Magnetic ball joint seat geometry... May the force be with you!

First I'd like to give due credits to Kees Koese who started with the idea of the magnetic ball joint and made the original design of the effector plate. Give his website a visit to have a look at his multispindle drilling machines. Technicians will like it :)

This post is a how and why about the geometry of the ball seat in the effector plate. Since the original effector plate was made with SLS printing we wanted to be able to print this with FDM printing (both being entirely different processes).

During the first test prints there was still manual rework needed. I wanted to end up having to do nothing at all regarding removing filaments, grinding etc. I think making geometry only with the printer without having to rework it is why 3D printing can be such a fantastic way to manufacture.

Before going into details I was amazed how rewarding and exciting it is to draw your thought/improvement in your favourite CAD program (mine is SolidWorks) and having it on your desk in 30 minutes... Every engineers dream.

The picture below shows the original geometry Kees drew a few (almost 6) months ago. It depicts the ball seat with the area (cylinder) where the magnet needs to fit. Because of the original being SLS printed there were virtually no limits regarding overhang or printing resolution. Notice the second (bigger and lower) sphere that ensures the ball will fit in the topmost seat.

magnetic ball bearing seat 1

When initially printing this test piece to check if the resulting dimensions were good enough to fit the cylindrical magnet (this situation needs a ∅10.5 mm diameter to result in a snug fit of the ∅10 mm magnet) I had a lot of problems printing because of 2 non-printing areas coming together in mid-air. Also the points where the 2 areas merged were sharp so the filament  kept sticking to the nozzle on it's way back. (a lot of words for saying the result was horrible...)

What to do... Make sure there were no pointy encounters so that meant closing the gap between the seat and the cylinder volume. See below:

magnetic ball bearing seat 2This was not the way to go. The wall thickness was so thin (wall thin-ness) that it could not be closed (and the ugly thin-ness needed to be drilled away manually... bah...). Because the overhang is at a 15 degree angle there is not enough previous layer to adhere to. Improving (but not really) solving) this could be done by adding wall thickness. But alas... Every solution frequently introduces it's own problem... The distance between magnet and ball was getting so big the magnet lost it's force-field on the ball... Thanks for nothing Master Yoda!

There had to be a way to ensure a minimal gap between magnet and ball without the overhang problem and without wall-thinness situation... If you have something you don't want: (re)move it... See below:

magnetic ball bearing seat 3

All that's needed is a defined sphere and a axial constraint preventing the magnet being pulled onto the face of the ball (creating friction when the ball is turning). If you add a partial ceiling coming from the side (instead of bottom or top) of the cylindrical hole then during printing you will see a very gentle alteration of the contour of the non-printing area. Then there is enough support from the previous layer to create the overhang.

Final measurement and test were done with calliper (measuring the distance over Ball and Magnet, subtracting the ball diameter and length of the magnet) and multimeter.

I measured a gap of 0.1 - 0.15 mm and when doing the beeeep testing for I got no beeeep. (measuring short circuit between ball and magnet).

This evening Master Simon helped me taking a picture of the actual force of the ball joint connection. Result: One connection can pull 1690 gram. Take that Master Yoda?

Below some pictures of the test, some prints and last but certainly not least the geometry of the connection.

Pulling water in Sigg bottles. I am holding the blue printed part.

pulling force of magnetic joint

Resulting weight being pulled:

Weight being pulled by magnetic joint

Seat of ball with magnet underneath (view from top):

Seat of magnetic ball joint (top view)Fit of magnet in area under steel ball (view from bottom):

Fit of magnet in area beneath steel ballMisprints being mentioned in this post, next to the result:

MisprintsFinally after having bored you to death the resulting geometry. First the cut-revolve of seat and magnet volume:

magnetic ball bearing geometry section view

Second the cut-extrude at ∅10.5 with a width of 8 mm. Seen from above. This prevents the magnet from getting in contact with the ball. I also added a radius of R1 at the inside of the cylinder making the sliced contour even more printer friendly.

magnetic ball bearing geometry axial stopENJOY !






Benchmark results of SolidWorks in Bootcamp and Parallels

My previous post was about how I installed SolidWorks on my MacBook Pro with Retina display. My main problem (before I bought my MacBook) was that there are no numbers for comparison between BootCamp and Parallels. Mostly it's "It runs very fast" and "The virtual machine is a little bit slower"...

I love numbers to compare so here is my setup and here are the results.

I used the following benchmark files from SolidMuse:

Situation 1: my new MacBook Pro with Retina display

  • Windows 7 Pro, 64 bit, SP1, running from BootCamp
  • 2.6 Ghz Intel Core i7
  • 8 GB Ram
  • Intel Graphics 4000 (shared) + NVIDIA GeForce GT 650M, 1GB

Situation 2: Same as Situation 1 but

  • Windows 7 Pro, 64 bit, SP1 running from Parallels
  • Parallels running 4 CPU cores
  • Parallels running with 4096 MB memory
  • Parallels video memory 1 GB

Situation 3: my Workstation XW6600

  • Windows XP Pro, 64 bit, SP2
  • 2x Intel Xeon Quadcore 2.5 GHz, E5420
  • 16 GB Ram
  • NVIDIA Quadro FX 4600, 768 MB


Punchholder benchmark
SolidWorks 2012 SP5.0, CRTL+Q for rebuild of model, results from feature statistics info.

Setup Rebuild # Time
1: MacBook with Bootcamp 1 57.89 s
2 60.14 s
2: MacBook with Parallels 1 69.85 s
2 74.91 s
3: HP XW6600 Workstation 1 114.61
2 119.66

Scoobie Doo Surface model benchmark
SolidWorks 2012 SP5.0, CRTL+Q for rebuild of model, results from feature statistics info.

Setup Rebuild # Time
1: MacBook with Bootcamp 1 12.70 s
2 11.86 s
2: MacBook with Parallels 1 13.91 s
2 12.78 s
3: HP XW6600 Workstation 1 22.41 s
2 22.40 s

Photoview 360 Lighter assembly benchmark
SolidWorks 2012 SP5.0, 1024x786 px, fileformet Tif, results from photo view 360 window.

Setup Good Better Best Maximum
1: MacBook with Bootcamp 19.8 s 50.8 s 1 m 55.0 s 4 m 42.9 s
2: MacBook with Parallels 11.9 s 1 m 2.0 s 2 m 20.3 s 6 m 6.3 s
3: HP XW6600 Workstation 13.0 s 1 m 3.7 s 2 m 26.3 s 6 m 0.4 s

Generally I noticed that when you run SolidWorks from your BootCamp disk inside Parallels that you do not have a very responsive display while modelling. For example it takes visual effort to draw the lines on the edges of the punch holder.

For doing real work for customers I would choose for the Bootcamp setup. Visual most attractive and fast. Only problem lies with SolidWorks itself. Though the big icons can be used in the command manager. All other icons (from menu and normal toolbars) are very small due to the incredible screen resolution. Another reason to use those keyboard shortcuts :) .


SolidWorks on a Mac

Last week my old mobile workstation laptop decided not to support me any more. I had some problems brewing, but it was still workable... However, after losing some work again I decided to invest in a new laptop.

I had my eyes on the MacBook Pro 15" with retina screen, but since I need to be able to run SolidWorks for my work (I am a self employed 3D engineer/consultant, see Luminize). I needed to be able to use SolidWorks from my future laptop-to-be. Not as before where my laptop was my main computing power because I have another workstation (big one, not mobile) to help me with the necessary CAD/FEM.

First I searched the Internet to see if there is experience with this, but it is hard to find some info on performance of SolidWorks on a Mac and the difference between Bootcamp and/or Parallels. After some reading I decided to take the plunge...

The Retina screen is absolutely fantastic. The display really is sharp as a knife, it almost makes your eyes bleed! OSX works so good and natural that while removing some files from my old laptop (I only used the MacBook for a few hours) I noticed the difference immediately. Also the flash disk is so fast that the startup from zero takes about 10-15 seconds...

But enough about this. I wanted to tell how I got SolidWorks running.

I downloaded parallels 8 and with that it installed the Windows 8 release preview automatically. I installed SolidWorks 2012 (no problems) and noticed that the display was slow in reacting. (if you hover over the front plane normally the edges show immediately, this was the case, but it was noticeably slow. I knew I would get irritated by that.) The reason for the lag is because Parallels acts as a video card, it works as an in-between, at the cost of the graphical performance. SolidWorks was setup to use "Use software OpenGL" and I wanted it to be hardware accelerating as fast as it could.

I made a backup of the Mac (DO IT, DO IT, DO IT, normally I don't, but really DO IT) before installing Bootcamp. I unknowingly made the mistake of not removing the external USB disk used for the backup when installing windows via Bootcamp. The result was that the Windows 7 installation(not Windows 8 because I need to be able to run older SolidWorks versions for some customers) failed because of the formatting of the Windows partition. Also I could not reboot to OSX... hot... cold... adrenaline... Luckily i had made the backup! Within 20 minutes I was back.

So if you have stayed reading until here, if you try this yourself, make the backup before attempting to install Windows 7 with Bootcamp and as a bonus also remove all USB devices except the USB thumb which has the Windows installation files.

Attempt 2 was successful, Windows 7 was up and running with full hardware capabilities. Installing SolidWorks went smooth. Pas de problèmes!

Booting in Windows to use SolidWorks to the max in my situation only takes 10-15 seconds before I can start modelling. Because of Bootcamp you must reboot if wou were working in OSX. I think if you look at the boot-up time versus the use of the hardware it is a good tradeoff to work from Bootcamp in these situations.

Don't be afraid to use your Mac without OSX. When you are modelling you don't need the programs that OSX offers? Don't let the excuse of email rule your (digital) life. You also have your phone or iPad for checking mail. If it is really important people can still use the phone, and checking your mail can still be done also from Windows. Or.... Do a re-boot when you go for a coffee. Good reason to press CRTL+S and save your work :) It's soooo fast you won't be annoyed by it.

The only drawback comes from SolidWorks itself. The icons in the are very, very, very small (not command manager icons, but the toolbars and menu's). I like small icons, but this is a little bit too much (or actually, too little).

Next post I will give you a performance comparison between my MacBook Pro with Windows 7 and SolidWorks 2012 + 2013 and my workstation with XP 64 Pro and SolidWorks 2012. I will use the benchmark files from SolidMuse: Punch holder, Scooby Doo Surface model and the Photoview 360 benchmark files.