Yazzo PolyBot - Cranberry Edition

by WilliamAAdams, published

Yazzo PolyBot - Cranberry Edition by WilliamAAdams Feb 1, 2011

Featured Thing!


After many little pieces, I am finally able to construct a delta robot that is mostly made of printed parts.

This thing is a delta robot frame. There is the basic frame, mounted motors, arms, tool holder, and the like. The arms closest to the motors are printed plastic notched spans. The longer arms are wood dowels, with tape on their ends to enhance the friction fit into their fittings.

Steel rods in this case are 5/16". The size doesn't really matter, you can select any size as long as the vertices match.

the size is 2' on edge. that's a pretty huge build area if you decided to use this robot as a 3D printer.

It's a good platform for experimenting with various delta robot ideas. The arms will be replaced with carbon fiber. The 'bearings' will be replaced with brash, with 1/4" axles, the motors will be replaced with servos, etc.

At any rate, at least a model can be built, with not too many different types of materials. There isn't a fastener on the whole thing. Of course if it were actually moving, you'd probably find places where you'd want some fasteners, but that's the whole point of the experiment anyway.

Some interesting aspects of the design:
This is an endoskeleton. It can take a 'skin', but it's not required for structural support.
The tetrahedron is a fairly stable self supporting structure.
No threaded fasteners, other than the threaded rods themselves
Use zip ties to hold the little motor board in place. That board could be gotten rid of entirely if the motors mounted to the rod instead

UPDATE: 05022011
Changed male and female clevis part counts. They were 24 ea, and they should have been 12.

UPDATE: 06022011
Created a page on RepRap as this is more of a development project than I think is appropriate for Thingiverse

UPDATE: 27022011
There have been many parts updates. I've included a new picture with the servos and new arms, as well as the base plate in place. I've added a video on YouTube of the thing in action: youtube.com/watch?v=VzWJDWvJvqo

Recent Comments

view all
I had no idea what a delta robot was before googling videos of it. My mind is seriously blown, I haven't seen anything move so fast. This project is amazing.
I really like the way you used three servos/motors in this design. I'm wondering if there is anything preventing this design from being flipped upside down. Raising the extruder rather than suspending it to eliminate the need for the metal bar frame. I'm trying to think of whether it would be able to cover the same area. Anyway, awesome design keep up the good work!
truncated tetrahedron... Hay, that looks pretty cool. I could model that fairly easily with the various types of connectors I have. Probably using ball/socket joints.

And don't forget to check out the latest edition: thingiverse.com/thing:9317

Liked By

view all

Give a Shout Out

If you print this Thing and display it in public proudly give attribution by printing and displaying this tag. Print Thing Tag


1) Decide how big you want your bot to be, and what diameter of rods you want to use.
2) Cut 6 threaded rods to the same size. If you pick a standard size that your supplier carries, you won't need to cut anything.
3) Print all the little bits and pieces of plastic you need.
4) Snap and slide them together
5) Put motors into motor clips
6) Put motor mounts on the triangle shaped board
== There are holes in the board so you can fasten to the rods
7) Use zip ties to fasten board to rods
8) Put arms on motors
9) Rejoice!

This is of course a very abbreviated instruction, as the actual instructions could take a while, and require some pictures. My hope is that the design is so obvious and symmetrical that you could figure it out from just looking at the picture.
I had no idea what a delta robot was before googling videos of it. My mind is seriously blown, I haven't seen anything move so fast. This project is amazing.
I really like the way you used three servos/motors in this design. I'm wondering if there is anything preventing this design from being flipped upside down. Raising the extruder rather than suspending it to eliminate the need for the metal bar frame. I'm trying to think of whether it would be able to cover the same area. Anyway, awesome design keep up the good work!
Ok - after seeing what you were doing here, I had to share an alternate polyhedron for the shape:



integrates the servo mount into the shape of the polyhedron while giving you greater Z displacement and room for a larger print bed. And the added bonus... your existing software
will most likely work. :)
truncated tetrahedron... Hay, that looks pretty cool. I could model that fairly easily with the various types of connectors I have. Probably using ball/socket joints.

