Last Friday, Professor Hod Lipson, who along with PhD student Evan Malone is the developer of the Fab@Home system, let us grill him about where home rapid prototyping is going.
Rapid prototyping has been industrially important for a while now, but the high cost of the machines -- $15,000 to $100,000 and up -- has kept it out of home use. A couple of recently announced projects aim to change that: Fab@Home, which we mentioned here, and RepRap, which we've mentioned at various times over the last few years.
Here at CRN, we are always interested in new technologies for programmable fabrication. Although it will be a few years before molecular manufacturing is working, near-future rapid prototyping systems may give us hints about some of the effects and implications of general-purpose manufacturing.
The first thing we asked Dr. Lipson was to compare Fab@Home and RepRap. The two have different goals: Fab@Home can be assembled with a soldering iron and a few screwdrivers, while RepRap needs actual machining skill. On the other hand, RepRap is expected to be considerably cheaper, in part because it can build many of its own parts (which Fab@Home doesn't try to do).
We spent a fair amount of time talking about the future: where technologies like Fab@Home are going. Dr. Lipson drew an interesting comparison between computers in the 1970's and rapid prototyping today. Mainframe computers had existed for quite a while, but the first hobby kit that allowed individuals to own a home computer made a huge difference. Even though that computer was very limited -- a few switches and lights -- it led rapidly to a number of commercial home computers, which rapidly sprouted keyboards, video displays, and other useful and user-friendly hardware. Thirty years or so from now, he expects material cartridges to be sold at Home Depot, active systems and electronics to be built into the products, and product ideas shared like music. (CRN expects molecular manufacturing to be more capable sooner than that. I asked him about molecular manufacturing and he said he wasn't in a position to have an opinion on it, which is a fair answer.)
They're currently studying incorporating an inkjet printing capability in the Fab@Home system. That would be useful for colored models, but also (in the future) for printing electronic circuits. They have already printed actuators (electroactive polymers) and batteries powerful enough to activate them. (For a description of this, and a look at near-future projections, see Dr. Lipson's paper "Homemade: The future of Functional Rapid Prototyping" (PDF).
A major problem that Dr. Lipson sees with today's rapid prototyping systems, aside from the cost, is that they are closed. They don't encourage experimentation -- substituting another material, to say nothing of tweaking the machine itself, can void the warranty. Fab@Home has been used to deposit everything from Play-Doh to chocolate, and the design is freely available.
One of the issues we're interested in at CRN is the availability of technologies that could lead to widespread capacity to develop or bootstrap molecular manufacturing. For this reason, I asked a general question about using hobby fabricators for building scientific apparatus such as microfluidics. The answer was that hobby-level RP probably will be limited in materials, accuracy, and surface finish, so would not be the tool of choice for making scientific equipment.
We're also interested in the ability of engineers to adapt to new materials and processes, since molecular manufacturing will certainly provide both. For this reason, I asked whether engineers experienced difficulty in designing products for rapid-prototyping systems. The answer was interesting in several ways. The different constraints of the process do indeed require "a few iterations" to adapt to. It didn't sound like an intractable problem for the average engineer. Dr. Lipson also said that engineers seemed likely to transition to debugging instead of thinking in advance -- building and testing rather than "think twice cut once." This is good in that it allows more complex products that can't easily be thought out -- but it's bad in that it encourages laziness. Rapid prototyping can allow a bad design to become entrenched, when designers continue to tweak and re-print it rather than thinking through the problem and abandoning the bad design. But rapid prototyping can also allow designers to explore new concepts more easily.
Thinking about the amount of operating system sophistication that was lost in transitioning from mainframes to PCs, I asked whether a similar loss of sophistication was a risk of rapid prototyping. Answer: Yes -- accuracy and speed will suffer, and low-cost machines will have even worse specs. On the other hand, a broad base of expertise and designs could be created. Dr. Lipson pointed out that BASIC, a computer language that was inelegant and limited, caught on because it was easy for beginners to use -- and people don't care about the accuracy of action figures.
A couple of final insights wrapped up the conversation. Dr. Lipson pointed out that as manufacturing has moved overseas, US manufacturers were keeping their business by turning to custom manufacturing. Rapid prototyping might continue that trend.
