Getting It Right By Getting It Wrong: RepRap And The Evolution Of 3D Printing

The beginning of the DIY 3D printing movement was a heady time. There was a vision of a post-scarcity world in which everything could and would be made at home, for free. Printers printing other printers would ensure the exponential growth that would put a 3D printer in every home. As it says on the front page: “RepRap is humanity’s first general-purpose self-replicating manufacturing machine.” Well, kinda.

Original Darwin. Photo: Adrian Bowyer, founder of RepRap.

Just to set the record straight, I love the RepRap project. My hackerspace put our funds together to build one of the first few Darwins in the US: Zach “Hoeken” came down and delivered the cut-acrylic pieces in person. I have, sitting on my desk, a Prusa Mendel with 3D parts printed by Joseph Prusa himself, and I spent a fantastic weekend with him and Kliment Yanev (author of Pronterface) putting it together. Most everyone I’ve met in the RepRap community has been awesome, giving, and talented. The overarching goal of RepRap — getting 3D printers in as many peoples’ hands as possible — is worthy.

But one foundational RepRap idea(l) is wrong, and unfortunately it’s in the name: replication. The original plan was that RepRap printers would print other printers and soon everyone on Earth would have one. In reality, an infinitesimal percentage of RepRap owners print other printers, and the cost of a mass-produced, commercial RepRap spinoff is much less than it would cost me to print you one and source the parts. Because of economies of scale, replicating 3D printers one at a time is just wasteful. Five years ago, this was a controversial stance in the community.

On the other hand, the openness of the RepRap community has fostered great advances in the state of the DIY 3D printing art. Printers haven’t reproduced like wildfire, but ideas and designs have. It’s time to look back on the ideal of literal replication and realize that the replication of designs, building methods, and the software that drives the RepRap project is its great success. It’s the Open Hardware, smarty! A corollary of this shift in thought is to use whatever materials are at hand that make experimentation with new designs as easy as possible, including embracing cheap mass-produced machines as a first step. The number of RepRaps may never grow exponentially, but the quality and number of RepRap designs can.

The Exponential Argument and the Economics of Replication

One of the key early design goals for the RepRap was to be able to replicate quickly, which would lead to the exponential growth of RepRaps around the globe. Adrian Bowyer, the founder of the RepRap project, gave a keynote address at the Seventh National Conference on Rapid Design, Prototyping & Manufacturing

All current engineering production generates goods in an arithmetic progression. Sometimes this is very fast − suppose an injection moulding machine makes plastic combs at the rate of 10,000 an hour. Suppose further that a RepRap machine can make one copy of itself a day, and also just one comb. After merely 18 days, the RepRap machines will be making more combs than the injection moulder, assuming people give them house-room.

Ten years later, the RepRap project is a phenomenal success — there are RepRap printers in the hands of hobbyists everywhere and the advances in the later models have come entirely from that community. But RepRap is by no means an exponential success. At any of Adrian’s proposed rates of replication, there would be a bazillion times more RepRaps by now than there are atoms in the known universe. I bet you even know some people who don’t own one. Proof by contradiction.

The bot does all the work. Simon Kühling just chills.
The bot does all the work. Simon Kühling just chills.

What happened? Almost nobody is replicating. The missing ingredient in the exponential-growth formulation is the number of people driving the machines (never mind the fact that you’re looking at a non-negotiable one-time outlay of a few hundred dollars for motors and electronics). Running a 3D printer of any design or make, as anyone who shepherds one knows, requires a deal of maintenance.

Almost nobody is printing RepRap parts because they are, like I am, trying to keep the machine working while printing the cool stuff that they wanted in the first place. Or they’re printing new and improved parts for their current machine. Or they’re printing the parts for their next, improved, machine. But they’re not wasting their time simply replicating.

Ironically, in that same keynote speech, Prof. Bowyer gets this part of the argument right but doesn’t follow its logical conclusion. He mentions that, as more people started printing parts for RepRaps, the price of these parts would drop to the material cost, which is nearly zero. This means free RepRap parts for everyone, right? Nope, it means that as long as anyone has anything else to do with their time that they value more than printing RepRap parts, they’ll do that because there’s just no money in printing RepRap parts.

A very nice set of parts from RepRapParts, Germany

And this is exactly what happened. Four years ago, when the Prusa Mendel design was busting out onto the scene, a set of printed plastic parts cost around $120. Two years later, it was more like $40 or $60. Now, even cheaper parts can be obtained for something like $10-$20, but those sets that are imported from overseas where the cost of one’s time, as measured in the average hourly wage, is a lot lower than it is here. At that price, for a few hours of running my machine, including the cost of plastic and maintenance, I’m not replicating.

The biggest misunderstanding of the RepRap project is that people’s time isn’t free and unbounded. A large number of folks just want to print something with their 3D printer, and they want it to print nicely without significant tuning. That’s why we see a wide variety of machines that will never be able to replicate themselves for sale in the $1200 – $2500 range, when anyone can build a RepRap themselves for around $500 or less. The expensive machines are being built for out-of-the-box usability, and the purchasers of these machines are trading money upfront for the time that they would otherwise have to spend tweaking and maintaining the machine.

