Some time back I stumbled across the reprap project (www.reprap.org) a worldwide open source project initiated by Dr Adrian Bowyer at the university of bath the goal of which was to design and build an affordable 3d printer that could self replicate ie print all it's own parts.
Now 3d printers are nothing new they've been around for quite a long time and will print in a variety of materials the most advanced ones are now able to do full colour prints or even print in metal. The down side is these things are huge and very expensive a basic model is typically in the tens of thousands of pounds and the size of a fridge freezer and you typically have to pay over the odds for the feed stock (the old ink cartridge trick where they get you not only for the printer but the ink it uses).
So along comes reprap a project to design and build something a little more affordable for use in the home and as things stand you can get a machine for under a thousand pounds depending on how much effort of your own you want to put in.
Reprap is a Fused Deposition Machine and uses thermoplastic as it's primary material (though there are plans for other materials in future) it takes in the plastic as a filament pushes it into a heated nozzle that melts it and extrudes it out of a small hole as a thin (usually 0.5mm or smaller) stream of plastic basically like a souped up hot glue gun (some of the initial prototypes were made from a hot glue gun). This extruder head is attached to a cartesian robot that can move it in three dimensions laying down the object one 2d slice at a time one on top of the other the plastic fusing to become a 3d object.
When I first started looking at reprap they had a series of prototypes but they were not at a version 1.0 yet plenty of people had tried various things but it was all a bit up in the air. As time went on they developed the first machine code named "Darwin" which was a sort of box shape made of rods with a bed in the middle that could go up and down on threaded rods and a print head mounted on the top able to move about on a belt driven X and Y stage. you can see an example here http://dev.www.reprap.org/bin/view/Main/RepRapOneDarwin.
It was still a lot of work to make one at this stage the electronics could be got in kit form but you still needed to machine the nozzle and somehow get hold of a set of printed parts to make up the frame. Eventually various outfits started producing kits a group called Makerbot in the US came out with their own design called Cupcake CNC and in the UK BitsfromBytes came out with the RepMan these machines were not technically repraps even though they derive from that same technology but what is called a repstrap a machine that is put together using traditional manufacturing techniques but that could be used to build a reprap so bootstrap the process.
I saved up some cash and got myself a Repman version 3 complete kit which included everything needed to put the machine together. The design is a fairly close variant of darwin the parts for the corners and so on are not printed in this case but made from laser cut acrylic and assembled by a series of nuts and bolts. I got the kit sometime last year but due to being away a lot and making a series of screwups I only now got the thing together.
It's driven by a pic microcontroller which runs the four stepper motors (X, Y, Z, and extruder) it will take an file written in GCode (a semi standard that is used on CNC machines) off an SD card to print out an object in a variety of thermoplastics. I'm currently using ABS which is what lego is made out of. You can basically take almost any model export it as an STL file run it through a program called Skeinforge and it will convert it to GCode that can be printed. It also has a nice little OLED screen that tells you what it's up to (the interface to which is very fragile basically a load of strips of tinfoil and caused me to have to replace it once before)
The extruder uses a screwthread driven by a stepper to pull in 3mm plastic filament into a metal nozzle which has been wrapped in nichrome wire. The wire, which is used in hair dryers and toasters and so on, heats up when a current is passed through it. A small thermistor on the tip allows the board to regulate the temperature and when it reaches 230 degrees or so (depending on what plastic you are using) it drives the filament into the nozzle extruding thin string of plastic.
Because of the way the device works laying down layer after layer it has a limitation of not having anything more than 45 degree overhangs because it can't deposit plastic into thin air there needs to be something holding it up. One way round this is to use a support material either the same plastic making thing supporting columns you cut away or some other material that is easy to remove (this is a future improvement to the machine). Usually some clever design can do without the need for support material.
So far I've been mainly calibrating and fiddling with the software to get a good quality of print out I've made a few objects with varying degrees of success a small mug, a little toy car, a dodecahedron. The software that generates the GCode is fairly complex and getting the settings right to get a good quality of print is tricky but I'm getting there.
