Wednesday, April 10, 2013

What cannot be 3D printed? Time to debunk some myths!

General 3D printing misinformation: review of facts and fiction!

Each technology has some fundamental limitations, and 3D printing is no exception.




As soon as I owned and wrote about 3D printers, people started "spamming" me with mainstream articles about it (no pun: still I like them and still I learn from them!). So even our grandparents would soon be printing their glasses, their bikes, their firearms, their houses and even their own legs!

For sure, I am not the one that will disenchant this emerging technology: I am writing this blog and I regularly post new designs on Thingiverse repository.

But as I write this (april 2013), you just cannot buy a 3D printer and print anything, nor even expect it to work as advertised if you do not understand very clearly how the thing works. It is just not reliable enough for the average user (unless you have no special expectation, in which case this pay doh printer may avoid frustrating experiences). It is even quite hard to reach an acceptable quality even for people born with a screwdriver in their hand.
Update 2021 : we're closer to reliable, out-of-the-box 3D printers. Still they often need to be properly tuned, especially the low-cost ones. And time will not fix their other drawbacks. Please read on!



First, let me state again that I do think 3D printing will become a major step/shift in the industry, and a huge one: I somehow agree with people that think it is even a new paradigm. However I realized that it could be useful to write down some of the misconceptions or inherent shortcomings instead of telling always people to clam down...

Also, let me say I still do write mistakes myself. But I just want here to tell what's real and what's plain untrue in my opinion in the mainstream media. There are lots of naive errors made by lots of writers that go and go uncorrected.

Actually, many reviewers just don't have a clue about what 3D printing at home really is. They also miss the fact that so-called fused deposition modeling exists in the industry since the late 1980s. Some brag about telling out loud that everything will be 3D printed in the near future - and especially firearms. Let me say I consider that there are just two kinds of journalists. The generalist usually relay crap, mostly redundant and unchecked. Sure, earning a life as a journalist is very hard and you cannot spend much time on writing articles because of that. But it is no excuse for badly reported information and less even for misinformation. like in my nearby newspaper recently.

There are too few journalists or writers that do a proper job on the matter, who are unsurprisingly usually more technologically-aware. The geekiest readers could have a look at wired or 3dprintingindustry for example. (update Jan 2014:) This article from Howard Smith is worth a read as it asks the broader and excellent question in the first place: do consumer really want to 3D print?

OK, back to 3D printing at home facts and fictions.


The peak of inflated expectations



Gartner did a brilliant analysis of how emerging technologies are perceived and talked about. They resumed it in their "expectation curve", or what they call a hype cycle themselves though it is no cycle at all (!)

It illustrates appropriately what is up here, even I you will have to shift everything significantly to the right because this one is obsolete:

Gartner's Hype "Cycle" for emerging technologies in 2010 (warning: this is really old given the content!)

See how 3D printing is (now) at the top of the expectations and will soon fall due to "disillusionment", before it really gets known, serious and mainstream? Remember how "virtual reality" would be everywhere and how "speech recognition" would discard your keyboard at their time? Virtual reality nowadays only start to appear on smartphones or Google glasses, while speech recognition is known enough to be used only where it works and make sense (once again mostly on smartphones, not on your desktop PC).


As I said last year, now are the times where everyone hears about 3D printing at home. No magazine can afford not to write something about 3D printers: I read it in "issues" ranging from gardening to religion! So/and this is no surprise that this is the year where most bullshit will be told also. :)

I guess that in half a year or so, there will be much less noise.

Existing 3D printers just do not match the average user expectation, seriously! It is no click-and-play technology, not at all, and this is why until recently you just could only get kits to mount yourself. Even if printers were reliable, you still would need to get the 3D model of object to print in the first place!

Now, 3D printing will necessarily make its way to the plateau of productivity of Gartner's curve. It will probably be quick to reach the market given the usefulness of the technology. Also, hey, a real lot of money could be made out of a completely reliable and user-friendly printer!


Quick factual data and hoax busters...

The good news: what's really new!