At the same time, inexpensive mass-produced machines based on the RepRap design compete with replicated machines at the lower end of the market. Simply put, economies of scale make the average cost of producing a RepRap lower as the quantity produced goes up. It makes absolutely no sense for me to replicate plastic printer parts when my potential buyer can get everything for the entire build cheaper from overseas. And I’m not complaining. This is a good thing because it means more machines, faster, in the hands of the DIY experimenters who will make the next breakthroughs.

Ecology and the RepRap

The original Mendel: Threaded rods and printed vertexes. Photo: Adrian Bowyer

Another part of the RepRap’s founding inspiration was biological, and this was implicated in the view the RepRaps had to replicate. After all, all biological life on the Earth perpetuates itself by reproduction. (Deep thought: reproduction is what life is.) So the RepRap in Bowyer’s ideal vision should be made out of parts that it would be able to make. This lead to early design choices including the use of the dreaded, threaded-rod frame, and an emphasis on simplicity and fab-ability over precision. The reliance on non-printable parts, “vitamins” in the biological metaphor, was to be minimized even if that meant making a less robust machine that was fiddly to calibrate.

In insisting on self-reliance, Bowyer misunderstood the ecology of the RepRap. Do bees produce everything they need to survive? No, the flowers help them out a lot. Flowers don’t exist in a self-reproducing vacuum either — they rely on bees and other insects to reproduce. The point is that nothing in nature is self-reliant, even for reproduction. Everything relies on something else. Why shouldn’t a RepRap evolve to be the cheapest and best possible device given the parts available, printed or otherwise? (A factory turning out inexpensive aluminum extrusions and shipping them across the globe is the flower in this twisted metaphor.)

MendelMax: A Mendel made of aluminum extrusion. Slightly more expensive, significantly more rigid. Photo: Kludgineer

Our DIY ecosystem has become rich in “vitamins” since the founding of the RepRap project. It’s now easy and cheap to get even relatively sophisticated parts like aluminum extrusions and linear motion slides direct from factories in China. One-off PCBs used to be prohibitively expensive to produce, but that’s no longer the case. Motor driver circuits, motors, and even niche parts have become cheaper. In a nutrient-rich environment, it makes sense biologically to evolve to take advantage of them.

Fortunately, insisting on building a frame yourself entirely out of threaded rod has become an anachronism — all of the new printer designs since the Mendel 90 to the Prusa i3 use an easy-to-calibrate design that’s based on large cutouts of metal or plastic for parts of the structure that require stiffness. One of the first salvos in this revolution was fired by the MendelMax, which is essentially a Mendel made out of extrusions. Then came the various delta bots, epitomized by the Rostock and its brother, the Kossel, that left replicability entirely behind.

I’m enthusiastic to see the directions that the community has taken over the last five years or so — relying as they do on more “vitamins” and resulting in higher quality and less time wasted in calibration, tuning, and maintenance. It’s to the point now that the use of aluminum extrusions isn’t even given a second though.


One of the important early ideals of the RepRap movement — universal replication and free printers for everyone — was pretty much a failure.  Still, to quote again from the website: “RepRap was the first of the low-cost 3D printers, and the RepRap Project started the open-source 3D printer revolution. It has become the most widely-used 3D printer among the global members of the Maker Community.” This is all true, and then some. RepRap has spawned a truly global community of enthusiasts all working on advancing the state of the art in DIY 3D printing. The number and quality of people working on the project is really amazing.

Rostock printer. Photo: Johann C. Rocholl -- its inventor.
Rostock printer. Photo: Johann C. Rocholl — its inventor.

But in my mind the biggest advances have been in the designs of the machines themselves. A decent 90-degree-frame RepRap-style bot today is significantly better than a Prusa of only four years ago, and infinitely better than the original Darwin of ten years ago. And it costs around half the price to assemble one. In addition, the ecosystem has grown to include out-there designs like the delta bots. All of these machines are awesome, but none of them are replicating to any serious degree, even the ones that could in principle. But who cares?

What’s awesome about RepRap is that, even though the founding replication idea behind RepRap has been a failure, the community and the machines that they’ve made have adapted to our “vitamin”-rich environment. Perhaps the utopian story of the replication-driven end to scarcity was useful or necessary in attracting people to the project in its early days. But ten years out, the community has already proved its value, and that’s not in making infinite numbers of printers, but in making shared, DIY design innovations: the nature of RepRap evolution has itself evolved.

Headline image courtesy [ccecil], design by [loubie]. And thanks to Freenode’s #reprap channel for the good discussion.

68 thoughts on “Getting It Right By Getting It Wrong: RepRap And The Evolution Of 3D Printing

  1. Another thing that astounds me is that the price keeps dropping but the specs don’t change much. It cost me exactly as much to build my first 3D printer – a Prusa Mendel from scratch as it did for me to buy a Stratasys Dimension on the used market five years later. FDM seems to have hit its limits in terms of resolution – where to go from here?

