Different Magnetic Joint Style... Is it any good?

This style of magnetic joint on the efexor is very interesting:

spiderbot magnetic joints

I think I might try this. Anyone have experience???? I assume I need the most powerful cylinder magnets I can find.... any type suggestions???

Looks interesting. I'm going to follow this.

This thread has info on another 3 ball effector: [forums.reprap.org]. There are probably quite a few others.

What holding force is necessary to keep everything together at high speed???? If I know the holding force necessary, I can select magnets and shaft diameter.

Also, since the Balls are 10mm spheres, what is the max cylinder diameter for 2 magnets on that ball without interfering with each other???

Quote
patrickrio
What holding force is necessary to keep everything together at high speed???? If I know the holding force necessary, I can select magnets and shaft diameter.

Also, since the Balls are 10mm spheres, what is the max cylinder diameter for 2 magnets on that ball without interfering with each other???

I am wondering if you can drill and tap Neo type magnets as I have a local company that make the things and do magnetic spheres as well as rod or disc magnets

there 9mm Diam spheres have 4.9kG Pull and there 5mm dia x 10 have 0.95 kG pull.

Would that be sufficient I wonder?

FYI

Doug

I have found some 3/16 inch dia 1/2 long cylinder magnets that pull nearly 3 lbs/1.4kg each just connecting to steel (non magnetic) surfaces. 1/4 inch dia magnets can hold up to 7 lbs.

I just don't have an idea how much pull I need..... or how big I can go in diameter.

Quote
patrickrio
What holding force is necessary to keep everything together at high speed???? If I know the holding force necessary, I can select magnets and shaft diameter.

You can compute this. Let's say your steppers have holding torque 44 Ncm and you use 16 teeth T2.5 pulleys.
Pulley radius: 16*2.5 / π / 2 ≅ 6.4 mm
Maximum belt force: 44 / 0.64 = 68.75 N
So if your magnet joint can keep about 6.9 kg then you are definitely good to go. If it can keep less than half of that then the joints will rattle at high accelerations.

If you care about high speed then rather get high quality ball joints.

I had trouble when the nozzle bumped a curled edge, it would crash. I ended up using springs to increase the tension, and z hops.
I used high temp 3/8 x 3/8 cylindrical magnets rated between 8 and 9 lbs from here, as I have a heated chamber.

---

My printer: Raptosaur - Large Format Delta - [www.paulwanamaker.wordpress.com]
Can you answer questions about Calibration, Printing issues, Mechanics? Write it up and improve the Wiki!

Quote
patrickrio
I have found some 3/16 inch dia 1/2 long cylinder magnets that pull nearly 3 lbs/1.4kg each just connecting to steel (non magnetic) surfaces. 1/4 inch dia magnets can hold up to 7 lbs.

I just don't have an idea how much pull I need..... or how big I can go in diameter.

It will depend on how much your effector assembly weighs. Mine weighs 110 grams.

Quote
Paul Wanamaker
I had trouble when the nozzle bumped a curled edge, it would crash. I ended up using springs to increase the tension, and z hops.
I used high temp 3/8 x 3/8 cylindrical magnets rated between 8 and 9 lbs from here, as I have a heated chamber.

I'm using the N52 version in the arms as well as this one on the effector.

So far I haven't been able to shake them loose. I'm still looking for a model that can really stress the ball joints.

It would be nice to have cylindric magnets with a hollow bottom that fits exactly onto the 10mm sphere.
Airgap is your enemy...increased magnet/metal contact area is your friend.

It's not possible to thread or drill magnets, they are to brittle. ( High temp magnets like N52 even more so )
-Olaf

Quote
etfrench

Quote
patrickrio
I just don't have an idea how much pull I need..... or how big I can go in diameter.

It will depend on how much your effector assembly weighs. Mine weighs 110 grams.

And the maximum accelerations and jerk used.
Notice that: F = m * a
You can have it working even with very crappy magnets which barely hold your platform weight if you select very low accelerations and zero jerk.

Paul Wanamaker and etfrench: Do you have pictures of your joints at the effector and towers so I can see how you did it?

I am trying to find a very small nozzle too, which should reduce effector weight.

Nozzle link

I am not going to have a heated chamber so N52 magnets are fine.

