Fiber Composite 3D Printing (The Bug)

Think of this machine as a very friendly sculpting spider.

For my last semester as an undergrad at IIT, I decided to give my education a twist and create my own research project. Fascinated by natures way to fabricate, I got inspired to create the Bug, a 3D printer-like robot that can create using composite fibers.

Step 1: Motivations and Inspirations

Fiber composites have been present for thousands of years now, they have evolved significantly into technologically advanced materials that today offer advantages like strength-to-weight ratio, flexibility or even insulating capabilities.

Today, we are living in a digital fabrication era where computers and prototyping machines are transforming our surroundings from mass-produced products into a world of endless customization. 3D printers have played a pivotal role in this desktop revolution, the speed of this additive process, as well as the low waste and relative low cost have made them a most-have machine for engineers, designers and researchers. 3D printers however are limited to layers and the strength of the bond between them, making parts significantly weaker than parts of the same materials made by injection molding, machining or casting. These days the market for 3D printers is getting crowded, and yet the market for fiber composites rapid prototyping has been barely explored.

Most of the inspiration was narrowed form nature, looking how animals and insects are able to create by mixing elements. My trip was also inspired by a trip to the Media Labs, where I saw a an amazing pavilion made by combining brilliant brains and hard working silk worms. If you don't know what I'm talking about you should.

Step 2: Research and Testing

With no time to lose I found myself spending hours trying to come up with ways to do this. Researching different resins I could use, different fibers I should start with, what sort of robot I would move all of these, how I would handle the extrusion of fibers and how they would bond and retain shape.

I decided to go for the UV cured type resins and tested a bunch of them until I found one that would cure fast enough and was able to retain shape. In the picture you can see my testing setup and an extruded thread holding horizontally once exposed to the process.

Step 3: IIT MMAE Department and IIT's IdeaShop

Realizing that I was not going to have enough time to design my own robotic CNC machine I decided I should try to get an existing one. I did some research into some repraps, the shapeoko and others, at the end Christmas came early with the donation by Ph.D Matt Spenko, a Fab at Home, an early 3D printer made by a group of students a couple years ago and left behind in the lab.

I would also like to take this opportunity to thank my advisor in this project, Ph.D. Matt Spenko and my unofficial advisor John Welin at the IdeaShop that not only provided me with tools, but more important, knowledge and challenges.

Thanks again!

Step 4: Design. Make. Test. Repeat...

Iterations. As you can see I did a couple until the time was up and I could no longer do more within the original time plan.

At one point I also changed my LEDs for 3W ones, huge improvement in curing time!

Step 5: Generating the Path

At the beginning, I didn't have an efficient way to generate the path, I had to select the points almost by hand. For this reason I came decided that the first print should be a simple 3 dimensional bell curve. I simplified the surface into a series of curves that intersected at the center and decided to give it a shot.

Step 6: First Prints!!!

The first couple of prints were very demanding and time consuming, a lot of monitoring was needed and problems with the Z-belt meant that I had pause the print to readjust the z-axis. The 3rd print however, was very rewarding as the object made looks somewhat to what I wanted to create!

Step 7: Code Generating the Path

Processing 2.0!

Using Processing I was able to grab any 3D model and extract a path for the machine to follow. In the images you can see some of the progress and how it went from a very chaotic randomness to a more smooth organized randomness, if that makes any sense. I also decided that it would work best if the machine would first make a supporting grid first and then it bonded the surfacing lines on top.

Here I would also like to thank John Mercouris for his help and trouble shooting while writing the code.

Step 8: Latest Print!

It has been a long way and I cannot yet see the finish, but the road looks amazing.

Comments

[DejayRezme]

Awesome. Great ideas!

I'd be more interested in 3D printing larger objects that are reinforced with fiberglass.

[Charlie Godfrey]

This is really good stuff! Recently I did something similar but used nylon as the matrix of the composite.
https://www.instructables.com/id/3D-Printing-High-Strength-Bio-composites/

[d'Waranv]

Sebastian, this is great work. I am trying something similar, with a thick paste. I have two questions,
1. How do you manage to keep your thread stiff and how do you cut the thread once the process is done.
2. how do you store the photopolymer ?

Please upload detailed design of the extrusion head if possible
Thanks in Advance

[Sebastian Morales]

As the thread was extruded it was coated in a uv cured resin, immediately after extruded it was radiated with uv light hardening the resin at the chosen position.

The code was design to have a single path to cover the entire surface, once done you could simply cut it with scissors. The point being that you would only have to cut once per print (simplicity).

The resin was contained in a lightproof reservoir (a photo film canister) and pumped to the mixer at the extruder.

[d'Waranv]

Sounds interesting. This brings up a couple more questions.
1. how do you load the fiber at the start and after every print ? wont the photopolymer cure or jam the nozzle ?

2. Also I was wondering if the thread doesn't fold or bend inside the extruder when it comes in contact with the polymer, how is the force applied to the thread to push it our through the nozzle?

[netgrazer]

Wouldn't a deltabot give you better range and agility, for what you're doing? I get that you're using the robot at hand, but if you're going to build v2 I'd consider going delta. Less parts, easier to scale up (i think), and the code needed probably won't scare you. Mobilizing it would involve pretty much the same arachnoid mechanisms and parts, plus the deltabot resembles that vectored orifice that spider web comes out of.