    1. Um, no FDM has not met its limits. If you think in the traditional RepRap train of thought, sure, but with an educated mechanical engineer, designing along the lines of absolute highest resolution, can do far better than even the $200,000.00 Stratasys machines. One caveat to high resolution FDM is speed. When it takes a magnitude greater time to print the same part, albeit much better looking , most dont want to bother.

      I have a machine based on the CoreXY system, with a cable drive system with repeat-ability down to .005mm, high end (NSK) ball screw Z , 1mm filament (custom) , a .1mm nozzle , and layer heights of down to .01mm. Capable of smaller parts than a lot of SLA machines , and parts are practically indistinguishable from SLA. No, FDM has NOT met its limits. Only a non-engineer would say that.

      However, even tiny prints take a very long time to print at such “resolution”. I am OK with that though, as it would still take longer to do it with any other method, but most people want faster , not slower , and the expense of quality seems to be OK with them. I laugh at what most people call a “great print”. If they only knew.

      Also, everybody knows you have a Dimension now. You mention it in just about every post. We get it, your special(certainly), but you don’t have to incessantly repeat it. Most people will not have that option, so suggesting, as I have seen you imply in other posts, that is what everybody should do , is silly.

      BTW, I agree with the article. The idea that 3D printer will/will have been able to print themselves, and the notion that they are the first machine able to do so is absolutely stupid. It cannot replicate itself , not-even-close. The only machine in existence today that can ACTUALLY replicate itself ENTIRELY, is a lathe. Yes , a lathe is the only machine that can make another fully functioning lathe all on that one machine.

        1. Which is exactly why I have been saying for years that FDM, while great to get people interested in 3D printing, does not approach the speed or accuracy/resolution of polymer based 3D printers. Plus, the maintenance is atrocious for anybody trying to “just print”, which limits the wide spread appeal. Get a single (ballscrew?) Z axis photopolymer machine that’s otherwise pretty much solid state (DLP or laser) with no or very few other moving parts and you can make extremely high resolution prints without much fuss. They are commercially available now and have been for years. But the patents on some of them are not expired. So instead, you have people still working on these somewhat junky FDM machines that require constant fiddling, a huge amount of software overhead and don’t even reach the speed or resolution as they are point based machines that can only print in one X and Y coordinate at a time and are trying to “paint” an object that uses cubes as pixels but instead uses round extrusions. Not even digital ones at that. There have been impressive improvements but the core technology remains the issue and I don’t see a way to overcome it as it is a fundamentally architecturally based issue.

          Why does “the community” at large seem so focused on trying to refine and push FDM technology? It’s an ok starting point but why has nothing better been embraced yet?

          1. Define better. SLA produces finer detail, but FDM machines can produce objects out of rubber, woody or stone-like material, chocolate or cookie dough or, with a little modification and/or fiddling multiple colors. Also for many uses detail or finish is not important (it is called PROTOTYPING) and so paying for it is a waste of resources. Finally, SLA and powder printers both use matierials which are messier and more hazardous than thermoplastic filament.

          2. Of course, better is in the eye of the beholder (FDM certainly wins the cost trophy at present) and I am not saying that FDM cannot produce a decent quality 3D object if correctly setup. The issue is that the quality isn’t really the same in terms of X/Y/Z resolution and layer thickness is still fairly thick in comparison and one is a point based technology so to produce smaller voxels, you need to take more time per layer if you are laying down less material per point. Waiting 3 days for a build is still an awfully long time if you want to really push the resolution or build volume and you have to have a machine that can actually hold that calibration and resolution, not just for the floppy liquid plastic but for the machine itself, across the entire build platform.

            The granularity / time tradeoff has to be considered as it is a core limitation of point based 3D printers. In theory, resin based technologies will eventually be able to vastly surpass FDM technology just by virtue of what and how it does it differently. The speed difference alone is getting ready to vastly change. FDM, by definition, is a first generation technology. It still feels very limiting for many applications and the lack of a remotely smooth texture is also offputting for many (but not all) applications. It’s not exactly replacing injection molded parts in most cases.

            Resin technology can also range from rubbery to hard materials, including nanofilled, high temperature ones that FDM just cannot do. It can also print hydrogels, for biocompatible applications. FDM cannot do that either. Some variants also can print any geometry, without needing support structures that FDM can’t really do. While I agree that some of the liquid photopolymers are not exactly food grade (though after polymerization, some are used in medical and dental applications), I also would not underestimate the hazards of ABS being heated up to melting temperatures, especially in an average home or dorm room setting, with limited air movement and possibly also with pets or small children around.

            These “cheaper / mass produced” FDM (or PLA only) machines that are coming out soon or are starting to be released now are designed for the average student/lay person/homeowner/etc who doesn’t know anything at all about 3D printing and just assumes it is the exact same “printer” as a normal 2D printer except it prints 3D objects instead of sheets of paper.

            Well, sort of. Sometimes. Only not really the same. Curious to see if people are discouraged by the upcoming crop of newer “lay user friendly” FDM machines or if they use it as a starting point to familiarize themselves with the technology until something better comes along.

