6x6x6 3D print: Delta CNC

As a runner up winner in the recent 'Real Challenge', I have been awarded with a custom 6x6x6 inch 3D print for my 'CNC machine for machining large pieces of polystyrene foam'. I realised what an opportunity this presented and thought of many things to do with it. Make a 3D print for something at work and impress the boss? Design a custom iPhone case, cram 30 cases into the print and flog them on Ebay? Print someone else's design?

I don't think so..

I desired a bit more of a challenge. How about using the 6x6x6 print to create all the custom bits for a simple, elegant and fully functional CNC machine? I am not talking about a conventional 3-axis gantry-type CNC mill, but something a bit more exciting. I set myself the challenge to design and build a machine based on the Delta Robot, capable of milling/drilling in soft materials, and have all the main custom components for the motion platform fit in the 6x6x6 inch envelope of the awarded 3D print.

This inctructable has initially been published after the completion of the design phase. The first few steps will discuss the objectives, background, software & electronics, mechanical design and conclusions. In addition there are several steps to be completed afterwards, showing the actual build of the machine. A change log is attached to this first step. Current version: V0.11

Enjoy,
Richard Tegelbeckers

Step 1: Objectives

Objectives:

• An important objective is to create a design for a Delta-robot based CNC mill, where the majority of custom components for the motion base will fit in a 6x6x6 inch envelope.
• The components in the 6x6x6 envelope must be printable on one of the Objet 3D printers at Instructables.
• Elegance is high on the agenda. Not only for the way the machine looks, but also in the way the motion base moves.
• As I already have a large CNC machine, I will initially try to 'borrow' as many parts as I can in order to keep the cost down.
• The motors must be powerful enough for the machine not to lose any steps, without having to resort to gearing.
• Speed is not the biggest priority.
• The accuracy of the mill should be sufficient for drilling small circuit boards (think free version of Eagle PCB) and milling small items.
• Last, but not least, it should be fun to design, build, and use the contraption!

Comments

[Pizzapie500]

This reminds me of this: http://shop.seemecnc.com/Rostock-MAX-3D-Printer-Kit-68398.htm

[RTegelbeckers]

The Rostock uses linear slides. A really nice example of another delta with linear slides, which is also a 3D printer, is the ifab by Festo: http://www.youtube.com/watch?v=txF78s-QczU

The original delta robot concept uses pivot points instead of slides, which is something I really like. However, as Johan Rocholl was so kind to develop his Rostock as Open Source, I might initially do a linear delta and use his firmware.

[Broom]

I've always loved these 3-arm controllers, since I saw the first one in a Caterpillar factory. I spent many years in the CMM industry, working on reducing errors, and these have a lot of advantages there - scales have inherent issues, and sliding parts create uneven sag. All of this can be accommodated with software corrections and repeated calibration, but a machine that avoids these problems altogether is pretty sexy!

Nice job so far. Looking forward to the update!

[RTegelbeckers]

Thanks dude!

It was from seeing a delta CMM a few months ago, that I first got thinking about this sort of thing. Although I do not like using slides, they do not have to result in sag if they are supported. See my my first instructable for an example: http://www.instructables.com/id/CNC-machine-for-shaping-large-pieces-of-polystyren/

[Broom]

No, they still sag. You may not notice the sag, but it's there.

[RTegelbeckers]

You are correct. For the main carriage on my big machine I was just going to have a couple of 50x50mm box sections spanning the entire width. When I made some calculations to predict the deflection, I established the need for more bending stiffness. The added 25x25mm box reinforcements restrict the max. deflection to acceptable levels.

[___]

This is really awesome and the mechanics are a lot simpler than it first appears.

[RTegelbeckers]

Thanks! From a mechanical point of view, this is the simplest 3-axis motion platform I could come up with. However, it is a bit more complicated in terms of software and it is not the best in terms of accuracy...

