“The lathe is the only machine tool that can make copies of itself,” or so the saying goes. The reality is more like, “A skilled machinist can use a lathe to make many of the parts needed to assemble another lathe,” which is still saying quite a lot by is pretty far off the implication that lathes are self-replicating machines. But what about a 3D printer? Could a printer print a copy of itself?
Not really, but the Infini-Z 3D printer certainly has some interesting features that us further down the road to self-replication. As the name implies, [SunShine]’s new printer is an infinite Z-axis design that essentially extrudes its own legs, progressively jacking its X- and Y-axis gantry upward. Each leg is a quarter of an internally threaded tube that engages with pinion gears to raise and lower the gantry. When it comes time to grow the legs, the print head moves into each corner of the gantry and extrudes a new section onto the top of each existing leg. The threaded leg is ready to use in minutes to raise the gantry to the next print level.
The ultimate goal of this design is to create a printer that can increase its print volume enough to print a copy of itself. At this moment it obviously can’t print a practical printer — metal parts like bearings and shafts are still needed, not to mention things like stepper motors and electronics. But [SunShine] seems to think he’ll be able to solve those problems now that the basic print volume problem has been addressed. Indeed, we’ve seen complex print-in-place designs, assembly-free compliant mechanisms, and even 3D-printed metal parts from [SunShine] before, so he seems well-positioned to move this project forward. We’re eager to see where this goes.
“We’re eager to see where this goes.”
Upwards?
B^)
This reminds me of climbing formwork used for building skyscrapers
Very cool idea and some novel solutions, but if the point is to make it fully self replicating and print in place, then a rack and pinion motion system might be a better choice than belts and pulleys. It eliminates some separate components at least, if you can dial it in so you don’t lose precision on every copy.
Just imagining how much of a mess it will make when the legs eventually get misaligned.
Or just bow inwards.
The thread profile is tweaked so that the legs naturally pull into the screws with gravity.
Neat idea, but seems like it would have progressively worse accuracy each generation as the printing errors add up. Hopefully over time that can be resolved. And larger printers will fail to scale due to issues of rigidity.
Also, spare us the hype about self-reproducing. This doesn’t print motors, electronics, or anything other than the frame. And as for dumping them on mars, are they stocking the filament at WallMars now?
“… as for Mars…”
Yeah, unattended fission-powered chemical plants a’la Zubrin are enough of a stretch. Unattended self-extending infrastructure is nuts. You would need an actual Von Neuman replicator that can use in-situ resources to make functional copies of itself. If we have self-replicating industrial production, we’re not settling Mars, we’re converting it. We will have terraformed Earth and then be working on exploiting Mars for whatever’s most valuable at the time, resources of offworld real estate.
All that said, multi-material 3D printing very definitely has a place in manned space exploration. I wouldn’t want to get on a slow boat to Mars without at least 3 fully-functional printers (each with differing architecture and software that are still somehow able to swap unprintable parts) and as many space-hardened (or Mars-hardened) microcontroller boards as can be brought along.
If something breaks or is less fit for purpose than we’ld like, I want to be able to convince a bunch of engineering post-grads to pull all-nighters to develop a replacement, implement it on Earth, and verify it works before shooting me the BoM, STL files, and firmware. Optimally, my spacecraft would be designed with clearly labeled breakout points where new computing hardware can be tied into the ships systems to permit substitutions and replacements beyond the standard triple redundancy.
An iteratively expandable 3D printer would have a place in setting up a settlement on a planetary surface. There’s a lot of things you almost never want to rely on if they’re printed, but a printer would be super-useful to be used as an intermediate manufacturing tool. As an example, if you can find clay on Mars, you can make “green sand”. If you can then smelt Martian sand or ore for molten iron, you can print an item you need out of something that burns away at temps below 1,600C and use it for lost-wax (plastic) casting. Bigger printers allow for bigger parts, up to several thousand pound objects. Not everything should be made this way, but you can make a lot of things, including a lot of parts for your first foundry, with which you can start making everything else.
Doesn’t print electronics yet, doesn’t mean it isn’t actually possible to print in place electronics. It has actually been done for the simpler level stuff, and if you are designing to be able to print you end up with much larger footprint motor to get the same result no doubt, but it could be done. Really its the silicon chip, vacuum tube, or whatever other logic method that so far has never been demonstrated even in concept (as far as I know).