      1. Im sure your printer is accurate. But i’m just as sure its not repeatable down to .005mm. A good fucking CNC (like my company’s half million dollar Mori) is going to have trouble holding .005mm. Even more so, I would be impressed if you had the gauging to accurately measure down to .0005mm which you would need to ensure .005mm.

      2. I’m happy to believe that, my Prusa i3 from an eBay kit happily prints 0.025mm layers. I haven’t tried a 0.2 or 0.1mm nozzle… yet. I should confess to being a hobby engineer so it was possibly built with more than usual attention to detail.

    2. to me this is entirely logical, the self-replicating nature of reprap is what forced prices of FDM down… business will always be willing to make money in mass manufacture, but the replicative aspect of reprap set a vague reference price

  2. > In reality, an infinitesimal percentage of RepRap owners print other printers

    I’m not sure about that. Other than test prints, all I’ve ever seen them used for is printing parts for another printer

      1. I’ve personally printed out 3 different machine’s each better than the previous. The motors bearings belts and hardware are the same.I had a friend buy a store bought one a few years ago and he printed out mine. I recently just used my most current machine to print out a router cnc machine.

    1. Sadly, I’m not sure either. :) All I can go on is people I’ve talked to. I haven’t (social) scienced it, and I don’t know of any good survey of the RepRap community. (If anyone does — let me know?)

      How many RepRaps are there out there? I’d guess a few tens of thousands. It’s still pretty darn impressive.

      1. I been printing for less than a year. I have printed hundreds of items from masters for lost-wax casting through scale models, screwed and bayonet adaptors for cameras, telescope accessories and things to use in my workshop like soft jaws and an illuminated bench magnifier. I can’t imagine living without a 3D printer now, buy to get the best out of one you need to be able to design the things you need in 3D.

  3. I guess I am one of the “infinitesimal percentage”, at least until recently. I built my first printer 8 years ago out of hardware store iron pipe and aluminum extrusions, along with some laser-cut acrylic from the now-defunct RepRap Research Foundation. From there I printed parts to make a 1st iteration Prusa Mendel. Since then I have printed parts for and built about a dozen printers, of which I still have three (the original, now-not-working printer; an over-sized MendelMax variant; and a newer h-bot style printer that is not quite finished).

    In fact, the vast majority of my prints have been to create other printers, with random other projects thrown in.

    However, in the last couple years, prices have dropped enough that it is no longer worth it for me to make more printers from an economics standpoint. I will continue making new printers for myself as I enjoy the iterative build-and-improve process.

    1. Still, this is a well-written article. I think it’s awesome that so many have access to this technology now. Just a few years ago I didn’t know anyone that had a 3D printer. Now I know quite a few.

      The problem I have with cheap machines from over-seas is that they are almost universally of inferior quality, resulting in many discouraged users. Granted, I do have a lot of experience with these printers, but once I have a machine set up my prints are pretty much set-it-and-forget-it. My materials cost alone for a printer that I build is generally more than what most printers sell for now-a-days though.

      When asked, I still recommend spending more and buying from a well-respected company rather than getting the cheapest Amazon or eBay printer out there. Money is not saved by spending less and having an unusable machine.

    2. I always wonder about this. I know a bunch of people who’ve built their own repstraps like that — out of whatever they can find. It’s hard to imagine that a pile of wood would make a worse starting point than a chinese printed set.

      But if you’re starting off completely DIY, you’re probably fairly committed. My guess is that your attitude / capability makes more of a difference.

      1. Like I mentioned, my first printer was built 8 years ago. At that time it wasn’t possible to just jump on eBay and buy a set of printed parts. The only option there was to get a set professionally printed at a cost of $1000+. And commercially made RepRap-type printers didn’t exist outside of a few hard-to-get or very shoddy products. In fact, the original MakerBot Cupcake hadn’t even been released yet. Skeinforge was pretty much the only option for slicing, and the only host software was the Java RepRap Host.

        I wanted a 3D printer, and the only practical option at the time was to build my own from hardware store bits. So that’s what I did.

        I also proved to myself that a tool or machine can produce parts for a second, more accurate, tool or machine. Which, of course, has to be true for the progression of technology. Some people still don’t believe it though.

  4. The problem with the original sentiment of Reprap was that it was based entirely on self replication, instead of availability of parts. in the beginning, 3d printed parts were hard to come by, and thus expensive. Despite what you may think, GOOD 3d printed parts are hard to come by. Anecdotal evidence: I was building my printer with parts sourced from spain. The 3d printed parts I received were bad – so bad that I couldn’t even use them. I had to search for new parts, and I thought I found them on Ebay. a few weeks later and I received my new parts. They were just terrible! just as unusable as the first set I got. nothing fit together, the material was brittle and several pieces broke. I had nearly everything I needed, but couldn’t get those few printed parts to put the rest of my printer together. Finally I was loaned a printer from the locally library and was able to calibrate and print the parts that I needed. Funny enough that the printer that I used was one with absolutely no printed parts in it’s construction.

