeds dramatically so that the arms don't come off. The magnets were advertised as N52 but of course you never really know.

Now what I'm thinking of is to double up on magnets. I know that the magnetic field becomes almost twice as strong. On the other hand, the weight of the effector and sliders will become heavier. Also, the pulling force might become 'too high' again, causing more friction of the magnets against the steel balls.

To counter this, I will add 0.1mm thick PTFE tape bought at the Action store.

Is it common practive to double up on magnets for these connections, or does it add so much weight that the benefits in holding power are negated by the effect on the motion (adding corner overshoot and other artifacts)?

---

http://www.marinusdebeer.nl/

Ohmarinus, what balls and magnets are you using? by the way, when you claim you never know (the holding force) - why don't you run a simple test with weights?