Hacker News new | past | comments | ask | show | jobs | submit login
3D metal printing is about to go mainstream (newatlas.com)
391 points by phr4ts on July 31, 2017 | hide | past | web | favorite | 158 comments

There are many services printing metal parts. Shapeways has been doing it since 2009. There are several workable processes. [1][2] This new machine competes with the ExOne Innovent.[3] That uses a single-step process (no oven needed) but is slower.

Desktop Metal's big claim is that they can lay down "up to" 8200 cm³/hr of metal. The "up to" weasel words are a problem. They're vague about the layer thickness. 3D printing has a basic trade-off between speed and precision. Most of the commercial vendors go for high enough precision that you can make working parts. Desktop Metal doesn't offer many pictures of their finished parts, but I did find one.[4] That looks like it was made with layers of about 0.5mm. The furnace step provides some surface smoothing. That's not bad for casting.

It's nice, but it's not clear that it's 100x, or even 10x, better than the competition.

[1] https://www.youtube.com/watch?v=rEfdO4p4SFc [2] https://www.youtube.com/watch?v=2vsaSzrhvcw [3] http://www.exone.com/Systems/Research-Education-Printers/Inn... [4] https://embedwistia-a.akamaihd.net/deliveries/5c8aec78d82aa1...

To be clear here, the ExOne machine does require sintering as a post-process.

The way I see it, Desktop Metal's main innovation is their furnace technology. Furnaces are typically large, expensive, industrial machines that cannot be operated in an office environment. Having a compact, easy to use furnace is a BIG deal. They also combine classic convection sintering with microwave sintering technology. Microwave sintering is a fairly recent technology (developed ~2 decades ago by folks at Penn State) that has a gigantic amount of promise (volumetric heating, finer metal microstructure, and more energy-efficient). MW sintering has seen fairly limited adoption in the industrial space so I am happy that DM is pushing this technology further.

The are pursuing two major technologies, an extrusion-based process (Studio system) and an inkjet-based process (Production system). I have seen parts from their Studio system and, from what I have seen, they are highly functional (comparable to casted parts).

Whether $1 billion is the right valuation is a question for the VCs and the market to answer. But, in my opinion, they are developing a very compelling set of technologies (if they end up working as advertised).

Any furnace capable of sintering metal parts is unsuitable for office deployment and will at a minimum need a fume hood and a way to vent the (very dangerous) fumes in a safe way. Typically that means you'll be operating machinery like this in an industrial setting.

That and desktop sized muffle furnaces have been around for half a century and can be purchased for under $1,000. I see a 12.7 x 10.8 x 15.2 cm one in my company's catalog for $870 so I don't consider this a huge breakthrough.

One advantage of microwave sintering over older methods (like muffle furnaces) is more even heating because energy is deposited throughout the volume of the part, not just at the surface.

Reduced thermal gradients -> more predictable post-sinter part dimensions and more uniform mechanical properties. Those are yield enhancers that could give significantly better production economics.

I think many of Shapeways metal offerings are castings from prints in wax, e.g. "For Silver, the minimum supported wall is determined by our ability to successfully print your product in wax and then cast it in a plaster mold. Walls that are too thin will break in the mold-making process." [1]

1: https://www.shapeways.com/materials/silver

Shapeways offers several different processes. The lost-wax stuff is for jewelry, so you can make hollow objects in expensive metals. Laser sintering is used for working parts. There's also MarkForged [1], which has their own process for 3D printing carbon fiber and metal with a sub $100K machine. It's kind of slow, but they're selling it to make dies for injection molding.

[1] https://markforged.com/

It would be pretty exciting to duplicate keys (especially proprietary and "do not duplicate" institutional keys) from the privacy of home. Might even hasten the end of the mechanical lock as a security device.

Sure, you can do that now with Shapeways or a crooked locksmith, but it would be fun to do independently.

It's certainly possible, but this kind of reminds me of discussions here on HN and elsewhere about way that the 3D printing of guns can create a crisis for gun control regimes. People in discussion threads reasonably pointed out that it's been possible to manufacture guns in home workshops for hundreds of years and that, for some kinds of weapons, it's not even toward the high end of challenging metalworking projects, and that indeed many people regularly do it either as a hobby or profession without 3D printing or even without CNC of any kind.

But this is kind of shocking for software-oriented people who might think that "manufacturing" "hardware" is a super-tough black art that can only be done by professionals in factories. And indeed that's pretty much my intuition as a software person who hasn't done woodworking since middle school shop class and never learned any of the other manufacturing skills that humanity has been working on for the last while. (Reading Bunnie's new book about manufacturing in China has been fascinating for me, because it's like "oh yeah, so all of these objects just come from people doing different tasks to fabricate them"!)

So people said about gun manufacturing that there was this funny intuition that 3D printing somehow allows people to casually manufacture complex objects at home which they otherwise simply couldn't do. But in fact, if people were moderately motivated, they could easily learn some of the other techniques that let them manufacture and/or duplicate objects. So we may tend to exaggerate the impact by thinking that other methods represent a huge, hard barrier while 3D printing represents true "push-button manufacturing". Neither side of this intuition is necessarily the reality.

