The Twelve Days Of… Self-replicating

Okay, it’s more like the 23 days but who’s counting? [Kliment] is giving the gift of self-replication this holiday season by uploading one new printable part a day. If you follow along and print each one you’ll have a Prusa Mendel by Christmas (this started on 12/2 so you’ll need to catch up). The Prusa is a variation on the Mendel that uses bushings wherever possible, reducing the need for bearings down to just two.

So yeah, you need to have access to a 3D printer in order to make the parts for this 3D printer, but that’s how it always works. [Kliment] has gone the distance to make this little exercise enjoyable. The parts that take longer to print are reserved for the weekends, some have been altered to include a holiday theme, and all of them have been optimized to fit on a Makerbot build platform.

[Thanks Christopher]

24 thoughts on “The Twelve Days Of… Self-replicating

  1. Well the idea is sound. kinda like P2P.

    1.Print one set of parts for someone who has no 3D printer
    2.person getting parts pays shipping
    3.person who just got parts puts it together and starts over at step 1 for someone else.

    or I could be missing something.

  2. And before some monkey comes in and (like usual) proclaims “this machine doesn’t replicate itself duuurhp”, yes it does. Maybe not to the extent that everyone would like, but technology isn’t always born perfect. They are working on making the extruder head print other materials so it will one day print the electronic parts too.

    I’ve always wanted a 3d printer, but the assembly price has always put me off. Even if you get the parts made for you the steppers and other materials will still cost you a pretty penny.

    – amazing still

  3. @Eirinn
    duuurhp, by that logic milling machines are self replicating.

    It’s still nice to have open source 3d printers but they will never be self-replicating in our lifetime, at least never to the point of making silicon wafers.
    Let’s save the term self-replicating for things that actually are.

  4. I dont understand why someone hasnt come up with a £199 3d printer. At the moment the plastic parts cost nearly that. Also the kits around at the moment can cost up to £1000 (makerbot) yet these dont even use linear bearings. If a laptop can be built for £99 then im sure we could build a £199 3d printer.

  5. First off comparing a 3D printer with a £99 laptop is rather ridiculous. Look at the quality (or rather the complete lack there of) of such machines. Also consider the cost of the components that go into them, as well as what is NOT inside of them.

    When you start pricing things out, there are a few snags with the idea of a commercial £199 3D printer. First off the electronics. All of the 3D printers require precise positional movement. This is either done by blind driving stepper motors, or setting up encoder systems to do the positional readouts. Either way they are expensive to build and drive. Then there is the control circuitry, most of the DIY systems rely on the arduino family and while some can run on your average 328, most use 644s or 1280’s (aka Mega). Even a clone mega from China will run you ~£25.

    That’s not including:
    3-4 Stepper motor controller(s)(or encoder systems with servomotors/steppers)
    3-4 Nema 14 or 17 stepper motors (depends on if you are going to use a stepper for the extruder or not)
    3-6 endstops (min & max)
    power supply
    misc wiring, and electronic components (heater controllers, temperature sensors, heater assembly for extruder, USB-TTL bub)

    Then there is all of the hardware, and unfortunately the part count is through the roof. There are those who are making great strides in simplifying the design, (the original Mendel uses 50+ bearings) but there are still quite a few parts in the designs.

    Then as you have pointed out there is the RP parts, remember that while these things can print themselves you still have to get the original set of parts from somewhere. For many that has meant sourcing parts through commercial 3D printing systems (myself included). Those people then try to offset the high price they paid for their parts by selling daughter sets to others. With each passing generation the prices for the DIY systems comes down, but good luck with all of the tooling involved for making a commercial system.

    The material itself isn’t that expensive but it is time consuming to print out a full set of parts for a printer, and time = money. Most of us have built these printers for our hobbies and passions, not so that we can endlessly print out more sets for others.

    Now don’t get me wrong, I don’t mean to be a scrooge, I AM helping in the creation of two other printers for others, who will then hopefully pass along the favor to others. But as a student I do not have time to make 10, 20, 30+ sets for other enthusiasts at nothing more than material cost.

  6. I dont get why these machines keep getting designed with such small working envelopes. With a larger print space, you could print not just frame connectors, but the frame spars as well. Assembling them from 2 inch sections would be silly.

  7. Hello. I am the Kliment mentioned in the article. I wanted to answer some of the points brought up here.

    @MRE – The machine has a 200x200mm build platform, which means you can build quite a lot of actual useful stuff with it. I have one part at a time on there because it’s more fun that way. If you wanted to be efficient you could print all those parts in five plates on even the smallest printers, and in one plate on the 200×200 ones. The reason the frame is steel is rigidity, not printability. We’re working on replacing that, but the print area is not what is limiting us, mechanical stability is.

