CubeSpawn on December 17, 2009

Hi Lewis,
Your comment is fairly long and addresses a divers range of issues, I’ve attempted to respond to them all, but if I miss something in my responses kindly forgive the oversight, if you bring it to my attention I’ll be happy to give my opinion ;-)

As originally envisioned, the cubes sizes were to be created in discreet steps at 250mm, 500mm, 1000mm, etc. everything will be metric, since that is the predominate system in use on THIS planet...

I'm strongly convinced this spec will change to put everything on 30cm increments, based on thinking done by Sam over at MakerBeam. This combined with standard conventions in the (metric) building trades having to do with divisibility of units, and mechanical considerations relating to the susceptibility of aluminum extrusion to deflection under load. 30 cm is a good choice for a Cartesian reference grid and attachment points.

So, a 300mm base dimension, making for a 600mm cube, 900mm, 1200, 1500 and so on…

While I will initially be completing the 500mm design, due to the expenses that have been incurred to construct it to its present state, (cheaper and quicker to finish than start a new one) Converting over to 600mm for the published design will be relatively trivial.

As to the distribution model: kits will be made available at the conclusion of the development of a set of plans to build the initial cube type: which is a light duty 3 axis CNC mill. It’s a very conventional gantry/router design: 2 ball screws drive the ends of the Y gantry which rides on ground linear bearings. X rides rails on the Y Gantry and Z is on one face of the Y gantry riding the X carriage.

These plans will be downloadable at no cost, or you may buy a kit from CubeSpawn – or you can bargain hunt components and build it yourself, your free to change the design any way you like, but are encouraged to conform to standards in 5 areas:

1) Build the cube to (one of the) standard dimensions as defined by the “official” plans, this can be chosen based on your requirements for a “working envelope”
2) Use the standard electrical interconnect at one of the specified locations, use the recommended ratings.
3) Use the standard data connection at the specified location.
4) Allow for the spacing of the pallet transporter, even if the cell doesn’t explicitly use it…
5) The project software will use the standards as assumptions to simplify its operations, especially in its earlier versions, so complying with the specs above insure your machine will work with the “suite” of management tools that are being designed.

Conformance to the standard is of course totally voluntary, there will never be a “CubeSpawn” compliance security force… ;-)

Everything else is wide open, what the cell does, how it does it, etc are not constrained by anything but the laws of physics.

The scenario you describe: of removing the precision drives from one machine and installing them on another, with .0005 repeatability is not included in the preliminary spec. It sounds like a difficult objective to achieve with aluminum extrusion frames. My thinking goes more along the lines of developing a ground fixture (your weaver rail) that will mount IN a cubes frame and provide that kind of precision, but the cubes themselves will very likely not achieve that degree of native accuracy, especially in their first generation.

I’m in full agreement it’s a desirable and essential capability, but we are walking naked in the forest at the moment, lets domesticate some animals and harness fire first ;-)

Oh yes, the coolant pump dilemma, and its associated sump/chips/bacteria/clogging/drip pan issues – not sure if I’m ready to marry that problem yet, given that I’ll be putting a halfwit robot in charge of the resolution of (some of) the issues it brings – again, not a gen 1 problem to solve, the 500mm MK1 cube design is based on Nema 23 motors, not going to be creating a lot of heating issues, unless its simply from stalling the whole motion stage while cutting ;-)

Although I’m totally in agreement that these are vital machine design issues that will have to be dealt with in the near term.

You go on to say
“Build a spindle head, an XY table, and a way of bolting them together that allows them the components of be upgraded and you have a winner.”

I couldn’t agree more.

One thing that’s potentially not clear is: the head changer AND the tool changers ARE designed and will be part of the Generation 1 solution. The initial prototype is a 3 axis mill cube, a tool-changer cube, and a head changer cube, plus a loading station, 2 pallets, and an unloading station, arranged in a cross.

So spatially: standing and looking at the machine:

On your left is the loading station, in the middle is the mill, on the right is the unloading station, between you and the mill cube is the tool changer and on the far side of the mill is the head changer.

So, no reconfiguration is required to change from milling to RepRapping, other than unloading the tool, initiating a head swap and loading a substrate on the pallet in place of tombstones, or whatever we are fixturing with.

I will make a spindle motor selection that’s appropriate to the light loading this machine will incur, but other compatibility issues, such as socket rating, type of power and so on will drive and constrain the choices down to an acceptable subset.
This will be an integral motor type spindle for simplicity. It may be a dc motor, to simplify its compatibility with the power for a RepRap extruder, but the final choice is not yet made.

This is not a trivial choice, in the sense it will influence what other as-yet-un-conceived interchangeable head design can also be used in this cube, but it is felt that all of these designs will evolve, as they should, so if I don’t take everything I should into consideration, a better version can still be built into subsequent designs.

As you mention a whole range of machines can be designed around the basic motion control module, although I would contend there are two basic designs: a “power” design for cutting and materials moving. And a “delicate” design for things like the CMM application.

One other distinct variant may be a low cost “Power” design for lower precision applications like routing.

The pallets will allow mounting of t-slot equipped plates as well as other surface types.

One fairly integral problem to design is an open source linear stepper motor (or other linear design) for the purpose of simplifying motion control in subsequent cubes.

Given they are mechanically simpler than ball screw designs, this would be especially well suited to the “delicate” positioning solution mentioned above and a linear stepper would be easier to manufacture on-site with a mill and a coil winder.

Under any scenario I can envision it will 2, 3, possibly 5 years until a collection of cubes can make all their own components from raw materials, and potentially much longer until they could make the stock from reclaimed materials, refining is potentially much longer still, but all the bright people who are interested in various aspects of this, working in parallel, may be able to totally invalidate that thinking – I hope so! ;-)

James

(I claim credit for all the typos, random assumptions, and omitted or incorrect answers entirely for myself!!)