Corex(y) me if I'm wrong.

Hi Folks

I've been thinking a lot about printing buildings over the last few days after a chance discussion with chap called David Fuchs on G+. The concept of an open source design for something that can accurately deposit concrete to facilitate an extremely automated and potentially very cheap means of building housing is really compelling and IMO a truly disruptive technology. I know that there has a been a bit of work documented in the forum in working out how to extrude concrete in a reliable way that isn't already patented but my thoughts have been focused on how you might move a heavy print head around a large area (say 100m^2).

So, what I have in my head is an core xy systems using lighting/stage truss like duratruss (shown below) as a external frame, laying aluminium channel along the Y axis lengths of truss and using that to support/guide wheeled carriages that are on either end of a separate truss that spans the frame along the x axis. The ‘X’ truss would then support another wheeled carriage from which the concrete extruder etc could be suspended. For the drive mechanism, I’m thinking big DC motors/servos and single geared sailboat/yacht pulleys with either a steel or braided rope.

For Z movement, I’m less certain of what might work – one option is 2 or 4 scissor lifts supporting the gantry but I suspect controlling the lift on these things accurately would be a challenge and they sway around like crazy when fully extended. Another option is to use more vertically mounted duratruss and rack and pinion (s) to provide lift at each corner using more DC motors.





Guiding principals for this design as follows:

- Modular
- Off the shelf parts
- Relatively easy to integrate into existing open source electronics and control software.
- Uses existing principals as much as possible

I’ve been looking for obvious flaws in the concept before I do any designs but I think I’m standing to close to see the problems. I’m also wondering if someone has already thought about this.

Incidentally, I've seen the work that contour crating has already done in this field but they've patented the whole thing and I'm unsure that moving the whole assembly along a y axis is a great idea on a building site.

So, please feel free to pick holes in the idea and thanks for taking the time to read this.

The z axis for large format machines is by far the trickiest part IMO.

You could use scissor lifts, but remember the response (increase in height per unit input) is non-linear. You will also need vertical guide rails to constrain the gantry in X and Y, while the scissor lift only provides Z motion.

It sounds like you are proposing to use the spines of the truss as the linear motion surface. This will only work if the tolerance on the spacing between them, and the diameter of the tubes, is controlled very well. You may want to look at a bolt-on linear motion track instead.

I would stay away from steel cables. They typically have a large minimum bend radius (relative to the diameter of the cable), are difficult to work with, and if they part, release a large amount of energy. If you choose some non-timing belt approach, the tricky part is going to be figuring out how to drive it with a sheave/pulley without slip. There's a good thread in the Development section on cable drive h-bots that is a good read for the challenges in this area.

EDIT

Regardless of what design you settle on, I would strongly suggest building a small scale proof of concept before you invest the money in a full scale build.

Edited 1 time(s). Last edit at 08/22/2013 10:15PM by crispy1.

crispy1 Wrote:
-------------------------------------------------------
> The z axis for large format machines is by far the
> trickiest part IMO.
>
> You could use scissor lifts, but remember the
> response (increase in height per unit input) is
> non-linear. You will also need vertical guide
> rails to constrain the gantry in X and Y, while
> the scissor lift only provides Z motion.
>
Firstly, thanks for the feedback

I think scissor lifts is a dead end unless you want them to move them in the y axis as well tbh.
On second thoughts, rack and pinion is pretty stupid too. What I'm thinking now is to have another larger structure that hoists the whole gantry up for z moves.
OR keep the gantry stationary and lower/raise the print head.

> It sounds like you are proposing to use the spines
> of the truss as the linear motion surface. This
> will only work if the tolerance on the spacing
> between them, and the diameter of the tubes, is
> controlled very well. You may want to look at a
> bolt-on linear motion track instead.

Nope, I think it makes more sense to have a channel or track you install on top of the flat ladder rungs of the truss. This should give a flat (enough) surface, to run a wheeled carriage. Particularly as you could pack it out if there is any sag in the truss. I think it’s also key to remember that if you’re printing concrete the accuracy tolerances will be a little more forgiving, maybe +/- 10mm?

> I would stay away from steel cables. They
> typically have a large minimum bend radius
> (relative to the diameter of the cable), are
> difficult to work with, and if they part, release
> a large amount of energy. If you choose some
> non-timing belt approach, the tricky part is going
> to be figuring out how to drive it with a
> sheave/pulley without slip. There's a good thread
> in the Development section on cable drive h-bots
> that is a good read for the challenges in this
> area.

Duly noted, I’ll have a look.

> EDIT
>
> Regardless of what design you settle on, I would
> strongly suggest building a small scale proof of
> concept before you invest the money in a full
> scale build.

Oh absolutely, I don’t think I’d ever get past a scale model but if I could prove the concept I’d happily leave it for others to turn into a working version. When I do the maths on the quantities of truss required, It’s an easy $10-15K before any of the other hardware so it’s not really a home project 

At the speeds in question there is no need to reduce the gantry mass too much (using the motor mass to hold it on guides is a plus)

A rack with a moving motor is possibly the easiest transport for larger equipment. A rack and drive on each side can be electrically synchronised likely cheaper than cables and tensioners in the modern age, good quality cable is not cheap and you will need a lot of it.

I would suggest a parallelogram Z axis sort of like the Wally printer. Think of a cube wire frame that is pressed flat. Then using a diagonal mounted linear actuator you slowly open the cube that has the XY gantry on the top (You loose part of the build envelope because it rises in an arc but you have a structure that is much more rigid than a scissor lift. You have very simple mechanics and what you are left with is a cube that you can pick up off the house with a jumbo crane and place nearby to start the next one.

There are all sorts of additives for concretes, plasters and grouts that can speed up or slow down the cure and could be mixed in before use in the pump or nozzle so it cures fast just after deposition. Adding expanded polystyrene beads (with surfactants so they mix better) is good for insulation and adding chopped glass or mineral fibre can reinforce pretty well. Leaving channels for drawing in cables and pipes will make fitting fast.

---

Kalle
--
Lahti, Finland
The only stable form of government is Open Source Government. - Kalle Pihlajasaari 2013

This is definitely too large for timing belts to be lifting the beams in Z. A roller chain would be the thing. (Imagine a motorcycle or bicycle drive chain.) There's plenty of variety in sprockets and chains to choose from, all of it defined by standards and not proprietary. Worm gear reduction will minimize or eliminate power needed to hold position.

This much weight will be a hazard, so plan a ratchet mechanism to make sure it can't come down until it's supposed to.

Vertical duratruss will probably be more stable than scissor lifts.