  • 3D printing at home is now possible because it is much cheaper than in the industry where it costs $10K+ or even $100K+ for a printer. Many thanks go to people like Adrian Bowyer, the father of 3D printing at home, and the "reprap" open way of life. And please, do consider twice before you buy an entry-level "profesional" 3D printer such as the UPrint or Cube from the two "major" 3D printing companies: you most certainly will end wasting tens of thousands dollars for nothing (full analysis here).


  • You can repair, design and produce complex objects at home (check this household category for example). But be aware of the limitations I talk about below.


  • The required software is free and functional. You need nothing to pay beyond the very hardware to build your printer and the filament you print. But it will not be that easy, because of the many options and tuning to achieve a great print.

  • It is a significant industrial change Together with micromanufacturing and fablabs, 3D printing could definitely be considered an industrial revolution 2.0 (actually the third one in history). Not only do you get a small factory at home to create new things yourself, but you also can repair thing that were ditched otherwise, which in turn fights planned obsolescence. Moreover, commons-based peer production is better for the environment as the carbon footprint is much lower from factory to the user (especially when we will be able to make and recycle our own filament). Now it will also kill some jobs, while it will create new ones as with any new technology. Oh and I really need to read MIT's Neil Gershenfeld's visionary book "Fab: The Coming Revolution on Your Desktop--from Personal Computers to Personal Fabrication" (2007).


The bad news: how limited is 3D printing?


  • Well... you need a 3D model of your object before you can print it!


Well, this is embarrassing. You just bough a 3D printer and, well... you would like to print a replacement for your broken part of your vacuum cleaner... Where to start from?!

Either it exists in public (or soon, private?) databases such as thingiverse, or you need to revert to the tricky field of computer aided design. I am sure that there are jobs to be created here and I would gladly be one of the first : "have your own special thing be designed by a professional for a 3D printer". I design it, you print it. Else, if you have no design you have nothing to print. And you cannot print it either if it is not well-suited to 3D printers (like when it has no flat parts, or big overhangs...).

One of my specific handlebar GPS mount, explicitely designed for a 3D printer.
You are lucky if yours exist, else you must design it yourself with CAD,
or adapt an existing source code, or subcontract a designer.

Low cost 3D scanners are also under development, but it is a long way before you can scan an object and print it without a lot of in-between work to fix the scanned mesh. And, once again, scanners will not get you all the intricate internal shapes of an object anyhow. Think about it: you just won't be able to scan a pair of scissors and get a functional printed copy!

And this is only one of the fundamental limitations of the technology...

  • Precision is limited


3D printers have a limited precision. The nozzle output is about 0.4mm, and trying to print details below 0.1mm is usually worthless. It is very well suited to many objects, but you will have hard times trying to replace the smallest broken nylon gear in your compact camera with a 3D printer. Even the professional or semi-professional selective laser sintering technique will not make it possible sometimes.
Check my fingerprints: I can't get really much finer than that,
which still will not meet technical requirements in some cases
(treefrog credits: MorenaP)

Precision can be increased by using a smaller nozzle. But you will suffer from excessively long prints. Think about like as if you were filling a 10cm square with a 0.1 pen tip and on 1000 sheets of paper! Even at a speed of 10 cm/s (which is fast), it will take ages.

Also a smaller nozzle will make it increasingly difficult to fine-tune the proper printing parameters. The quality and homogeneity of the filament being extruded will also impact the extrusion much more.

  • Printable material is limited


Talking about material, well... 3D printers just print a very few. Check my review of printable materials. Beyond some low-temperature melting plastics, you have a few pastes. Then, professional printing techniques offer metallic powders to be fused by high power lasers, but the results still may be brittle (or at least quite expensive).

Do not think you could print anything that will stay close to an engine that heats (ie. most of them), as your object will just met. For example, the common PLA we use becomes soft as low as 50°C (~120°F), and no plastic you can print will stand more than 200°C (~400°F).