My motors are small 0.9 degree/400 step Nema 14 that are 20mm thick with a torque of 4.903325 Ncm (500 gm cm). My pulleys are 15 tooth 2.032mm. So, by your calculation:

Pulley radius: 15*2.032 / π / 2 = 0.485104266544cm
Maximum belt force: 4.903325 / 0.485104266544 = 10.1077754581 N

So I need magnets with a strength of about 1 Kg. Since the ends of the magnets are flat they will have a slight air gap with the steel balls, so I need higher than 1Kg/2.2lbs magnetic pull for maximum acceleration with this motor. Since there are 2 magnets per motor, the forces would be half this...

The magnets I am considering are:

Magnet 1

Magnet 2

Magnet 3

Magnet 4

These have holds of 1.5, 2.2, 2.7, and 7 lbs. I think I will order sets of all of them just in case. Shipping is $14 regardless....

I can't counsel you on what configuration would be best, I just warn you that I and others have had some problems with magnetic joints.
Up till now I hadn't published my method because it had failed on occasion under worst case scenarios, but perhaps others can learn from it or improve it.

Since my printer is large I used 12mm x 10mm internal carbon fiber rods (near 400mm long), and embedded the magnets in the ends using JB-Weld (not JB quick - it's not as strong) to quickly create a perfect pocket and bearing surface. These connect to 0.5 inch steel bearing balls. I use a squirt of graphite in the pocket to lubricate.

I chose JB-Weld because I needed an adhesive that would bond metal and plastic, create a good bearing surface, and withstand elevated temperatures (my printer has a heated chamber). You may find another hard setting epoxy works better (I could not obtain one that was rated for the temperatures).

One "feature" of JB-Weld is that it has iron particles in it. This can be good or bad in this situation, as I'll touch on below.

The first step was to clean everything very well with solvent. You do not want any oil on the surfaces.
Then I put some rigid foam down into the carbon fiber tube to the depth the magnet would reach in order to contain the adhesive.



Then I put in a bit of JB weld in the tube, with a bit on the bottom of the magnet, and put the magnet in the tube. The magnet has a bit of tape so it will be a tiny bit away from the bearing (not rubbing), probably not needed.



Then I placed a heavily waxed (with shoe polish) bearing on top, and taped it down. The magnet sucks up to it. I did have some trouble getting some of the bearings off after it cured, not a perfect wax job I guess, but the pockets were perfect.



For the connecting the bearings to the towers and effectors I used a virgin set of bearings (had not been waxed), put tape on them where I wanted the adhesive to stop, and roughened the surface lightly with a sanding drum in my Dremmel. This was pretty easy. I cleaned them up with acetone, and cleaned the brackets with alcohol.



Then I used JB-Weld and clamped them in the brackets that I had 3D printed at my makerspace. This allowed me to have precise spacing without having to machine anything. Having the right clamps is a must. Do not use JB-Quick (a faster drying variety) as it is not as strong. I had some fail with JB-Quick when the chamber got hot, but not with JB-Weld.



And this is how they connect:



The strength of the joints between the rods and the bearings seemed good and I thought it would be fine, however it was not strong enough if the hot end hit a curled edge on a large angled piece when the effector was extended (worst case scenario), and it would come apart. I added some springs and stainless wire from the effector to the top brackets to increase the tension and haven't had any failures since. This is the common fix. It doesn't help that the rods I'm using are longer than fashionable so it can print out to an extended distance at a lower anger, but I need the distance...

If I was going to do this again, with just magnets, I would do a couple things:
- Double up the magnets
- If I had extra magnets and a pull test rig I would test using a circular dam to keep the JB-Weld away from the center of the magnet in the pocket next to the bearing, to force more flux to go through the bearing, since JB-Weld is ferrous. No way other way to know if it helps or hurts. Or find a different high temp epoxy.

Hope that helps...

The ball joints on mine look very similar to the Spiderbot joints as do most other magnet ball joints. I have an additional ring magnet under the ball. If you have a ball with a stud, then it may be harder to find a magnet to fit under the ball.

I got some M2 rod size brass ball joints as a back up just in case. They should work fine in my small delta too.