For the problem of duplicating proprietary keys, it seems like anyone could already do this at home without especially expensive equipment and without especially extensive training. I remember reading a Mickey Mouse cartoon from many decades ago where a key was supposed to be duplicated from a negative impression taken in some clay (or something), and this was presented as a basic skill of a generalist mechanic!

On the other hand, maybe this intuition is partly right if some significant population of would-be home gunsmiths or key-copiers is intimidated enough by hardware and manufacturing that they're sort of waiting around for the pushbutton solution.

Edit: looks like other people in this thread have said this a lot more concisely. :-)

> But in fact, if people were moderately motivated, they could easily learn some of the other techniques that let them manufacture and/or duplicate objects.

That's actually a substantial barrier. People are lazy, and pressing 'print' on some design fed into a cornucopia machine (of which a 3D printer is a rudimentary fore-runner) substantially lowers that barrier. It requires no special knowledge beyond the feeding of raw materials into hoppers.

That's a lot less than what would be required to safely turn on a lathe, let alone making something with it.

the thing is that there are laws about some of these sorts of things, they applied before 3D printing came along, and they usually apply to 3D printing too.

I help run a Makerspace, a couple of times a year I get a teenage boy (and it is always a boy) come thru the door wanting to print a gun .... after I've given him the lecture about how stupidly dangerous that is, I then remind him that a) here in NZ hand guns are illegal, and b) he would still need to be police vetted and obtain a firearms license - the law doesn't change because you used a 3D printer

Keys are pretty easy even with "under the kitchen sink" home materials, even easier with a little alginate powder for the mold and craft-store quality binary epoxy to cast. It's not going to do much good with mid/high sec locks that utilize floating pin keys, but then neither would the 3d printed route.

You can do that with a file and a jeweler's saw or hacksaw. Certainly with a milling machine.

But how do I find a "blank" key to start from?

The point of these proprietary institutional key systems is that you can't. The blank is patented, and the authorized manufacturer will only sell it to authorized representatives of an institution which has a contract to use that key system. It will further guarantee that only one institution gets to use that system per geographical region.

It's "security through the inordinate difficulty of acquiring an object with a particular shape." Cheap 3D printing of metal might put an end to that strategy.

There is an image processing script, that will take a picture of the key and deduce the key number, or bitting code, or "key cut code", which is an expression of the heights for each pin. [1] [2]

Then you can plug that bitting code into a script which will generate an STL model for the key. [3]

I did this, except I measured the key with calibers and back-calculated the key number. I printed the key on a cheap Prusa I3 in ABS, it was not impressive in terms of resolution or strength, but it did work in my house. I carefully tested it because I thought it might break off in the lock, and it felt weak, but worked -- definitely good enough for a single covert access, or even use as an emergency backup key to hide somewhere.

No fancy metal printer needed, and you could write a script to pull pictures of keys out of twitter or whatever and automate the process.

[1] http://hackaday.com/2009/09/22/photographic-key-duplication/

[2] http://webcache.googleusercontent.com/search?q=cache:kxuH9jR...

[3] https://www.thingiverse.com/thing:52761/

If you can carve the teeth of the key, you can certainly carve the blank.

i'd start by plugging "key blank" into your favorite search engine. if that fails you horribly, you can buy brass all over the internet: http://www.onlinemetals.com/ https://www.metalsdepot.com/ mcmaster-carr, or amazon. i suggest a pair of calipers, too.

I have a key that claims "copyright" and "do not copy". I've always assumed the copyright claim was bogus, on the grounds that functional items like keys aren't within the scope of copyright law.

I figured that "do not copy" label was for the key cutting people to see and deter them from copying it. But I'm not sure if anyone really follows those labels.

I get my keys cut at a convenience store down the street. Key cutting is not exactly a sophisticated service market run by specialists. So I doubt there is much stringent ethical procedures or industry reputations to lean on.

Meanwhile someone copying fobs in Toronto for people was publically shamed, which is silly because anyone can buy an RFID copier on Amazon for $30. http://www.cbc.ca/news/canada/toronto/condo-key-fob-copy-1.4...

Amazingly the building I live in charges $100 for replacement fobs.

  I'm not sure if anyone really follows those labels.
You're probably right.

When people's security needs are great enough to justify paying $150+ per lock, high-end locks offer:

(a) Patented keys that the manufacturer only sells to authorized distributors (who are contractually obligated not to cut keys without a key duplication card)

(b) have complex designs that can't be cut on standard key-cutting machines and

(c) have special 'key control' pins that mean different locksmiths get incompatible locks and key blanks.

Needless to say, a sufficiently good 3D scanner & printer (or indeed a sufficiently patient person with a file and a pair of calipers) could bypass these protections.

Several years ago, I lived in an apartment with "do not copy" keys. I took one to the hardware store and they refused to copy it, so I took it home, ground out the words with a Dremel, filled the gap in with solder and polished it flat. Same store happily copied it.

If you're important enough you can get the lock makers to make you your own custom blank that they don't sell to anyone else.

I ran into this at college when trying to issue a bunch of keys to club members.

