HP 3D metal printers are 50X more productive for final parts

HP Metal Jet printers are up to 50x more productive, delivering low-cost, high-quality final parts. New Metal Jet Production Service opens up world of applications to global customers; Partnerships with GKN Powder Metallurgy, Parmatech, Volkswagen, Wilo and more.

HP today also launched the Metal Jet Production Service, enabling customers around the world to rapidly iterate new 3D part designs, produce final parts in volume, and integrate HP Metal Jet into their long-term production roadmaps.

“We are in the midst of a digital industrial revolution that is transforming the $12 trillion manufacturing industry. HP has helped lead this transformation by pioneering the 3D mass production of plastic parts and we are now doubling down with HP Metal Jet, a breakthrough metals 3D printing technology,” said Dion Weisler, CEO and President, HP Inc. “The implications are huge – the auto, industrial, and medical sectors alone produce billions of metal parts each year. HP’s new Metal Jet 3D printing platform unlocks the speed, quality, and economics to enable our customers to completely rethink the way they design, manufacture, and deliver new solutions in the digital age.”

HP Metal Jet is a groundbreaking, voxel-level binder jetting technology leveraging more than 30 years of HP printhead and advanced chemistries innovation. With a bed size of 430 x 320 x 200mm, 4x the nozzle redundancy and 2x the printbars, and significantly less binder by weight, HP Metal Jet delivers greater productivity and reliability at a low acquisition and operational cost compared to other metals 3D printing solutions. HP Metal Jet will start with stainless steel finished parts, delivering isotropic properties that meet or exceed ASTM and MPIF Standards.

Transforming Industries With HP Metal Jet Technology

In an industry-first collaboration, HP is partnering with GKN Powder Metallurgy to deploy HP Metal Jet in their factories to produce functional metal parts for auto and industrial leaders including Volkswagen and Wilo. GKN Powder Metallurgy is the world’s leading producer of materials and products using powder metallurgy technologies and includes the brands of GKN Sinter Metals, GKN Hoeganaes, and GKN Additive Manufacturing. The company produces more than three billion components per year and expects to print millions of production-grade HP Metal Jet parts for its customers across industries as early as next year.

“We’re at the tipping point of an exciting new era from which there will be no return: the future of mass production with 3D printing. HP’s new Metal Jet technology enables us to expand our business by taking on new opportunities that were previously cost prohibitive,” said Peter Oberparleiter, CEO of GKN Powder Metallurgy. “Our DNA and our expertise in powder production and metal part processing using digitally networked systems will enable us to drive industrialization across the whole additive manufacturing value stream. By combining the forces of HP and GKN Powder Metallurgy, we will push the productivity and capability of our customers to unprecedented levels based on the economic and technical advantages of HP Metal Jet technology.”

Volkswagen, one of the largest and most innovative vehicle makers in the world, is integrating HP Metal Jet into its long-term design and production roadmap. The collaboration between Volkswagen, GKN Powder Metallurgy and HP has resulted in the ability to move quickly to assess the manufacturing of mass-customizable parts such as individualized key rings and exterior-mounted name plates. Volkswagen‘s multi-year plan to use HP Metal Jet also includes the production of higher performance functional parts with significant structural requirements, such as gearshift knobs and mirror mounts. As new platforms such as electric vehicles enter mass production, HP Metal Jet is expected to be leveraged for additional applications such as the lightweighting of fully safety-certified metal parts.

“The auto industry is being revolutionized – not only do customers now expect personalization, but by 2025 the brands of Volkswagen Group will have introduced 80 new electric models,“ said Dr. Martin Goede, Head of Technology Planning and Development, Volkswagen. “A single car consists of six thousand to eight thousand different parts. A big advantage of an additive technology like HP Metal Jet is it allows us to produce many of these parts without first having to build manufacturing tools. By reducing the cycle time for the production of parts, we can realize a higher volume of mass production very quickly. That’s why HP’s new Metal Jet platform is a huge leap forward for the industry, and we look forward to raising the bar on what is possible to deliver more value and innovation for our customers.“

GKN Powder Metallurgy is also leveraging HP Metal Jet technology to produce cost-effective industrial parts with higher hydraulic efficiency for Wilo, a global leader for pumps and pump system solutions. Wilo is looking to HP Metal Jet technology to produce initial hydraulic parts such as impellers, diffusors, and pump housings with widely variable dimensions that must withstand intense suction, pressure, and temperature fluctuations.

Reinventing Healthcare With HP Metal Jet

To serve the medical industry, HP is also partnering with Parmatech, an ATW Company, to expand mass production of Metal Jet parts for customers including OKAY Industries, Primo Medical Group, and others. Parmatech is a world leader in metal injection molding and has been a metals manufacturing pioneer for more than 40 years, specializing in producing low-cost, high-volume metal parts for the medical and industrial sectors.

“HP Metal Jet represents the first truly viable 3D technology for the industrial-scale production of metal parts. Our customers demand extreme performance, quality, and reliability and HP’s advanced technology and heritage of market disruption give us the confidence to deliver beyond expectations,“ said Rob Hall, President of Parmatech. “We are excited to deploy HP Metal Jet in our factories and begin manufacturing complex parts, such as surgical scissors and endoscopic surgical jaws, and new applications and geometries not possible with conventional metal fabrication technologies. HP Metal Jet technology will play a key role in our mission to develop innovative solutions for the unique challenges of our customers.“

Designed for Mass Production: HP Metal Jet Pricing and Availability

In first half of 2019, customers will be able to upload 3D design files and receive industrial-grade parts in large quantities from the new Metal Jet Production Service. The parts will be produced by HP partners GKN Powder Metallurgy and Parmatech to ensure the highest standards of engineering and production quality. For more information and to register for access to the HP Metal Jet Production Service go to HP.com/go/3Dmetalparts.

Complete technical information on HP’s new Metal Jet technology can be found at HP.com/go/3Dmetals.

Commercial HP Metal Jet solutions will be offered at under $399,000 and begin shipping in 2020 to early customers and with broad availability in 2021. Reservations for customers to pre-order HP Metal Jet systems are available today.

171 thoughts on “HP 3D metal printers are 50X more productive for final parts”

  1. OK… so… Not ONE single specification related to actual throughput or performance. Nothing about powdered-metal “inks” costs. Nothing about “voxel size / volume”. Nothing technical beyond the ‘printable’ volume itself. 400 mm (long) x 350 mm (wide) x 200 mm (high/thick?) But WITHOUT specifics … then things like 4× printheads, and 2× this and 50× greater throughput… Are kind of mendacioius. Disingenuous. I would imagine that the tech is quite speedy though. After all, HP has made quite a reputation for fast inkjet process. And a low-gum sinterable metal is a good thing. Just would be nice to know how quick it it might be. … I followed the linkie to the HP website. It was no different… nothing specific. And — pet peeve — the only photo of a part on the main website page? An 8 speed manual transmission shifter knob. Its like advertising motor oil… With pictures of sexy-curvy Mazerati cars. Or tooth whitening toothpaste… With perfectly photoshopped greco-asian female headshots… Just saying. A $400,000 metal printer for … shifter knobs? … There’s another part of me that likes the idea. Shifter knobs. Small decorative parts that are hard to mold. From old’ fashioned pot metal or newer magnesium-aluminum alloys. Maybe if the [i]“per knob cost”[/i] is low enough, once the machines are going 24 hours a day and demonstrate near-zero operational shutdown problems in production, well … setting up a line of 20, 30 … 100 machines, working in parallel, one might be able to make quite a bit of “output stuff”. Good for HP. GoatGuy

    Reply
  2. OK… so…Not ONE single specification related to actual throughput or performance. Nothing about powdered-metal inks”” costs.Nothing about “”””voxel size / volume””””. Nothing technical beyond the ‘printable’ volume itself. 400 mm (long) x 350 mm (wide) x 200 mm (high/thick?)But WITHOUT specifics … then things like4× printheads”” and 2× this and 50× greater throughput…Are kind of mendacioius. Disingenuous. I would imagine that the tech is quite speedy though.After all HP has made quite a reputation for fast inkjet process.And a low-gum sinterable metal is a good thing. Just would be nice to know how quick it it might be.…I followed the linkie to the HP website.It was no different… nothing specific. And — pet peeve — the only photo of a part on the main website page?An 8 speed manual transmission shifter knob. Its like advertising motor oil…With pictures of sexy-curvy Mazerati cars.Or tooth whitening toothpaste…With perfectly photoshopped greco-asian female headshots…Just saying.A $400000 metal printer for … shifter knobs?…There’s another part of me that likes the idea.Shifter knobs. Small decorative parts that are hard to mold. From old’ fashioned pot metal or newer magnesium-aluminum alloys.Maybe if the [i]“per knob cost”[/i] is low enough once the machines are going 24 hours a day and demonstrate near-zero operational shutdown problems in production well … setting up a line of 20 30 … 100 machines working in parallel”” one might be able to make quite a bit of “”””output stuff””””. Good for HP.GoatGuy”””””””

    Reply
  3. Yes, and this “main limitation” is so profound that it severely curtails the usefulness of the entire field of manufacture. In 2005 I saw jet engine combustor “swirl cups” printed in stainless steel/binder which were wicked solid with braze material in a second operation. They were good enough to test the geometry and make relative comparisons between designs. The next step might be micro-peening of the layers as they are “laid-down”, but that product could only hope to match cast material properties. Now, physical vapor deposition of metal atoms, analogous to how pyrolytic graphite is made, might be the alien technology everybody wants to use to build their O’Neill cylinders. Nickel laid down by physical vapor deposition with zero porosity and extremely fine grain size – now that might be something we can work with.