And don't forget to check out the latest edition: thingiverse.com/thing:9317
OK - we are watching. Your silent audience is in suspense. When is the next iteration on the evolution for the Polybot?
Oh, you mean like this thing: thingiverse.com/thing:9317 =-O
I wonder how difficult it would be to make one of these:


It seems like it would make a fine input device for 3d modeling...
another possibility is that you put a tool on the effector, like a laser line pointer, and scan your model that way.
I think it would be fairly straight forward. For positioning the PolyBot, I'm just using inverse kinematics. To do it the other way, you'd just use forward kinematics. I think a key though would be having fairly decent encoders, whether they are part of the servo/motors, or separate. If you want to pay $50/encode, the MakerBot guys have those new magnetic encoders. Otherwise, you could pay $40-$50 and get something like the Dynamixel servos, or HiTec, and get some decent encoders embedded in the servo.

Totally doable. I'm sure those magnetic encoders will come down in price over the coming year.
Please don't tell me that this is going to be an other design for a RepRap. I mean who thought that an A-Frame would work for a 3D printer after a box. But now there's a pyramid out! 8-)
I think of it more as my experimental personal manufacturing robot tool base. you could probably put a pick and place vacuum attachment as an end effector as easily (if not more easily) as a plastic extruder. A dremel flexi shaft might even be possible with servos strong enough to handle the lateral forces.

My hope is that people will riff on the design and come out with cool things.
I've been making a lot of minor changes to the beast, and I finally got servos attached with some properly sized arms. The video can be found here: youtube.com/watch?v=VzWJDWvJvqo

I have not updated the parts list. I'll do that as a derivative as there are enough differences.
Are you going post files to print this? This is sweet.
The parts list below are all the files needed to construct this thing. I've been working on new parts that reduce the parts count, but haven't released the next version as yet.

If you use the parts listed below, it's mostly plastic parts which will allow you to experiment. Then you can replace different parts one by one as you want to improve your bot.
Hmm. Ok, sorry but I'm not very wise in the thingaverse. I see the parts list down below, but when I click through it doesn't really go anywhere. Just to a page that it says it's used in this (delta robot) and there's no suppliers. I don't see a file link or something, like I've seen on other thingaverse objects (usually some stl file or set of them)...I don't have a 3D printer personally, but was going to try to send them over to ponoko and get it printed. I've been screwing with a delta robot a fair bit, but getting the setup you have looks like an upgrade. Hence the interest :-)

BTW, what size motors are those in the clamps?
Hmmm, a bit of a conundrum there. All of those parts listed are things here on thingiverse. The "Part" column is their actual name. So, you could type that into the thingiverse search bar and find each part individually.

OK, that's a total pain in the butt. I guess I should add the things here directly, but that kind of defeats the purpose of the parts system here.

I think what really needs to happen is we need the ability to add ourselves as vendors, with the parts that we make being listed for zero cost, or so

I'll put together a .zip with all the parts and add it on here. That way I can at least version it easily.
Oh, I didn't realize that the link at the bottom and the links returned when searching the part *name* would be different, but I see what you mean (just tested it). Well, a bit of work, but yeah, I can get it this way too. Don't want to seem lazy.

BTW, what type of motor are you using? I asked what type of servo you were using (as that's the normal for deltas), but those look like steppers...? Are you using those instead?
I was playing with nema 17 motors just to get started. My intention is to use servos. I've just received some today, so I can get started on the movement!

