Hi, Im a newbie here and to RP as well, i think this project is something quite special or at least could be if given enough support and progress but it seems the project has been in the works for a few years now so im wondering when you lot reckon its likely to really take off?

I've not read up on everything yet but am i right in thinking these machines currently take ages to make even small objects?

If so is this just early design problems or some limitation to the heating and cooling of the plastics?

It will take off once we achieve self-replication - the 1.0 release.
Until we have machines making machines making machines, the project doesn't seem important to most people. When we have 1-2-4-8...2^N of them out there, people will start to care, and every engineering student or hardware geek who wants one will build one.

--------

Regarding taking ages to make even small objects, our depostion rate is roughly ~7 cc / hr (.4 inch^3 /hour). I think we've improved that figure, but we haven't doubled it. The extruder has to force thick gooey plastic through a small orifice, doubling or quadrupling the pressure doesn't double or quadrupling the amount extruded, but it does burn out the motor or pop the extrusion nozzle off. There are inherent limitations to a plastic-extruding mechanism, I think, especially one built with some plastic components.

We've talked about a larger-orifice bulk extrusion head, but no one has mucked around with that yet.

My suspicion is that there is a barrier to entry right now for a lot of people because not everybody who is interested in the applications of the reprap are interested in the process of making a repstrap. I think once there is a kit of parts that can be built with a set of tools that you would expect a reasonable person to own, it'll start gaining momentum.

Some people enjoy the journey, some the destination.

The answer to the "when will this take off" question depends primarily on what you mean by "take off." The Wright Brothers took off in 1903 but it took another 25 years for Lindbergh to cross the Atlantic and then 50 years after that for the 747 to make it widely affordable.

I've been trying to make heads or tails of RepRap, Fab@Home and their ilk since micro-manufacturing is an area I'm very interested in, and I have a lot of conflicting thoughts. At one end of the spectrum we have molecular fabrication, which, given the present state of technology, is not unlike Jules Verne writing about a manned trip to the moon in 1860. All we can do today is say why we can't do it, it's going to take some decades of basic research and many billions of dollars before it becomes a reality, assuming that it goes the way of the microchip rather than the nuclear-powered car.

At the other end, you have conventional CNC lathes and mills, which are by now a mature and well-understood teachnology. However, they are very, very limited in terms of their capabilities however you look at it. FDM and EDM prototyping machines are, for all practical purposes, a small step above these, and share the vast majority of their limitations. The problem with RepRap right now is that it's incapable of manufacturing anything capable of justifying the cost of building the machine to begin with. This means you either need to make it a lot cheaper to build, or capable of building much more interesting things.

The problem that I see is that true self-replication is infinitely far away at this point, at least if you look at it in terms of cost, which is what matters most. Look at the parts that account for the cost of a RepRap or Fab@Home device, and you'll see things like stepper motors, power supplies, or lead screws, to name a few. If you need $2000 worth of parts to build a machine, and you can self-replicate $200 of those parts with $10 of raw material, then the best that you will ever do is to lower the cost of the machine to $1810. I'm pulling numbers out of the air, and it would be interesting to see a more rigorous analysis.

Ultimately, the hope that I have for RepRap & co. is that they will help us to discover unlikely applications for devices like this, which will in turn inspire investment in the sort of engineering needed to advance the state of the art more dramatically.

As an aside, it's interesting to consider the example here of digital photography and printing. Five years ago a high-end inkjet would equal or exceed the quality of store prints, often at an equal or lower cost per print. But now, especially for larger prints, even your local Target or Wal-Mart store likely has a very good large-format printer in the store that can beat all consumer devices on both quality and cost. It's really tough to compete with a machine that works for a living.

"I'm pulling numbers out of the air, and it would be interesting to see a more rigorous analysis. "

Yeah, I think that you are pulling numbers out of the air, all right. There've been some pretty rigorous analyses done around here which have convinced most people working on RepRap that replication of much of a 3D printer is quite feasable indeed.

Colin K. wrote:
-------------------------------------------------------
> The problem with RepRap right
> now is that it's incapable of manufacturing
> anything capable of justifying the cost of
> building the machine to begin with. This means you
> either need to make it a lot cheaper to build, or
> capable of building much more interesting things.

Are all research projects expected to have a positive ROI?!

For "right now", you may be correct, since it seems to be costing the first few builders about US$2700, using commercial RP for the replicable parts.

