Hacker News new | comments | show | ask | jobs | submit login
The Achilles’ Heel of 3D Printing (iijiij.com)
29 points by aniijbod 1665 days ago | hide | past | web | 56 comments | favorite

I happened to be reading Make magazine's 3d printing guide, and I was struck the similarity to late 70s/early 80s personal computers. Lots of different technologies. Kits. (Mostly) terrible cases, even wooden ones. All sorts of "solution looking for problem" kinds of examples - 3d printing a coat hook is the "key all your recipes into a computer" of this decade.

To outsiders, 1970s computing looked like a bunch of kooky hobbyists, and it wasn't far off. I think in 10 or 20 years, 3d printing is going to be the solution to lots of problems we haven't considered yet.

The difference is that 3D printing is already 20 years old. It's maturing a bit more slowly than did software.

That's very misleading. In the late 1970s computers were already 30 years old.

Yes, the "PC" was new. And that's the whole point - we now have cheap (but rubbish) desktop 3D printers. It's the equivalent of the transition from mainframe to minicomputer to PC.

While you're right that both were old technologies. 3DP is noticeably slower in many variables. If we look at the "quality" of output from the mainframes -> Apple II, we see that the calculations would of course be the same and the programming largely the same - the discrete nature of programming being what it is.

However, the dimension of cost for 3DP (which is what the personal 3D printer movement is excited about) is only one dimension of many, and I'd argue not the biggest limitation.

3DP materials and process quality have certainly improved over time, but all of the current technologies are dead-ends in regards to competing with traditional processes - i.e. FDM/SLA/ProJet/PolyJet don't physically seem able, ever, to produce parts as nice as injection molding can for a myriad of reasons - even without cost as a concern. The situation for SLM/ProMetal processes is little better.

Perhaps the recent investment in 3DP will increase R&D spend on quality, but it doesn't seem to be the case if you look at the major or "maker" players.

That list of criticisms isn't really so harsh but rather honest constructive suggestions and realistic assessments of where 3D printing is now.

That being said, there are a lot of comparisons to 70s era computers and it seems to be a valid observation. However, past performance is not indicative of future results.

In 10-20-30 years after the 70s hobby home computing club, applications are obvious.

So, maybe this is hard to ask, but what applications do you think are going to be obvious in 30 years time? What is the Visicalc of 3D printing going to be? What if there is no killer app?

The killer app is "just in time" manufacturing of single parts from a gigantic library. This will create a "long tail" type business that can offer a huge line of obscure parts that otherwise aren't available, as the original manufacturing lines will have been closed down long ago.

Someday, 3D printing will probably be fast enough and improved enough to displace a variety of low-volume manufacturing applications. "Improved enough" to my mind would be manufacturing in metal with much more precise tolerance.

>The killer app is "just in time" manufacturing of single parts from a gigantic library. This will create a "long tail" type business that can offer a huge line of obscure parts that otherwise aren't available

Have you ever held a McMaster-Carr catalog?

I have. That is where I order parts for my 3d printer. Interestingly, you are making op's point because I frequently have to order from other places because as long a tail as Mcmaster-Carr can cover, it doesn't come close to being all that I need. It is essentially a comparison of a long, but finite tail vs. a practicaly infinite one (ie, anything I can create in a cad file).

> That is where I order parts for my 3d printer.

And you make my point :)

some of my parts :)

I have their catalog and order from them frequently. Did you have some point you were trying to bring up? Because the "long tail" type of business I envisioned in the previous post would be for stuff that's more obscure and, more to the point, discontinued.

I'm just trying to say that a company relying on "traditional" manufacturing methods can still offer almost half a million different, and mostly obscure, parts. I'm pretty sure they don't have half a million different production lines for that, and most of their stuff is CNC machined. So, if the demand for even more obscure or even completely custom parts ever becomes economically significant, they can probably adapt to supply that need as well, without using any 3D printing whatsoever.

Actually, McMaster doesn't manufacture anything, as far as I know. They are a distributor. (They tend to deliberately obscure this by refusing to tell you the brand name or actual manufacturer of any item they sell in advance.)

But back to the main point. The "long tail" type of business I envision is one that manufactures one-offs, or quantities so limited that the volume demand is below the level needed to interest a distributor such as McMaster.

The problem is the number of parts you're going to be able to produce is limited by the materials properties of whatever you're accreting.

Sure, you'll be able to produce anything that's mostly cosmetic, but how about something like the bearing housing for an old washing machine? I have to believe that will still get milled because it has to have certain mechanical properties a 3D printing process can't guarantee.