"Laser cut" keys have a second set of teeth inside one of the low spots in the blank that normal machines can't copy. Luckily, the locks that have the bar to fit the inner teeth are expensive and only important things get them so the hardest part of copying the key was convincing the guy at the hardware store that I just needed copies with the same profile and not the inner teeth.

Institutions can buy key systems where the blanks are patented, and only sold by authorized dealers to authorized buyers as part of the institutional contract. This makes it very hard to come by usable blanks. Your average locksmith won't have them.

It's a piece of metal. Even if the locksmith doesn't have the blank you can simply make your own on a copying mill. It will take a while because you'll need a fine bit to make a usable key and the registration when you flip it has to be perfect but this is absolutely doable.

The blanks that locksmiths have are an optimization in time and cost, usually not in technology, they are there so you can walk out with your new key in 5 minutes and at $10 rather than an hour or two and $200. The trick is that the blank has all the lengthwise grooves pre-cut and the copying grinder then merely has to slot the blank to the required depth and cut off any excess. This is so that some $7 / hour person can make your keys and not a trained machinist with a very expensive piece of gear.

Keys with tricks in them (magnets, embedded RFID chips, bearings, springs and so on) are a lot harder to copy than keys that are simply steel.

Why not use plastic?

It'd work for a few times, at least. In the 90s, BMW gave out plastic wallet-sized cards with flip-out emergency keys with at least some of their cars. Not super durable for repeated use, but handy to have. Looks like something similar continued into the 2000s for a while, at least: http://www.bimmerboard.com/members/q/original/BMW%20Transpon...

My '96 Chevy Cavalier came with one of those.

For occasional or one time use, that's actually fine. Just need a tension wrench to turn the cylinder so the plastic doesn't bend/break.

The part quality can be impressively good - the test parts I've seen match or exceed the resolution of typical FDM prints (from ~$2k FDM machines).

It's like the brogrammer world just discovered powder metallurgy.

This exact technology has existed for decades but hasn't seen widespread use because it has serious problems. I don't see any evidence that Desktop Metal solved these issues either. The resulting parts have high shrinkage, poor dimensional tolerance, and poor mechanical properties. The sintering process leaves voids inside the material that serve as stress concentrations, causing the material to fail well below its rated strength. Also sintered parts tend to fail catastrophically rather than yielding since the adhesion between particles is much weaker than the yield strength of the metal.



I see a bunch of PhDs with backgrounds in material science (among other things). It's just possible they might have a clue.

I'm referring to the hyperbolic headlines talking about a "metal printing revolution," not the actual company. There are already off the shelf solutions to do nearly the exact same thing that everyone here is getting all worked up about. What's more, you could have purchased those solutions before the internet was even a thing.

What does that have to do with brogrammers?

I think OP is referring to the software-oriented HN crowd who have highly upvoted this article despite little evidence that desktopmetal is doing anything revolutionary.

How has it not seen widespread use? Buy just about any consumer-grade power tool today, and inside will likely be some powdered metal components. They're in some of the high-end tools, too, and can provide excellent quality with decent design.

I have no clue whether these prototyping machines will produce parts of that quality, but the important observation is that 'good enough' is the key threshold. Many times you don't need the full strength of a machined steel part, just something a bit better than plastics can provide.

Those parts are produced via Metal Injection Molding (MIM). MIM parts have similar issues to printed (laser sintered) parts but injection molding is a far more efficient process for mass production.

I thought the gears are done via a simpler sintering process: https://www.youtube.com/watch?v=-K6_wbBnXmc

Is that still considered MIM?

I'm reserving judgment until I get my hands on parts - which aren't yet publicly available. I've heard that their parts have passed rigorous materials testing, but until I get make some, machine them, and physically test, will I start to believe.

With $200M+ they better be doing something new! So far, we know they are very good fundraisers and marketers, but I do think it could be different this time (i.e. maybe they'll crack it).

Yeah. That is a bit scary. People got used to think that plastic and wood are pretty weak and fragile materials (compared to metals). But people believe that metals have much more strength (even if in reality they don't - in case of some metals and alloys), so I wonder if they start using printed metal parts in their cars (or similar), because it is much cheaper (but "it is made of metal, so must be strong enough!").

Will this cause more accidents?

$120k for the prototype printer, $360k for the production printer... still about two orders of magnitude away from being practical for me to set up a microfactory in my garage. Maybe by the time my car is self-driving and earning me money instead of sitting in my garage, I can get that microfactory set up affordably.

By the time you can afford that printer, there are others that cost $100k that do better job. For most tinkerers it will be better investment to place order in local metal workshop with has good printer than have your own cheap lathes, CNC mills and 3d printers in the garage.

>Maybe by the time my car is self-driving and earning me money instead of sitting in my garage, I can get that microfactory set up affordably.

Have you ever thought why there are no cheap capital assets sold in mass markets for public that provide good ROI for no additional effort? (this is good question for armchair economists)

> Have you ever thought why there are no cheap capital assets sold in mass markets for public that provide good ROI for no additional effort? (this is good question for armchair economists)

Pretty much for the reason you covered in paragraph one to be honest.

If I can achieve good with 10k and someone else can achieve awesome with 100k then the person with 100k will win (assuming they can handle demand).