Would it increase holding force to create a "closed loop" magnetic path?

Say, you have two rods on one steel ball. Would it be better to use the attracting poles of the rod magnets or the repellent?

Would it help to connect the other ends of the magnets with a flexible steel tube to close the magnetic path? ( read: no air gap )
-Olaf

What is this flexible steel tube you speak of?

If you have two magnets going to one steel ball, or to two balls touching (which I have) then the poles of the two rods that attach to the same point must be opposite.
If you are stacking magnets to double them up, then stack them the way they want to go.

When you make your rods be sure that every other one has a different pole on the end that attaches to the effector.

To check the pole of a magnet, just set a compass nearby, then mark the magnet.

Quote
etfrench
What is this flexible steel tube you speak of?

I thought of something like this:



The relative movement between two rods close to the effector is pretty low. The woven steel won't have much influence on the printer, but could almost double up the magnet force.

@ patrickrio
I know you want to build a small, portable delta printer, but can't find the appropriate thread, so I comment here:

1. IIRC you want to use 24V stuff. I 'd use a 5-6S LiPo battery to be fully independent of a wall plug.

2. Did you consider using 28BYJ-48 steppers? They are tiny/lightweight and have a inbuilt gearing that makes them pretty powerful and they have a high stepcount /rev ( 4096 steps/rev )
I just learned, they can work with our beloved stepstick drivers with a little modification.
They would be small enough for a direct drive extruder, too.
Found a vid that explains the mod and says, it has 800g/cm torque afterwards.


Sorry for OT
-Olaf

Edited 3 time(s). Last edit at 11/28/2015 06:53AM by o_lampe.

Yes, I have been planning to use 24v to increase the effectiveness of cheaper/smaller steppers. I bought some of those steppers to test, but my primary plan is with a more standard Nema 14 0.9 degree stepper motor 20mm thick.

The thought of using Li-Po had occurred to me, but I figured I can add those any time.... 70w of 24dc is 70w of 24dc regardless of source....

I bought some of those geared motors and knew about the modification, but I shied away from them because of the unknowns... 1. is there play in the gearing? that would really screw up delta prints.. 2. is it fast enough? deltas need relatively fast motor movements when the arm angles are low.... 3. similar to the previous, is the torque lost at high speeds? these will have to turn pretty fast, so torque could drop....4. I didn't like that the motors only had 2 mounting screws.

I haven's seen anyone else try to use these, so figured there must be a reason.....

However, the price is rediculous at $1.50 each delivered.... just add an appropriate gear!!!!. I had to buy some.....

Edited 1 time(s). Last edit at 11/28/2015 03:43PM by patrickrio.

I started to play around with one of them and found no play in the gears. Even 0.1mm moves back and forth are clearly noticeable on the shaft.
But over time with the "help" from bipolar torque enhancement, there will be play.
I would try to use external bearings on belt-pulleys to keep the load from the stepper.

I tried it on a Ramps with bridged microstepping jumpers, so I could only test it in 1/16 mode ( stupid board )
Will change that today and see, how fast they can be. IIRC I had more than 30turns/min with the standard ULN2003 driver.
It will definitely help to raise the voltage because on 12V the max. current is limited to 220mA ( ~54 Ohm coil resistance )
They got barely warm at 12V.

I started to design a NEMA 17 to 28BYJ adapter, its in the print ATM.
I'll try to replace my extruder motor on the MK8 direct drive with it.
Or use two to replace the z-steppers on my Prusa I3 ( sure a winner )
-Olaf

edit: replaced the files. New ones have deeper recess for the mounting lugs and different hole spacing.

Edited 1 time(s). Last edit at 11/29/2015 07:41AM by o_lampe.

if there is any play in the gears that develops, using them in a Z axis where you have essentially automatic backlash control via gravity may be good... and slow movements are not a problem either.

They may not be good for extruders that need quick reverse control for retraction though.....

Edited 1 time(s). Last edit at 11/29/2015 08:57AM by patrickrio.

Prusas Z-axis seems to be the only option at 12V. I tried to run a direct drive extruder and the hobbed gear wobbled on the shaft, due to its uncommon shape.
Turning a ACME or M5 leadscrew would only load the bearing in axial direction.
-Olaf