That's why I note that improvement would involve manufacturing in metal. This may be possible already to a limited degree, for example, with sintered metal, but that's an expensive process that isn't economical for washing machine parts.

Yeah, sintering could probably be done. But getting the right shape is only half the battle - you also have to get the right properties, which can mean getting the right kinds of dislocations and grain growth, which you normally only get by controlling the (high) temperature profile.

Laser printers are everywhere now. There's one in every office. But there are still printing presses.

Everyone expects that the "killer app" will be replacing injection molding or something. It will be something much more unexpected.

Exactly, and this is the example I use most frequently. Home printing let us make flyers for lost cats and party invitations. You could feasibly print out your own book, but no one with a brain would do it.

3D printing (which should realistically be called what is has been for years-additive prototyping) doesn't replace any part of the major manufacturing industry, it compliments it and makes it available to do a few things at home.

What about the ability for an AI to design appliances, furniture, etc for you based on what it knows anout you and the answers to any questions it thought relevant. Essentially your own personal architect, interior designer, industrial designer, etc wrapped up into one simple and quick interface.

Cool idea, but look at the unit cost. Right now, the per-unit cost is way too high to make that economical, especially for something large like furniture. Also, the value of your idea lies in the "human responsive AI," not the 3D printing.

Oh, I totally agree that unit cost right now is way way too high. That said it is many orders of magnitude cheaper to do a one off plastic part with a 3d printer than an injection mold. Also, 3d printers are just one small piece of the one off automated factory.

My company has over a dozen 3D printers in the our design office in SF and our manufacturing facility in Shenzhen. They range from $250K professional units to $1K consumer models. I've seen the progression from old 2001 models to the the state of the art models purchased this year.

12 years later, and the latest printers are good for 1 purpose.


Sure, a hobbyist may find this tech somewhat useful. For example to print a case for his raspberry PI. Even that takes 4 hours to print, and if he drops it, it will shatter. Oh and finish is TERRIBLE. If a mass produced version is available it will always be better in every way.

I've never held any 3D printed part in my hand that compares to a nice soft touch painted part, or even an injected molded piece.

This tech is like where the semiconductor industry was in the 50s. 60 years from now, we'll be printing some cool shit. But in a few years time, we're not going to be printing out shoes, headphones, and furniture like people here are dreaming about here.

>if we suddenly discovered a way to do 3D printing (probably using 3D technologies other than those in current use) that was anywhere near as fast as conventional manufacturing, the resulting revolution could [...]

And that is 3D printing hype in a nutshell: yes, the technology and its conceivable developments are not good enough to replace any significant part of mass production. But if some miraculous new development happened, if only this technology could develop at a pace as fast as some incredible outlier such as integrated circuits, then it would be a revolution!

In fact there is nothing that revolutionary about 3D printing, and it is very closely related to manufacturing technologies that were simply developed earlier because they held more economic potential. At its core, a 3D printer is a working head that is computer-controlled to move in three dimensions with an attached tool that deposits small amounts of material. A CNC milling machine is also a 3D-moving working head with an attached tool that removes small amounts of material. This is far more useful because you can work with any kind of material you want.

"Ah," the 3D printing evangelists will say, "but that is so wasteful! We need an accretive process to economize material." Except we already have another widely-successful industrial process that also works by adding material: industrial robotics. An industrial robot arm is a computer-controlled 3D moving working head (notice a pattern here?) that can add new parts to an existing workpiece. The parts are standardized and mass-produced separately, at a ridiculously low price, out of ridiculously strong materials.

So between CNC milling and industrial robotics, which are basically the same technology with different working heads, there is just not that much of a niche for 3D printing to make a dent into. Yes, that could change tomorrow if aliens came down from heaven and brought us a miraculously advanced 3D printer. But at that point the discussion reaches ground previously reserved for messianic religions, a discussion that I'd rather stay out of right now.

You nailed it other than the bit about robot arms adding parts. 3D printing and reductive fabrication methods both make parts, it's not about assembly.

Still, 3D printing is in its infancy, already the food industry is getting more and more of them, they're being used in different parts of the process (for instance: mould making, which is a very delicate art) and they're getting more precise, faster and cheaper. Let's give it a few years before we declare it a failure, the camera obscura wasn't a very good DSLR either.

You can sometimes make the same product by either shaping it directly or by making smaller, simple parts and binding them with e.g. screws. That's why I feel industrial robotics should be part of the discussion.

I do actually agree that 3D printing has pretty convincing use cases: food industry, small aluminum parts for airplane engines, customized cartilage replacement etc. I just think these are relatively small and non-revolutionary niches.