It's pretty obvious (if you look at..well every economic system ever really) that capital begets capital.

It's not a huge thing, the US had this issue at the turn of the 20th century.

Well, the other obvious reason is if everyone can achieve "good" for 1k, everyone will make the investment, leverage the asset, and the ROI will disappear because it was easy & accessible to everyone and becomes highly competitive.

Bitcoin mining was a good example. Everyone has a graphics card, and mining was pretty trivial. Great example of a cheap capital asset with low effort to monetize. Quickly, the ROI slimmed until only large scale operations with low costs and good economies of scale were profitable.

Yep but even there the people with the large amounts of capital won once GPU's stopped been effective since they could afford ASIC's.

I think on some level society has always known this, having access to large amounts of capital doesn't guarantee you'd win but if someone offered me the choice between starting a game of monopoly with $100 or $100,000 I know which I'd pick.

Fun trivia aside, Monopoly was created to demonstrate that activities that promote wealth creation are more beneficial than ones that allow monopolists to run amock.

I'm assuming you mean physical assets that you can keep local? Otherwise historically investing in stocks provides a good ROI with no additional effort.

Well, if you are okay with some nonzero additional effort, then kitchen supplies like an Instant Pot or a Sous-vide heater fall into the joint categories of

A) < $100

B) Saving $1000/year relative to a certain level of cuisine consumption.

Because cheap capital assets produce lousy results, and good ROI requires trained, professional operators.

But not two orders of magnitude too expensive for a high end machine shop with a small warehouse. You can see this as something that you may dream about as a 'maker', but more importantly this could be something that allows small companies to make parts locally rather than overseas.

I'd say the two go hand-in-hand. The holy grail of the maker movement is microfactories where individuals or small teams can produce local-scale consumer products out of their garages/basements to sell to their communities. That by definition is a small business with a small warehouse. :)

From a marketing standpoint alone the ability to rapidly produce metal cases for things would give products an air of quality that you could never match with a plastic 3D printer and might cost too much to CNC. Between this and the costs of pick-and-place machines coming down I'd love the ability to design/build my own consumer electronics products and sell them out of my garage.

You can already get into a small-scale CNC mill for under $10k to make parts, look at the Tormach offering.

For better or worse it's a long ways from "machining/printing a part" to having something ready to sell. Even just dealing with aluminum (easy to machine), you have deburring, polishing and/or tumbling, and (probably) anodizing standing in the way of selling your new consumer part. Some of that is fairly easy to get going with - small scale anodize is problematic.

Other 3D printing technologies is already well into this territory. I've used companies like Sculpteo and Xometry numerous times for professional 3D printed parts in various technologies (mostly SLS nylon). The per unit cost is more than having a Makerbot running in an office, but the quality is way better. Even for FDM sometimes I prefer to have someone else make the part.

I can imagine those companies buying these machines and passing on the cost savings -- I would definitely start to think about using it instead of CNC.

Even if only the price of the prototyper comes down a bit more, it'll improve things significantly.

The ability to rapidly prototype in low cost, low latency iterations will be a boon across many industries - regardless of whether or not you end up outsourcing production.

And of course, you may just end up outsourcing down the road.

Yeah, but at <$500k, they can rent it out for $60/hour and recoup expenses in a year. Even if we bump it up to $100/hour to cover administrative costs, that's cheap enough to be accessible for a lot of average citizens to use.

I can think of a few things where printing 50 of them in an hour would be adequate to recoup a $100 equipment rental.

Hm, like what? I'm trying to think of some applications for metal printing.

Hard to source parts for old cars? Those cost a lot more than $2 each.

Car parts, computer cases (built-in heatsinks), rock climbing gear (not sure printed metal is strong enough for this one), sculptures, gun receivers (again, there are strength problems), matching door knobs and faucets for houses, prosthetic frames (to be later encased in silicone)... I'm sure there's more.

I would consider using it anywhere I would be looking at CNC-ed metal, especially for one off parts/prototypes. A lot of the cost of CNC is programming and fixturing, which 3D printing technologies save.

Much of the toolpath programming/selection/etc could be "automated" the same way that its automated for FDM printers. But then you would lose some of the "quality" advantage and likely increase tool wear.

But the reverse is also true of 3d printing, frequently I've wanted more control of some particular aspect of the printing job, but existing slicing packages make that very difficult. For example, just slow down the printing along one edge (rather than a whole layer), or change the infill for one particular area, or more detailed temp control of "inactive" nozzles, or just more options during nozzle switch (aka pause for 20 seconds to cool the inactive 10C while heating the active, then wipe both).

Does seem competitive vs CNCs. But what are its benefits? CNCs might have higher set-up costs but are incredibly accurate.

Correct on the accuracy, but you can design parts with features like undercuts and cavities on 3D printing that are either not possible, or much more expensive, to CNC.

Not every metal part requires the 0.001" tolerance you can get with CNC as well.

I look at it as another technology that can be selected when the parameters make sense for it. I'm more familiar with plastic technologies, for them I switch between FDM, SLS, SLA, Polyjet and Injection Molding depending on:

- Forecasted quantity

- Time to market needed

- Price constraints

- Accuracy and surface finish required

- Design requirements

Each technology has it's own advantages and disadvantages.