Also I don’t think once this is a more finished product/concept and as long as you are printing the next generation within the realm of sanity for the materials in question, so as to avoid floppy and warped parts, that you will really find much if any loss of accuracy with this method. It is all leveraging that common theme of thread pitch turns a fairly big rotation in small linear movement – a method that is fairly tolerant of imprecision in the production. And once you do that initial bed mapping stage to get all 4 corners happily co-planer after the early bed adaptive layer(s) even if the frame you printed is crooked the motion it produces will be barely changed from the ideal, if affected at all, and that same ‘bed mapping’ concept could also be applied to the frame itself – it can map its own lack of perpendicularity for the x-y axis and correct for it while printing its own Z or any future part…
The real problem with scaling up ever bigger generation will be the mass – that frame gets heavier, the amount of filament it must carry or lift from the supply increases, but the materials in use don’t become more suitable to this new load, and the energy requirements grow rather rapidly. Repeating the same scale each generation probably ends up with you having a heap of not quite perfectly compatible z frames and printers should you ever separate the two (wear pattern issues I expect), and a heap of waste z axis rails…
Capacitors, resistors, coils , motors, in theory every of these parts could be printed with different printable metal and plastic materials the obvious part which seems currently totally impossible in this build volume seems the microchip that could take any STL file and the complete program from the previous printer as input and the run and print it. if this impossible looking part could be figuered out the other would be relatively easy to add as a topping.
You aren’t going to achieve specs anywhere near standard components. A 3D printed capacitor wont have enough capacitance.
It isn’t just the processor that would be an issue, you need mosfets for the hot end and bed, unless you find a way to print wire that increases resistance as it heats and you are fine with a fixed max temperature. You also need some form of semiconductor to drive the motors, either a stepper driver or FOC motor controller, both of which would be impossible to print unless you can print semiconductors.
You talk about the processor as if with a bigger build volume you could print it, but that just isn’t true, especially not when only working with plastics and metals.
Slow but possible:
https://youtu.be/8BOvLL8ok8I?feature=shared
https://youtu.be/ZHqXe-ttbZM?feature=shared
But if you have volume and power to play with it doesn’t matter if the 3d printed version has worse energy efficiency and must be 20x or even 2000x the volume – it will be possible to create the functional passive electronics from a 3d printer with the right input materials.
But as far as I know the nearest anybody could claim to have come to a processor to run the show out of a 3d printing tech is a very small mechanical logic machine. The sort of thing that can add some short binary numbers…
ESL (equivalent series inductance) is one of the most important specs for a capacitor and this spec gets worse with size. I don’t think an FDM printer can print anything with the same specs as a ceramic capacitor that I can buy from LCSC for $0.005 regardless of power/volume limits.
Another comment linked a video showing capacitors made with additive PCB manufacturing, but that’s a totally different and expensive process. Not comparable to FDM
But when you have near infinite volume etc it doesn’t matter if the circuit ends up huge with more parts to overcome the relative weakness of the production method. As it stands the demoed 3d print in place electronics are terrible, but they are also functional enough you can design around their weakness.
I doubt any time soon it will really make any practical sense to do so for anything much beyond signal wire 3d tetris though a part and the simplest of print in place circuits. But that doesn’t stop it being worth developing or being a currently plausible if impractical technique.
Where the heck has [SunShine] been since the turn of the 21st century? Self-replicating (in-part) 3D printers were available right at the start of this whole subject years ago. Take RepRap for-example.
Quoting from [1]:
“RepRap (a contraction of replicating rapid prototyper) is a project to develop a low-cost 3D printer that can print most of its own components. As an open design, all of the designs produced by the project are released under a free software license, the GNU General Public License.”
The problem is today’s 3D printer cannot make all the parts needed for a stand-alone 3D printer or even a printer kit. Parts made of metal for example still cannot be easily or affordably made today. Time has proven this complication is what kills this romatic notion of 3D printers exponentially replicating more 3D printers.
The steps of generating and maintaining a reliable Bill-of-Materials (BoM) for the parts that cannot be 3D printed and therefore must be sourced from third-party manufacturers or distributors always seems to fall apart over time. As for kits: The added complexity of making a reliable BoM for kits plus sourcing and packaging the kits seems to only add to the added cost, delay, and ultimately – failure-rate.