    Reply
  4. You point out the main limitation with this process. It’s not a 3D printing”” process. It uses a laser to build up a part by fusing thin layers from a supply of metal powder. It might be suitable for a few limited applications. But as noted”””” the metallurgical quality of the finished part is not optimum.”””

    Reply
  5. 3D printing or additive manufacturing yields chrap parts with poor fatigue properties poor crystal structure etc. Cast properties at best. It’s fundamental.

    Reply
  6. When you need only 2,500 of them, is when. Thing is tho’, I see this tech competing square-on with modern 5 axis CNC machines. At about the same per-unit cost. A 5 axis CNC machine could mill that pretty shifter knob out of hex-stock aluminum (or bronze for that matter!) in a couple of minutes. Including the shaft threading, and these days, even the finish spit-and-polish. Given the usual automated bar-stock feeding, programmed for step-and-repeat until stock runs out, said CNC would make 20 an hour, about 500 a day. 2,500 of them in just a standard work week. By comparison, by my calculations of 10,000,000 voxels/sec, each shifter knob is what, about 40 mm x 40 mm x 60 mm “as a block”? → 96,000 mm³ per, 768,000,000 voxels ea, or perhaps ±75 sec … 1¼ minute a knob. THIS WOULD BE FASTER than CNC. But unless your looking for that curious pixelated surface roughness, and you don’t need to post-machine the knob’s spiral threading for tight shaft match, then there is likely more steps involved. Unpacking from the white “salt” that holds the printed knobs in place. Post-machining. All that. I’m betting comparable 3 minutes a knob, or 20 an hour, by the time the paint dries. So, very, very similar timing. Thing is (which shows I’m an advocate, not a hopelessly hard-headed goat), one can make parts substantially more complicated than any CNC machine might be conjured to perform. Parts with loose-things inside. Parts with fantastic geometries and inside convex curvatures that defy machining. Parts down to the millimeter scale (CNC machines really don’t like that scale), or long thin AND complicated parts more akin to wires than blocks. CNC can’t handle that. IT HAS A PLACE, without a scintilla of doubt. Just not for sports car shifter knobs per se. (Thing is, I get it: from a marketing perspective, what better to hand a prospective client than a shifter knob mounted on a cool hardwood plate to put on the client’s desk or wall-of-sports-tro

    Reply
  7. Still, I don’t understand why you are so impressed or unimpressed (can’t tell) or moved by this… In what limited applications are “> 93%” dense metal parts of use?

    Reply
  8. Outstanding find. Thank you. 1200 dpi is actually almost 50 µm a dot. Hence why layers are either 1 dot or 2 dots thick. Makes sense. I can imagine that there are many parts where 50 × 50 × 100 um is just fine for voxels. Like sports car gear shifter knobs. (Did you notice how rough the shifter-knob was optically? No part of it was mirror-smooth. I’d expect such from 50³ voxels.) Using “my HP printer” as a likely scale-sizing estimator, it supposedly has 1170 nozzles just for the black ink. And in its one-swipe-high-res mode, it outputs a sheet of 8.5 × 11 paper every 10 seconds. (It is MUCH faster in “draft” mode, and quite a bit slower in “photo-quality” multipass output though.) It also prints 1200 × 1200 dots per square inch. So, assuming that HP gins up the ol’ print inkjet thing to where 3 or 4 thousand nozzles are squirting their stuff out with radical precision, yet giving up little in speed, then I’d imagine the printing is multipass, but still more or less within a (higher) order-of-magnitude close to my desktop printer. 430 × 320 mm per layer. 50 µm thick. 3,000 printhead nozzles. 5× the dot-lay per nozzle speed over my printer. Mine: 54,000 mm² ÷ 10 sec → 5,400 (about) mm²/sec. Dots are 50² µm² ea (1200 dpi). 5,400 × (1 ÷ 0.05² µm²) → 2,200,000 pixels/sec. 2,200,000 ÷ 1,170 nozzles → 1,800 dot/sec per nozzle. Theirs (projection): 3,000 nozzles × 1,800 dot/s × 5 (better thru-put guess) → 28,000,000 voxel/s 28,000,000 × 0.05³ mm³/voxel → 3,400 mm³ per second. 430 mm x 320 mm ÷ 0.05² → 55,000,000 dots/layer 55,000,000 ÷ 28,000,000 → 2 second a layer. 200 mm thick ÷ 0.05 → 4,000 layers 4,000 × 2 sec/layer → 8,000 sec/full scan. → 2.2 hours a block. That’d pretty good, if it were true! If you note the graphic of the printers, the closest (left) one shows 4+ hours. Seems like they’d not choose the worst-case to show… So the printing is probably slower than that, for a “full block”. So my nozzle-count estimate, or the 5× perf

    Reply
  9. ⊕1: brief and to the point. Sophisticated musing… I think THE ‘problem’ with Von Neumann casted machine actors (so far) is that none have come even remotely close to building “the brains”, the “nervous system” and the critically important sensing metrology to (ideally) build replicas that have such significant productive capacity that ‘building more’ is only a wee fraction of their hourly enterprise. For instance, chipmaking. Here we are, 60+ years since the invention of the transistor (74 years, 1947, Bell Labs, Murray Hill NJ… and 59 years for Integrated Circuit // Noyce) and today it takes dozens of different purposed, fabulously precise equipment, scores of exotic materials, and an invisible “feeder supply chain” that is just as remarkable, if in many ways more-so. My point? When I think of Von Neumann self-replicators, I tend to think of them as “whole cities”: like the Silicon Valley, hundreds of key manufacturing centers effecting the supply chain of “stuff” that allows chips to pop out the far, far end. E.g. enterprise specialized in making wire, for them others specialized in making wire-drawing dies, yet others in solvent borne shellac insulation; yet another industry of copper mining, another of its smelting, yet another in its electrolytic purification, and yet-yet another in turning blocks of it into giant spools of wire-drawer’s feedstock. And that’s to make the magnet wire in a hundred grades, to make just the WINDINGS for the motors which power just about every aspect of the Von Neumann collective. Yet others making the iron, more reälloying the pig into magnetically hard steel, roll stock, lamination punchers, motor assembly fabbers. Very likely independent specialists in fabbing all the itty-bits: bearings (oh, the ball bearing specialists!), housing casters, … and the METROLOGY behind keeping these plants accurately working. Never mind… I could go on endlessly in this chain. But the metrology behind it is what I feel is the vital t

    Reply
  10. Here’s the information that should have been in the article (from h20195(dot)www2(dot)hp(dot)com/V2/getpdf.aspx/4AA7-3333ENW.pdf ). • Multiple parts produced at the same time, or large parts, in a powder bed 430 x 320 x 200 mm (16.9 x 12.6 x 7.9 in). • Parts can be arranged freely in multiple levels in the powder bed to optimize packing density, productivity, and cost. • No build plate required, compared with selective laser melting (SLM). • Low-cost, high-quality final parts for serial production up to 100,000 parts. 4 • Best-in-class price-productivity. 3,4 • 1200 x 1200 dpi addressability in a layer 50 to 100 microns thick. • Finished parts with isotropic properties that meet or exceed ASTM and MPIF Standards.5 • High reusability of materials can reduce materials cost and waste without compromising part quality. 6 • Density after sintering > 93%, similar to MIM.

    Reply
  11. Of course if you are laying it down atom by atom, then there is a slight chance you can build up the various carbide inclusions and other intermetalic microstructures that give the high performance to any metal stronger than a simple pure metal. Good luck trying to fabricate the strained and distorted microstructures like work hardening or martensite steel.