It's the running costs. If you want to make a lot of the same part and need a 5 axis mill, they have to be operated by technicians which are extremely expensive and they can't work on anything else.

Make a lot of parts on a printer and you spend a little more time designing the part and time to tune the settings (which might be done depending on the material you use) and you're off and printing and just paying for print time. Not cheap, but cheaper than paying techs at mid-scale.

That's an order of magnitude cheaper than they were 5-10 years ago (at least for the prototype printer). Noncommercial pilots can easily spend that much money on a plane + training whereas seven figures is far rarer unless you're getting a supersonic military jet (like an F5) or planning on flying clients around to finance a small corporate jet.

The rate at which the price of this technology is falling is fantastic.

That's the cost of two quality CNC machines, it's definitely within the realm of access for smaller machine shops everywhere.

That's if you get used or cheap CNCs - the kind that are largely sold to Chinese factories or mom and pop shops for low value add manufacturing. A good (brand new) horizontal machining center with zero point fixtures, tooling, and training costs a minimum of 200-300k if you want to make high value add parts for aerospace/medical, not to mention the costs of metrology scanners and other quality assurance gear which easily adds another 100-200k.

At 360k, most machine shops can easily afford it if they can justify the purchase with demand. Also, almost no one buys these outright because they can easily finance it with a loan secured by the machine itself. We're talking a midlevel machinist's salary and overhead for five years. Since the majority of domestic metal manufacturing is for high margin industries with poor turn around times, these machines would pay for themselves very fast.

And then you're making parts with tolerances less than 0.001". This technology is unlikely to be doing that.

yeah, this isn't for the hacker. The thing that pulled me into 3d printing was a 400 dollar machine that opened up amazing possibilities.

For that price its more fun to buy a cheap 3d printer and use printed models as cast model. I even think you can use printed pla for 'lost-pla casting'.

Yeah, apparently even "10x cheaper" would still require me to take out a mortgage for a printer.

Still, doing so and offering "3D metal printing as a service" (3DMPaaS) might actually be a useful investment.

Would be nice to have a video of the process

This is another Ric Fulop company, notoriously the founder of A123 Systems [1]. That company also raised "a ton" of money but ultimately blew up, filing for bankruptcy. I'm skeptical of this new endeavor because of both economics and technology.

[1] https://en.wikipedia.org/wiki/A123_Systems

Another huge concern is the fact that they are trying to push to market two completely separate 3D printing process technologies, and are valued at over a billion dollars before shipping a product. A few of their claims and marketing materials are too hyped and there's a lot of skepticism about their actual technology benefits. Limitless hype has been an ongoing problem in the 3D printing field, and they seem to be capitalizing on it.

> The company has raised a ton of money in the last few months...

> ...Desktop Metal's Studio machines are also a ton more practical to have in an office.

> But there's a ton of metal options...

I'm guilty of this too, but I think there are more ways to describe a plethora of items than "a ton". Unless, of course, there are actually 2000lbs worth.

This article coincidentally is an exception! A ton is the right order of magnitude:

The US Treasury says that $1 million in $100 bills weights 22 lbs[1].

The $115 million that they raised in the last few months would weigh about 2200 lbs.

[1] http://answers.google.com/answers/threadview?id=441929

edit: grammar

Epic win on the weight conversion :-)

Strictly speaking, a "plethora" should be an excessive amount of something.

It still might work in this context though, if you think the company has raised too much capital...

Well technically, a plethora is a 'large or excessive amount of something', so I think in this case it works, though it feels a little sesquipedalian.


The entry you linked to agrees with my definition:

"Strictly, a plethora is not just an abundance of something, it is an excessive amount."

We are in the superlative bubble.

"Epic win"

Would you say I have a plethora of pinatas?

I don't think they make enough of this point:

> Depending on the nature of the part, it might be necessary to do some post-print surface finishing like sanding or bead blasting to smooth out the layered surfaces

If this is anything like the powder-bed parts I've handled, the layers are going to be pretty rough. I wouldn't be surprised if they need some degree of post-machining. Don't sell your CNC mill just yet.

Furthermore, 15% shrinkage during sintering? What's the dimensional tolerance on the finished part then? I'm guessing it's not great.

Casting stuff has a ton of overhead cost and is very impractical for low volume production. EDM is expensive as all heck. For stuff that doesn't need to be made out of exotic material this is great.

For low volume production and prototyping this technology is great and cannot become mainstream soon enough. So what if it has to be finished. A casting or forging is the same way.

Forging costs a ton. The die has to be machined by (you guessed it!) a tool and die maker. Those guys are expensive. Once the die is made, however, bashing out the parts with it is fairly cheap per item.

Back when I worked in the aviation industry, sinking a die for a part cost a cool quarter million.

As long as the shrinkage is predicable, you could still have very good tolerances

Somehow I doubt it. The shrinkage must be dependent on geometry/density and their method of compensating can't possibly be good enough to hold tight tolerance over an entire complex part.

Forgings and castings all shrink as well, because metal shrinks substantially when it cools. The dies and molds are made oversized.

