Extremely low-cost fabricators?

Has anyone looked at lower-cost fabricator concepts? Eg, there is an ULCPC notebook (or eventually a cell phone), USB dongle(s) controlling multiple fabricators, a couple of DC motors in each, and that's it. No other electronics. Replace electronics with software, fabbable mechanics, and simple wire circuits.

Hand-waved examples:
* Do shaft encoding with gears and spring switches, dumb DC motors, and software extrapolation between clicks.
* Instead of shaft encoding, use a non-drive system to get precision position.
** A clockwork pendulum bounces between head and wall, several clicks per swing, providing velocity and position info. The computer provides high-precision timing for free. Software extrapolates between samples.
** Webcam/cell photo chips are cheap. One might be able to get high-precision position by image processing of a sheet-of-paper with a complex printed target.
** Replace drive shafts with gears, and use software modeling of gear behavior and defects to recover accuracy.
** Replace a horizontal shaft with a rotating stage.
* Simplify the z-axis.
** Float the stage or writer. Valve water in or out. Or simply drip it at a known rate, and use time as a proxy for position.
** Human turns a screw.

Any other ideas?

Basically, a bootstrap fab package would consist of a head, a dongle, some click-together fabbed structural pieces, a couple of cheap motors, wires, assorted fabbed mechanics, and some springs. That's it.

Off topic:
* Bone-like largely hollow structural pieces.
* Adding microfibers to improve strength.

Thank you for your inspiring work.

Interesting ideas. Some random comments in the text.


Mitchell Wrote:
-------------------------------------------------------
> Has anyone looked at lower-cost fabricator
> concepts? Eg, there is an ULCPC notebook (or
> eventually a cell phone), USB dongle(s)
> controlling multiple fabricators, a couple of DC
> motors in each, and that's it. No other
> electronics. Replace electronics with software,
> fabbable mechanics, and simple wire circuits.

Could we power the motors over USB (in that case we'd probably need to keep the laptop plugged to the grid while printing)?

We'd also still need something capable of electrical input and output signals - an Arduino comes to mind (around 25 euros).


>
> Hand-waved examples:
> * Do shaft encoding with gears and spring
> switches, dumb DC motors, and software
> extrapolation between clicks.

Seems possible.

Also, end stops naturally with touch contacts instead of optoswitches.

> * Instead of shaft encoding, use a non-drive
> system to get precision position.
> ** A clockwork pendulum bounces between head and
> wall, several clicks per swing, providing velocity
> and position info. The computer provides
> high-precision timing for free. Software
> extrapolates between samples.
> ** Webcam/cell photo chips are cheap. One might
> be able to get high-precision position by image
> processing of a sheet-of-paper with a complex
> printed target.

This might get a bit tricky - webcam quality is not necessarily so good, lighting conditions and cast shadows change the picture, and the picture analysis programming might be heavy for a low performance laptop. I suspect it would also take a lot of programming effort, that could be put to better use. But in the long run this kind of high-information feedback loop would probably be good to have.


> ** Replace drive shafts with gears, and use
> software modeling of gear behavior and defects to
> recover accuracy.

Small errors could accumulate. Some kind of feedback will probably be needed.


> ** Replace a horizontal shaft with a rotating
> stage.
> * Simplify the z-axis.
> ** Float the stage or writer. Valve water in or
> out. Or simply drip it at a known rate, and use
> time as a proxy for position.

My gut feeling is that water would complicate things. But on the other hand, it could enable all kind of pneumatic fun.

> ** Human turns a screw.

Sounds like too much wasted time from the operator. That's what computers and machines are for, for tasks that require turning a screw every ten minutes.

>
> Any other ideas?

Could we replace the steelframe and fastenings with plastic too? If they have to be bought as new, they add up to some money too (maybe around 50 euro). Probably threaded rods and nuts are needed for the three axes at least, though.


>
> Basically, a bootstrap fab package would consist
> of a head, a dongle, some click-together fabbed
> structural pieces, a couple of cheap motors,
> wires, assorted fabbed mechanics, and some
> springs. That's it.
>
> Off topic:
> * Bone-like largely hollow structural pieces.

Sounds fun. I want to computer generate these with genetic algorithms

> * Adding microfibers to improve strength.
>
> Thank you for your inspiring work.

Hi all

interessting! two of your ideas i had a few month ago to:
o using timing capability of computer to extrapolate position of a simple, determistic, area scanning mechanics
o use water (solenoid valve) flow out to control a z axis
i did a few work on that. cheap means that you have to avoid linear bearings, linear movement and electronics as much as possible. i was thinking about a lissajous figure based mechanic consisting of two orthogonal wheels which are driven by the same motor in constant speed with a constant ratio (ie 1:4.0001). the wheels act as excenters and drive two bars forth and back. those bars are connected to each other. the connectionpoint, lets call it toolpoint will scan an area. depending of the (constant) gear ratio and the timing resolution a very good resolution may be achived.
i wrote a quick and dirty demonstration in ruby and sdl and built a techdemo in LEGO technics (tm).

a picture is available under

i could provide code or videos too

b.abu

... go through the links in this thread: [forums.reprap.org] - especially the high-accuracy parallel-scara-concept in the harmonic-drive-paper in the last post ...

Viktor

hi viktor

not quite what i meant. my first atempt was parallel kinematics too. i implemented forward and inverse kinematics for a five-bar mechanism like that on [www.parallemic.org] but then i had the other idea which is far more easyier to build and "control".

b.abu

Hi b.abu,

maybe some of the concepts here [forums.reprap.org] or here [forums.reprap.org] are interesting too?

Viktor