Now that I think of it, if you're going to mount it under a hexapod/octopod, the legs can do what the deltabot would be doing. So mobilizing it that way solves the entire deposition-coordination issue. That gecko "velcro" stuff would be great on the leg tips for stability, even if the small surface area of the spider's sticky "feet" won't accomodate wall-climbing (but it might, cue ground-breaking battery tech :).

The machine could print it's own support structures to climb on, like some type of specialized disposable scaffolding. The object being built could itself have small holes on its surface for the robot's legs to lock into. In the finalizing stage those holes could even be filled as the robot climbs down. Support structure placement could be realized procedurally of course.

Nice work on the Processing side btw! Can't wait to see where this is going, other than up :)

A nice isolated hut would be cool, or an ultra light bird wing made of movable segments, or a drone skeleton, or how about a lamp with varying degrees of translucency. Seal a hollow structure with foil, and you can pour concrete in it. Should I go on? :)

Several spiders could be cooperating to build one big structure, like a bridge that needs to be built overnight, or emergency housing. AI swarm tech is booming right now, so the timing is perfect.

[netgrazer]

tl;dr - way to bury the compliment dude.

Also possible upgrades: multiple strands of stronger/lighter materials like glass, aramide, kevlar or carbon fibre for filament. You could combine it with a suitable polymer like epoxy with some type of accelerant, or a monomer with quick-working activator introduced in extrusion chamber resulting in a high density polymer on the spot. Sci fi stuff. But if the end product only needs the tensile strength of the filament (like when you're going to fill up the empty spaces with something else, for example: concrete, bacterial excretions or acrylate) UV resin works just fine.

Prices for UV resin have gone down so what you're doing now will probably remain the most sensible method for prototyping this way for years to come.

To conclude, sorry for my ramblings (your amazing project got my gears spinning, and I still think a deltabot makes sense with all your Z-axial activity).

Keep up the great work. And thanks for sharing it!

[Sebastian Morales]

Hi Netgrazer,

First of all, thank you for taking the time to write this comment. Regarding the Deltabot, I think it is a good idea, I still think the next iteration should be a robotic arm that gave the extruder some extra degrees of freedom, that way the thread could always extrude tangent (or close to tangent) to the surface.

I love your idea of having this be able to move around, specially how if many of these could work together, then they could really start weaving, reinforcing the structure some more (and build a lot faster!). Then you could even have different materials giving selected volumes distinct properties.

Stronger materials are coming next, I started with cotton due to is low price, availability and ability to absorbe the resin but I recently got a carbon fiber spool.

Thanks again,

[netgrazer]

The degrees of freedom needed to print tangential to your printable surface could be provided by mounting the whole thing under a hexapod (yes, I'm still at it). They can tilt their bodies in pretty much any direction. Your printable area (action radius) would instantly be sized up towards power cable length :)

Coding the thing to maintain awareness of its position could be a bitch, though. IR beacons defining the print bed (your entire room) might help, but the necessary computing power will be increased even more.

Nice to see your work pop up on blogs, you're making everyone giddy with anticipation!

[Sebastian Morales]

Ha! Talking about robotic friendly spiders... I think there might be still some developing before I can reach that point...

Yeah! the blogs covering my work have been really cool surprises. Some of them do a great job on setting high expectations :)

[netgrazer]

No pressure... no pressure at all :) Just make sure you're having fun.

[Ottoclav]

I'm a sophomore working towards a ME degree as well. Amazing stuff, and you have given me some amazing ideas about this project. Wish I was knowledgable enough to help you work on this!

[servant74]

This is a great project. If something seems a little 'hair brained' that probably means it is still doable given a bit more thinking and experimentation (It has been 40 years since I was in college, but it was still on of the best times of life to learn and experiment! I am sure you are seeing great posibilities, so keep going!)

[Sebastian Morales]

Hey Otto(clav),

Are you also at IIT? It is an amazing school, even more so if you dare to go a little outside of the curricula. Don't worry about being knowledgable, knowledge comes with interest and drive.

[servant74]

Great work. I hope you get a chance to develop it more.

In the fiberglass industry (and possibly carbon fiber) they have a device called a 'chopper'. It cuts the fiber into short pieces as it is being sprayed with resin onto a mold. This might be a good addition after the initial structure is layed up to help complete the process. The shorter pieces help the fibers conform to curves more readily.

[cteixeira85]

Excellent approach... I like it. I own a conventional 3D printer.. which I don't mind to tune to work with your system and help on testing. Some troubles you got was more for path finding and hardware failure... taking you out of the aim.. composites and other materials for 3D printing...

[power000]

i like your invention

[samern]

This is so very interesting. Like you I was bitten by this bug some time ago and about 3 years ago built and ran a Fab@Home. I still have it and recently added a Sanguinololu board to run it. However the belt system leaves a lot to be desired and I am converting to leadscrews. I had looked at turning it into a large format printer with dual heads and even a light duty CNC machine (I originally had it running with a dremel head) but this is most intriguing. I would love to see how this evolves if it does any further.

[Sebastian Morales]

The belts! ahhh! what this project really needs is an arm. I am sure you have seen the shapeoko, but I think it could be ideal for what you are trying to do.

[samern]

At one point I debated that the entire printer actually be an arm and in fact I saw Lockheed Martin's industrial sized printer in action that does exactly that. It's a robotic arm with an extruder at the end. What can't be simpler! No rails, no gantries, no rods, no belts, no leadscrews. Just 3 steppers with arms and the extruder. Hmmmmmm.....