On Replication

Replication, and specifically self-replication, is a big topic here in the RepRap community.

I've written a bit on the subject, as part of clarifying the goals for the RepLab project. I believe these distinctions have some relevance to the RepRap project also.

Specifically, I'm saying that for a recursive, non-self-assembling replicator, the degree of automation of the replicating act matters far more than the degree of recursion.

In other words: Conveyor belt!

On Replication

I'm already planning on trying a conveyor with my mendel once I have the basics working.. just so I can do really long runs of parts and have them drop into a bin when complete

Your argument is convincing, but I think it doesn't capture the whole picture. You make a good argument for the importance of automated self-replication, but of course the degree of recursion matters too.

For example, a type 3 commercial printer might be able to automatically print and bind books without human intervention. But the farthest it will get toward self-replication is to print and bind its own instruction manual... so that's not very helpful to people who want to make more commercial printers!

Especially as someone who has received some printed parts from another Reprapper (Wade), I am very appreciative of the fact that Reprap can print a good deal of its own parts. But I'm sure that Wade is also very appreciative of the amount of human effort that goes into printing a set of Reprap parts, so you're right for sure that it would be good if the process was more automated.

But basically, I think the conclusion is to aim for both a high degree of automation *and* a high degree of recursion.

Interesting read though! I like your categorizations. And I never thought of kilns as self-replicating machines before.

It is fun, and I'm going to have to read it more carefully.

Generally, I try to be careful never to underestimate dedicated tooling. Industry is currently is working on making solar cells using normal 'printing on paper' technology, and, while the initial molds are expensive, you can cast Darwin (or Mendel?) parts somewhat quickly and with negligible effort once you've made some molds.

On the other hand, dedicated tooling is 'useless' to the small shop or solitary user/builder, who can't afford a huge printing press. That user/builder can make plastic parts cheaper than a injection molding machine can, at least at first. More generally, it depends ...

Edited 1 time(s). Last edit at 11/28/2009 05:38AM by SebastienBailard.

Recursion matters, it's crucial in fact, but recursion is also old hat. Machines make all products including other machines. It's automated recursion that has interesting consequences, with Type 3 replication the most flexible kind. It's the Type 3 general replication that is new, and combining them is already changing everything. Consider: all one needs to write software is a computer and some software. If that's not changing everything, what is?

The goal of RepLab is to build a suite of machines that together display more recursion than is currently possible with a single, desktop machine. For instance, a RepLab could cut threads for leadscrews, manufacture bolts from round stock, print out Mendel parts and assemble all the motor circuitry automatically. That's quite the RepStrap!

Some of our members, particularly Marcin, are thinking ahead to melting metal, casting billet, and building an extruder and roller plant that can make basically any raw material for the RepLab that doesn't fit in one's pocket, from scrap. Marcin also wants to extract aluminum from clay, He's quite dedicated!

Interesting.

I drew this up:
http://objects.reprap.org/wiki/RepLab#Economic_Model_.28Quick.2C_to_the_Spreadsheet.21.29

I know that a small bronze foundry can pay for itself, if you're a decent amateur sculptor and master at casting/moldmaking. I have heard that hobby aluminum foundries can pay for themselves doing sandcasts of things like housenumbers and plaques. This assumes decentish entrepreneurial and graphic arts skills. I have not heard that hobby steel/iron foundries pay for themselves.

I'd try filled epoxy aka epoxy granite for everything I could before falling back to steel or aluminum to make findings for a CNC router. I'd carefully price aluminum billet before casting my own.

Since we've got bloggy name recognition amongst geek fabricators, I'd suggest working on something like a benchtop CNC router. I feel it would have the greatest potential for self-funded development. And the highest chance for success, since it's an extension of our existing research and communities. Not that a little cnc-ed grinder for making end-mills wouldn't be cool, for example.

Edited 1 time(s). Last edit at 11/30/2009 05:07AM by SebastienBailard.

Marcin's emphasis is different from that of the RepLab project at large. His goal is permafacture, replicable village-scale creation of all the comforts of a high-tech lifestyle, from agriculture and housing to machine tools and their products. I think it's a beautiful vision and one where even partial realization can produce great economic and social benefit.

RepLab is more focused on general replicators, with the expectation that the tools will display significant recursion potential. It isn't necessary nor automatically economical to make our own leadscrews, bolts etc. but the capability to make those pieces is emergent from having a decent CNC lathe for more complex jobs.

Leo Dearden is championing filled epoxy construction for the heavy machine tools in the suite, along with graphite air bearings and a few other fascinating prospects. We're expecting to see a few approaches tried for most of the tools we'll need.

The three most promising and immediate tools, from my perspective, are all extensions of existing open source work. We're talking about a plastic extruder based on the RepRap project, with a chassis and conveyor built with long duty cycles and continuous automatic operation as primary goals. Also a ceramic, glass and metal powder printer based on CandyFab and retrofitting some readily available cheap inkjet printer.

Last but not least would indeed be a router table, probably based on Marcin's RepTab. Something which could be set up for routing with a mill or for plasma cutting. Eventually we'll add modular automatic hold-downs and a sheet feed for real automatic production. Possibly a laser as well; who knows what people will come up with?

There are many other tools being discussed, particularly a lathe, a mill, and a PCB assembly line. The three I mentioned have the great advantage of already existing in some form or another which we're able to access and adapt.

I thought your economic model was well thought out; I'll forward it to the RepLab list.