[chuckels]

Richard,
sweet looking design. very interested in following it thru.
one thing I'm having a hard time getting my head around is what keeps the cutter mounting that is suspended by the linkage rods with swivel ends from trying to rotate?
Chuck

[RTegelbeckers]

Hi Chuck,

It simply is not easy to work out! Have a look at the video file in step 4, I suggest playing it back with 'repeat' on. There are also plenty of vid's on youtube for delta robots. Still, I guess the only way to really understand it, would be to play around with a physical machine in front of you... I hope I do not have to wait to long, to be in that situation myself!

[RTegelbeckers]

Forget my last comment! See the attached pic - it shows the motion platform as seen from below. Imagine trying to rotate the moving tool holder. In order to rotate around the drill axis, some of the links would need to become longer and some would need to become shorter. As the links do not change in lenght, the tool holder will be kept from rotating.

[eecharlie]

This looks ambitious and awesome. I've built one delta robot, years ago, so I can't help but chime in. Apologies if you've done your homework and this is not new information:

You're going to want super high stiffness linkages and no slop in your ball joints. What I did was get ball joints used in high-end R/C toy cars, and I used tubular carbon fiber stock for the linkages. That still may not have been good enough, but my servos were standard hobby so that was what killed me.

For your servos, I really would think about an indirect drive mechanism like a lead screw so that you reduce the number of sensor counts per degree of rotation in your joint axis - rather than be put off by the prospect of gear drive slop, engineer it out of your system. In high-precision robot systems like the Barrett WAM or surgical robots like http://brl.ee.washington.edu/laboratory/node/12 , counter-tensioned cable drive systems are used to eliminate slop. Remember that the Delta robot has non-uniform position accuracy across the workspace, so you have to design against the worst-case performance at the edges. You might also consider a base platform DOF in rotation so that you can sweep your workpiece through the best part of the Delta's workspace.

There are a couple master's theses and other academic papers on the kinematics/dynamics math for driving these guys. One big shortcut for you might be to grab the code for libnifalcon, the open-source driver for the Novint Falcon which is a modified Delta manipulator. In fact, you may want to look into whether the Falcon's modified geometry provides better kinematics, or maybe they did it that way so it was easier to spatially arrange the motors and encoders. Either way, design elements to steal - maybe even buy one to pilot test your control software & get a better feel for accuracy across the workspace?

Good luck!

[RTegelbeckers]

I have done a bit of homework, but thanks for your comments and info nevertheless!!! I am especially glad being introduced to the Novint Falcon, which I never heard of before.

I am hoping to counter the slop in ball joints by tensioning them with rubber bands or tension springs, see step 10 for a photo of the principle.

There will be a low level of accuracy (guessing up to +/- 0.3mm in the main range) due to the step size of the steppers, despite microstepping to 1/16th. For an initial CNC 'toy' with a small workspace, this will be ok for me to play around with. As soon as I would progress to achieve better accuracy, I already have some ideas I could try to eliminate slop.

The software side of things is the only thing I have not yet looked into too far, but I do know there is plenty info around. For now I will just try to get the hardware together and after that I will worry about the software...

[eecharlie]

Cool. I think the way you want to tension your ball joints is by putting your spring/rubber band pulling along the axial direction of those linkages so there's no slop in pull. The spring tension will need to be greater than the greatest force you expect on the cutting tool, since there's a cosine factor in how much of the linkage force translates to force in the x-y plane.

[RTegelbeckers]

The primary function of the rubber bands or tension springs will be to keep the rod ends in place - the links would fall off without them! As a secondary effect, keeping sufficient tension on will remove slop in the spherical part of the rod ends. With the total mass of the moving platform (incl. drill) being well below 1 kg, there should not be too much tension needed to make it work Also note I am using silver steel shafts, which will form a pretty good fit in the rod end bores.

[1-big-dog]

Awesome! I look forward to seeing more of your build!

[RTegelbeckers]

Thanks!!!

[samern]

I have to say I am very, very intrigued! I have built a RepRap, and a Fab@Home and both presented some challenges. I am now going with a Printrbot (just for giggles) and an upsized Prusa. In every one of those machines, getting the slides square was super critical. In your design, it's all about pivot points. Genius! From what I see, your head is a dremel tool like the rest. While this is purely CNC, it seems not too far fetched to eventually add the ability to make it a 3D scanner (for small parts) in that you would go around the item being scanned and park a small camera on an arm . I am looking forward to see this operational, and if you end up marketing it, I will be first in like to buy a set.