It is mis-titled, its a survey of what 3D printing is currently good for vs what it isn't good for. It notes that it's not good for punching out billions of simple parts.

The article is written for investors who are looking for the 'next big thing' and gives some interesting metrics for when 3D printing is worth investing in, vs now when it is a niche. Similar to other stories we've talked about here, the 3D revolution is coming in 2020 not today.

Look at this picture: http://www.reprap.org/wiki/File:2012-12-15-clone-wars-geneal...

It is about a project that started 1 year ago in Spain making 3d printers and now have more than 100 of them(the graph gets outdated fast).

There is today more new people coming to this technology every single year than the entire population dedicated to them before.

3D printer is not about what could be done with it today, but about what these people is going to do in the future from that.

E.g Inexpensive(1/10 the price of filament) plastic printing from pellets: http://makibox.com/details/product/makibox_a6_ramen_bundle

PS:I have a 3d printer and love it.

I have a 3d printer (makerbot), and I love it.

But I also produce low-volume injection-molded plastic parts.

and building the molds and then producing the parts is still faster.

I'd love to read a post on your mold making and injection molding adventures.

Well, the product is a LEGO-compatible case for iPhone 4/4S and iPhone 5, available at smallworks.com. It was my then 12 year-old son's idea.

The first mold, for the iPhone 4/4S) were made in China. The second was for the iPod Touch (4th gen). We've modified the second here in Austin for the iPhone 5.

We use Solidworks to design the part, and then outsource the actual mold design. If it was a two plate mold (just cavity & core), then we would probably do it ourselves, but all three (actually, 2.5, I guess) molds have 3 sliders as well. These require a semi-complex mechanism, and need to be hand fit (with bluing, very old-school/traditional) in order to hold the pressure of the injection molding process.

The pieces for the third mold were all done with EDM, rather than traditional CNC machining.

We do the injection-molding just outside Austin, though I'm considering moving it to Houston.

Duplicating LEGO is about as difficult as it gets in the injection molding world.

Though my degree was in ME, I'm really more of a software guy.

(We're about the same age, btw.)

Super stuff man, really. And that's a great idea, your son really got that one right. Amazing story.

I took the liberty of sending out a tweet about this, hope it will get you some sales.

I've worked with a professional moldmaker where we had molds made for the casting of the base/table for a CNC rig (very hard to make a stable rig, lots of interesting postprocessing for hardening).

Molds are fascinating, especially molds for high pressure/high speed injection, the precision and the amount of trickery to deal with warp and cooling are a lifetime of study.

I never quite realized just how much work goes into making molds until I saw it up close, those are very complex devices, far more complex than I ever imagined. Part of the reasons for using EDM rather than regular machining is probably the finish, the precision and the hardness scale of the materials used.

You might enjoy this:

I worked for a company that built CNC conversion rigs and I never really lost the bug for metalworking.

Sounds like the same old saw that the computer cognoscenti said about desktop computers. From my point of view, I don't give a damn about any of their quibbles---they are either no longer a problem or subject to on-going R&D. What I want is something that allows me to proto-type my ideas. Plastic of any sort is just fine for testing. If it works, I can consider small production runs in the material of my choice. Good heavens, I can even have it cast in gold or sterling silver if I want :) This technology allows me to move from day-dream to reality. If the 'industry' scoffs let them, I'll be busy hacking away...

I've been talking to Josef Prusa and Kliment about going from print to cast. There's a few other guys who exactly that. The 2 makers' suggestions are to cast using natural PLA. It burns off the cleanest of all the current plastics.

Along with that, I am very interested in Prusa's print head. It will be able to print PEEK, nylon, polycarbonate, and other hotter (and noxious outgassing) plastics.

Perhaps 3D printing will never be quite as fast, but it won't end up mattering very much? The article did alude to one possibility - that demand for mass customizability would alleviate any speed disadvantages.

A practical example might be: Let's say I need some chairs. I can go down to Ikea and pick up one of 5 mass produced chairs they have and be done with it in half a day. Or I could browse through designs and reviews online, pick something I like and have it arrive a week and a half later when it's done printing (or have my neighbourhood or home printer take that long...)

A bigger advantage for 3D printing has more to do with onsite, personal manufacturing, of which "mass customization" is a secondary benefit.

Many of the objections the industry insiders use to describe additive printing are more or less the same arguments people have made about homebrew computing, back before the Steve Wozniac showed off the Apple I and IBM got into the game. Why would you use personal computer? They are slower than the minicomputers and mainframes. They are "craft" machines for hobbyists.