Of course, which servos you use completely depends on the arm length and material you're using for the arms. In my current prototyping, I've been playing with some carbon fiber, so things are fairly light and strong. I'm going to see if the standard hobby servos AdaFruit currently sells will do
the trick. I've always got some Dynamixel 12s on order. A bit pricey, but a different type of control.
You may find some existing code useful, specifically the genhexkins (general hex kinematics ) module of EMC2 (http://linuxcnc.org/)

imac.unavarra.es/~cnc/ or in english:

Not EXACTLY the same appication, but perhaps a starting point.
Bad ass...
All pointers are appreciated.
Hey, this design looks good! I was startled to find someone building a nearly identical design to my own. I'm not quite as far along in the build as you are as I hit a couple months of a rough patch. But I designed an almost identical concept for my own Gada Prize entry... Perhaps we could collaborate and work together?

Check out the FabPod page in the Gada Prize page on the reprap wiki. (http://www.reprap.org/wiki/FabPod)

Feel free to contact me if you wish to discuss further. There is a link to my blog site on the FabPod page on the reprap wiki, and from there you can contact me. (don't want to post my email h
ere to help minimize spam lol)

And good job! The robot looks awesome!


is a very cool related rapid prototyper, festo also has a bunch of other cool videos in their bionic learning network section

Its a pretty rudimentary FDM machine, looks like it has a double stack nema23 sitting on top of the effector to drive the extruder which is a bit counterproductive.

Either style is a good candidate for a bowden extruder or perhaps a 1.75mm filament extruder that could be driven by a nema 14

I t
hink the big challenges will be to get a host software/firmware combination that will work. Then getting good joints that are accurate, smooth moving, lite weight.

It might be possible to get the host software to generate gcode that is just joint angles and then you could have pretty standard fir
mware. If we had arbitrarily small segment lengths that would work perfectly. I know people have figured out some of the speeds for minimum segment length because of the throughput of the firmware side. Somebody who knows some math could probably figure out how linear the movement would be with segm
ents corresponding to those lengths. That would tell us if we can use standard firmware that just linearly interpolated between fixed positions or if that wouldn't allow sufficient accuracy with current controllers.
Not sure what a "double stack" NEMA 23 is... NEMA just deisgnates the bolt pattern (not length). Torque can be increased roughly linearly with an increase in motor length... Since this was probably a one off machine, the designers may have just went for broke, got a high torque motor to ensure results... Existing Open source hardware was no doubt off the table for them :p
I like the Festo. When I first saw that video a few months back, I thought "wow, very sleek". Of course the music probably helps.

You're right. It's all about the math, weight, and tolerances in joints. I think having a platform like this helps you experiment with all of these aspects, assuming you've got the software part of it nailed.

At the moment I'm playing with some carbon fiber, bronze bushings, smooth axles, and g
enerally making things lighter, simpler, and more accurate.

I think a reasonably accurate and repeatable machine can be constructed at this level, and the nice thing is you can focus on any aspect of the design you care to improve, without having to change the whole design.
very cool! I actually just did a whole project on one of these delta arms for my numerical methods class. Have you figured out how you're going to control it yet? I suspect there must be some good closed form models for the inverse kinematics somewhere.

You can see my report/code here: docs.google.com/viewer?a=v
amp;srcid=0B8AGtzEPL1XpNGIwMjM0NzQtZmMyMS00Ym U2LTkwMjItNTkzNzA0MDJkZTAx
For me it's exploratory. I don't have a fixed model as yet. I was looking at some of the considerations of other designs on RepRap, and around the net. I have some reverse kinematics code from doing graphics, and I was going to apply that. I'd like to get a better physical model than the idealized model of the graphics world though.

This is obviously one of the more interesting areas of hard work to be done.

I do want to go with servos though, which should make a certain amount of ease in terms of programming (vs steppers). At least that's my current thinking.

I'll make it a RepRap Wiki thing so people can experiment as th
ey see fit and not be constrained by what I'm doing.

The basic bot is a good platform to start with I think.
Awesome. I'll be sure to poke my head in from time to time to see how it goes ;)
This looks really exciting, I can't wait to see more
Please Document that over at the RepRap Wiki, I know a lot of us would love to help in this dev.