Once the thing can replicate those parts itself, we're looking at a US$400 parts cost. That's a 6.75x cost reduction. Not too shabby.

> If you need $2000 worth of parts to
> build a machine, and you can self-replicate $200
> of those parts with $10 of raw material, then the
> best that you will ever do is to lower the cost of
> the machine to $1810. I'm pulling numbers out of
> the air, and it would be interesting to see a more
> rigorous analysis.

I'd say for Reprap Darwin (once replicating the RP parts) you're looking at maybe US$400 of parts, about US$200 of which is stepper motors. So we're looking to self-replicate about US$2300 of parts for say US$20 (or so). Plug the RP-able .stl files from the RepRap project into an online RP quote setup such as the one at [www.alphaprototypes.com] and you'll see the US$2300 or so number. I haven't plugged them all in myself, but EricM did; I'm using his number here.

Lowering the remaining US$400ish cost significantly *is* going to be a challenge, but a US$400 investment to start a little cottage industry "printing" small items such as cups, plates, doorstops, bookends, containers for food and water, ... hmmm, suddenly it doesn't seem *too* far from being viable, as least as a hobby rather than a primary income source. Many homes in affluent nations own US$400 (or more) PCs already. They are not expected to turn a profit, either.

The Altair home computer in 1970 was very hard to justify in ROI terms too, but hobbyists bought and constructed them. By 1977 there was the Apple II, by 1981 the IBM PC... rapid evolution leading quickly to positive ROI for real, practical use, both in business and in the home. Right now, Reprap Darwin is a research project. No more, and no less. A decade from now, there is a possibility that its descendants will be much more widespread and more significant than that.

> It's really tough to compete with a machine
> that works for a living.

Only true if the machine has low cost widespread alternatives that drive down the cost per item produced the commercial outfit can charge (home inkjet printers, in your example). [Furthermore, during the last five years, faster home Internet connections and higher resolution screens mean that fewer images *need* to be printed on paper; people share them online instead, and archive them on CD or DVD, so lowering the market value of paper prints, at least for mass market "snapshot" photos! Flickr [www.flickr.com] competes very well with the Walmart printers on cost!] Yes, if Stratasys machines were in every Walmart and the massive markup on their filament material were absent, the economics, at least for those living in the West near a Walmart (!), would be very different. But that is not the current situation, even for the small fraction of the world's population that does live near a Walmart! :-)

Jonathan

Hmmm... uses for a RepRap that (may) simultaneously drive an industry to its knees or create one

-Any game that uses collectible figures
-Robo-One and related
-Architectural conceptual models (my wife would kill for this)
-Kids toys (and not quite kids toys, ever see what people are doing with Nerf these days?)
-That blasted battery door to my TV remote that broke and now I can't even find it
-Custom party favors (do NOT underestimate the power of a hostess who can't find EXACTLY what she wants)

Nothing HUGE or earth shattering, but each could be a somewhat healthy market, too. Once there's a foot in the door, who knows what else may show up. RepRap made hardware has to be looked at in a specific light. It's somewhat specialized and should mechanically be looked at more as a special alligned plastic-plastic filament composite in the current incarnation. It's not going to be perfect for everything, but there's nothing saying it can't be used to make a different machine that can correct these deficiencies.

Rapid prototyping is also somewhat of a misnomer. Nothing can beat the speed of a 1000 ton injection molder for plastic creation. However, it's when changes or just one-ofs are needed that it will be faster. It takes a lot less money, time, and effort to use a RepRap to build, say, a custom mobile phone casing than it would to do injection molding from start to finish. Doubly so if there is a change that's needed. Heck, if you just wanted a mug with raised lettering saying "World's Greatest ... Something" it would be worth something in this kind of field.

SOI Sentinel Wrote:
-------------------------------------------------------

> Rapid prototyping is also somewhat of a misnomer.
> Nothing can beat the speed of a 1000 ton injection
> molder for plastic creation.

If you have a mold built for that 100 ton injection molder you have gone far beyond building prototypes.

~Seth

Forrest, perhaps you could point me to some of those analyses because while I saw a lot of great experimenting and research, I did not find anything that changed my opinion.

I did look at this, though, which I think suggests I'm at least not wholly out of place:

[blog.reprap.org]

Sure, you can fab 55% of the parts count, but what do those parts represent in terms of cost? You could probably manufacture all the plastic parts from raw stock in the garage pretty cheaply, so the Von Neumann efficiencies are not of primary significance, though not unimportant.