I corresponded with a guy who was making his own intake manifold for his mopar. (A very cool project.) He made his own molds. I did some calculation, and said he had to make the mold about 10% oversized. He told me I didn't know what I was talking about, he couldn't believe it would shrink that much. His cast manifold wound up 10% undersized :-)

Anyhow, dimensions that require tight tolerances get machined to spec after the forging/casting.

> Forgings and castings all shrink as well, because metal shrinks substantially when it cools.

Forgings and castings also require post-op machining to bring them into tight tolerances, which isn't something that fits the mental model of a lot of people when it comes to 3d printers.

People who don't own 3d printers. There is plenty of cleanup work on a 3d print of any complexity. I find myself even drilling a fair number of holes in some of my prints because the minimum consistent hole size my printer can generate is a few mm.

If you cared about dimensional tolerance, you'd hire a machinist.

What is the strength of such a part, compared to a regular cast part? Say I 3-D print a spanner. How well will it hold up against a spanner that was cast and heat treated?

The process in the article is effectively sintering. If [1] is any indication, you can get to within an order of magnitude of strength of cast/forged parts, if sintering variables are controlled for properly. Which may not be the case for a diyer, metallurgy is a complex art.


Yeah and I also worry about shrinkage - they're saying ~15%, which would be a massive issue with some profiles. I don't see how a lot of parts would work very well without distortion.

good spanners are not cast and heat treated. They cast a blank which is roughly the right size, but then they forge it - much like an old fashioned blacksmith, but there are a ton of jigs involved (probably literally 2000 lbs though I first wrote that as an expression).


Heavy use metal parts are often milled/shaped into a blank, then forged.

Forging involves heating the part, then hammering it into shape using specialized jigs (reverse imprint of the part). Most metals have a grain structure, similar to wood. The combination of heat and pressure bends the metal's grain structure around the contours of the part. This imparts a tremendous amount of strength compared to milling or casting. Since milling generally cuts into the grain at arbitrary angles, while casting often creates a disorganized grain structure (there are newer casting techniques which give better results).

The parts are often milled again, to ensure exact dimensions. Then finish/treatments are applied.

Obligatory video of Chinese workers forging a ring: https://www.youtube.com/watch?v=ZWXFhdeOjMY

And Swedish ones: https://www.youtube.com/watch?v=tEF2erBBVZ4

From the video:

The production system is cloud-connected

Seriously, that's a selling point nowadays? I have to buy a hyper expensive piece of hardware and if the company goes under I might not even be able to use it anymore?

Not everything needs to be on a bloody cloud.

> but the only affordable printing materials are cheap ABS plastics

Not very true at all. You can get PLA for $20/1kg or less. Even resin for SLA printers is often possible to find for affordable prices, especially considering that you can sometimes use less material without the need for infill.

I'm really not even sold on the idea that so many people "need" metal printers. Seems like most people would be way better off with the incredibly cheaper plastic options.

Isn't PLA less structurally sound than ABS? I thought the reason it was used over ABS was that it could be formed at a lower temperature.

Yes, PLA is somewhat weaker, but still more than strong enough for most desktop printing applications. It's also much less toxic to print with than ABS.

Can we get a citation for the toxicity of ABS? My Google-fu is failing me as I can only find articles that say the fumes can cause minor irritation when inhaled or in contact with the eyes.

"The higher temperature ABS-based printers had total UFP emission rates nearly an order of magnitude higher than the lower temperature PLA-based printers (1.8-2.0 10^11/min compared to 1.9-2.0 10^10/min)."

"Primary gas-phase products of ABS thermal decomposition at very high temperatures have been shown to include carbon monoxide and hydrogen cyanide, as well as a variety of volatile organics (Rutkowski and Levin, 1986). Exposure to thermal decomposition products from ABS has also been shown to have toxic effects in both rats (Zitting and Savolainen, 1980) and mice (Schaper et al., 1994)."

from http://ac.els-cdn.com/S1352231013005086/1-s2.0-S135223101300...

Doesn't it put off styrene gas?

PLA is also susceptible to heat. Leave a PLA model out in your car with the windows rolled up in the summertime (especially here in Arizona where you can easily cook a meal in your car!) - and you'll come back to a model in not-so-good shape (it won't melt, but it will deform and sag).

Doesn't it also seriously degrade over time in UV light?

Only minimal delamination, solvable by increasing print temperature. For SLA resin prints though UV exposure is a big problem.

Perhaps that's what I was thinking of.

Any reason not to set default skepticism to high for Loz Blain and NewAtlas?

>...it's going to compete with traditional mass manufacturing

>...the hype is real

The team, tech, results, deals already signed, all seem really impressive in their own right. No hyperbole needed to get a more views.

It's not my area, if someone tells me this really has a shot at competing with mass manufacturing in the next 5 years I retract everything.

A forging is 3x the strength of a casting of the same part from the same material. That's why when upgrading the power of your muscle car, forged parts are the way to go.

What's the relative strength of 3D printing?

I'm told the metal powders are still more expensive than equivalent traditional materials, and that in some cases (Ti) can be explosive. Anyone know what the real economics are in terms of materials and energy costs?

Aren't all metal powders explosive given sufficient dispersion in an oxidizer and a source of ignition?