One of the interesting things about the reprap project is that each machine can print parts for other machines even the next generation it can also make spares for itself or incremental upgrades and new parts for it. So people have already designed most of the spare parts for my machine as well as a couple of different print heads like one that takes a dremel tool to do lightweight milling. There are also sites dedicated to making things on the printers like http://www.thingiverse.com/ who offer designs other people have made. It's a very community orientated project with a lot of contribution from general tinkerers. One of the current problems is plastic warping on larger builds, this is due to the fact that as plastic cools it shrinks slightly and this tends to mean the lower layers of a print curl upwards. Various people have come up with various modifications to alleviate or fix this like a heated print bed that keeps the plastic warm enough that it doesn't warp or new ways of designing your objects to distribute the stress as it sets.
While I've been building my "Darwin" variant the reprap people have developed a version 2 of there printer code named "Mendel" (http://objects.reprap.org/wiki/RepRap_Version_II_Mendel) and should I want to I could use my printer to print out the parts to make one of these I would just need to add some electronics and a few rods and screws to it. Or I could take parts of it I like adapt them to my existing printer. It's wonderfully flexible in this manner since it is all open source people take the designs improve and modify as they see fit and ideas and improvements feedback into the design of the next generation machine.
As to what I'm going to do with this thing now I've got it the possibilities are endless I can make almost anything I could want out of plastic using it and plastic is fairly cheap. I have a few ideas for projects I'd like to try but at the moment I'm just learning how to configure the thing and get the best out of it. I'll probably print off some spares first of all since it's structure is made of laser cut acrylic which is not the most robust of materials there are already a couple of bodged repairs on some of the stressed parts. There are also some potential improvements I could make most of the extruders have moved to pinchwheel driven systems rather than the screwthread mine uses so perhaps I could replace that.
You can see photos and a couple of timelapse videos of the machine and stuff I've thus far made with it on my picasa album http://picasaweb.google.com/warriner.matt/Repman#
Comments
Fascinating stuff.
I remember seeing something on Tomorrows World absolutely years ago about this sort of thing, though I think that was using lasers and UV (which half an hour on Wikipedia has told me was probably 2-Photon Polymerisation)
I really like how simple the concept is...I'm guessing the motors are super-accurate (so I won't be re-creating this with Mindstorms anytime soon). Can you get 3D scanners? Surely then you are 1 step away from a replicator...
How long did it take to make the car? Good use of timelapse by the way :-)
That is awesome... Just plain awesome. I barely understand it, but I think its awesome.
Yeah that's one of the many way to do 3d printing anther is laser sintering where laser is used to basically spot weld dust together (plastic or metal) then another layer is added and so on. I've even seen paper and a laser cutter used so it cuts paper with a laser adds another sheet glues it to the previous one and then cuts that. The FDM technique the reprap uses is one of the simplest to do.
You could in theory tie it in to a laser scanner and use it to produce replica's of objects, people have already done this with preexisting data files for instance this chap made a copy of his head http://www.thingiverse.com/thing:749.
A home made laser scanner is not impossible to do you just need a line-beam lens (stem of a wine glass will do at a pinch) a laser pointer a webcam. A rotating platform and then measured distances from your camera and laser aiming the line beam so it crosses the exact centre of your plaform. Then you capture images of the object at various rotations use edge detection and various other tricks to filter out all but the laser use some math to work out how the laser beam is distorted vs how it would be with no object there given the distances and angles between camera and laser you can get a distance from your 0,0 point. Use these points to produce a point cloud a set of 3d coordinates that make up your object feed that into a program that can turn that into a mesh export the mesh in STL and the feed that into Skeinforge an then you have a file you can print out. I've not seen anyone who has done this complete sweep tho.