Ideally, we are moving to a future economy where you "buy" a product by purchasing a license to a design, and then having your household printer produce the product. If it is for something larger or requires more exotic materials, you might place a print order at the neighborhood Kinko's, Ace's Hardware, or print co-opt.

Ideally, we are moving to a future economy where you "buy" a product by purchasing a license to a design, and then having your household printer produce the product.

Speaking for myself, I don't want to live in a future where you've done what it's going to take to enforce those licenses.

Generally, I think ubiquitous, cheap, personal manufacturing is a good thing. If we have a number of Open Source products we can print out. That will substantially lower the cost of living for everyone. Hopefully, the vast majority of retailers will turn into service providers.

There may still be folks who want to license designs out.

Nobody trusts a corporation to enforce their DRM mechanism because they get to be the final arbiter of what is "fair". What is fair is always good for the corporation, never mind the user. They may die and kill off access to those goods. The other extreme is hypocritical, simply because we run our digital goods through hardware -- all resting on the notion of allocation of resources. Besides, content creators want to eat, too.

A reasonable compromise is to have the DRM enforced by someone other than the seller. An escrow service is an improvement, but is not much better. And no one wants to depend on the government to host DRM servers. A peer-to-peer DRM might work. Bitcoin is an example of a technology that can be modified for this: http://codinginmysleep.com/exotic-transaction-types-with-bit...

The mediation is done peer-to-peer. You can, for example, sign a bitcoin contract saying you are lending an ebook to a friend, and have it published elsewhere. Granted, this would all be on an honor system -- if you can play back media, then you have access to the content, no amount of clever algorithms or coding can change that -- but so is much of the enforcement for civilized interaction. A small group will "cheat", the same way there are breakage in a retail store. If the voluntary participation of such a scheme benefits everyone involved and is not too inconvenient (centralized DRM servers that crash or gets taken down when the company dies, or requires a constant internet connection), I think most people will go for it.

It's best we experiment this kind of transaction with digital products, before 3D technologies get good enough that people start buying design licenses for goods.

Better yet, we make sure we have Open Source versions of products that people can download and print out.

This seems obvious. A generic solution can do everything, but there's no reason to expect it would be the best at anything; a specialised, single-function machine can always beat a 3D printer for performance, for the simple reason that printers have traded away performance for versatility. And in software, abstraction gives you flexibility while degeneralisation gives you efficiency. It's the same old story.

" a specialised, single-function machine can always beat a 3D printer for performance"

But there is no more specialized thing than a 3d printer. You can print glasses for your face, not for a general model of faces but for your unique face, your specific sizes in all dimensions.

Getting to performance on the big machines is cost prohibitive for lots of things because you need big runs.

You misunderstand my use of the term "specialised". A 3D printer can print anything, barring some of the limitations outlined in the article. Most machines are intended to serve much more specific functions, and do so with great efficiency: I can print a hole in any shape I like, but if all I want is a hole, why shouldn't I just use a drill press?

I dont think this article, which is interesting, is about what it says it is about. I don't think anyone seriously expects 3d printing to replace mass production any time soon, but i do see it as part of a series of related trends:

We need fewer, smaller physical things, and less stuff, to enjoy life. My new TV is has a larger screen and weighs less than its predecessor, and even so is largely redundant given i have an ipad. I used to need a laptop and a desktop computer, a games console, and so forth. Most or these now gather dust. I dont need a dedicated "movie" room, so i need less space and less furniture. And so on.

When we do want things, we often want them highly customized. Consider the insane variety of phone cases selling in popup stores all over the place.

I suggest to the author that anyone poo-pooing 3d printing today is in the exact same boat as, say, hot metal typesetters looking down on photo-typesetting and laser printers for their comparatively low quality back in, say, 1985. First they lost their low end customers, then all their customers, and then the people who had replaced them lost most of their customers (ok that's in progress).

> We need fewer, smaller physical things, and less stuff, to enjoy life.

This isn't an axiom. It's a personal preference. Don't imagine it applies to everyone else.

I don't think it's an axiom, i think it's a consequence of convergence. I used to own a camera, a video camera, a personal organizer, a landline phone, a cell phone, a discman, a gameboy, etc. I now have all these things in one device that's smaller and lighter than any of them used to be.

> I used to own a camera, a video camera, a personal organizer, a landline phone, a cell phone, a discman, a gameboy, etc. I now have all these things in one device that's smaller and lighter than any of them used to be.

And it isn't as good of a camera as a modern dedicated camera is. In fact, it might not even be as good as the camera you replaced it with, especially when it comes to ergonomics. Similarly, a dedicated gaming device often has better ergonomics for gaming than trying to play the same games on a phone.