Not all, but many. Particularly metals that would likely be used for printing applications, e.g., iron and aluminum.

However a ton of materials are explosive if made into a fine powder then dispersed, so this is more of a general fabrication issue, than something unique to 3D printing.

Other than the temperature a relatively compact oven is potentially reaching, what's the breakthrough here other than the successful funding?

Production time.

Production is always judged by how much time it takes for raw material to become finished parts. Additionally that is split between 'operator time' (person required) and 'machine time' (just the machine running).

'Operator time' is a function of how skilled the operator has to be (machinists cost a lot more than technicians).

'Machine time' is a function of operating cost and depreciation.

So anything that shortens those times, or cuts those costs lowers the cost per part. Parts "have" to cost less than a threshold amount to meet the sales price of the final assembly + margin.

That is all basic manufacturing. The key is that costs have typically been reached by doing things in volume with 'tooling' (investing in dies and jigs to configure the machine to easily make the one part). That essentially makes the production line 'single use' while it is tooled that way.

These guys are proposing that they can make metal parts at the same cost as the tool and die folks in smaller quantity and with no setup costs. That changes several things;

1) You can make warranty/repair parts "on demand" so cut the cost of making a million widgets and storing them in a warehouse.

2) You can make 'small runs' of products for more specialized markets at a price that the market will accept.

3) You can support more variations of a given product without your spare parts inventory exploding.

4) Production can be parallelized from small scale shops so a large 'mega factory' isn't needed, instead you can get a dozen shops with this gear to work in parallel to meet your production target.

If they can pull it off, it really does change a lot. If they can get the costs down further it opens the possibility of domestic delivery of parts from a certified vendor rather than a warehouse somewhere.

In terms of second order effects, I think one interesting angle is that low-setup manufacturing makes ip / copyright in the STL files more valuable.

If you don't have the design then a human needs to spend a bunch of time dialling in the geometry.

I can imagine manufacturers wanting to find ways to "DRM" spare parts. One way to do that might be to make it more costly to produce designs from first principles. So having for example specific complex bits of geometry that ultimately force you to license the authorised design if you want to "economically" fab the part at low volume.

Absolutely. They are going to want to charge a 'tax' for a third party manufacturer to make the part. Something that will be hard to enforce however.

The fact that the company shipping them wants to be in the news?

More seriously: the main innovation here seems like accelerating the print volumes and decreasing the per-item cost to the point that this could be used in production rather than just for prototyping. That's a kind of "mainstream", though still on the high-end business side, not consumer.

Mostly cost & benefits over existing metal 3D printing technologies.

Current metal 3D printing machines cost well over $1 million, and have some significant caveats for part design. Due to the residual stresses in the parts, supports structures are required for many part geometries. This means your parts that have been designed to be unmachineable, now have support structures that then need to be machined off. The easily removable supports are a huge win for Desktop Metal here.

Repeatability has also been an issue in metal 3D printing. This is partly due to the nature of the sintering/melting process, but mostly because the majority of parts produced on metal 3D printers are actually designed to be manufactured using another process. I have higher hopes for the production system than the studio system for improvements in that area, but remains to be seen at this point.

As a layman, I am skeptical. This is very similar to the promises that were made regarding 3d printing of other materials, and those weren't quite fulfilled.

Like many hyped new techniques, they end up as techniques that are almost good enough to be practical.

4 furnaces per printer... reminds me of factorio

This is where I learnt the word from.

While (potentially) impressive, it's not clear to me that this will really replace production. I mean, it's faster than other printers, but still far away from regular production speeds.

I've watched a few DIY foundry videos on YouTube where makers melt down aluminum cans and scrap metal into chunks of metal, ready for re-use. I wonder if these 3d printers will be able to use reclaimed metal created from a similar type of process.

If you're already doing this, you probably have the skills and the tools to make your own molds (3D printed in plastic even), mold a rough version, then mill it down to proper tolerances. If your molds are good enough, and your pour goes correctly, and your tolerance needs aren't that high - the part that you get might only need a minimum of cleanup (sprue removal, some filing or sanding, sand/bead blasting).

They would need to be able to create a powder with consistent grain size.

Printing with improved PMC (which is what this sounds like) doesn't seem that revolutionary to me... I know there was a Kickstarter a while back for metal-based PMC-like filament to use in regular 3-D printers (for kiln firing later), and I think somebody already makes a device to print using PMC itself. While doing so precisely, strongly, and cleanly enough for mechanical applications, and with a much broader spectrum of metals, is great, the prices seem rather far from the headline hype. (Not that the current options I mentioned don't leave much to be desired.)

MiniMetalMaker was extruding a metal clay for sintering: https://www.minimetalmaker.com/ mainly for jewelry etc. There is also the Markforged Metal X https://markforged.com/metal-x/

Filamet is the Kickstarter'd filament I was thinking of. It looks like they've got a decent product, though it also looks like the campaign became a bit rocky (typical Kickstarter delays, etc.). (I didn't back, so can't speak to it beyond appearances.)

It's been about 10 years since I left Caterpillar --- but I don't think you can weld (reliably) on PM / sintered parts. This was one of the concerns the crotchety old manufacturing engineers brought up when I proposed replacing some expensive machined bosses with a much cheaper PM part.