  • Printing/production speed


Printing all this carefully took me ~6 hours: it is really slow!
As its alternate name tells it, "Rapid prototyping" 3D printing is just not meant for mass production. As we already said it, slicing a bit object into 0.1mm layers and filling them sucessively with a 0.4mm nozzle just take ages. In fact, as of today, printing something quicker than 20 minutes is rare, and any owner experienced hour-long prints (if not day-long). And the longer it takes to print, the more chances are that a failure happens at some point in the process...

Unless you have very specific requirements and unless you can pay a lot for it, a 3D printer is NOT a productive factory... This is even true for professional 3D printers, except for some limited fields like in the prosthetic or the space industry, or when deeply mixed material is needed, or when intricate shapes just cannot be molded or carved. Past a few items per day, you will find it is really not appropriate.

By the way, when you want to switch to a mass-production, the prototyping design may not be compatible with the industrial requirements (eg. molding a formerly 3D-printed object will require a complete redesign if it was not though about at the very start of the design).

Seriously, just stop at once when you reading or listen to someone that tells you that everything will be printed in the future: this is plain false and naive. Once again, think about how quickly you can mold something compared to fused deposition modeling. It even explains why some people use a 3D printer to print a mold (!) when they aim for a small production.

  • Not all 3D printable shapes can be printed


For now, 3D printing at home relies on a filament pushed through a melting nozzle. For this reason, it will not print on top of nothing, which in turns impacts what you can print and how you shall design the object.

Consider the letter "M". The inside is a so-called "overhang", and it requires support to be printed because it floats in the air. Support structure may be added automatically by the software, but it must be carefully and manually removed by hand afterwards (unless you use esoteric and complex dual-head setups with water-soluble PVA plastic for example). As soon as the overhang angle is above 45°, the printer will start pouring plastic in the air, which generally ruin the print. Hence the 3D object designer must take this into account.

The flat tab on the bottom right just
helps printing reliably, else the thing
usually pops off the bed before the end.
Also, the shape must stay in place during the print. If the part of the object in contact with the bed is too small, as it happens for a sphere, it will most probably pop off before the end of the print (the plastic retracts a bit when cooling down). Once again, no definitive answer exist to make sure the printed part stay on its the support: special tape, glue, heated beds, wood, PCB...

As a consequence, we take this into account and add sometimes explicit tabs or "rafts" to counteract the issue, as shown on the design on the left. It must be removed manually when the print is over.


Note that there are other professional printing technologies that avoid the overhang and popping off issues, such as those based on fusing the top layer of a powder with a laser, then adding another layer of power and cycling. Some other use a liquid polymer that is cured with hard light or lasers. Except for a very few attempts such as the Form1 on kickstarter, none of them made their way to our homes yet, and they require quite messy powders or polymers with their own sets of limitations.


  • 3D print a gun? Oh stop talking about it to me!

Why not print a better politician or a better journalist instead? :)

I just hear this so often. As usual any new technology triggers fears. Of course, humans are curious and quite aggressive, so it is just plain obvious that some try to do "nasty" things with a printer.

Now I have a better advice: just go to your nearest hardware shop and buy some plumber pipes! It will make a much more robust gun that will not fail after six rounds and will not less likely blow in your hand... Now, of course, this would not make big titles, nor will the terrorist-fearful elderly be interested in fused-deposition modeling technologies...

And for sure, soon or later someone will really manage to print a "fully" usable gun that will even get undetected at the metal detectors (and get you in bigger trouble even, as a possible illegal weapon exporter or maker). But what about printed ammo then? Actually, you can also print a bomb if you are a fool and manage to shape explosive as a melt-able filament. And nobody ever printed a bazooka?

I just have mixed opinions on this. I can understand that people and government don't like it (especially when they do not want third-party resellers...). But declaring the printed guns and their design illegal is obviously not enough to bar terrorists from using the printers. And once posted it is too late, ie. the war is lost in advance...
Just write "3D printing" in the box, Randall Munroe (XKCD) it so right once again!