I understand convergence. I understand satisficing. I also understand why people still make SLRs, and it isn't just to keep dust out of the various lenses. If you really want a good camera, you aren't going to be fully satisfied with a phone that has a megapixel count. The gaming world is more fluid because games themselves can adapt to new hardware and interface styles, but specific games often really do go better when you have buttons as opposed to a touch screen.


And, finally, my little quirk: I refuse to believe that decluttering is a valid end in itself. Having a closet full of stuff I enjoy owning isn't a personal failing or a personality flaw.

The primary objection to 3D printing in the article seems to be about speed in transforming raw materials into a finished product, but that's not the only thing that needs to happen to sell a product.

The entire process is:

1. Transform raw materials into a finished product

2. Transport product from factory to store or warehouse

3. Transport product from store or warehouse to customer's home

Looking at it like this, it seems to me that the speed of 3D printing relative to other ways to transform raw materials into a finished product is irrelevant.

What really matters is the speed of 3D printing relative to other ways to transport the product from the store or warehouse to the customer's home. In other words, how long does UPS/FedEx take to deliver or how long does it take to visit the local mall?

All of these "bullet points" about 3d printing sound suspiciously like people 30 years ago talking about why personal computing wasn't going to be a big deal. Makes me believe in it even more.

3D printing is disruptive, not to industrial manufacturing, but to the retail value-chain. As mail-order is currently disrupting retail, 3D printing will disrupt both retail and mail-order.

"Ink" will be sold as a commodity at supermarkets/gas stations (perhaps eventually piped directly as a utility, like gas or water - though our consumption would need to increase dramatically to justify that).

Profits will be made on product design, and the industry will look just like the downloadable information-based industries of today: software/games/music/movies/education/news. Some business models will be a service related to the product, not the product itself. Although long-tail diversification will be possible, it mostly will be be dominated by a few products and corporations (due to the same factors as today: concentration of resources, quality assurance of trademarks, wish for community support (e.g. SO questions), marketing reach and herd mentality). Although initially suppliers will be diverse, the industry will soon consolidate - the same pattern as every other new technology.

Components that can't be 3D printed - e.g. silicon - will become addable modules (a cheap commodity, bought like screws/nails/pet-food pellets that you add to the hopper). Designs will modularize around this interface.

There'll be open source versions too - also consolidated, for the above reasons, as they are today.

Looking at the bigger picture, it's fun to see the Hype Curve in action. 3D Printing is just passing the peak of the first hill of the curve. So we should expect the hype/expectations to drop quickly for a while, bottom out and stay low for a couple of years, then recover to a useful level where it will stay.

All part of the cycle.

As a layman, I thought the main issue with 3D printing was the material (the article only briefly mentions this). Plastic goes a long way, but still, when I look at almost non-trivial thing around me, I tend to see more-than-just-plastic.

Same comments were made by printshop owners when Apple introduced the first Laserwriter. Sure indication of the big guys well on their way to being referred to as dinosaurs.

Wonder if there will be elemental composites akin to RGB with 3d printing (e.g. rubber/metal/plastic -- layman's thinking) to which [most] all things can be created?

I consider 3D printing to be mature technology once a 3D printer can clone itself! I bet at that point von Neumann smiles in his grave for self-replicating systems.

I may be wrong, but there are a few problems with a self-replicating printer of this type:

- The size of the stage determines the largest size of part that can be printed. The stage is necessarily smaller than the largest part of the printer.

- The print head has to have a melting point greater than the material being printed. Meaning it can't be made of the material being printed.

RepRap is a project with the state goal of creating a self-printing machine like this. But looking at the current designs, they mostly stick together common materials with some 3-D printed bits. This is much more attainable and useful than really being able to recursively print a new machine, because the common parts are produced at scale through existing channels and are stronger/cheaper than the machine-printed parts.

The Achille's Heel of 3D printing: it isn't tomorrow yet.

Few of these downsides are actually fundamental, most of them will change dramatically as technology changes.

Will 3D printing be the foundation of 21st century manufacturing? Almost certainly not. But will the fundamental techniques drive a revolution in manufacturing and become key elements in next-generation fully automated, fast-turnaround, fully configurable manufacturing? Probably.

The list is also padded pretty heavily. In reality, something has a killer app or it doesn't. 3D printing's weaknesses in speed and materials don't seem likely to go away any time soon, and in fact they seem intertwined in a way—faster printing would seem to require ever more specialized materials. It's obviously an exciting technology that will carve out a few niches, but I don't see what all the hype is about.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact