Sky Printer - Prototyping an Adaptable Large-Scale 3d-Printer




Introduction: Sky Printer - Prototyping an Adaptable Large-Scale 3d-Printer

The Sky Printer is an adaptable, cable-based delta-gantry system that can span large areas with a minimal footprint. The Sky Printer is designed for large- scale applications in potentially remote locations. It can be understood as the evolution of the conventional 3d-printer.

We are a team of two architecture students from California College of the Arts in San Francisco. (team members: Thomas Monroy, Taole Chen)The project was developed in the FA14 Creative Architecture Machines Advanced Studio taught by Jason Kelly Johnson and Michael Shiloh (also check out the Digital Craft Lab which is the overarching research department at CCA).

We think the project has immense potentials that we were only able to graze in the two months we had to develop it and were surprised that no one has really attempted it before us. Although the cable-based system inherently has issues with stability, rigidity and may not be easily adapted to very precise applications, there are distinct advantage that point to an exciting future for computer-aided construction:

  • minimal material costs: compared to conventional 3d-printers and CNC-routers, the Sky Printer doesn't need a platform, it hijacks existing structures and topography to construct its own coordinates system. Thus, it can scale up effortlessly without adding exponential material costs relative to its size.
  • minimal maintenance: mechanical parts are minimized, therefore possibilities of failure are minimized. Also, it has the potential to be remote-controlled, allowing it to be deployed in hard to access locations.
  • ability to print on uneven surfaces
  • adaptability: We see tons of possible applications, as the adaptable system allows it to be set up in any environment with vertical surfaces, such as mining pits, abandoned cities, canyons, mars, etc

Within the constraints of the studio class, we focused on additive clay printing as a method to test the system, but the Sky Printer is really much more than just a printer, as we envision it to have an exchangeable tool set that would include grapplers, drills, spades, analysis tools, etc.

We'd like to mention and thank the people who have helped us, either directly or indirectly, realize this project. Being architecture students with no practical knowledge in engineering, we relied a lot on available online and offline resources:

  • Jason and Michael for providing us with a solid and extensive foundation in electronics and programming, and of course continued guidance throughout the development process.
  • Taylor Fulton and Max Sanchez, who developed a delta-printer that prints clay towers in the previous year as part of the Creative Architecture Machines studio and who provided us with a working clay formula and bits of code we could hijack. In extension, Jeff Maeshiro, who gave us an even more detailed clay formula from years past.
  • Andrew Maxwell Parish (electricslim from the Artist-in-Residency Program at Instructables) who helped us fix a broken 3d-printer and let us steal equipment from the Hybrid Lab :) (don't worry, we returned all of it!)
  • This post on a obscure robotics website that was seemingly the only one that contemplates about cable-based robots., but whose author gave up in the idea due to lack of time. (We hope he will read this instructable eventually)

The Sky Printer is an easy to replicate system and we will publish our findings in the following steps, in the hope that others will start picking up on the idea and develop it even further.

Step 1: Material List

Things you'll need:

  • 3 stepper motors (size will depend the weight of the tool head, we used Nema 16s which worked perfectly fine for us) .
  • A TinyG - probably the most expensive investment on this list. It's around $130, but, in our opinion, well worth the money. It is very stable and smooth and you can operate up to 4 stepper motors. Alternatively, you can use a Quadstepper from Sparkfun. OR: any other motor control shield if you are well versed in programming (because you'll need to eventually alter firmatas and what not)
  • A DC power supply that corresponds to your steppers. The TinyG can handle power supplies between 12V-30V, so make sure yours falls within that range (you dont want to fry your most expensive component).
  • Something to construct the winches and tool head with. We developed prototypes using laser cut acrylic, 3d-printed PLA, but also just hand-cut foamcore and plywood in the beginning stages.
  • A computer with Rhino 5.0 installed.
  • Grasshopper, which is a Rhino Plugin and Firefly, which is a Grasshopper Plugin (confusing, I know).
  • potentially hardware such as bolts, nuts, screws. Also, probably glue.
  • monofilament. We used 15lb clear filament, worked great.

If you are planning on printing with clay:

  • Air compressor
  • PVC pipe + adapters and other components to connect to the compressed air
  • silicone tubing (or better, surgical tubing with high pressure resistance but flexible material. we had problems where the silicone would expand at hotspots as the pressure passed a certain threshold)
  • clay (more details later)
  • high torque drill and mixer bit for clay mixing
  • a bucket
  • spatula
  • lost of patience

Step 2: The Tool Head

The tool head, at this point, is open for iteration. Ours simply had to be able to hold the silicone tube with a nozzle and be able to attach to strings. The images are ordered in reverse chronology, so the first one is the most recent one. Notice that in the later iterations the strings are attached to a ring, rather than a fixed point. We hypothesized that this would resolve an issue we had of the tool head not being level when moved too much in certain directions. Theoretically, it should balance out the differing heights at the three anchor points. It worked to some extent, but there's still a lot of room for development.

The bolts were used for weight. Acrylic and PLA are both very light weight, so we had to increase the weight for stabilization.

We put up process images of several rather than one single tool head, because we want to demonstrate that you don't have to rely on a specific machine to make it. Fortunately, we had a Type-A Series 1 and laser cutters at our disposal, but the first extruder was entirely hand-cut in a matter of minutes.

We included one .stl file for 3d printing and one .dxf ready for laser-cutting, but we do encourage to experiment with and design your own tool head, as ours is by no means perfect.
For example, at one point we bought a 6 DOF IMU (Six Degree-of-Freedom gyroscope+accelerometer) so we could potentially neutralize tool head wiggling, but did not have time to implement it. Due to the cable-based construction, the tool head wiggles whenever it is moving. It was also not imperative for us to solve this quirk of the system, as the wiggle actually helps when printing clay.

Step 3: The Winch Casing

Similarly to the tool head, the winch casing evolved through the course of the studio. The last iteration has two slots for bolts that you can slide around so it can hang onto a variety of ledges. You can also just bolt it down to a straight ledge or to an anchor that you can bolt/hang onto the wall.
Point is, it's actually very flexible. All you need is a method to fixate it the winches somewhere high up in a space. At one point, we just used a few clamps to clamp the stepper motors to unistruts that were hanging around in our studio.

Step 4: Wiring

The wiring is pretty simple. There is a USB port, terminals for the steppers, connection for a regular computer fan, and power.
Notice our power supply says it's an AC supply. Don't mind it, it's lying! We checked it with a multimeter and it is in fact DC. Speaking of which, always check your power supply with a multimeter before connecting to anything, especially when you're dealing with expensive electronics like the TinyG!
As for the stepper wiring, the colors will be different with every model. But Michael taught us a simple trick using a multimeter that should also help you find the right pairs every time (Bipolar steppers always have 2 pairs of wires, if it has more than 4 wires, it's not bipolar).

  • Switching the multimeter to read out resistance, take two wires and press one to each end of the measuring pin.
  • If it shows any number other than 0, switch one of the wires with another one.
  • If the resistance shows 0, it means you have found a pair. (the pairs on our steppers were GREEN-RED, YELLOW-BLUE if that helps anyone)
  • Simply connect 1 pair to pin 1/2, the other pair to 3/4 on each stepper terminal.

You are now connected!

Step 5: System Setup

The setup is pretty straight forward. Find a suitable space with three fixation points somewhere high up and a safe place to keep your TinyG. We were able to complete the entire setup in under an hour.

The hard part is calibrating the system inside grasshopper. Because we did not include any feedback sensors, the Sky Printer doesn't know where its tool head is located. So we need to manually measure out the height of the winches, the distance to each other, and their respective distance to the center point. The center point is important, because it acts as the homing origin. We needed to first find it by measuring it out physically, and then move the tool head above it and set it as origin in Gcode to achieve accurate movements.

Step 6: Understanding Software

A crucial part of the project was the coding which was all carried out within Grasshopper (it's a visual programming language). The program itself requires you to have Rhino 5, a NURBS-based modeling software which is commonly used in architecture, but also industrial design. You can download a free 90-day trial, after which you have to purchase the software.

To install grasshopper, you can download it from their website.

You will also need the Firefly plugin.

Further instructions can be found inside the grasshopper script which is included in the attachment.

Step 7: Printing With Clay

We are in a love-hate-relationship with clay-printing. The results are beautiful when they succeed, but the complexities of the material make it a very challenging project to tackle.

Step 8: Word of Caution

Compressed air can be dangerous. You will absolutely positively need a pressure regulator when working with compressed air. We didn't have a clue in the beginning, so we connected the PVC chamber straight to the air compressor. Fortunately, we didn't glue down the end pieces (which we should have done, another mistake) and once the 150 psi shot into the tube, it only transformed the tube into a cannon instead of a pipe bomb which would have caused some serious damage as we were holding the contraption in our hands. Another lucky coincidence was that we happened to point it towards the corner between the floor and a wall, so it didn't hit anyone (it only left a huge dent in a piece of MDF and a slightly bruised finger).
Also, eye protection, gloves and gas masks are recommended, as the clay can be damaging to the human body when exposed for too long.

Step 9: Clay Formula

The science of mixing clay is an entirely different beast altogether and we are nothing but kids lost in a strip club. We are absolutely clueless regarding the optimal composition of clay and found a working formula by trial and error, and with a lot of help from others who came before us. We also used a measuring cup, which measures volume, whereas clay is usually measured in weight. But it worked. (to some extent, our clay artefacts are still prone to cracking, and we were not able to fire them in a kiln: instant crack)
Here's the formula:

  • 400ml water
  • 400ml Grog
  • 400ml Kentucky Ball Clay
  • 400ml Silicone 325 Mesh
  • 400ml EPK
  • 400ml Custer Feldspar
  • 1 bottle cap of Darvan 7 (or 175 +-10 drops)

You would want a less viscous mix (more like a clay slip) so it can move through the hose effortlessly. We also used photo lights (flood lights or heat guns work as well) to bake the prints while they're printing, to decrease their drying time and increase stability. With this method, we were able to achieve a maximum angle of about 30 degrees, after which the artefacts started collapsing.

If you're in the Bay area, the ingredients can be purchased at Ceramics & Crafts Supply, which is located in SOMA, San Francisco.

Step 10: Material Chamber

The material chamber we put together by going to the hardware store and finding the right fittings on site. We are not putting up a component list as (a)we don't know the names for more than half of the components, and (b)the exact form doesn't matter as long as it is an air-tight chamber. For the second chamber, we have a cylinder made out of acrylic in the center, so we could see how much clay we had left in the chamber.

Step 11: The Results

Payoff for our blood, sweat and tears.

Step 12: Potential and Improvements

Technically, there's a lot to improve. We have identified a few points that would be our next steps in the development process:

  • automated homing system: right now, we're adjusting everything by hand. In a "consumer-friendly" version, we envision an automated homing system that would automatically calibrate the system when installed at a new location. For that, we think that potentiometers can be installed on the winches that could read the angle of the string at each winch. Also, included somehow there would be laser-measuring devices that would automatically read the distance between the components. This would give the system a feedback loop for it to calibrate itself.
  • Wiggle: as discussed before, the tool head wiggles due to the construction logic of the system. Depending on the application, one might want to neutralize the wiggling, which could be achieved by installing a leveling mechanism using a gyroscope and accelerometer.
  • ceramics expertise: we should have probably consulted with people who work with ceramics to make the artefacts more durable and maybe even stretch the material constraints of our system.
  • Tool heads: we have only experimented with clay extrusion and 2d pencil drawings so far. What else could the system be applied to? Imagine a CNC router, a grappling transportation system for warehouses, etc.
  • Architecturally, we designed a project along side the actual machine. The project, which we call the Regenerative Bio-Catalyst, proposes the hijacking of contaminated mining sites and turning them into experimental ecologies using a gigantic Sky Printer. But this shall not concern this instructable, The point is that conceptually, we barely scraped the surface of a very rich technology.

Be the First to Share


    • Anything Goes Contest 2021

      Anything Goes Contest 2021
    • One Board Contest

      One Board Contest
    • Raspberry Pi Contest

      Raspberry Pi Contest



    5 years ago

    Is there a way I can order the parts for this


    Reply 5 years ago

    I am not sure I understand your question. Are you asking if these items are available online or if they exist as a kit that can ordered?
    If you are asking the latter, then no, the parts do not exist as a kit. But I am certain you can order these parts online, although I wouldn't know where to get all of these. We kind of just used whatever we could find. Many parts we found in the studios, or borrowed from teachers/classmates. Others we bought locally, since we didn't know what we were looking for. Point is, the mechanical parts are really simple and you don't have to use the exact ones we had. The complicated part is the code and getting the clay mixture right, which can all be found on this instructable. Hope this helps.