Samer

[RTegelbeckers]

Since getting interested in CNC & 3D printing, I have been looking at a way to have a very simple 3-axis motion platform without slides. It aleady existed in the form of a delta robot. Being able to have bits printed in plastic, I am now in the position to design and build the Delta based CNC. I purchased the mini drill tool from RS Components, long time time ago, their part no. being 182-584. I am planning to publish the design files after completing my own machine, so you might be able to print the bits for one yourself on one of your printers...

Thanks,
Richard

[samern]

Then sign me up! I can't wait to see how you resolve the electronics/motion control issue and the software related stuff with it. I have the same hardware on my large CNC machine (which I am using to cut parts for my Experimental airplane), but I would love to see what you turn up.

Samer

[RTegelbeckers]

Planning to use my large CNC (see my 1st instructable...) for constructing an electric car. Doing this machine for fun, as a good use for the free 3D print...

Good luck making!

[alcurb]

Brilliant! I can hardly wait to see the finished product.

Have you estimated the cost of the build minus the software?

[RTegelbeckers]

I already have a lot of the bits, such as stepper motors and drives. So far I have spent GBP52 on rod ends, studding, silver steel and Al tubing. Yesterday I did also get a Polycarbonate corrugated roof sheet panel for GBP24, to be used for the transparent doors. If to be build from scratch, incl. everything apart from a PC/laptop AND the 3D printed bits, it could cost upwards of GBP300.

[amandaghassaei]

this is great! glad to see you are making the most out of your prize! what is your timeline for this project like? when do you think the first working prototype will be ready?

[RTegelbeckers]

I do not really have a timeline. In a way I was given a design timeline by noticing the 3D Design Contest: I knew what I wanted to do and used the contest as a deadline for getting the design ready. Right know I wil just wait for the moment I can pass on the STL file to the Instructables dude (m/f), responsible for doing the magic on the 3D printer. From the moment the file will be taking shape into ABS, it will take a good few days for the green bits to make their journey to the UK. At that stage I will already have prepared all the other parts, ready for assembly. However, the software side of things is a completely different thing altogether...

[Rob311]

I like the idea of a delta bot vs a gantry, nice a simple. Given that you don't need the added strength if its a its a 3d printer (rather than the massive rigidity needed in a gantry system to make a cnc router cut accurately). The software is going to be a bit more complicated, but it should just be a matter of figuring out the formula and then its just plug and chug. makes me wonder why these aren't more popular?

[Rob311]

also- I would try to modify the design to run off of standard servos. Should make controlling the damn thing much easier and cheaper for us people that will undoubtedly try to make one...

[RTegelbeckers]

For now I will stick to stepper motors & stepper drivers as I can 'borrow' them from my existing CNC machine... Thanks for your comments!

[jlp6k]

Nice design but if speed isn't a priority, you may take a look at http://parallelrobots.blogspot.fr/2012/04/linuxcnc-hexapod-machine-tool.html
It is a 6 DOF delta CNC where the levers have been replaced by ballscrew. It should lead to cheaper (weaker) motors and improved accuracy.

[RTegelbeckers]

Thanks for the link! The 6DOF looks very impressive, but is in a completely different league. Whereas my Delta will have cost me GBP52 on top of parts I already have or will borrow from my larger machine, I can imagine a complete single lead screw axis of the 6DOF would have cost more then all I will have spent on both of my machines put together. Of course, the accuracy of the Delta will be fairly poor. I guess accurate in the order of +/- 0.3mm in certain ranges. I just want to get something together to play with for now and hope to improve on this machine afterwards. First of all I would like to improve the accuracy by introducing some sort of motor reduction. Once I get accuracy I would look into including an extruder for 3D printing. I will see if I can borrow a 3D printer somehow, as I am anticipating having to experiment.

[ketirock]

Very clever, well done.

[RTegelbeckers]

Thanks, very kind of you! :-)