And just as the conversation was ending, it occurred to me that, just as growing vegetables for your own consumption is not counted in Gross Domestic Product, so manufacturing a product for your own use probably wouldn't be counted either. Dr. Lipson agreed, and said that the GDP measure would just have to change. He pointed out that intellectual property would also have to adapt. We have written frequently about the problems posed by intellectual property and other forms of information, and I hope that the intellectual property system will be capable of fostering innovation rather than protectionism once product designs can be developed by hobbyists. (It's worth repeating again: The US became world-class in software without any software patents until the mid-1980's, and rapid-prototyped product innovations will probably flourish best under a similar copyright-centered IP policy.)
The conversation did not make me re-think any of my opinions about the likely development trajectory or implications of molecular manufacturing. It was encouraging to see several of my projections echoed in a different but related field.
Tags: nanotechnology nanotech nano science technology ethics weblog blog
The machining requirement has become a thing of the past for both the RepRap and the spinoff Clanking Replicator project. I recently redesigned the extruder barrel for RepRap's Mk 2 extruder so that it could be made out of hard copper tubing with a bit of braising. That has worked out very well and the Mk 2 can handle several engineering plastics like HDPE, ABS and polypropylene in addition to the polycapralactone that it was able to extrude before.
Posted by: Forrest Higgs | March 05, 2007 at 09:43 PM
Forrest, thanks for the info. That's good news.
Do you know if anyone has yet tried to build Fab@Home parts with RepRap or vice versa? Is anyone even talking about producing a combined design?
Chris
Posted by: Chris Phoenix, CRN | March 06, 2007 at 08:31 AM
***Do you know if anyone has yet tried to build Fab@Home parts with RepRap or vice versa? Is anyone even talking about producing a combined design?***
LOL! I think that the question would be more one of why a person possessed of a RepRap or one of its spinoffs like Tommelise WANT to use their machine time to make a fab@home? Fab@home is considerably less sophisticated in both concept and technology, never mind much more expensive.
Posted by: Forrest Higgs | March 12, 2007 at 07:36 PM
Well, the Fab@Home project was started in Oct. '06, and it's already shipping. RepRap hopes to announce self-replication in 2008, according to its home page. Which means there's no way I can get one for another year or more.
Fab@Home, since it doesn't depend on FDM, might be a useful "seed" for RepRap's. A hybrid design approach might bring the cost down incrementally, or produce models that would be buildable with less skill.
If RepRap is as great as you say, then it should be able to undercut the current price of some Fab@Home parts, or increase their performance. In other words, RepRappers should be able to make money selling e.g. FDM heads on the Fab@Home website. That's a very practical reason to be interested in building F@H parts.
Chris
Posted by: Chris Phoenix, CRN | March 12, 2007 at 08:37 PM
***RepRap hopes to announce self-replication in 2008, according to its home page.***
The director of the RepRap Project is a very conservative person who doesn't like to raise expectations. Truth be known, though, a RepRap prototype in New Zealand, Zaphod, replicated it's first part way back in September at the Paraflows conference in Vienna. About a week ago Zaphod printed up a full set of parts for a Mk 2 FDM extruder. Since then they've gone into a new build of an extruder. It's been tested and it works fine.
***If RepRap is as great as you say, then it should be able to undercut the current price of some Fab@Home parts, or increase their performance.***
Again, the question is one of why anybody with a working RepRap want to make a fab@home. Why on Earth would anybody use a $400 machine to make parts for a less capable machine costing several thousand dollars?
While Darwin will be a very capable machine I'm at about the point with Tommelise that Vik Olivier down in New Zealand was back in September. Tommelise is a bootstrap self-replicating printer that anybody with a spot of woodworking skill and a few hand tools can make up for about $150-175 in parts. I've redesigned the RepRap FDM printer to use HDPE, ABS and polypropylene. When I get around to it I'll be qualifying it for PVC as well.
Allowing that there might be a market out there for fab@home after self-replicating printers like Darwin, Zaphod and Tommelise are operational I'm sure that fab@home enthusiasts will be able to buy parts from anybody who has a self-replicating printer and the spare machine time to make them. Who knows.
My feeling is that the fab@home froze their design process far too soon. Cornell should have spent more time qualifying a proper extrusion system and less time on flash and pizzaz. Full specs for the RepRap Mk 2 FDM extruder were published in November of 2005 as open source hardware. They certainly had the technology at their disposal.
Mind, that's just my pedestrian opinion. Fab@home touted as "innovative technology" puzzles me greatly. It could have been so much more with so little extra effort. :-s
Posted by: Forrest Higgs | March 13, 2007 at 07:24 PM
Is Fab@Home design frozen? It's open source, and I assume they wouldn't reject improvements. They have a wiki, after all.