It uses fairly standard stepper motors nema 23's which have a resolution of about 1.8 deg. I did consider building a repstrap using the mindstorm kit as the base but it needs more motors than that. You can get round that by adding your own controller and motor to it using the freely available hardware developer kit for the nxt box but if you get to that stage it seems like you might as well have just swapped it out for a pic or arduino based microprocessor instead.
The car took a while maybe an hour or so to make the speed of object production depends greatly on the quality of output you want and how much infill. With solid objects it doesn't completely fill the shell with plastic it uses a pattern (criss cross or triangle or hexagon) to give the structure strength and save on plastic it then fills the final couple of layers completely to give it a solid shape. You can tweak the amount of infill and the layer density (you can get thiner layers by moving the head faster so it stretches the filament out as it extrudes).
Currently the Y axis is giving me alsorts of grief the pullys that drive it and the driveshaft that extends the motor output to the otherside keep coming undone in the middle of prints leading to the output being completely screwed up as the axis shifts during print. You can see that a bit in the first car I printed. I reckon a bit of duck tape will fix that up. I also had a problem with the zaxis jamming which some adjustment seemed to fix but that threw out the leveling I had done so I've had to relevel the head and readjust the print height.
Both of these issues make me tempted to try and print the parts for a mendel or reprap version 2 printer as its design is such that this sort of thing is not so much a problem. Still it'd be nice to get some decent quality prints out of this thing first.
I've been thinking about making a printable Rubiks cube I'm pretty sure I've figured out how they work and with some fiddling I reckon it would be printable.
Surely your challenge should be to ship over to America with just this, and tonnes and tonnes of plastic filament, and go from there? Start with a house (3 hours a brick?), and move onto furniture etc.
If I had half a clue about 3D modelling I'd be tempted by one of these, I can see all sorts of funky uses...
This is by far the coolest project I've seen in a long time. Thanks for sharing, EMW!
I wonder how easy it would be to print a model from Lightwave... or blender?
It's not too difficult to print from blender I've done it with some test objects
the main issue is making sure the object is manifold (blender has a select non manifold option that highlights any issues ctrl+alt+shift+m) then you just export it to STL which the Skeinforge program can read and turn into tool paths in GCode.
Currently I'm printing some spare parts as I've noticed a few cracks in the acrylic that makes up the corner blocks. One of the issues of this design is that acrylic is quite fragile in these sorts of tensions. But then it's a machine that can make it's own repairs so that resolves itself :D
I also need to sort out the filament reel it keeps jamming so when unattended the extruder runs out of polymer which isn't so great on long prints.
Managed to fix the issue with my filament reel it seems to feed correctly now so I can leave it printing unattended. Spent last night fiddling with the zaxis discovered that the crack in one of the corner blocks is causing the bearing that front left corner of the bed rides on to slip out and jam the z stage as it goes up. This means the bed slowly over a few prints goes out of alignment being too high on one corner that only increases the jamming problem.
Did a makeshift repair with duck tape (is there anything it can't do :D) left it printing a replacement corner block over night. The print came out pretty good but the print head was a little close to the bed and so the part was glued down so tightly I broke a stanley knife trying to get it off the bed when it was finished and eventually removed it only when I broke the bottom layer off the part. It still feels pretty sturdy so it'd probably do at a pinch. It also had a lot less z wobble in it which may be down to the bearing not moving so much.
I'll see if the thing can manage to print another one that I'll hopefully be able to remove easier. Ideally I'd prefer replacing as many bits with printed solid blocks as possible since they will likely last longer than the acrylic plus disassembling the thing and reassembling is likely to lead to more chance of failure.
Hopefully today I'll get the calipers I ordered off of amazon the other day that will allow me to better measure the filament thickness and wall thickness to better fine tune the parameters in Skeinforge. I did have this idea of a sort of automagic calibration print that inkjet style generated a load of objects with all sorts of settings you could then pick the best of and set that in a sort of one shot configuration thing but that will have to take a back seat while I get this broken corner block sorted out.
glued down so tightly I broke a stanley knife trying to get it off the bed
The resultant piece - if removed safely - must be a bloody strong piece. I think it's superb that the first things you're creating with your printer are spares.