Then again, those guys really loved to say stuff like "no, that's not how it's done." - so maybe they're wrong / tech has improved significantly.

> Each production printer can produce up to an incredible 500 cubic inches of complex parts per hour.

That is 124 iPhone 6 sized solid blocks per hour. Incredible indeed!

Your sarcasm is noted and appreciated!

Of course, this isn't supposed to print easily stamped iPhone backs, it's supposed to print fuel injectors, retention brackets, integrated linkages, etc. Parts that are complex and costly to produce through traditional machining but trivial to print.

I don't know much about how Desktop Metal works; how does this compare to things like ExOne?

In case anyone else was wondering about the power requirement for the sintering furnace: 208V 3-phase, 30A. The 3-phase requirement may be an impediment to some hobbyists; they should probably offer it with a buying option of their own branded inverter.

The $300k + might be a bigger impediment.

a threephase supply is much cheaper.

But could it print a CNC machine ?

First it would print a locking mechanism to prevent the humans from unplugging its power supply..

This may bring the per part costs closer to what prototyping costs come in at for CNC machining or Metal Injection Molding.

I don't see this being used with a lot of exotic materials yet, but for stainless steel this is great.

Doesn't oven treatment induce slight deformations?

I think I can safely say that the parts won't be flat to 1000 of an inch (.03mm) - or whatever your tolerance is. However this is something that any existing process can achieve with arbitrary shapes either.

I think this process replaces sand casting, a process that also has many slight deformations. You just make the parts a little big and then machine the important sides to the exact side you need. The downside of sand casting is it often leaves a little bit of sand embedded in the metal which will destroy your tools. If the internal metal quality is as good as sand casting, the ease of doing arbitrary shapes and lack of sand in the metal make it a winner. That you have to have a lathe/milling machine to finish the part is not a change.

If the above is right (it seems reasonable, but I don't know if it is), then beams and and flat stock will continue to be made with existing process. However odd shaped things like an engine blocks it could be a winner.

Engine blocks will almost certainly remain as a cast part, simply due to the production capacity of casting is orders of magnitude higher than additive manufacturing, as well as being relatively cheap. Metal additive manufacturing is currently used in aerospace to reduce complex assemblies of parts into a single piece, or to make small complex shapes that can't be made by a traditional machining or casting process. Additive manufacturing, for the foreseeable future will not replace load bearing parts as well, mainly due to the different processing steps load bearing parts undergo. It will also most likely stay with aerospace for the time being, since the price per part is prohibitive for automotive use currently.



> Parts also shrink up to 15 percent during the debinding and sintering process – but again, that's all automatically managed by the system.

Given this change in volume (which I assume would be highly nonlinear depending on design density and geometry?) I wonder what repeatability is like.

Perhaps for the manufacturing (non prototyping) printer they have a closed loop feedback mechanism to build a few outputs and feed any dimensions out of tolerance back to the input stage.

It might also be possible to model the shrinkage in software and compensate for it automatically, assuming you know what materials you are going to use.

Could be a minor surprise when you check that you have a 200x200mm work area, design a 100x180mm part, and then get told that it won't fit on the build area.

It's not just materials, but the geometry of the part that plays a role in the shrinkage. Though I wonder if some form of finite element analysis could overcome this (that software won't be cheap).

The sintering process is the hard part of printing metal and where a lot of Desktop Metal's core technology is. They basically use traditional heating combined with microwave heating controlled via infrared temperature sensors:


The idea of an "office-friendly sintering furnace" would have been comedy a decade ago.

What advantage does metal printing offer over casting a 3d printed wax/resin?

How does this compare to a metal CNC machine? Cost and Speed?

CNC machines will probably continue to be the king of things they can machine. The catch, though, is that you can 3d print a lot of structures you cannot machine by any means (see: SpaceX SLS's their rocket nozzles).

You have to design for the manufacturing process. Parts designed for CNC machining will always feel like a square peg in a round hole for 3d printing. Once you start designing for 3d printing, though, things get more competitive quickly. You can design parts to have hard to machine shapes, or parts that would have too much waste to machine from a solid billet of material and 3d print them without the waste.

Your computer case is bent up from sheet metal instead of machined from a solid block of steel for a lot of obvious reasons, and it makes no sense to compare the processes for a design like that. 3d printing will enable similar shifts of design.

The two fill different rolls. I expect you will use this to create a rough part and then put it in the CNC machine for final finish.

The game will radically change when metal printing arrives at the point where it can print coils and make stators and rotors.

Wake me up when metal printing reaches that point.

So, about those good manufacturing jobs...

So now anyone will be able to print a real gun with complete anonymity.

As apposed to casting and cncing one yourself? People can already build guns anonymously if they want (and many do).

People can also use current 3D printers to (mostly) print unconventional guns anonymously. Do these printers also concern you?

Reaction to technology should be with reason, not fear mongering.

3D printing is much more convenient than casting. Also I assume metal printing would make better weapons than plastic ones.

No fear mongering, just making an observation.

Is this an ad?


Guns for everyone!


Registration is open for Startup School 2019. Classes start July 22nd.

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