I admire [SunShine’s] optimism and enthusiasm for the concept of self-replicating 3D printers. Yeah but… I’ve been around awhile (years), and even today I’m still skeptical. I wish him lots of luck, even though it’s a concept that has been tried yet failed many times in the past. But just because history is littered with failure, there is still hope for the future – provided something new is introduced.
Yes, I can buy a cheap 3D printer from China these days. But there are hidden costs in that cheap printer from:
1) Human cost. Are the people that source the printer and/or its parts paid a living wage and/or do they have a respectable quality of life (e.g. not a Slave)?
2) Future cost. What’s the point of buying an affordable 3D printer solution if the country you are buying it from is obsessed with going to WAR with you for egotistic-bully reasons?
3) Longevity. Given 1 & 2 above do you want to become the owner of a printer who long-term you have zero insight into whether it has long-term support or (better-yet) open documentation and source code?
* References:
1. RepRap
https://en.wikipedia.org/wiki/RepRap
Consider that there might be an enormous gap between the subject’s knowledge and goals vs. the author’s depiction of the subject’s knowledge and goals.
I posted a similar comment on the video. Sunshine does know all of this stuff, and talks about it in earlier videos of his. I had to edit my comment to acknowledge that.
My hope is that this mostly-nonfunctional self-extending Z-axis isn’t the final word we hear about this project. Someone, somewhere, needs to take the next step in designing a printable printer and replace the X/Y carriage assembly, with it’s reliance on rods and precision bearings, with some kind of printable compliant mechanism with decent longevity.
Man, that comment veered into insanity at the end there. Completely ignoring your China rant at the end, to address your concern about printing metal parts, RTFA where it says “[SunShine] seems to think he’ll be able to solve those problems now that the basic print volume problem has been addressed.” The video also addresses your concerns about metal components at a high level and says the topic will be addressed in future videos.
All of the components you’re concerned about have been demonstrated by other people using various types of printers. Sure, never all in the same printer, but that’s the project, isn’t it?
Your comment about BOM and kits falling apart, this is also as true for any available product.
They ofcourse pay someone to maintain this BOM, but either way.. prusa still sells printers.
But it might actually be more expensive to sell kits, since then you need to support all the schmucks who can’t even assemble shoes onto their own feet.. and you will probably get bad reviews either way! If they pay you to assemble you at least have a baseline to go off.
A huge problem I don’t see him being able to solve is making the main control board for the printer. Just printing plastic and metal I can’t see him making a microcontroller, mosfets, thermistors, etc.
I don’t know why people are so fixated on the current plastic output of this printer. Nothing about RepRap is specific to FFM or plastic printing the focus here is on the *mechanism*, which could ostensibly also apply to metal printers.
Impracticticalities aside, I love the engineering going into this.
One thought is to start with a short step Z axis. Allows the next layer of the leg to be printed before moving the gantry up.
For capturing accuracy and precision, using a standard base set is advisable. Some part of the frame that is stationary to calibrate to as layers are added. Touch the stationary base every so often and it should be able to compensate forbthe error and print the new set of layers to get back to acceptable tolerance as it moves up. Like anything, the better it is at keeping position and tolerance, the more costly it can get. That I s a separate engineering challenge.
For the set base, i would try a plumb bob. Spool out a wire in a straw over the weight. The weight hovers in a hall effect sensor that shows measures the wire is in relation to the straw. I’m happy to help with this concept.
“What’s the point of buying an affordable 3D printer solution if the country you are buying it from is obsessed with going to WAR with you for egotistic-bully reasons?”
Well first off you get an affordable 3d printer. duh.
Lol, exactly what I was thinking, especially since all the parts are easily replaceable, it makes 3D printing more affordable than 2D color printing
Reminds me a lot of self-erecting tower cranes that lift themselves up by adding segments to their own tower. As far as a self-replicating manufacturing machine goes, though, remember that the RepRap project was founded 18 years ago with that goal. Its open-source designs gave rise to today’s consumer FFF 3D printers, most famously the Prusa i3 line, and the components making this printer possible.
Adding rollers to those threads could be the z solution