    There are still parts of the world where people have trouble sourcing the parts to make a printer. The availability is just not there. Instead, designs should be made out of what is locally sourced and available (and cheap). Aluminum extrusions can be bought in most countries, whether t-slot, tubes or bars. Plasma cut frames might be a good idea if you can design around the inaccuracies of the cutting process. There are a few components that will be difficult to source, like stepper motors, electronics, extruder and hotend.

    The biggest issue of the reprap project and open source hardware is this two steps forward, two steps back dance that happens as new people enter the field.
    For example, it has been known for a long time that using off the shelf threaded rods for your z axis results in a phenomenon called “z wobble”. This is particularly apparent in the older designs that use an 8mm linear guide smooth rod, and an 8mm leadscrew to drive the z axis. the 8mm linear rod might be made to high tolerances, but that 8mm screw certainly is not. a slight bend in those rods translate into a sinusoidal z aliasing artifact because those 8mm linear rods are not strong enough to counter the bend in the screw. This effect is worsened by adding a guide bearing at the end of the screw.
    Yet people still use it, to this day there are new printers designed that use this exact setup, whose symptoms and remedies were fixed years ago. There are plenty of other examples like this, like non orthogonal belt paths along the linear axis, insufficient support for a linear rail, or not properly taking acceleration or belt tension into account. Like your previous article of H bot versus CoreXY. CoreXY has been deemed a vastly superior mechanism to H-Bot, yet people still try to develop new designs for H-bot without understanding the implications.

    Even the fabled Prusa i3 design has many design flaws that could be easily fixed with some proper engineering. Many of the same mistakes made on the i2 design were brought to the i3 design, and some new mistakes were made.

    The manufacturers of non-RepRap machines are doing most of the innovation. Printrbot has their Big-E printer, Ultimaker has a very solid design, even those new HobbyKing printers are quite good for their price. The Reprap community is having a hard time getting it’s act together and collaborating on new designs that make sense, are engineered well, and have a certain amount of innovation.

    1. I’ll give you some of those design flaws. I run an 8/8 Prusa 2. The secret to killing Z wobble is to have the smooth rods stiffer than the drive thread. So you leave your motors loose or you use aquarium tubing as the coupling, or similar. Firmly anchoring the smooth rods, going to 5mm threaded, or going to 10mm smooth can work too. There’s a million ways, but still I agree that you see folks tightening it down to “help”.

      I don’t know who’s doing most of the innovation.

      Ultimaker is awesome. But Ultimaker spun out of RepRap. Ditto Printrbot: And for innovation, you can’t beat the deltas — also out of RepRap. Lulzbot? RepRap. Makerbot? Originally RepRap.

      Most of what you think of as “commercial” printers have their roots in RepRap, or were actually RepRap projects. So it’s hard to say where the innovation is. It may just be that people in RepRap who have viable innovations often build companies around them. Prusa.

      1. The problem is overconstrained z screws. just having rods that are stiffer reduces it but it’s still there because an over constrained screw is still exerting forces on the frame. A better idea is to have the mechanism being raised slide on the rod but “float” on the Z nut isolator. It has an added advantage of not crashing your machine hard if the Z switch is wrong or you have parts on the bed.

        1. Exactly, JRDM. Another solution is to let the entire Z screw float (radially, not axially), which is actually easier than keeping it rigid.

          And this makes it indeed a typical RepRap / Open Hardware design flaw. People see the stuff on Github and assume it to work perfectly on the spot. Which it sometimes doesn’t. The designer has no interest in fixing it, because his own printer is fine and he doesn’t get a penny for trying to fix other printers remotely. Owners of the cheapo shops have no clue (if they try the stuff they sell at all), so no solution from there either. Other shops bring out an entire new printer release instead of a fix for the old one.

          In some cases a proper solution actually appears at some point, but usually just drowns in the mass of fruitless tries of the clueless.

          1. Well said.
            A lot of ‘improvements’ out there are snake oil. The worst ‘improvement’ is fitting over engineered-axis systems. The Darwin had anti-backlash z-nuts, the Prusa doesn’t as gravity gets rid of backlash without creating excessive friction. But people fit anti-backlash nuts and convince themselves their prints are better…

        2. A floating Z screw is one of the best improvements that can be made to a 3D printer. It takes the engineering error and production error out of the Z rods, if you are driving using threaded rods. It’s a very cheap and quick solution. Been doing it to every printer since I started back in 2011. If any of you have any questions about printing, the #reprap irc channel is a great resource. You will find many people on there who have been fundamental to the roots of the community. If it wasn’t for them, I wouldn’t have gotten as far as I did. I wouldn’t call myself a subject expert but I definitely can answer people’s questions and help troubleshooting. If you’ve had an issue with something, I’ve likely been there and solved it. Otherwise those great folks can help.

      2. May be rooted in RepRap, but are no longer independent entities that design printers that can be manufactured and assembled by the average hobbyist. take a look at the current ultimaker, without a lasercutter or waterjet you won’t be able to make one. same as makerbot, printerbot, etc. Most repraps can be made with a drill press being the biggest tool investment. These companies you listed have moved onto milled parts, lasercut parts, waterjet,, etc – requiring huge investments into the tools required to make them.

        And now that these companies exist and have the tools, they innovate with those tools and resources they have. printrbot has a full pcb manufacturing line. Makerbot is owned by one of the largest 3d printer companies in the world. Ultimaker develops their own slicer software (cura) and models website (youmagine).
        In fact, the developer of Octoprint is paid by a company to continue development. A few of other developers have been picked up by companies.

        The Chinese made printers are getting even better, with original designs that are cheap and functional. What I’m saying is that Reprap has bootstrapped all of these companies, who have taken repraps and in some cases have made them better. The companies have done what an open source project could not do: Focus it. They could iterate and refine designs faster, they would build experience and brand. Just like Prusa has built his brand. There are very few Reprap printers that have a brand, that have focus, that have iterative and progressive designs. This is opposed to the chaos right now where the same failed designs re-appear because of a lack of collaboration. sure this chaos can breed creativity, but when you do a google search of Prusa i3 you will find hundreds of me-too designs that lack innovation, that lack fixes to the engineering challenges, that lack something that makes it better then the other designs. Prusa’s own i3 design is no different, his innovation was to make it cheaper, use less parts, but lacks mechanical and reliability improvements from the designs before.

        1. Laser-cut parts maybe used to be hard to get — I’m not so sure that’s true anymore. (My experience, big city, nerdy circles.) The local hackerspace is a good start. Some companies may be using custom-machined parts to prevent hobbyists from making their printers at home, but my guess is that a lot of them are doing so b/c it makes the printer better or cheaper to manufacture. They’re allowed to design, IMO.

          Buying into the RepRap evolutionary metaphor, the chaos is normal. The good will (hopefully) emerge and be copied more often. The point you’re making is that evolution is wasteful compared to knowing the right path and putting all our resources behind it. But nobody knows / agrees on the right path a priori, though. Everyone wants to try out their own thing — even if that means repeating mistakes. But who knows, maybe repeating the mistake will lead to a solution this time?

          But returning to your original point: I see most of the innovation happening in the free/open, and then slowing down as it enters corporate production. The Ultimaker _was_ innovative when it was a RepRap — they’ve focused on tweaking and refining that innovation since then. It’s still the same basic machine. Makerbot is basically Cupcake++, and they still can’t get their extruder as reliable as a Wades + J-head. Wanhao is just tweaking existing designs.

          In this context, it’s worth mentioning Stratasys, who invented and then patented FDM anyway. In some sense, all the Reps are just copying that original idea. So maybe you’re right about corporate innovation. :)

          But even Stratasys was idea first, company second. They’ve only been incrementally improving their original design.

          Show me a company that comes out with an idea as radical as a delta. We’ll see who’s innovating.

  5. I really don’t understand the right / wrong fixation in this article. As seen in the comments, some people print other printers. I print parts with my printer, not other printers. Neither case proves the other wrong.

    In fact, I find this sort of article exactly what was wrong with the printers must print other printers thing. It is telling others that they are “doing it wrong” when they are doing what they want. There you go. I will tell you (the author) that you are doing it wrong by telling others that they are doing it wrong!

    1. Fair enough! I think we can all agree that it’s “right” to help other folks build up printers, either by printing them or by sharing knowledge. Don’t let me get you down if you want to design for replicability.

      But you had to be there witnessing the heresy that was Mendelmax in 2011-2012. Folks were getting crucified (not literally) for using other than threaded rod. Metaphorical pitchforks!

  6. I had a hard time sourcing motors for my scrap build. That pretty much prevents me from building another printer, unless some more motors find their way to me.

    I’m in $40 and can sort of booger out prints at .1mm height .5mm nozzle width.

      1. Ok so if my math is right, then I’d have a $76 build. Right? That’s moving the wrong direction for me.

        My goal was free. I made a hot glue printer for $0 and I was making good progress on a nylon weedeater line printer until I accidentally miswired a transistor (frying my arduino) while trying to get 2 wire greycode output from teacup to drive a stepper hanging off an l293d, and had to start over with a ramps kit. I also had to purchase thermistors since the one from my shucked jeep coolant sensor melted and the one I stole from a powermac case fan turned a very nasty shade of black after a while. Also purchased ecig kanthal heater wire (which replaced the soldering iron after it broke) and kapton tape because _safety first_!

          1. Dude I drilled out and heater wire wrapped my own printer hotend. After a few tries, I now know how a thermoplastic extruder works. I can build them. That skill is now a part of me. How is that not completely fucking rad? The level of experience I attained is so totally worth the hours.

  7. Now if only commercial insurgent companies like MakerBot, Ultimaker, etc. would realize the RepRap ethos IS NOT FUCKING APPLICABLE TO YOUR COMMERCIAL OFFERING.


    1. Additional note to ‘consumer 3D preinters” – if your marketing shows architects, engineers and consumers using your printer to 3D print parts, but does not focus on them replacing $3 chinese parts, printing their own upgrades through necessity not desire, and other actual use of your 3D printers, THAT MAKES YOU A LIAR. LIARS. THIEVES AND LIARS.

      “We made some $! I know.. Lets rebrand some open source software!” is not entrepreneurial. It is theft.

    2. I understand the Makerbot inclusion, but I don’t understand why Ultimaker deserves your ire. Yes, they make a commercial product. But they do post all their designs and software.

      1. And their machine, like all the others, is made of the lowest possible quality bits and bobs from china, costs 2500$ + and their marketing implies you can actually print 3d models with it in a professional setting, something that simply isnt true!

        1. 80% of the parts are sourced in Europe. And while we had some quality issues when the requested volume increase. But that’s fixed now. Marketing is marketing, I agree with you on that one. But I personally really think the UM2+ is a really good machine that is the closest to “just working all the time” that you can get.

          Software is still open source and free.

          (I’m the Cura guy. You know, who made one of the leading Open source 3D printing software pieces while being employed by Ultimaker)

  8. If you double the number of printers (and combs) every day (one replica a day per printer, right?), it comes to 262,144 printers after 18 days (and the same number of combs).
    At a rate of 10,000 combs an hour, it takes 26 hours and 8 minutes to make 262,144 combs, not 18 days…

    The break-even point is a bit farther than that: it takes a little over 23 days.

    I’m bald…

    1. You are correct but maybe forgetting the combs that have been produced already? eg at end of day 3 you have 4 printers and 7 combs.
      All this is just one comb per printer per day and at that rate the articles point stands, after 18 days you now have enough printers to outperform injection moulding at one print per printer per day.

      If we assume the comb is a one hour print and you switch to 24 hour comb production on day 16 you’d be producing 393216 with 16384 printers vs 240000 a day for injection. It would then take five days at that rate of production to equal the cumulative number of combs that have been injection moulded.

      1. You wouldn’t use a 3D printer to make combs, you would use it to make the machine that makes the combs. 3D printers excel in making specialty hardware not rapid reproduction. That’s why injection molding is so cost effective because it does one job quickly and one job only.

  9. There were a few problems with the RepRap idea. Only about 20% of the parts were 3D printed and they tended to use a LOT of hardware. Add to that the fact that plastic isn’t a good part to use for frame joints. And even back then, RepRap adherents sometimes were pretty derisive to things like metal parts.

    But RepRap movement was a necessary step in the process. I don’t know what to call it know since the 3D printed content has gone down.

    1. The alternative interpretation that was current even in the early history was that all of the hard-to-get or expensive parts should be printable. But there was never a workaround for motors or driver boards. (Though they’ve gotten much cheaper now.)

      Or that the parts that embody significant design or engineering thought should be printable, and that’s closer to right on. But then it’s just open-source hardware.

  10. Excellent article. I had to chime in and say, self replicating has not been all that bad, even with delta. I bought my first vitamins printed in PLA for my delta. I then redesigned my whole machine and used it to 3d print its own upgrade parts. It now prints at higher resolution and much faster than before. I will be using my 3d printer to print certain parts for my upcoming cnc mill project. The 3d printer has been excellent tool. The problem as I have been having is that no one seems to want one. I offered to 3d print the vitamins and build a 3d printer for friends, they all declined when I said the price of the parts was going to be a couple of hundred dollars. I certainly paid way more than that. But what I paid for parts and much more in time I learned more than I imagined I would. Learning something by working with my hands is somewhat of a therapy for me. Countless hours invested = fun in my book.

  11. While this article expresses some quite correct observations, it also misses just as much as it found “failures” in the original RepRap concept.

    One thing with the above picture of the economy is, a few years from now on there will be no longer a point in assembling (stitching a few parts together and calling it “DIY” is a bit of a stretch for me) a printer at all. Cheapo-chinese will learn quickly on how to deliver printers doing a job just fine. 3D printers as a whole will become the vitamin.

    With your cheering on industry parts, how would you develop considerably different designs in the future? Ones which require different, not yet available industry parts? The more you rely on industry parts, the more you lock down certain design aspects. In a few years you’re where you were before RepRap: each considerably new design costs huge investments.

    Also, the more the world moves over to ready-to-use printers made entirely of industry parts, the less hackable they become. Before too long 3D printers are just as hackable or rather unhackable as 2D printers today. That is, if you stick to the “more vitamins are better” picture.

    Last one for now, people don’t build printers because it’s economic. Printers are built because it’s fun to assemble, fun to design, fun to make something. Else these thousands of variants of already proven designs can’t be explained. People enter the forum saying “Hey, I like this design, what can be changed?”. This requires low reliance on industry parts as well as replicability.

    RepRap is more than getting hands on affordable printers. It’s a way of desigining, perhaps even a way of living.

    1. I think there is a sliding scale. In some ways, it seemed RepRap tried to require for 100% DIY with the most sophisticated tools needed was a hand drill. But you don’t have to own the manufacturing technology yourself to have a non-proprietary part design made on prototype volumes in materials better than using plastic for frame joints.

      Things like laser cut steel or water jet cut aluminum often wasn’t considered, because the technologies might seem inaccessible. They could allow for one-off prototypes or small batches for hobbyists and enthusiasts and still be reasonably priced. Those technologies may seem out of reach by many, but I bet you’ll find a job shop with them in most major cities now, even in countries that might not be perceived to have that kind of sophistication. Or even if only only 1 in 100 makers might have a CNC router, but there’s still enough of them that anyone could get a custom panel design cut from any of those that do without having to own the machine themselves.

  12. I think the critical contribution of the RepRap project was to prove/expand the size of the market. When the project started I would have to explain to other geeks what 3D printing was. RepRap birthed thousands of such conversations and most ended with “I need one of those!”. Once the idea entered the culture and became a geek obsession it was obvious there was a useful market there. the first working models and the pell mell improvements (and the open source license) made it inevitable that commercial exploitation would occur. It is still possible to print the parts for a printer as well as buy a turnkey machine and that is all good for me.

  13. I’d consider money and time as a resource too. Once a 3d printer is set up, i can produce about 1€/hour of value with a 5 minute effort. 1€/hour is just an average price for what someone is willing to pay to have it, considering myself, minus the operating costs. A machine that makes me enough money to buy the vitamins and pays me the hours to get another working(to get even more profit) sounds enough self replicating to me, even if it is limited by the environment.

  14. Just a recommendation for a Hackaday article…
    Could you do something on the REAL start of the DIY 3D printing movement?
    The RepcRap project simply coincided with the internet boom. Ten years prior THOUSANDS of tinkerers (what hackers were called before media outlets like Make magazine and Hackaday came up with their own names) were making “Santa Claus Machines.”
    We’re talking late 70s/early 80s using discarded pen plotter components. How many readers know what a pen plotter is?
    It just saddens me that Mr. Bowyer and his silly project gets any recognition since it was an ill-conceived concept piggybacking a movement that would have happened completely without his involvement. Making matters worse is the general ignorance of the big media outlets that basically attribute his project to the “invention of 3D printing”.

    1. The Don Lancaster link you posted is full of dead links, and very light on documentation of late-70s FDM machines.

      #citationneeded :)

      I don’t think we ever said RepRap invented 3D printing. Perhaps your comment is more appropriate elsewhere.

      1. So you’re implying you’re a big media outlet now? Seriously though, every mention of RepRap’s history implies they are somehow the start of it all…. like this article. This is how myths get started. If not corrected, they turn into fallacious legends… like Alexander the Great or Steve Jobs.

        The link actually mentions stuff Hackaday has never featured. It was intended to be the first step in a process occasionally called “Investigative journalism”. It’s a little different than to that which you’re accustom.
        #work >:~o

        I realize it’s easy money to simply direct traffic to websites with cool projects. All you have to do is write a sloppy synopses, kick back and profit, I get it. I was just suggesting an original article to pay homage to those who actually started the DIY 3D movement and mostly developed everything RepcRap ‘created’.

        WARNING It might require clicking more than one link.

        … or you could just write painfully bad short stories?

  15. OK, if we admit that the replication of replicators depends on supply of necessary constructive elements, then why aren’t they made as automated assembly units for certain set of primitive elements (massively produced by other “symbiots”), instead of being made as FDMs? Imagine if we had a robot that constructs its own copy from some sort of Meccano set parts, or from LEGO elements? Now imagine that the parts are orders of magnitude scaled down (existing construction sets are created to fit our hands … well, kids’ hands anyway) and that they can be assembled into macroscopic objects that appear almost smooth on the surface.

    Then we would need one machine to churn out primitive elements in quantities from raw material, and another machine to manipulate them into their places. And of course, a method of fastening the elements together.

    But to be able to make that, we need design of a machine whose each iteration is capable of creating a slightly smaller machine, without loosening tolerances. Is that possible?

  16. It is hard to criticize open-source project openly, but its good to finally read it.
    There where too much silly ideology under the reprap project. I don’t care about the ratio of 3d printed part on my 3d printer, I just want a 3d printer that works!!!
    When I started, I was in search of a usefull robotic project. I knew little in electronic and wanted to learn. I benefited that everything was explained in the reprap project so it could fill the holes in my knowledge to finish this project.
    I never bought a 3d printer, I never used a design made by another. Althought I designed 5 different 3d printer models of my own and build 3 of them.
    Back in 2012 what motivated me to build my own 3d printer was the poor mechanical design of the reprap, plus my mechanical engineering degree.
    I made my first frame out of plywood panel laser cutted. Some parts where custom made for that specific design and impossible to buy for a decent price at the time. I did not had access to a 3d printer so I managed to build temporary cheep wood parts attached with tiewrap (like a reprap :-) to let me print the remaining 3d printed parts. I guess it was the first 3d printer that 3d printed itself! ;-) After that, the first part I tried that was organic (yoda) came out very very good!! Even better than I thought it could be. I was amaze. That’s a funny memory.

    But I was not alone thinking that I could do better than a mendel and a lot of people jumped in and now there are so much variations… I guess none of the first ideology worked, but it changed the world anyway so the reprap movement is finally right.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.