In any case, self-replication is not inherently magical. A stone chisel can be used to make copies of itself, but not an MP3 player. But, if a RepRap can be used to make a better tool, then it, like the stone chisel, will move the state of the art forward, and that is what counts. I would *like* to have my opinion changed, just like I'd like to have my flying atomic powered car, thank you very much

Think about all the extra effort and time that is required to mine the steel, transport and refine it, then machine the tool, transport the plastic material and make the parts, pack the parts and ship them, then maintain and repair the tool and finally dispose of it somehow.

With a reprap, you just transport roughly the right amount of raw plastic material directly to the point of use - a significant saving of time, effort and cost.

Once you forget about trying to apply the rationale for economies of scale, the options appear a litle different...

"Lowering the remaining US$400ish cost significantly *is* going to be a challenge"

Not hardly. I'm looking at a $150-175 parts cost. I'm getting there basically by using one controller board instead of four and gearmotor/shaft encoders instead of steppers. I've also brought the energy consumption way down, too.

"-That blasted battery door to my TV remote that broke and now I can't even find it "

My pet peeve is the little HDPE snap-on lid to the tank that holds the windscreen washing liquid on my jeep. Mine cracked. It's about 50 mm diameter, 10 mm high and weighs maybe 5-6 grammes. Replacement cost at the dealership? $5.50! One of my first printing projects will be to print one of those.

"Forrest, perhaps you could point me to some of those analyses because while I saw a lot of great experimenting and research, I did not find anything that changed my opinion. "

Colin, the developer group have passed around numbers and calculations in the internal email round robin for over a year now. As time has gone on and we have fewer imponderables to guess at the numbers have got more and more reliable.

The analyses aren't that difficult to do.You can do them yourself if you don't trust what we say.

For Darwin, the non-printed parts list is running about US$400. For Tommelise, the last time I checked, it was about US$134. While I haven't really formalised a bill of materials (BOM) for Tommelise, Darwin's BOM is pretty easy to construct from the information on the RepRap site.

Keep in mind that the parts estimates don't include the tools you need to put them together. That's an important number that isn't usually included in the cost estimates.

Now, let's get to the printed parts. For Darwin, if I remember correctly Ed was talking about it taking about 1.25 kg of plastic. For Tommelise, a complete plastic replacement of wood would take maybe 3 kg. I don't use as much steel rods as Darwin does in my Tommelise.

Pardon me if I convert things to pounds and cubic inches, because that's how things are figured here in the US.

Darwin requires about 2.75 pounds of printed plastic while Tommelise requires about 6.6. Those numbers translate to 80 cubic inches of printed plastic for Darwin and about 190 cubic inches for Tommelise.

First off, if you go to a commercial 3D printer they're going to charge you about $30/cubic inch. That means that the printed parts for Darwin are going to cost you about $2,400 while Tommelise's will cost you about $5,700.

From that it's pretty clear that the number of people who trundle off to have the STL's for Tommelise and Darwin made at commercial printers is going to be pretty small. What IS happening is that people like EricM are bootstrapping printers and having the parts kit for the Mk II extruder printed commercially. That is just about cost-effective.

What both designs depend on is having a Darwin or Tommelise to make parts. When that is the case you are talking about a cost which asymptotically approaches the cost of the plastic filament.

Right now you can buy HDPE filament for about $4.50/lb. ABS costs about a dollar per pound more, if I remember correctly. That means that the filament to make the printed parts for a Darwin will cost you about $12.50 while for a Tommelise the cost will be closer to $30.

One big caveat to the whole RepRap project is that we do not know whether we can reliably print big objects at ambient temperatures using the Mk II fused deposition modelling extruder. 3D printers like those Stratasys makes depend on the printing taking place in an oven kept just a few degrees below the melting point of the plastic. That means that you can print a whole object before you cool it off and makes the handling of shrinkage and warping considerably simpler although it massively increases the cost of the printer.

Vik and I have been slowly building up the know-how to be able to print things at ambient temperatures. That kind of know-how hasn't been coming easily. I'm beginning to think that it is going to take me longer to figure that out than it took me to design and build Tommelise.

Edited 3 time(s). Last edit at 05/28/2007 01:14PM by Forrest Higgs.

As some one else has said, as soon as People can get there hands on the rp parts, it'll start growing fast, then when people can buy a complete kit assembled or un-assembled it'll take of fully.

"Are all research projects expected to have a positive ROI?!"

No, absolutely not. But if they don't have a positive ROI then it's very difficult to motivate private individuals to get very involved. Since a lot of us are hobbiests, it's important to acknowledge the hobbiest budget.

In this case, I think the RepRap crew has done an excellent job of doing just that... And of course the project itself lends itself exceptionally well to this goal.

~D

Forrest: I am inclined to agree that this is a materials science problem first and an electrical/mechanical engineering one second. I've been debating with myself over whether a volunteer effort like this is well-suited to take on these sorts of challenges. The interesting thing is that a large part of science is simply rolling marbles down ramps until something interesting happens.

To the extent that a lot of this marble-rolling doesn't require anything more than good record-keeping and experiment design, it is very much within the reach of serious hobbyists to do significant work. But, I do think it requires a lot more discipline than the average OS software project, as you can't necessarily just keep nudging the code until it sort of works. The tolerance for bugginess is simply going to be much lower working in a physical domain.

Now, as for the price of various machines, the RepRap-Fab@Home comparison is interesting. Raw parts for Fab@Home look like ~$400 if you order their kit. The cost is low as it's simply waterjet-cut acrylic sheet, which is a much lower cost process than conventional rapid prototyping.

Another interesting difference is that F@H uses a lot more precision mechanical parts, especially ball screws for linear positioning. No matter how you slice it, these will be somewhat costly. But, they are somewhat of a sure thing from an engineering perspective. The question is whether they are overkill for the application. If RepRap's drivetrain proves sufficiently capable from a precision/repeatability perspective, it will enjoy a major advantage here.

In terms of cost, I wonder how much you could cut if you would accept a smaller working size? 300mm^3 is a huge volume, especially considering the rates of material deposition we're working with. Dropping down to a 150mm cube, you can maybe use cheaper steppers and gear them, and get rid of a lot of hardware without sacrficing accuracy.

michaeljoffe Wrote:
> Once you forget about trying to apply the
> rationale for economies of scale, the options
> appear a litle different...

Yes and no. My father does architectural woodworking, and I can have him build me anything I want for the just the cost of materials, so it's sort of like having a universal furniture replicator

And yet, nearly all the stuff in my apartment is from Ikea, because I can buy finished pieces from them for less than it would cost my father just to buy the wood and hardware. Of course, it's child's play for my dad to make a piece an inch shorter or wider or whatever. If you don't mind getting your car in black, the global supply chain is astoundingly efficient.

In terms of where a RepRap-type device can first achieve real utility, it's going to be in areas where customization is valuable. I still think the killer app for micro-manufacturing is clothing, because (a) none of us are the same size and (b) there is a high value placed on individual styling. The point here isn't to make a pair of jeans cheaper than ones made in China, but to make a pair that fits much more nicely and looks exactly the way you want, and is therefore worth more money.

"But, I do think it requires a lot more discipline than the average OS software project, as you can't necessarily just keep nudging the code until it sort of works."

Do I get the impression that you haven't actually ever automated an instrument? That is EXACTLY what you have to do. I did my first one, turning an beam clipping infrared thermometer into a sky mapping system for the 12-14 micron band, in 1980.

The only difference between 1980 and now is that whereas it cost the Swedish gov't more than US10K for that sky mapper then I could do the same thing now for maybe $250 or less using the sorts of dirt cheap microcontrollers that you can buy over the web these days.

Forrest, I think Colin K. is correct. RepRap is more complicated than a software project because the "make" process involves drilling, soldering, and so on. This is much more difficult than installing and running software "configure ... make ... make install ... run", which usually does not involve problems like hunting down itermittent electrical faults or having n-1 bolts and n bolt holes.

Mind you, for an experienced fabricator and tinker, it's not too bad.

"nudge it until it works" means different things to different people. In software, it often means fiddling with the code until the exception message disappears and the user interface does something reasonably logical. Much less frequently, it means "now let's go back and run the suite of tests to see if that accounts for all the known possible scenarios."

We tolerate bugginess in software because the consequences in terms of lost time/money/sanity/human lives are _usually_ relatively low. More so than many OSS projects, bugs in RepRap will cost people non-trivial amounts of time or money. If you have to wait three hours to test out a part, and it's flawed, that's a much bigger setback than running "make test" and seeing a failure message within a few minutes.

To Sebastien's point, we are also getting into more complex problems with weird variables. I recently built a piece of audio gear which had perhaps eight parts, only one of which had more than three leads. It took me rebuilding the thing about ten times to get it right because I was misreading the schematic and because until I picked up a used o-scope, I was working in the blind.

From the perspective of the original question--"When is RepRap likely to take off"--a big part of that will be driven by reliability, when a hobbyist can spend most of his time focusing on making things with their RepRap, as opposed to making their RepRap work.

Has anybody tried using the reprap in the oven?

Haven't got an oven that big and making everything in Tommelise oven proof would make it very expensive indeed.

I've been thinking an oven could probably be built around a reprap, but I don't think I'd want to try it. The idea of putting the corner brackets in an oven, for the express purpose of softening/keeping hot the very material they were made from, seems less than wise.

What about infra-red lamps? could they heat the piece enough?
Directing heat from a light source is always easier than with an oven. It would save power and make it more manageable than an enclosed oven.

Although all that heat close to electronic components... And the room would ebventually get hot also.

Never mind your electric bill.

SebastienBailard Wrote:
-------------------------------------------------------
> Forrest, I think Colin K. is correct. RepRap is
> more complicated than a software project because
> the "make" process involves drilling, soldering,
> and so on. This is much more difficult than
> installing and running software "configure ...
> make ... make install ... run", which usually does
> not involve problems like hunting down itermittent
> electrical faults or having n-1 bolts and n bolt
> holes.
>
> Mind you, for an experienced fabricator and
> tinker, it's not too bad.

Mind you, it had become so simple only due to lot of effort to write the tools like configure and make. Without them, compiling a source code distribution was a nightmare.

You are comparing apples to oranges here.

If RepRap could reliably print plastic and metal, had robotic manipulators to assemble stuff, this comparison would be more fair.

Also, note that software runs on a computer, which is an incredibly complex piece of engineering. Computer needs a compiler, which is another complex thing, that took insane effort to get it right.

You can compare a modern computer+ a good bunch of system code to the tools needed to assemble a RepRap and see which is simpler

Don't be discouraged by any of the complexity/cost factors. It's hard to get the ball rolling, but it is unstoppable in the end.

It will be rolling since tinkering is much addictive than TV, sex and computer games to exactly the people most capable and suited to get this done. This and similar projects are doomed to success.

The hurdles you encounter are the same FOSS movement had. The dynamics are the same, and you know the result. Just need to get your analogies straight to see this.

You mention heat lamps.

I was wondering, actually, if down the road the thing could be made in a heat mover.
Since the thing will eventually override the insulation, and heat up the room, it'd be nice if that's where it got its heat from.
Think of a refrigerator...built backwards. The hot coils on the inside.

Still, I think this is something to consider only after we're using materials that don't deform under the conditions being used. Ceramics, perhaps. Metals, maybe.

It seems like there are two visions of "take off" here. One in which the device becomes readily available to people with basic technical knowledge as a hobby, and one in which the device becomes an appliance for widespread distribution of physical objects. Frankly, it's difficult to envision the latter, even if repraps are given away for free, such that the cost of producing objects is reduced to the cost of material and the time required to print. It's just too easy and fast to mass produce plastic parts by injection molding, at unit costs (in time and money) barely more than the cost of plastic. If you ask people to choose between RP'ing four plastic plates (in one color and not dishwasher safe) over six hours at a cost of $15/pound, and driving 15 minutes to WalMart to buy four plastic plates (in an attractive array of patterns) for $1.50 each, few people will have to think very hard. I can imagine greatly improved RP being useful for people in extremely isolated locations, where it might be convenient to stockpile raw material rather than specific parts.

The former sense of "take off," to become widely availabe for hobbyists is much more reasonable. There are a lot of creative people out there, and the ability to share the fruits of that creativity, in ones or twos, is very powerful, and people will pay a premium in time or money for that creativity. $150, even $500, is not unreasonable as a hobbyist tool-your computer was more, it's the same range as hobbyist power tools, less than a lot of specialist machinery. There's always going to be a limit to the size of object that's practical to produce this way, though, and it's always going to be more expensive to produce 100,000 by RP than by conventional manufacturing. These fundamental limitations restrict in-house RP to fabrication of small-batch, small-size, probably customized parts.

keep this in mind though: when the PC was starting out in the 80's it was slow, clunky, and couldn't do much. now look at how amazing these things are. i imagine the same thing will happen with our machine as it grows and evolves. its hard to predict what will happen with disruptive technology like this (see also: internet)

also, if we end up using HDPE as our main material (it looks like a good candidate...) then the things *will* be dishwasher safe =)