    6 years ago

    Hey guys,

    first, I was part of one of the initial studios for this, great to see how far its come in the past three years!

    I wanted to ask how you're writing to the arduino board and steppers, as I'm working on a project currently and was attempting to use the quadstepper component but was having no luck. Jason had recommended I get in contact with you, and I'm hoping you can shed some light on your use of the data write component and how that was working out.

    Let me know if you'd like me to get more detailed, or would like to email directly. Any words would be appreciated.


    Reply 6 years ago


    It's been a while since I looked at the project, so I'd likely need to go back to refresh my memories if you want to go more into detail. But to answer your question briefly, we decided to acquire a TinyG for controlling the motors as it simplified the coding part tremendously - you can feed it Gcode directly and it has built in acceleration control. Plus, it's a good investment as you can always repurpose it for future projects. We had some issues with the quadstepper too mainly in terms of coding. If I remember correctly, the firefly components were not sufficient to do what we needed it to do, and we would have had to code our own components. Back then we simply didn't have the time to tackle that challenge and decided to go the easier (but pricier) route.
    I don't know if that helps. But maybe if you reveal some details of your issue we can figure it out? Either here or you can shoot me a pm.


    6 years ago

    You could add some rotary encoders and to find out where the head is, you could time how long it takes for the head to get to each individual winch using a bump switch on each winch. cool idea thou


    7 years ago

    Very innovative. he's right. The use of non static cable would be more precise. have you though about using a camera or ultra sonics for positioning the tool head? I can can see an upscaled version of this extruding a house.


    7 years ago on Introduction

    So awesome, I can imagine that the buildings of future will be blobby organic structures like yours that constantly evolve and grow to peoples needs and the advances in technology. A colony on Mars in a crater or maybe even under the sea could have structures fully fitted and ready for them when they arrive then grow with the colony as it grows without being out of date or irrelevant to future developments. The Sky printer could 3d print water pipes and electronics(laser sintering or another metal printing method) into the walls (must be reachable for maintenance though). SO MUCH POTENTIAL!

    I wonder if an expanding foam/plastic froth(by pumping air through the plastic as it is printed) printer combined with a normal 3d printer would be successful in creating walls with insulation built in to all the complicated structures that it may be needed in without having to squeeze celotex into there. One possible head would be a particulate filling head, for sand, loose insulation in cavities, aggregate, earth, gravel.... obviously varying in size and power but for a work site this could really help with moving the material from the truck to the place it is needed then applying it directly, negating the need for a whole host of people! The future is a place of desk based engineers and designers, with a few maintenance crews and no workmen!

    I really hope you continue to develop this as you could earn a living from it if you can get copyrights in. I guess the print head design would be a major part too copyright, if milled or 3d printed from metal with durable parts that are interchangeable, upgradable and most importantly replaceable ...something that designers/engineers seem to forget about now. Its the most unique part of the project as everything else is pretty much already made.

    This is really inspiring to me as I am interested in designing something just haven't decided what to do/specify in yet!


    Reply 7 years ago on Introduction

    This is also the most exciting part to us! Glad someone else can see the potential in the system. We have developed an entire project speculating on printing experimental ecosystems for wildlife in contaminated areas, exploring how the system itself could start informing the shape of habitat structures, and also how the "controller" would potentially be able to design the artefacts at his home desktop and press an upload button, which would send the information over satellite to a remote location. We only included a few of the images of the second project in this instructable (it's on the last step), but I will upload a more complete set of drawings to my online portfolio and can give you the link then.

    As for designing something, don't wait! The one thing you learn in architecture school is that no one ever knows that specific thing they want to design. Just set a rough direction/topic and start sketching/researching/building up a hypothesis, you will eventually get to a good design project in the process.


    7 years ago

    Very interesting and creative. Suggestion: fishing line is relatively elastic. I would wonder if steel bicycle cable would give much tighter positioning?


    Reply 7 years ago on Introduction

    Yes, thank you for the suggestion. We only kept with the monofilament wire because precision was not a main concern for this project. Steel cable or maybe even a more rigid belt could minimize the wiggling potentially.


    7 years ago on Introduction

    So cool! I'm a student at CCA also and I think I remember seeing this project. Awesome job!