[nzlug] reprap

[Simon Bridge]

On Thu, 2008-04-10 at 09:44 +1200, Vik Olliver wrote:
> On Wed, 2008-04-09 at 14:35 +1200, Simon Bridge wrote:

> > A self-reproducing machine only needs to be a bit bigger than the
> > largest of it's own components. (Note - anything which *needs* a
> > circular hole can be made in two parts.)
> 
> Yeah. I've made the rods 500mm long on the next one just because it's a
> nice round number.
> 
You'd probably get away with 300mm for, say, height. There will be a
practical limit on these earlier machines due to the need to be able to
get your hand in there.

> > A smaller machine won't need as much metal to maintain it's strength -
> > so it can make it's own support struts. That leaves the z-axis screws
> > and the x-y rods.
> 
> Tricky. The rods also get used as bearing surfaces. It might be possible
> to use something cheap, like thick welding rod - maybe multiple rods, or
> back them with plastic.
The bearing rods are the easy part - it's the screws that will be
trouble as they need to be fairly precise. A future design will have to
find another way.

> 
> At present deposition is so bloody slow that it's quicker to drive to
> the hardware store and buy rod.

Yeah well... I'm thinking in terms of solving this sort of problem in
principle. After which, we can take shortcuts in the name of pragmatism
with a clear conscience.

Speeding up extrusion would be a worthwhile project, yes. As would
working out a way to coordinate two heads on a single build. Right now
you have the man-month problem.

Presumably you could speed up extrusion with a hotter element, longer
tip and shoving the material in there harder. Mostly a viscous flow
problem - probably been solved already.

> > Q. What is the shortest time for the existing machine to build all it's
> > own components, if run continuously?
> 
> About 2 weeks.
> 
So - if it takes a day to construct and test one of these things (doing
nothing much else) then you get linearity in about 65-70 days (making
one a day, and the stock of spares builds exponentially).

Of course, assuming each machine makes parts for all of one machine. If
you share workload so all machines work on building one machine, the run
time per machine ends up close to the time to make the largest
component. (Sectioning the components so all parts are about the same
mass would provide a more exponential scaling, but weaken the structure
and add to the assembly time.)

So... 
1. speed up extrusion
--- need figures and the math - want to know how viscosity changes with
temp. preferably from direct measurement. Though you could figure
extrusion rate as a function of temperature quite slowly :)
--- motor torque vs supplied power I guess, as well as the maximum
torque that can be applied before the delivery mechanism fails. Don't
you really need low gears on that motor anyway?

2. reduce component mass
--- need material properties of the room-temp material used
--- must see if I can find those organic design notes

Yep - definitely a thesis in this.
Perhaps I should go talk to ECE people.