It seemed fairly strong the raft and the bottom layers are full infill so it makes it stronger there. When it failed and came apart it was on the first hexagon layer infill layer.
Still it was pretty robust, the hexagon internal structure makes for some fairly strong objects.
Before printing that I did a corner bracket that holds the diagonal ties in place that came off a lot easier (much smaller surface area stuck down) it was very solid and the holes lined quite nicely after a bit of clean through with a drill bit. I suspect the whole thing will be a lot stronger with the acrylic laser cut flat parts replaced with solid abs objects.
that is very cool mate.
It seems like there should be something you could apply to the bed (a spray, preferably) to stop it adhering quite so much - maybe some sort of mould release compound?
Seems a bit odd that they use a material which isn't very resilient... although I guess it is meant just to bootstrap you into builiding the parts...
When I was reading about these a while ago MDF seemed to be a popular bed.
It's one of those fine line things adhering too much and you can't get it off the bed adhering too little and best case it will either just warp like crazy or worst case it comes loose during the print and then that's that what ever you are printing is ruined. Couple of people have been experimenting with vacuum beds I was considering using the Kapton tape to lay down a layer that I could peel off or perhaps use like those battery release things you get in toys.
Yeah the acrylic is fragile but only in certain situations where a sheering force is applied and they've mostly avoided that sort of thing it's just a few parts of the machine that by design or accident get these kinds of forces applied to cause issues.
They did used to do fully cast parts but it was very time consuming and expensive laser cutting a few sheets of acrylic is a lot cheaper. They might have been better off using wood which is better at handling those sorts of forces.
I've now got a couple of replacement corner blocks just need a few corner tie pieces and I'll be away.
I've not replaced all but one of the base corner blocks and fair number of diagonal ties. The z axis bearings now are much more comfortably seated and so far the zaxis slipping seems to be fixed. Of course fix one thing and something else invariably needs looking at and at the moment it's the y axis giving me jip.
I've begun printing out parts that I could use to build a reprap v2 I figure this machine would be unlikely to survive boxing up and shipping to the US so if I use it to make the parts I need I can pick up the mechanics rods screws nuts and the few bits of electronics locally. How ever with the y axis going haywire during prints the output I'm getting is not very usable the y slipping more and more until it goes completely.
I tried retentioning the y axis belts that move the carrige back and forth but that didn't seem to help. I guess I'll have to have another look at that drive shaft try and get the thing printing square again.
Had a few set backs over the weekend one was I threw my back out on friday which meant spending most of the weekend lying down dosed on painkillers but I also issues with the printer it decided to ram the print head into the bed and managed to melt its way 6mm in before I found it and extracted it from the acrylic.
This buggered up the nozzle, I managed to open it up again with a little filing but this widened the output from 0.5mm to 0.7mm which reduces its resolution a bit and throws off some of the numbers. I ordered a replacement hot end kit and tried it printing one of the sets of parts for mendel, naturally the y axis came loose again, and then I ran out of polymer.
I do have some new polymer on order but mean time I thought I'd have a play with the reel of HDPE that I got with the machine and before had no luck extruding.
This time I got the stuff to extrude relatively easilly but the problem with HDPE is it doesn't stick to anything. It doesn't adhere to the acrylic print bed even roughed up, kapton tape is even worse, supposedly the best option is using a house hold chopping board as they are made of HDPE or PP which it will stick to. I didn't have one handy so I tried the old standby Ducktape it did stick to it (or more accurately fuse to the plastic) but it seemed to provide a base it stuck to and indeed allowed it to be removed afterwards. I then failed to print the dodecahedron shape a few times and eventually tried a square calibration shape which demonstrated hideous levels of warping. Much worse than ABS this stuff was warping in the early layers deforming the shape and ruining the print.
I suspect with some settings tweaks you might get a useful print out of it but for anything really good I suspect a heated bed would be needed it just warps too much.