    @ cost discussion – The Prusa Mendel is the cheapest replicating (in the limited sense of replicating all the custom-made parts) 3d printer yet. The nonprinted parts are available for about 300€-350€, and the printed parts are at around 150. I and a number of others are participating in a loaner program, where you get the printed parts at no cost on the condition that you print two sets to give to others. Once the number of printers reaches critical mass (and the Prusa can replicate in a day, the standard mendel took nearly 60 hours of print time), the cost of parts will drop. The electronics are already around the 130€ level, and dropping. The hardware store stuff is not going to get any cheaper, but we are replacing it with printable parts at an astonishing rate. The motors and the parts that need to handle high temperatures are the main expense.

    @ everyone wanting one but having no access to a printer – have you considered a repstrap? A repstrap is something you build out of hardware store parts and use as a minimal printer, good enough to print parts for a better one. The wolfstrap is the current favorite. Instructions are on the reprap wiki in English and German.

    Finally, in true holiday spirit, is running a contest that will give free electronics to the person who prints the holiday prusa mendel as a gift and submits the best documentation of the entire process. Have a look at it. We’re really making an effort here to get as many people printers as reasonable.

    If you want to know more about the project, join us in freenode IRC at #reprap or checkout


  8. I’m thinking of building a rep-strap style machine from broken printers. By broken I mean thrown out because of high ink costs. I have found the newer printers all use servo driven carriages. Is the resolution of the plastic encoder strip good enough for position control of the build platform? If we could do away with the steppers the cost would drop considerably.

  9. @ HMav

    There’s a number of people trying stuff like that, but so far nobody I know of has gotten it to work as accurately as steppers without having it end up costing more. If you want cheap motors, pull apart an old scanner. The problem with servos is that you need to coordinate extrusion motor speed with movements in both the X and Y dimensions. Inkjets only need to set the nozzles on the printhead to the state that matches the encoder position, they do not actually have to move in a precise, coordinated way. The inkjet, as far as I understand it, adapts the ink emitted to its position, rather than using feedback to control the position. Doing that is fine for one axis (the second axis is moved only when the printhead is stationary), but breaks down when you need coordinated motion (as in drawing a given shape). Modern 3d printers coordinate three axes (x, y, and amount of plastic deposited), and move the Z axis only when the head is stationary. I have no idea how that can be reasonably done with servos. That said, it would be very cool if you can prove me wrong on this one.

    (Incidentally, I posted a response to several others in the thread, but it appear to be “awaiting moderation”. Is this normal?)


  10. re: Sea Shadow comments

    I dont think that a £199 3d printer is impossible. The main problem with the reprap movement is that it HAS to make itself. Why?
    A 3d printer is basically a cnc machine with a extruder head. My first cnc machine used a very simple control board which cost a few pounds. Motors were unipolar and its used, as do the reprap and makerbots, threaded rod for the drive system. It probably cost less than £100 pounds, maybe £50.

    And it worked, just. The difference between a cnc machine and a 3d printer is that a cnc machine needs to be more solid as its cutting into materials.
    A 3d printer is depositing materials and because of this doesnt have to have very powerful parts, hence the nema 14 and 17 motors.

    Why dont we say we are going to build a £100 3d printer instead of having it build itself.

    Also why does it need a arduino?

  11. Id say at the moment the reprap and makerbot are over engineered. I dont think that the makerbot needs to be anywhere near the cost that it is now given it uses a bearing+threaded screw design.

    The part that may need the most work is making the extruder design simpler with off the shelf parts. In at the moment experimenting with using mig welding parts for the nozzle.

    By using off the shelf parts it becomes much easier to build one or supply a kit. I think the main aim is to get the cost down so that people are able to use it with minimal fuss.

    I also think that maybe it should be basically a simple cnc machine.

  12. @Sea Shadow,
    you should google again and you will find Arduinos below 15€, or a much more powerfull 32 bit ARM on Stm-discovery below 12€ (including the jtag like programmer/debugger).
    Electronic and plastic is really NO-cost in such a project.
    The real costs are in mechanical parts, motors, connectors and labor.

  13. @Kliment: EMC handles PWM drives with quadrature feedback fairly adequately and I’ve made a mill with the linear strips salvaged from printers. You only get basic PID control with EMC though last time I looked so to get decent accuracy at speed you would need to reduce backlash and flex right down to nothing.

    I reckon though with proper digital filters in the loop, printer style mechanisms would work. Perhaps the belt drives would need to be switched for steel cables, but that would still be way cheaper than the cast iron frames, acme threads and ball nuts you need on a decent mill.

  14. @nes

    Cool, see if you can get it done. There’s already people using EMC as a printer controller. Of course the mass being moved and the drag forces are greater than on a printer, but not as big as with a mill. Steppers start at 14 euros new in Europe, and decent stepper controllers at 13. So if you can really get something working for less than 27 per axis, that would be quite an improvement. The other issue of course is that EMC needs a low-latency parallel port, which few machines have these days. Thus the entire fooling around with microcontrollers.

    But hey, if you get something working, throw it on the forums or the wiki and I’m sure people will be quite happy to try it out.


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.