Need help just pm me :)
I went and created a RepRap page, and entered for the Gada Prize while I was at it.
that sounds great! When I get a chance I'll pester you for help and get it going.
Correct me if I'm wrong but a delta robot is no good for reprap because it does not follow a straight path between points. It is only really good for pick and sort machines, no?
It might be fun to find out. The potential payout is quite high in terms of mechanical simplicity - and hence would significantly lower the cost of getting started.

The new, lightweight extruder that Adrian put on the blog recently might be a good sample payload. Perhaps we should ask how much it weighs?

Then replace the stepper with a continuous rotation servo :)
I think it would be a matter of getting the right weight ratios so the motors can handle the load. Carbon fiber tubing, smaller lighter end effector, strong servos (possibly with gearing), tightly toleranced joints, and I'm sure this design could handle a little ol' extruder head. Especially if that extruder head was driven by a light weight continuous servo.
The nice thing about servos..... They don't cost much :) That's why you're hard pressed to find stepper motors in new inkjet printers - a servo made from a DC motor and encoder is fast and cheap (but requires a little more software) :)
I don't know. I'm not a professional delta botter. I think it's worth an exploration to find out whether it can go beyond the typical pick and place usage.

For me, doing plastic extrusion is only one thing. I can imagine a bot that can both extrude, pick and place the electronics, and then continue extruding, to form whole systems, all in one place.

I think the path part can work out, I don't really see a physical limitation there. There are limit
ations in terms of available torque at certain points in the build envelope, and you have to deal with slop, and whatnot. But again, it seems worth the investigation.
I went back and found this article for you. Check the comments for some of the known issues standing in your way. I am not trying to say you can't do it just that it is not ideal and you will have many hurdles to overcome.

Hay! I had just read that Hackaday thread a couple of days ago. I've also been looking at this tutorial: forums.trossenrobotics.com/tutorials/introduction-129/delta-robot-kinematics-3276/

Some of the issues they were discussing in the Hackaday article had to do with using the robot as a CNC with a router tip. that might require a fair bit of lateral integrity, which may be hard to achieve.

I think for something like a 3D printer, if you go with a "liquifier only" head, the load m
ight be low enough that it's achievable, without to much distortion.

It probably comes down to how much slop can be taken out of joints, how stiff they can be at low enough weight to not require too much torque from the motors.

I could imagine using the same platform to do 3D scans as well. Jus
t stick a laser line in as a tool, with a webcam attachment, and put your scan device on a pedestal in the middle, or whatever.
I stand corrected. Can't wait to see the finished bot.
Nice! Do you have a video?
The video can be found here: youtube.com/watch?v=VzWJDWvJvqo
No video. There's no automated movement yet, so that exciting piece will have to wait until my servos arrive. Probably next week. I could do a video of me moving the mechanism by hand.

This weekend I'm going to have a couple of friends assemble them from scratch to see how long it takes them and what to improve. I can video some of that.
I have actually ordered some more bearings and carbon fiber tubing, and Dynamixel servos. I won't really work on the motion until I have the servos because that makes a huge difference in the electronics. The bushings and tubing will make a much smother movement.

the challenge continues. It is a fun project.
Would you mind trying bamboo as a form of tubing? It's lightweight, cheap and available widely (occasionally in the form of skewers and chopsticks).
Yah, that's a very good idea. Similar to this thing: thingiverse.com/thing:5284

Those vertices are fairly straight forward. The one that is a passthrough (that holds the shelf) might be a bit tricky if the bamboo is not smooth. Otherwise, it's easy, and you can just expoxy the bamboo into the vertices.

Same goes for any material. Just change the vertices, including using
them inside a tube instead of outside.

To go even simpler, you could take the bambook, or metal, or whatever, and put a screw through them at the vertices, or tie them together with string, or rubber bands, or whatever, and you don't even need the vertices. This would work well in rural communitie
s who already do pole binding, particularly with bamboo.
This looks great... I would love to see a youtube video of it in action. I am curious how it figures out the z axis.