When the cavemen discovered the club, it was most probably already the same issue: owners did not want others to have one, neither did the tribe rulers wanted uncontrolled people to start bashing randomly. And any new technology was tried on the war field before it proves "useful" or not (according to your point of view).

Once again everything would probably be easier with more education, culture, tolerance...

  • Reliability and tuning is hard

Printing an object requires extensive use of software and algorithms. And they are not yet quite user-friendly because of the many settings that can ruin your print from the software point of view -- oh, did I tell about the hardware setup also?

Oops. Non fatal over-extrusion in the (not so) many parameters of Cura
Since you need to "slice" your 3D objects into layers that are printed one of top of each other, it is easy to understand that there are about 10 parameters at the minimum to be taken into account (in reality the software just hides hundreds of parameters that nobody wants to tweak!).

Most often, there is no way to recover from a bad setup like above. And claims of raw abstract art will most probably not stand long!

In fact, The Art of 3D Print Failure is a dedicated flicker group! -- this one is from wolf e twain



Caveats to know about before buying a 3D printer


Also, if you are about to buy a 3D printer yourself, I warmly recommend Hrvoje's Guide to Not Buying a 3D Printer. There are more and more corporate bullshit that makes it more and more difficult and he gives some very useful hints.

His post also refers to this one by Daniel Brown who explains in detail the risks with low cost and/or crowd-funded printers (with a lot of data for the $300 "hard" limit).

My own guess is that below $600 you must first be a tinkerer. The cheapest and the more certain you are to run into issues, so gettig a dirt-cheap printer "just to try" and 3D print is a very wrong strategy. No printer is 100% reliable as now (update: still even in 2014!), and all the cheapest ones do rely on risky trade-offs.


Conclusion: please take care about what's being said about 3D printers!

In my humble opinion, writers and reporters should at least read about 3D printing first in the wikipedia or the reprap website, or personal experiences like this one. Better even: go and check by at your nearest fablab, spend a few hours there, talk with people, and only then make a worthwhile article!

Actually, a local newspaper recently sent a journalist at our French fablab. He spent a few hours there and then wrote about "3D laser printer" in his paper, mangling names and technologies by the way. Another one stated that Dell was building 3D printers, after he went to a meeting organized by Dell and where a booth was given to a third party which featured 3D printers. This newspaper is very well-known and it hurts me that they are so bad: I would not trust them whatever they will be talking about now.

So please readers, keep on reading and whenever you find a piece of information valuable, just check it against different sources: never take something from granted, even if it comes from this blog! :)


2 comments:

  1. To add one more point: there is some free software, but not all is mature. You might do some things in one, and have to jump to another for some other things you want to do with your model, and then you first will have file format conversion problems, and then have to learn a completely different interface in the second program. And then you might find that the software in principle should be able to do what you want, but in practical life it does not ever seem to get ready with your specific problem.
    Some knowledge in mahematics could be very handy, together with some basic understanding of 3D coordinate system transformations.

    ReplyDelete
    Replies
    1. Thanks & true! Actually, I managed to use almost exclusively Cura so far, while some other use Repetier and other netfabb, and so... None is far superior to the other, but users tend to focus on only one as they are all hard to master besides the "basic/newbie" option (which still fails often).

      And yes, there are serious differences (and sure, technical ones). The biggest come from the 3D slicers: skeinforge / kisslicer / slic3r... Really, what's nicely sliced with one may sometime be poorly done with another.

      E.g. the annoying "thin hollow double walls" appear when the inside is left unfilled because the outer walls are too close to each other for the slicer to add material (given the nozzle diameter). In theory you can detect it by looking at the head "path" in the slicer before sending the sliced files to the printer (kisslicer is great at showing this, while it fails at slicing things more than skeinforge/cura -- which in turn is very slow, oh...). But checking the path is often overlooked and not very readable. And anyway, it is usually quite difficult to get rid of such holes in the print. Using another slicer is a good solution here... or tweak the design slightly accordingly.

      Still a messy process indeed. I guess/hope this year will show huge improvements in this field.


      Delete