Good news that RepRap is probably closer than 2008 and there's a seed version.
In my talk today, I mentioned that RepRap could do 3 or 4 types of plastic, and one of the audience members said, "Yes, but I think it should be able to do metal too. The practical problems should be solvable..." Turns out that he's been researching FDM metal.
Chris
Posted by: Chris Phoenix, CRN | March 14, 2007 at 09:39 PM
***Good news that RepRap is probably closer than 2008 and there's a seed version.***
It's a little better than that. Adrian Bowyer is working on a release version of a RepRap machine that he plans on releasing called Darwin. It's a pretty complete design, not a bootstrap like my Tommelise design. I'd guess it will be running nicely about May-June, but it could well take the rest of the year before on-line documentation reaches a point where a fairly ordinary enthusiast could make one up at home. Mind, one guy is bootstrapping the basic connector blocks out of wood, I think, so Darwin could spread pretty quickly.
The Mk II extruder parts kit was made out of wood successfully last year by a guy named James Wilkins.
***In my talk today, I mentioned that RepRap could do 3 or 4 types of plastic, and one of the audience members said, "Yes, but I think it should be able to do metal too.***
That's not precisely true. Tommelise, a RepRap spinoff (Clanking Replicator Project) has extruded ABS, HPP, HDPE and polycapralactone extrusion plus some halting first steps at extruding common 60/40 tin/lead solder for circuitry and spackling compound for structural support.
RepRap per se, to the best of my knowledge, has extruded polycapralactone (ages ago), Wood's metal for circuitry, and spackling compound (Polyfilla) for support material.
I sent some HPP and HDPE filament over to the hard-core RepRap folks at Bath University and in New Zealand so that they could test it in their Mk II FDM extruder, but to date I've not heard if they've had time to try it out.
***Turns out that he's been researching FDM metal.***
COOL! I'm hammering that idea, too, in my very limited spare time. Did you get his name? :-)
I think a takeoff on the RepRap Mk II extruder could FDM metal traces. I've scratched out a new extruder barrel and thermal barrier design that I hope will let that happen.
Posted by: Forrest Higgs | March 15, 2007 at 02:09 PM
The products of fabbers I've seen so far don't look terribly precise. Would it really be worse to think in terms of a milling machine, for significantly more precise results? Even if you're thinking recycling plastic, you could melt it down into a block to be milled.
Also, in order to make things of metal, maybe they should be concentrating on something like making casting molds out of clay or plaster. It's not a one-step process, but clay is pretty universally available, and melting down scrap aluminum isn't terribly high-tech. A "milling tip" for clay could be made of aluminum, allowing the loop to be closed.
Posted by: Tom Craver | March 15, 2007 at 07:53 PM
Several people working in the various teams at RepRap are hot to use RepRap to make molds either directly or to make plastic mockups that can be used to make sand casting molds.
Personally, working with orange-hot crucibles of molten aluminum isn't my preferred pastime. I'm quite interested in high strength ceramics like silicon nitride. They can be extruded as slurries directly or cast in plastic molds. The people at Stanford use a hacked kitchen microwave oven to fire them. I can get right alongside that sort of thing. :-)
http://www-rpl.stanford.edu/files/paper/2000/m2000a.pdf
Posted by: Forrest Higgs | March 15, 2007 at 11:45 PM
dear sir
This is Dr. V. Veeranna, M.Tech in manufacturing technology and Doctareate in Flexible manufacturing system. Presently working as Associate Prefessor in Engineering college. Please give the information for
1. Type of components that can be manufacture at home using Rapid prototyping.
2. Is it possible for addopt smale scale industry
3. Is it gives monitary benifit
Posted by: Dr.V.Veeranna | March 16, 2007 at 10:50 PM
***1. Type of components that can be manufacture at home using Rapid prototyping.***
You can manufacture components that you would be able to make on a commercial fused deposition modelling 3D printer like those made by the firm Stratasys.
***2. Is it possible for addopt smale scale industry***
I'm fairly certain that that will be more than possible.
***3. Is it gives monitary benifit***
For small production runs I think that it definitely will be cost effective. If you are looking to make hundreds of thousands of something, however, I do not think that it will compete with commercial injection moulding machines.
What I see the technology doing is enabling virtually anybody to be able to do product design and small production runs out of their home or small workshop. This, I think, will change the whole nature of how products are designed, who designs them, and which ones see large scale serial production runs.
Posted by: Forrest Higgs | March 17, 2007 at 12:56 PM