    Reply
  12. ⊕1… Emoji (smiley face). I’m neither overtly impressed nor pessimistic. I think HP may well have hit a sweet spot of utility for printed-sintered-metal-parts. There is plenty of applications for 93{22800fc54956079738b58e74e4dcd846757aa319aad70fcf90c97a58f3119a12} dense metal parts. Mmmm… like sports car shifter knobs (Goat flees from fusilade of tomatoes!)No actually I have some modest experience with sintered powder molding. Granted it was back when dinosaurs walked the Berkeley streets but still … sintering is sintering. We found that while sintered parts are nowhere near as strong in tensile strength and softer in Young’s modulus compared to bulk-metal they do have some pretty interesting and useful engineering properties themselves. They can be remarkably tough: there is no voidless matrix to support crack propagation. Bang on them with a hammer and they tend just to dent not crack apart. They also tend to be good”” with repetitive stress deformation — repeatedly flex a sintered part”” and it tends to immediately work-harden then stabilize. Of course if your strains are too great they fail. At lower stresses than solid-metal parts for sure. Thing is that there are a lot of parts that are better in metal but don’t utilize even a small fraction of solid metal’s tensile strength and Young’s modulus stiffness potential. Complicated stuff that goes into automobile passenger seats. Stuff that goes into the making of auto doors latches. Sure… conventional manufacture definitely delivers good cheap durable reasonably precise stuff for the assembly line by the tens-of-thousads without a hitch. But for shorter runs the TOOLING needed to gin out these parts can be an excessively high overhead”” ultimately suppressing innovation.So yah. ⊕1. I see a spot for these HP parts-printers. Not everywhere. But “”””a place”””””” right on part with CNC machines and flatbed waterjet cutters. Or plasma cutters. Just saying””GoatGuy”””””

    Reply
  13. 3D printing or additive manufacturing yields chrap parts with poor fatigue properties poor crystal structure etc. Cast properties at best. It’s fundamental. “” “””

    Reply
  14. When you need only 2500 of them is when. Thing is tho’ I see this tech competing square-on with modern 5 axis CNC machines. At about the same per-unit cost. A 5 axis CNC machine could mill that pretty shifter knob out of hex-stock aluminum (or bronze for that matter!) in a couple of minutes. Including the shaft threading and these days even the finish spit-and-polish. Given the usual automated bar-stock feeding programmed for step-and-repeat until stock runs out said CNC would make 20 an hour about 500 a day. 2500 of them in just a standard work week. By comparison by my calculations of 10000000 voxels/sec each shifter knob is what about 40 mm x 40 mm x 60 mm as a block””? → 96″”000 mm³ per7680000 voxels ea or perhaps ±75 sec … 1¼ minute a knob. THIS WOULD BE FASTER than CNC. But unless your looking for that curious pixelated surface roughness and you don’t need to post-machine the knob’s spiral threading for tight shaft match”” then there is likely more steps involved. Unpacking from the white “”””salt”””” that holds the printed knobs in place. Post-machining. All that. I’m betting comparable 3 minutes a knob”” or 20 an hour by the time the paint dries. So very very similar timing. Thing is (which shows I’m an advocate not a hopelessly hard-headed goat) one can make parts substantially more complicated than any CNC machine might be conjured to perform. Parts with loose-things inside. Parts with fantastic geometries and inside convex curvatures that defy machining. Parts down to the millimeter scale (CNC machines really don’t like that scale) or long thin AND complicated parts more akin to wires than blocks. CNC can’t handle that. IT HAS A PLACE without a scintilla of doubt. Just not for sports car shifter knobs per se. (Thing is I get it: from a marketing perspective what better to hand a prospective client than a shifter knob mounted on a cool hardwood plate to put on the client’s desk or wall-of-sports-trophi”

    Reply
  15. Still I don’t understand why you are so impressed or unimpressed (can’t tell) or moved by this… In what limited applications are > 93{22800fc54956079738b58e74e4dcd846757aa319aad70fcf90c97a58f3119a12}”” dense metal parts of use?”””

    Reply
  16. Outstanding find. Thank you.1200 dpi is actually almost 50 µm a dot. Hence why layers are either 1 dot or 2 dots thick. Makes sense. I can imagine that there are many parts where 50 × 50 × 100 um is just fine for voxels. Like sports car gear shifter knobs. (Did you notice how rough the shifter-knob was optically? No part of it was mirror-smooth. I’d expect such from 50³ voxels.)Using my HP printer”” as a likely scale-sizing estimator”” it supposedly has 1170 nozzles just for the black ink. And in its one-swipe-high-res mode”” it outputs a sheet of 8.5 × 11 paper every 10 seconds. (It is MUCH faster in “”””draft”””” mode”””” and quite a bit slower in “”””photo-quality”””” multipass output though.) It also prints 1200 × 1200 dots per square inch. So”” assuming that HP gins up the ol’ print inkjet thing to where 3 or 4 thousand nozzles are squirting their stuff out with radical precision yet giving up little in speed then I’d imagine the printing is multipass but still more or less within a (higher) order-of-magnitude close to my desktop printer. 430 × 320 mm per layer. 50 µm thick. 3000 printhead nozzles. 5× the dot-lay per nozzle speed over my printer. Mine: 54000 mm² ÷ 10 sec → 5400 (about) mm²/sec. Dots are 50² µm² ea (1200 dpi).5400 × (1 ÷ 0.05² µm²) → 2200000 pixels/sec. 2200000 ÷ 1170 nozzles → 1800 dot/sec per nozzle.Theirs (projection): 3000 nozzles × 1800 dot/s × 5 (better thru-put guess) → 280000 voxel/s280000 × 0.05³ mm³/voxel → 3400 mm³ per second.430 mm x 320 mm ÷ 0.05² → 550000 dots/layer550000 ÷ 280000 → 2 second a layer. 200 mm thick ÷ 0.05 → 4000 layers4000 × 2 sec/layer → 8000 sec/full scan. → 2.2 hours a block.That’d pretty good if it were true! If you note the graphic of the printers the closest (left) one shows 4+ hours.Seems like they’d not choose the worst-case to show…So the printing is probably slower than that”” for a “”””full block””””. So my nozzle-count estimate”” or the”

    Reply
  17. ⊕1: brief and to the point. Sophisticated musing…I think THE ‘problem’ with Von Neumann casted machine actors (so far) is that none have come even remotely close to building the brains”””” the “”””nervous system”””” and the critically important sensing metrology to (ideally) build replicas that have such significant productive capacity that ‘building more’ is only a wee fraction of their hourly enterprise. For instance”” chipmaking. Here we are 60+ years since the invention of the transistor (74 years1947 Bell Labs Murray Hill NJ… and 59 years for Integrated Circuit // Noyce) and today it takes dozens of different purposed fabulously precise equipment scores of exotic materials”” and an invisible “”””feeder supply chain”””” that is just as remarkable”” if in many ways more-so. My point?When I think of Von Neumann self-replicators”” I tend to think of them as “”””whole cities””””: like the Silicon Valley”””” hundreds of key manufacturing centers effecting the supply chain of “”””stuff”””” that allows chips to pop out the far”” far end. E.g. enterprise specialized in making wire for them others specialized in making wire-drawing dies yet others in solvent borne shellac insulation; yet another industry of copper mining another of its smelting yet another in its electrolytic purification and yet-yet another in turning blocks of it into giant spools of wire-drawer’s feedstock. And that’s to make the magnet wire in a hundred grades to make just the WINDINGS for the motors which power just about every aspect of the Von Neumann collective. Yet others making the iron more reälloying the pig into magnetically hard steel roll stock lamination punchers motor assembly fabbers. Very likely independent specialists in fabbing all the itty-bits: bearings (oh the ball bearing specialists!) housing casters … and the METROLOGY behind keeping these plants accurately working. Never mind… I could go on endlessly in this chain. But the metrology behind it is what I fe”

    Reply
  18. Here’s the information that should have been in the article (from h20195(dot)www2(dot)hp(dot)com/V2/getpdf.aspx/4AA7-3333ENW.pdf ).• Multiple parts produced at the same time or large parts in a powder bed 430 x 320 x 200 mm (16.9 x 12.6 x 7.9 in).• Parts can be arranged freely in multiple levels in the powder bed to optimize packing density productivity and cost.• No build plate required compared with selective laser melting (SLM).• Low-cost high-quality final parts for serial production up to 100000 parts.4• Best-in-class price-productivity.34• 1200 x 1200 dpi addressability in a layer 50 to 100 microns thick.• Finished parts with isotropic properties that meet or exceed ASTM and MPIF Standards.5• High reusability of materials can reduce materials cost and waste without compromising part quality.6• Density after sintering > 93{22800fc54956079738b58e74e4dcd846757aa319aad70fcf90c97a58f3119a12} similar to MIM.”

    Reply
  19. Look Matteo, there are numerous applications that are simply unsuited for laser fused (i.e. cast) materials – that’s just mechanical engineering 101 stuff. It’s not going to change just because the process that cast the material is different. If 3D printed items are replacing cast parts, then that is about the limit of the tech.

    Reply
  20. So, we are now replacing single crystal directionally solidified single crystal first stage turbine airfoils with 3D printed ones now, eh? Evidently Siemans has successfully tested such a thing – no mention of how long it lasted though – or what stage, but still remarkable. Mea culpa.

    Reply
  21. Well, every shift knob I’ve ever operated was leather wrapped plastic or urethane wrapped with a thread insert – I’m downmarket.

    Reply
  22. When I think of Von Neumann self-replicators, I tend to think of them as “whole cities” ” Yes I remember in one of the ‘Berserker’ stories that point was made & the plot involved the Berserker machines setting up such a city in a fresh asteroid belt to make & repair the war machines, & human efforts to wreck the ‘city’.

    Reply
  23. Taken the “wrong way”, this is remarkably funny: “Germans have been beating their head against the gear shift knob” Just saying, GoatGuy

    Reply
  24. ⊕1… Emoji (smiley face). I’m neither overtly impressed nor pessimistic. I think HP may well have hit a sweet spot of utility for printed-sintered-metal-parts. There is plenty of applications for 93% dense metal parts. Mmmm… like sports car shifter knobs (Goat flees from fusilade of tomatoes!) No actually I have some modest experience with sintered powder molding. Granted it was back when dinosaurs walked the Berkeley streets, but still … sintering is sintering. We found that while sintered parts are nowhere near as strong in tensile strength and softer in Young’s modulus compared to bulk-metal, they do have some pretty interesting and useful engineering properties themselves. They can be remarkably tough: there is no voidless matrix to support crack propagation. Bang on them with a hammer, and they tend just to dent, not crack apart. They also tend to be “good” with repetitive stress deformation — repeatedly flex a sintered part, and it tends to immediately work-harden then stabilize. Of course if your strains are too great, they fail. At lower stresses than solid-metal parts, for sure. Thing is, that there are a lot of parts that are better in metal, but don’t utilize even a small fraction of solid metal’s tensile strength and Young’s modulus stiffness potential. Complicated stuff that goes into automobile passenger seats. Stuff that goes into the making of auto doors, latches. Sure… conventional manufacture definitely delivers good, cheap, durable, reasonably precise stuff for the assembly line by the tens-of-thousads without a hitch. But for shorter runs, the TOOLING needed to gin out these parts can be an excessively high overhead, ultimately suppressing innovation. So yah. ⊕1. I see a spot for these HP parts-printers. Not everywhere. But “a place”, right on part with CNC machines and flatbed waterjet cutters. Or plasma cutters. Just saying, GoatGuy

    Reply
  25. Look Matteo there are numerous applications that are simply unsuited for laser fused (i.e. cast) materials – that’s just mechanical engineering 101 stuff. It’s not going to change just because the process that cast the material is different. If 3D printed items are replacing cast parts then that is about the limit of the tech.

    Reply
  26. So we are now replacing single crystal directionally solidified single crystal first stage turbine airfoils with 3D printed ones now eh? Evidently Siemans has successfully tested such a thing – no mention of how long it lasted though – or what stage but still remarkable. Mea culpa.

    Reply
  27. Well every shift knob I’ve ever operated was leather wrapped plastic or urethane wrapped with a thread insert – I’m downmarket.

    Reply
  28. When I think of Von Neumann self-replicators” I tend to think of them as “”whole cities”””” “”””YesI remember in one of the ‘Berserker’ stories that point was made & the plot involved the Berserker machines setting up such a city in a fresh asteroid belt to make & repair the war machines”””” & human efforts to wreck the ‘city’.”””

    Reply
  29. Taken the wrong way”””” this is remarkably funny: “”””Germans have been beating their head against the gear shift knob””””Just saying””””GoatGuy”””

    Reply
  30. Sheepishly (particularly hard for “the goat”), AS A SCIFI avid first-edition-signed collector, I have to admit never having read or collected the Berserker series… apparently to my loss. Thanks. GoatGuy

    Reply
  31. Sheepishly (particularly hard for the goat””)”” AS A SCIFI avid first-edition-signed collector”” I have to admit never having read or collected the Berserker series… apparently to my loss. Thanks. GoatGuy”””

    Reply
  32. Look liar, you’re simply FoS. There are roles where the microstructure of sintered parts is superior to cast or forged, because it is produced without aging or work hardening (requiring relieving). It is it’s own thing, not a variety of casting.

    Reply
  33. Look liar you’re simply FoS. There are roles where the microstructure of sintered parts is superior to cast or forged because it is produced without aging or work hardening (requiring relieving). It is it’s own thing not a variety of casting.

    Reply
  34. Coincidentally I was going for a test drive for the first time last night with my brand new shifters. Carbonfibre donchaknow? Which means largely plastic, of course, but more expensive. The impression I was getting from this story was that they wanted one-off or 5-off personalized car bits. So the gear knob doesn’t just have 8-speeds marked on the top. It has “Scaryjello” embossed down both sides once the dealership has given you the $2000 personalisation option (higher retail price, lower resale price, what’s not to love?) If you are really tasteless the car you’ve just bought from Goat’s BMW dealership has “Goat BMW and Beer Emporium” embossed in the custom parts, because the average car dealership couldn’t resist that. And maybe that will encourage you to fork out the $2000 to get it replaced with “Scaryjello”. And I guess it’s possible that a modern fancy shift know has enough internal structure (for the 6 different fingertip controls) that making it CNC would actually be more difficult. An additional factor is that the car dealer might just have their 3D printer in the back of the shop. While they aren’t going to have a CNC mill.

    Reply
  35. The alternative method of making turbine blades is… casting. So Scaryjello saying that the printing is at best as good as casting is hardly a contradiction is it? You get really excited about all these cool things, and then you see not everyone agrees. And you launch personal attacks on anyone who isn’t 100% with the cool new tech. Maybe you should just settle down a bit. Someone disagreeing with an article that you like isn’t a personal attack on you, so stop behaving like it is.

    Reply
  36. Coincidentally I was going for a test drive for the first time last night with my brand new shifters. Carbonfibre donchaknow?Which means largely plastic of course but more expensive.The impression I was getting from this story was that they wanted one-off or 5-off personalized car bits. So the gear knob doesn’t just have 8-speeds marked on the top. It has Scaryjello”” embossed down both sides once the dealership has given you the $2000 personalisation option (higher retail price”” lower resale price”” what’s not to love?)If you are really tasteless the car you’ve just bought from Goat’s BMW dealership has “”””Goat BMW and Beer Emporium”””” embossed in the custom parts”””” because the average car dealership couldn’t resist that. And maybe that will encourage you to fork out the $2000 to get it replaced with “”””Scaryjello””””.And I guess it’s possible that a modern fancy shift know has enough internal structure (for the 6 different fingertip controls) that making it CNC would actually be more difficult.An additional factor is that the car dealer might just have their 3D printer in the back of the shop. While they aren’t going to have a CNC mill.”””

    Reply
  37. The alternative method of making turbine blades is… casting.So Scaryjello saying that the printing is at best as good as casting is hardly a contradiction is it?You get really excited about all these cool things and then you see not everyone agrees. And you launch personal attacks on anyone who isn’t 100{22800fc54956079738b58e74e4dcd846757aa319aad70fcf90c97a58f3119a12} with the cool new tech. Maybe you should just settle down a bit. Someone disagreeing with an article that you like isn’t a personal attack on you so stop behaving like it is.

    Reply
  38. Very much so, the short stories were especially good. But now seem a bit…dated. in some ways. Anyways to the topic-I am excited, this is the first step, at close to half a million its not for most peoples garage or office. But its not unreasonable to foresee a time in the future where it could be. And once that occurs I think we will see rapid use of them in many innovative ways. Including some boot strappy ways. The major impediment will be all the other non structural materials needed, primarily electronics. But I could easily see electronic kits being sold separately from the structure of things. And someone will make a home made rocket engine once the materials are there, and a matching printing method that is affordable is there.

    Reply
  39. Very much so the short stories were especially good. But now seem a bit…dated. in some ways.Anyways to the topic-I am excited this is the first step at close to half a million its not for most peoples garage or office. But its not unreasonable to foresee a time in the future where it could be. And once that occurs I think we will see rapid use of them in many innovative ways.Including some boot strappy ways. The major impediment will be all the other non structural materials needed primarily electronics. But I could easily see electronic kits being sold separately from the structure of things. And someone will make a home made rocket engine once the materials are there and a matching printing method that is affordable is there.

    Reply
  40. 3d printed metal is nice for complex internal shapes. Impellers in rocket engines might be the most infamous one there you want to mix fuel and oxidizer trough interlocking holes in the bottom at +100 bar and some ton/ second. Low volume production, very hard to cast.

    Reply
  41. 3d printed metal is nice for complex internal shapes. Impellers in rocket engines might be the most infamous one there you want to mix fuel and oxidizer trough interlocking holes in the bottom at +100 bar and some ton/ second. Low volume production very hard to cast.

    Reply
  42. numbers crunching is logical.. but need to think hp team might have also done this numbers crunching… But ultimately being hp there must be some strategy about claims which will become clear in coming time once end users and service providers come up with idealistic case studies and real numbers… and then there will be new hype design for hp 3d printing… And in this time muscle of hp in terms of global footprint and finance may create dominant place … Still wait and watch game..

    Reply
  43. numbers crunching is logical.. but need to think hp team might have also done this numbers crunching… But ultimately being hp there must be some strategy about claims which will become clear in coming time once end users and service providers come up with idealistic case studies and real numbers… and then there will be new hype design for hp 3d printing… And in this time muscle of hp in terms of global footprint and finance may create dominant place … Still wait and watch game..

    Reply
  44. numbers crunching is logical.. but need to think hp team might have also done this numbers crunching… But ultimately being hp there must be some strategy about claims which will become clear in coming time once end users and service providers come up with idealistic case studies and real numbers… and then there will be new hype design for hp 3d printing… And in this time muscle of hp in terms of global footprint and finance may create dominant place … Still wait and watch game..

    Reply
  45. numbers crunching is logical.. but need to think hp team might have also done this numbers crunching… But ultimately being hp there must be some strategy about claims which will become clear in coming time once end users and service providers come up with idealistic case studies and real numbers… and then there will be new hype design for hp 3d printing… And in this time muscle of hp in terms of global footprint and finance may create dominant place … Still wait and watch game..

    Reply
  46. numbers crunching is logical.. but need to think hp team might have also done this numbers crunching… But ultimately being hp there must be some strategy about claims which will become clear in coming time once end users and service providers come up with idealistic case studies and real numbers… and then there will be new hype design for hp 3d printing… And in this time muscle of hp in terms of global footprint and finance may create dominant place … Still wait and watch game..

    Reply
  47. 3d printed metal is nice for complex internal shapes. Impellers in rocket engines might be the most infamous one there you want to mix fuel and oxidizer trough interlocking holes in the bottom at +100 bar and some ton/ second. Low volume production, very hard to cast.

    Reply
  48. 3d printed metal is nice for complex internal shapes. Impellers in rocket engines might be the most infamous one there you want to mix fuel and oxidizer trough interlocking holes in the bottom at +100 bar and some ton/ second. Low volume production very hard to cast.

    Reply
  49. 3d printed metal is nice for complex internal shapes. Impellers in rocket engines might be the most infamous one there you want to mix fuel and oxidizer trough interlocking holes in the bottom at +100 bar and some ton/ second. Low volume production, very hard to cast.

    Reply
  50. Very much so, the short stories were especially good. But now seem a bit…dated. in some ways. Anyways to the topic-I am excited, this is the first step, at close to half a million its not for most peoples garage or office. But its not unreasonable to foresee a time in the future where it could be. And once that occurs I think we will see rapid use of them in many innovative ways. Including some boot strappy ways. The major impediment will be all the other non structural materials needed, primarily electronics. But I could easily see electronic kits being sold separately from the structure of things. And someone will make a home made rocket engine once the materials are there, and a matching printing method that is affordable is there.

    Reply
  51. Very much so the short stories were especially good. But now seem a bit…dated. in some ways.Anyways to the topic-I am excited this is the first step at close to half a million its not for most peoples garage or office. But its not unreasonable to foresee a time in the future where it could be. And once that occurs I think we will see rapid use of them in many innovative ways.Including some boot strappy ways. The major impediment will be all the other non structural materials needed primarily electronics. But I could easily see electronic kits being sold separately from the structure of things. And someone will make a home made rocket engine once the materials are there and a matching printing method that is affordable is there.

    Reply
  52. Very much so, the short stories were especially good. But now seem a bit…dated. in some ways.

    Anyways to the topic-I am excited, this is the first step, at close to half a million its not for most peoples garage or office. But its not unreasonable to foresee a time in the future where it could be. And once that occurs I think we will see rapid use of them in many innovative ways.

    Including some boot strappy ways. The major impediment will be all the other non structural materials needed, primarily electronics. But I could easily see electronic kits being sold separately from the structure of things. And someone will make a home made rocket engine once the materials are there, and a matching printing method that is affordable is there.

    Reply
  53. Coincidentally I was going for a test drive for the first time last night with my brand new shifters. Carbonfibre donchaknow? Which means largely plastic, of course, but more expensive. The impression I was getting from this story was that they wanted one-off or 5-off personalized car bits. So the gear knob doesn’t just have 8-speeds marked on the top. It has “Scaryjello” embossed down both sides once the dealership has given you the $2000 personalisation option (higher retail price, lower resale price, what’s not to love?) If you are really tasteless the car you’ve just bought from Goat’s BMW dealership has “Goat BMW and Beer Emporium” embossed in the custom parts, because the average car dealership couldn’t resist that. And maybe that will encourage you to fork out the $2000 to get it replaced with “Scaryjello”. And I guess it’s possible that a modern fancy shift know has enough internal structure (for the 6 different fingertip controls) that making it CNC would actually be more difficult. An additional factor is that the car dealer might just have their 3D printer in the back of the shop. While they aren’t going to have a CNC mill.

    Reply
  54. Coincidentally I was going for a test drive for the first time last night with my brand new shifters. Carbonfibre donchaknow?Which means largely plastic of course but more expensive.The impression I was getting from this story was that they wanted one-off or 5-off personalized car bits. So the gear knob doesn’t just have 8-speeds marked on the top. It has Scaryjello”” embossed down both sides once the dealership has given you the $2000 personalisation option (higher retail price”” lower resale price”” what’s not to love?)If you are really tasteless the car you’ve just bought from Goat’s BMW dealership has “”””Goat BMW and Beer Emporium”””” embossed in the custom parts”””” because the average car dealership couldn’t resist that. And maybe that will encourage you to fork out the $2000 to get it replaced with “”””Scaryjello””””.And I guess it’s possible that a modern fancy shift know has enough internal structure (for the 6 different fingertip controls) that making it CNC would actually be more difficult.An additional factor is that the car dealer might just have their 3D printer in the back of the shop. While they aren’t going to have a CNC mill.”””

    Reply
  55. The alternative method of making turbine blades is… casting. So Scaryjello saying that the printing is at best as good as casting is hardly a contradiction is it? You get really excited about all these cool things, and then you see not everyone agrees. And you launch personal attacks on anyone who isn’t 100% with the cool new tech. Maybe you should just settle down a bit. Someone disagreeing with an article that you like isn’t a personal attack on you, so stop behaving like it is.

    Reply
  56. The alternative method of making turbine blades is… casting.So Scaryjello saying that the printing is at best as good as casting is hardly a contradiction is it?You get really excited about all these cool things and then you see not everyone agrees. And you launch personal attacks on anyone who isn’t 100{22800fc54956079738b58e74e4dcd846757aa319aad70fcf90c97a58f3119a12} with the cool new tech. Maybe you should just settle down a bit. Someone disagreeing with an article that you like isn’t a personal attack on you so stop behaving like it is.

    Reply
  57. Look liar, you’re simply FoS. There are roles where the microstructure of sintered parts is superior to cast or forged, because it is produced without aging or work hardening (requiring relieving). It is it’s own thing, not a variety of casting.

    Reply
  58. Look liar you’re simply FoS. There are roles where the microstructure of sintered parts is superior to cast or forged because it is produced without aging or work hardening (requiring relieving). It is it’s own thing not a variety of casting.

    Reply
  59. Coincidentally I was going for a test drive for the first time last night with my brand new shifters.

    Carbonfibre donchaknow?

    Which means largely plastic, of course, but more expensive.

    The impression I was getting from this story was that they wanted one-off or 5-off personalized car bits.
    So the gear knob doesn’t just have 8-speeds marked on the top. It has “Scaryjello” embossed down both sides once the dealership has given you the $2000 personalisation option (higher retail price, lower resale price, what’s not to love?)

    If you are really tasteless the car you’ve just bought from Goat’s BMW dealership has “Goat BMW and Beer Emporium” embossed in the custom parts, because the average car dealership couldn’t resist that. And maybe that will encourage you to fork out the $2000 to get it replaced with “Scaryjello”.

    And I guess it’s possible that a modern fancy shift know has enough internal structure (for the 6 different fingertip controls) that making it CNC would actually be more difficult.

    An additional factor is that the car dealer might just have their 3D printer in the back of the shop. While they aren’t going to have a CNC mill.

    Reply
  60. Sheepishly (particularly hard for “the goat”), AS A SCIFI avid first-edition-signed collector, I have to admit never having read or collected the Berserker series… apparently to my loss. Thanks. GoatGuy

    Reply
  61. Sheepishly (particularly hard for the goat””)”” AS A SCIFI avid first-edition-signed collector”” I have to admit never having read or collected the Berserker series… apparently to my loss. Thanks. GoatGuy”””

    Reply
  62. The alternative method of making turbine blades is… casting.
    So Scaryjello saying that the printing is at best as good as casting is hardly a contradiction is it?

    You get really excited about all these cool things, and then you see not everyone agrees. And you launch personal attacks on anyone who isn’t 100% with the cool new tech.

    Maybe you should just settle down a bit. Someone disagreeing with an article that you like isn’t a personal attack on you, so stop behaving like it is.

    Reply
  63. Look Matteo, there are numerous applications that are simply unsuited for laser fused (i.e. cast) materials – that’s just mechanical engineering 101 stuff. It’s not going to change just because the process that cast the material is different. If 3D printed items are replacing cast parts, then that is about the limit of the tech.

    Reply
  64. Look Matteo there are numerous applications that are simply unsuited for laser fused (i.e. cast) materials – that’s just mechanical engineering 101 stuff. It’s not going to change just because the process that cast the material is different. If 3D printed items are replacing cast parts then that is about the limit of the tech.

    Reply
  65. So, we are now replacing single crystal directionally solidified single crystal first stage turbine airfoils with 3D printed ones now, eh? Evidently Siemans has successfully tested such a thing – no mention of how long it lasted though – or what stage, but still remarkable. Mea culpa.

    Reply
  66. So we are now replacing single crystal directionally solidified single crystal first stage turbine airfoils with 3D printed ones now eh? Evidently Siemans has successfully tested such a thing – no mention of how long it lasted though – or what stage but still remarkable. Mea culpa.

    Reply
  67. Well, every shift knob I’ve ever operated was leather wrapped plastic or urethane wrapped with a thread insert – I’m downmarket.

    Reply
  68. Well every shift knob I’ve ever operated was leather wrapped plastic or urethane wrapped with a thread insert – I’m downmarket.

    Reply
  69. When I think of Von Neumann self-replicators, I tend to think of them as “whole cities” ” Yes I remember in one of the ‘Berserker’ stories that point was made & the plot involved the Berserker machines setting up such a city in a fresh asteroid belt to make & repair the war machines, & human efforts to wreck the ‘city’.

    Reply
  70. When I think of Von Neumann self-replicators” I tend to think of them as “”whole cities”””” “”””YesI remember in one of the ‘Berserker’ stories that point was made & the plot involved the Berserker machines setting up such a city in a fresh asteroid belt to make & repair the war machines”””” & human efforts to wreck the ‘city’.”””

    Reply
  71. Taken the “wrong way”, this is remarkably funny: “Germans have been beating their head against the gear shift knob” Just saying, GoatGuy

    Reply
  72. Taken the wrong way”””” this is remarkably funny: “”””Germans have been beating their head against the gear shift knob””””Just saying””””GoatGuy”””

    Reply
  73. ⊕1… Emoji (smiley face). I’m neither overtly impressed nor pessimistic. I think HP may well have hit a sweet spot of utility for printed-sintered-metal-parts. There is plenty of applications for 93% dense metal parts. Mmmm… like sports car shifter knobs (Goat flees from fusilade of tomatoes!) No actually I have some modest experience with sintered powder molding. Granted it was back when dinosaurs walked the Berkeley streets, but still … sintering is sintering. We found that while sintered parts are nowhere near as strong in tensile strength and softer in Young’s modulus compared to bulk-metal, they do have some pretty interesting and useful engineering properties themselves. They can be remarkably tough: there is no voidless matrix to support crack propagation. Bang on them with a hammer, and they tend just to dent, not crack apart. They also tend to be “good” with repetitive stress deformation — repeatedly flex a sintered part, and it tends to immediately work-harden then stabilize. Of course if your strains are too great, they fail. At lower stresses than solid-metal parts, for sure. Thing is, that there are a lot of parts that are better in metal, but don’t utilize even a small fraction of solid metal’s tensile strength and Young’s modulus stiffness potential. Complicated stuff that goes into automobile passenger seats. Stuff that goes into the making of auto doors, latches. Sure… conventional manufacture definitely delivers good, cheap, durable, reasonably precise stuff for the assembly line by the tens-of-thousads without a hitch. But for shorter runs, the TOOLING needed to gin out these parts can be an excessively high overhead, ultimately suppressing innovation. So yah. ⊕1. I see a spot for these HP parts-printers. Not everywhere. But “a place”, right on part with CNC machines and flatbed waterjet cutters. Or plasma cutters. Just saying, GoatGuy

    Reply
  74. ⊕1… Emoji (smiley face). I’m neither overtly impressed nor pessimistic. I think HP may well have hit a sweet spot of utility for printed-sintered-metal-parts. There is plenty of applications for 93{22800fc54956079738b58e74e4dcd846757aa319aad70fcf90c97a58f3119a12} dense metal parts. Mmmm… like sports car shifter knobs (Goat flees from fusilade of tomatoes!)No actually I have some modest experience with sintered powder molding. Granted it was back when dinosaurs walked the Berkeley streets but still … sintering is sintering. We found that while sintered parts are nowhere near as strong in tensile strength and softer in Young’s modulus compared to bulk-metal they do have some pretty interesting and useful engineering properties themselves. They can be remarkably tough: there is no voidless matrix to support crack propagation. Bang on them with a hammer and they tend just to dent not crack apart. They also tend to be good”” with repetitive stress deformation — repeatedly flex a sintered part”” and it tends to immediately work-harden then stabilize. Of course if your strains are too great they fail. At lower stresses than solid-metal parts for sure. Thing is that there are a lot of parts that are better in metal but don’t utilize even a small fraction of solid metal’s tensile strength and Young’s modulus stiffness potential. Complicated stuff that goes into automobile passenger seats. Stuff that goes into the making of auto doors latches. Sure… conventional manufacture definitely delivers good cheap durable reasonably precise stuff for the assembly line by the tens-of-thousads without a hitch. But for shorter runs the TOOLING needed to gin out these parts can be an excessively high overhead”” ultimately suppressing innovation.So yah. ⊕1. I see a spot for these HP parts-printers. Not everywhere. But “”””a place”””””” right on part with CNC machines and flatbed waterjet cutters. Or plasma cutters. Just saying””GoatGuy”””””

    Reply
  75. 3D printing or additive manufacturing yields chrap parts with poor fatigue properties poor crystal structure etc. Cast properties at best. It’s fundamental.

    Reply
  76. 3D printing or additive manufacturing yields chrap parts with poor fatigue properties poor crystal structure etc. Cast properties at best. It’s fundamental. “” “””

    Reply
  77. When you need only 2,500 of them, is when. Thing is tho’, I see this tech competing square-on with modern 5 axis CNC machines. At about the same per-unit cost. A 5 axis CNC machine could mill that pretty shifter knob out of hex-stock aluminum (or bronze for that matter!) in a couple of minutes. Including the shaft threading, and these days, even the finish spit-and-polish. Given the usual automated bar-stock feeding, programmed for step-and-repeat until stock runs out, said CNC would make 20 an hour, about 500 a day. 2,500 of them in just a standard work week. By comparison, by my calculations of 10,000,000 voxels/sec, each shifter knob is what, about 40 mm x 40 mm x 60 mm “as a block”? → 96,000 mm³ per, 768,000,000 voxels ea, or perhaps ±75 sec … 1¼ minute a knob. THIS WOULD BE FASTER than CNC. But unless your looking for that curious pixelated surface roughness, and you don’t need to post-machine the knob’s spiral threading for tight shaft match, then there is likely more steps involved. Unpacking from the white “salt” that holds the printed knobs in place. Post-machining. All that. I’m betting comparable 3 minutes a knob, or 20 an hour, by the time the paint dries. So, very, very similar timing. Thing is (which shows I’m an advocate, not a hopelessly hard-headed goat), one can make parts substantially more complicated than any CNC machine might be conjured to perform. Parts with loose-things inside. Parts with fantastic geometries and inside convex curvatures that defy machining. Parts down to the millimeter scale (CNC machines really don’t like that scale), or long thin AND complicated parts more akin to wires than blocks. CNC can’t handle that. IT HAS A PLACE, without a scintilla of doubt. Just not for sports car shifter knobs per se. (Thing is, I get it: from a marketing perspective, what better to hand a prospective client than a shifter knob mounted on a cool hardwood plate to put on the client’s desk or wall-of-sports-tro

    Reply
  78. When you need only 2500 of them is when. Thing is tho’ I see this tech competing square-on with modern 5 axis CNC machines. At about the same per-unit cost. A 5 axis CNC machine could mill that pretty shifter knob out of hex-stock aluminum (or bronze for that matter!) in a couple of minutes. Including the shaft threading and these days even the finish spit-and-polish. Given the usual automated bar-stock feeding programmed for step-and-repeat until stock runs out said CNC would make 20 an hour about 500 a day. 2500 of them in just a standard work week. By comparison by my calculations of 10000000 voxels/sec each shifter knob is what about 40 mm x 40 mm x 60 mm as a block””? → 96″”000 mm³ per7680000 voxels ea or perhaps ±75 sec … 1¼ minute a knob. THIS WOULD BE FASTER than CNC. But unless your looking for that curious pixelated surface roughness and you don’t need to post-machine the knob’s spiral threading for tight shaft match”” then there is likely more steps involved. Unpacking from the white “”””salt”””” that holds the printed knobs in place. Post-machining. All that. I’m betting comparable 3 minutes a knob”” or 20 an hour by the time the paint dries. So very very similar timing. Thing is (which shows I’m an advocate not a hopelessly hard-headed goat) one can make parts substantially more complicated than any CNC machine might be conjured to perform. Parts with loose-things inside. Parts with fantastic geometries and inside convex curvatures that defy machining. Parts down to the millimeter scale (CNC machines really don’t like that scale) or long thin AND complicated parts more akin to wires than blocks. CNC can’t handle that. IT HAS A PLACE without a scintilla of doubt. Just not for sports car shifter knobs per se. (Thing is I get it: from a marketing perspective what better to hand a prospective client than a shifter knob mounted on a cool hardwood plate to put on the client’s desk or wall-of-sports-trophi”

    Reply
  79. Still, I don’t understand why you are so impressed or unimpressed (can’t tell) or moved by this… In what limited applications are “> 93%” dense metal parts of use?

    Reply
  80. Still I don’t understand why you are so impressed or unimpressed (can’t tell) or moved by this… In what limited applications are > 93{22800fc54956079738b58e74e4dcd846757aa319aad70fcf90c97a58f3119a12}”” dense metal parts of use?”””

    Reply
  81. Outstanding find. Thank you. 1200 dpi is actually almost 50 µm a dot. Hence why layers are either 1 dot or 2 dots thick. Makes sense. I can imagine that there are many parts where 50 × 50 × 100 um is just fine for voxels. Like sports car gear shifter knobs. (Did you notice how rough the shifter-knob was optically? No part of it was mirror-smooth. I’d expect such from 50³ voxels.) Using “my HP printer” as a likely scale-sizing estimator, it supposedly has 1170 nozzles just for the black ink. And in its one-swipe-high-res mode, it outputs a sheet of 8.5 × 11 paper every 10 seconds. (It is MUCH faster in “draft” mode, and quite a bit slower in “photo-quality” multipass output though.) It also prints 1200 × 1200 dots per square inch. So, assuming that HP gins up the ol’ print inkjet thing to where 3 or 4 thousand nozzles are squirting their stuff out with radical precision, yet giving up little in speed, then I’d imagine the printing is multipass, but still more or less within a (higher) order-of-magnitude close to my desktop printer. 430 × 320 mm per layer. 50 µm thick. 3,000 printhead nozzles. 5× the dot-lay per nozzle speed over my printer. Mine: 54,000 mm² ÷ 10 sec → 5,400 (about) mm²/sec. Dots are 50² µm² ea (1200 dpi). 5,400 × (1 ÷ 0.05² µm²) → 2,200,000 pixels/sec. 2,200,000 ÷ 1,170 nozzles → 1,800 dot/sec per nozzle. Theirs (projection): 3,000 nozzles × 1,800 dot/s × 5 (better thru-put guess) → 28,000,000 voxel/s 28,000,000 × 0.05³ mm³/voxel → 3,400 mm³ per second. 430 mm x 320 mm ÷ 0.05² → 55,000,000 dots/layer 55,000,000 ÷ 28,000,000 → 2 second a layer. 200 mm thick ÷ 0.05 → 4,000 layers 4,000 × 2 sec/layer → 8,000 sec/full scan. → 2.2 hours a block. That’d pretty good, if it were true! If you note the graphic of the printers, the closest (left) one shows 4+ hours. Seems like they’d not choose the worst-case to show… So the printing is probably slower than that, for a “full block”. So my nozzle-count estimate, or the 5× perf

    Reply
  82. Outstanding find. Thank you.1200 dpi is actually almost 50 µm a dot. Hence why layers are either 1 dot or 2 dots thick. Makes sense. I can imagine that there are many parts where 50 × 50 × 100 um is just fine for voxels. Like sports car gear shifter knobs. (Did you notice how rough the shifter-knob was optically? No part of it was mirror-smooth. I’d expect such from 50³ voxels.)Using my HP printer”” as a likely scale-sizing estimator”” it supposedly has 1170 nozzles just for the black ink. And in its one-swipe-high-res mode”” it outputs a sheet of 8.5 × 11 paper every 10 seconds. (It is MUCH faster in “”””draft”””” mode”””” and quite a bit slower in “”””photo-quality”””” multipass output though.) It also prints 1200 × 1200 dots per square inch. So”” assuming that HP gins up the ol’ print inkjet thing to where 3 or 4 thousand nozzles are squirting their stuff out with radical precision yet giving up little in speed then I’d imagine the printing is multipass but still more or less within a (higher) order-of-magnitude close to my desktop printer. 430 × 320 mm per layer. 50 µm thick. 3000 printhead nozzles. 5× the dot-lay per nozzle speed over my printer. Mine: 54000 mm² ÷ 10 sec → 5400 (about) mm²/sec. Dots are 50² µm² ea (1200 dpi).5400 × (1 ÷ 0.05² µm²) → 2200000 pixels/sec. 2200000 ÷ 1170 nozzles → 1800 dot/sec per nozzle.Theirs (projection): 3000 nozzles × 1800 dot/s × 5 (better thru-put guess) → 280000 voxel/s280000 × 0.05³ mm³/voxel → 3400 mm³ per second.430 mm x 320 mm ÷ 0.05² → 550000 dots/layer550000 ÷ 280000 → 2 second a layer. 200 mm thick ÷ 0.05 → 4000 layers4000 × 2 sec/layer → 8000 sec/full scan. → 2.2 hours a block.That’d pretty good if it were true! If you note the graphic of the printers the closest (left) one shows 4+ hours.Seems like they’d not choose the worst-case to show…So the printing is probably slower than that”” for a “”””full block””””. So my nozzle-count estimate”” or the”

    Reply
  83. ⊕1: brief and to the point. Sophisticated musing… I think THE ‘problem’ with Von Neumann casted machine actors (so far) is that none have come even remotely close to building “the brains”, the “nervous system” and the critically important sensing metrology to (ideally) build replicas that have such significant productive capacity that ‘building more’ is only a wee fraction of their hourly enterprise. For instance, chipmaking. Here we are, 60+ years since the invention of the transistor (74 years, 1947, Bell Labs, Murray Hill NJ… and 59 years for Integrated Circuit // Noyce) and today it takes dozens of different purposed, fabulously precise equipment, scores of exotic materials, and an invisible “feeder supply chain” that is just as remarkable, if in many ways more-so. My point? When I think of Von Neumann self-replicators, I tend to think of them as “whole cities”: like the Silicon Valley, hundreds of key manufacturing centers effecting the supply chain of “stuff” that allows chips to pop out the far, far end. E.g. enterprise specialized in making wire, for them others specialized in making wire-drawing dies, yet others in solvent borne shellac insulation; yet another industry of copper mining, another of its smelting, yet another in its electrolytic purification, and yet-yet another in turning blocks of it into giant spools of wire-drawer’s feedstock. And that’s to make the magnet wire in a hundred grades, to make just the WINDINGS for the motors which power just about every aspect of the Von Neumann collective. Yet others making the iron, more reälloying the pig into magnetically hard steel, roll stock, lamination punchers, motor assembly fabbers. Very likely independent specialists in fabbing all the itty-bits: bearings (oh, the ball bearing specialists!), housing casters, … and the METROLOGY behind keeping these plants accurately working. Never mind… I could go on endlessly in this chain. But the metrology behind it is what I feel is the vital t

    Reply
  84. ⊕1: brief and to the point. Sophisticated musing…I think THE ‘problem’ with Von Neumann casted machine actors (so far) is that none have come even remotely close to building the brains”””” the “”””nervous system”””” and the critically important sensing metrology to (ideally) build replicas that have such significant productive capacity that ‘building more’ is only a wee fraction of their hourly enterprise. For instance”” chipmaking. Here we are 60+ years since the invention of the transistor (74 years1947 Bell Labs Murray Hill NJ… and 59 years for Integrated Circuit // Noyce) and today it takes dozens of different purposed fabulously precise equipment scores of exotic materials”” and an invisible “”””feeder supply chain”””” that is just as remarkable”” if in many ways more-so. My point?When I think of Von Neumann self-replicators”” I tend to think of them as “”””whole cities””””: like the Silicon Valley”””” hundreds of key manufacturing centers effecting the supply chain of “”””stuff”””” that allows chips to pop out the far”” far end. E.g. enterprise specialized in making wire for them others specialized in making wire-drawing dies yet others in solvent borne shellac insulation; yet another industry of copper mining another of its smelting yet another in its electrolytic purification and yet-yet another in turning blocks of it into giant spools of wire-drawer’s feedstock. And that’s to make the magnet wire in a hundred grades to make just the WINDINGS for the motors which power just about every aspect of the Von Neumann collective. Yet others making the iron more reälloying the pig into magnetically hard steel roll stock lamination punchers motor assembly fabbers. Very likely independent specialists in fabbing all the itty-bits: bearings (oh the ball bearing specialists!) housing casters … and the METROLOGY behind keeping these plants accurately working. Never mind… I could go on endlessly in this chain. But the metrology behind it is what I fe”

    Reply
  85. Here’s the information that should have been in the article (from h20195(dot)www2(dot)hp(dot)com/V2/getpdf.aspx/4AA7-3333ENW.pdf ). • Multiple parts produced at the same time, or large parts, in a powder bed 430 x 320 x 200 mm (16.9 x 12.6 x 7.9 in). • Parts can be arranged freely in multiple levels in the powder bed to optimize packing density, productivity, and cost. • No build plate required, compared with selective laser melting (SLM). • Low-cost, high-quality final parts for serial production up to 100,000 parts. 4 • Best-in-class price-productivity. 3,4 • 1200 x 1200 dpi addressability in a layer 50 to 100 microns thick. • Finished parts with isotropic properties that meet or exceed ASTM and MPIF Standards.5 • High reusability of materials can reduce materials cost and waste without compromising part quality. 6 • Density after sintering > 93%, similar to MIM.

    Reply
  86. Here’s the information that should have been in the article (from h20195(dot)www2(dot)hp(dot)com/V2/getpdf.aspx/4AA7-3333ENW.pdf ).• Multiple parts produced at the same time or large parts in a powder bed 430 x 320 x 200 mm (16.9 x 12.6 x 7.9 in).• Parts can be arranged freely in multiple levels in the powder bed to optimize packing density productivity and cost.• No build plate required compared with selective laser melting (SLM).• Low-cost high-quality final parts for serial production up to 100000 parts.4• Best-in-class price-productivity.34• 1200 x 1200 dpi addressability in a layer 50 to 100 microns thick.• Finished parts with isotropic properties that meet or exceed ASTM and MPIF Standards.5• High reusability of materials can reduce materials cost and waste without compromising part quality.6• Density after sintering > 93{22800fc54956079738b58e74e4dcd846757aa319aad70fcf90c97a58f3119a12} similar to MIM.”

    Reply
  87. Of course if you are laying it down atom by atom, then there is a slight chance you can build up the various carbide inclusions and other intermetalic microstructures that give the high performance to any metal stronger than a simple pure metal. Good luck trying to fabricate the strained and distorted microstructures like work hardening or martensite steel.

    Reply
  88. Of course if you are laying it down atom by atom then there is a slight chance you can build up the various carbide inclusions and other intermetalic microstructures that give the high performance to any metal stronger than a simple pure metal.Good luck trying to fabricate the strained and distorted microstructures like work hardening or martensite steel.

    Reply
  89. Yes, and this “main limitation” is so profound that it severely curtails the usefulness of the entire field of manufacture. In 2005 I saw jet engine combustor “swirl cups” printed in stainless steel/binder which were wicked solid with braze material in a second operation. They were good enough to test the geometry and make relative comparisons between designs. The next step might be micro-peening of the layers as they are “laid-down”, but that product could only hope to match cast material properties. Now, physical vapor deposition of metal atoms, analogous to how pyrolytic graphite is made, might be the alien technology everybody wants to use to build their O’Neill cylinders. Nickel laid down by physical vapor deposition with zero porosity and extremely fine grain size – now that might be something we can work with.

    Reply
  90. Yes and this main limitation”” is so profound that it severely curtails the usefulness of the entire field of manufacture. In 2005 I saw jet engine combustor “”””swirl cups”””” printed in stainless steel/binder which were wicked solid with braze material in a second operation. They were good enough to test the geometry and make relative comparisons between designs.The next step might be micro-peening of the layers as they are “”””laid-down”””””” but that product could only hope to match cast material properties.Now physical vapor deposition of metal atoms analogous to how pyrolytic graphite is made”” might be the alien technology everybody wants to use to build their O’Neill cylinders. Nickel laid down by physical vapor deposition with zero porosity and extremely fine grain size – now that might be something we can work with.”””

    Reply
  91. You point out the main limitation with this process. It’s not a 3D “printing” process. It uses a laser to build up a part by fusing thin layers from a supply of metal powder. It might be suitable for a few limited applications. But as noted, the metallurgical quality of the finished part is not optimum.

    Reply
  92. You point out the main limitation with this process. It’s not a 3D printing”” process. It uses a laser to build up a part by fusing thin layers from a supply of metal powder. It might be suitable for a few limited applications. But as noted”””” the metallurgical quality of the finished part is not optimum.”””

    Reply
  93. Look liar, you’re simply FoS. There are roles where the microstructure of sintered parts is superior to cast or forged, because it is produced without aging or work hardening (requiring relieving). It is it’s own thing, not a variety of casting.

    Reply
  94. Sheepishly (particularly hard for “the goat”), AS A SCIFI avid first-edition-signed collector, I have to admit never having read or collected the Berserker series… apparently to my loss. Thanks. GoatGuy

    Reply
  95. Look Matteo, there are numerous applications that are simply unsuited for laser fused (i.e. cast) materials – that’s just mechanical engineering 101 stuff. It’s not going to change just because the process that cast the material is different. If 3D printed items are replacing cast parts, then that is about the limit of the tech.

    Reply
  96. So, we are now replacing single crystal directionally solidified single crystal first stage turbine airfoils with 3D printed ones now, eh? Evidently Siemans has successfully tested such a thing – no mention of how long it lasted though – or what stage, but still remarkable. Mea culpa.

    Reply
  97. “When I think of Von Neumann self-replicators, I tend to think of them as “whole cities” ”
    Yes
    I remember in one of the ‘Berserker’ stories that point was made & the plot involved the Berserker machines setting up such a city in a fresh asteroid belt to make & repair the war machines, & human efforts to wreck the ‘city’.

    Reply
  98. ⊕1… Emoji (smiley face).

    I’m neither overtly impressed nor pessimistic. I think HP may well have hit a sweet spot of utility for printed-sintered-metal-parts.

    There is plenty of applications for 93% dense metal parts.
    Mmmm… like sports car shifter knobs (Goat flees from fusilade of tomatoes!)

    No actually I have some modest experience with sintered powder molding. Granted it was back when dinosaurs walked the Berkeley streets, but still … sintering is sintering.

    We found that while sintered parts are nowhere near as strong in tensile strength and softer in Young’s modulus compared to bulk-metal, they do have some pretty interesting and useful engineering properties themselves. They can be remarkably tough: there is no voidless matrix to support crack propagation. Bang on them with a hammer, and they tend just to dent, not crack apart. They also tend to be “good” with repetitive stress deformation — repeatedly flex a sintered part, and it tends to immediately work-harden then stabilize. Of course if your strains are too great, they fail. At lower stresses than solid-metal parts, for sure.

    Thing is, that there are a lot of parts that are better in metal, but don’t utilize even a small fraction of solid metal’s tensile strength and Young’s modulus stiffness potential. Complicated stuff that goes into automobile passenger seats. Stuff that goes into the making of auto doors, latches. Sure… conventional manufacture definitely delivers good, cheap, durable, reasonably precise stuff for the assembly line by the tens-of-thousads without a hitch. But for shorter runs, the TOOLING needed to gin out these parts can be an excessively high overhead, ultimately suppressing innovation.

    So yah. ⊕1. I see a spot for these HP parts-printers. Not everywhere. But “a place”, right on part with CNC machines and flatbed waterjet cutters. Or plasma cutters.

    Just saying,
    GoatGuy

    Reply
  99. ” 3D printing or additive manufacturing yields chrap parts with poor fatigue properties poor crystal structure etc. Cast properties at best. It’s fundamental. ” <-- Except 3D printed turbine blades do fine, so as usual you are FoS.

    Reply
  100. When you need only 2,500 of them, is when. Thing is tho’, I see this tech competing square-on with modern 5 axis CNC machines. At about the same per-unit cost. A 5 axis CNC machine could mill that pretty shifter knob out of hex-stock aluminum (or bronze for that matter!) in a couple of minutes. Including the shaft threading, and these days, even the finish spit-and-polish. Given the usual automated bar-stock feeding, programmed for step-and-repeat until stock runs out, said CNC would make 20 an hour, about 500 a day. 2,500 of them in just a standard work week.

    By comparison, by my calculations of 10,000,000 voxels/sec, each shifter knob is what, about 40 mm x 40 mm x 60 mm “as a block”? → 96,000 mm³ per, 768,000,000 voxels ea, or perhaps ±75 sec … 1¼ minute a knob.

    THIS WOULD BE FASTER than CNC. But unless your looking for that curious pixelated surface roughness, and you don’t need to post-machine the knob’s spiral threading for tight shaft match, then there is likely more steps involved. Unpacking from the white “salt” that holds the printed knobs in place. Post-machining. All that. I’m betting comparable 3 minutes a knob, or 20 an hour, by the time the paint dries.

    So, very, very similar timing.

    Thing is (which shows I’m an advocate, not a hopelessly hard-headed goat), one can make parts substantially more complicated than any CNC machine might be conjured to perform. Parts with loose-things inside. Parts with fantastic geometries and inside convex curvatures that defy machining. Parts down to the millimeter scale (CNC machines really don’t like that scale), or long thin AND complicated parts more akin to wires than blocks. CNC can’t handle that.

    IT HAS A PLACE, without a scintilla of doubt.
    Just not for sports car shifter knobs per se.

    (Thing is, I get it: from a marketing perspective, what better to hand a prospective client than a shifter knob mounted on a cool hardwood plate to put on the client’s desk or wall-of-sports-trophies? Shifter knobs embody What Men With Bâhlls want in life. Trophies.)

    Just saying,
    GoatGuy

    Reply
  101. Outstanding find.
    Thank you.

    1200 dpi is actually almost 50 µm a dot. Hence why layers are either 1 dot or 2 dots thick. Makes sense. I can imagine that there are many parts where 50 × 50 × 100 um is just fine for voxels. Like sports car gear shifter knobs. (Did you notice how rough the shifter-knob was optically? No part of it was mirror-smooth. I’d expect such from 50³ voxels.)

    Using “my HP printer” as a likely scale-sizing estimator, it supposedly has 1170 nozzles just for the black ink. And in its one-swipe-high-res mode, it outputs a sheet of 8.5 × 11 paper every 10 seconds. (It is MUCH faster in “draft” mode, and quite a bit slower in “photo-quality” multipass output though.) It also prints 1200 × 1200 dots per square inch.

    So, assuming that HP gins up the ol’ print inkjet thing to where 3 or 4 thousand nozzles are squirting their stuff out with radical precision, yet giving up little in speed, then I’d imagine the printing is multipass, but still more or less within a (higher) order-of-magnitude close to my desktop printer. 430 × 320 mm per layer. 50 µm thick. 3,000 printhead nozzles. 5× the dot-lay per nozzle speed over my printer.

    Mine: 54,000 mm² ÷ 10 sec → 5,400 (about) mm²/sec. Dots are 50² µm² ea (1200 dpi).
    5,400 × (1 ÷ 0.05² µm²) → 2,200,000 pixels/sec.
    2,200,000 ÷ 1,170 nozzles → 1,800 dot/sec per nozzle.

    Theirs (projection):
    3,000 nozzles × 1,800 dot/s × 5 (better thru-put guess) → 28,000,000 voxel/s
    28,000,000 × 0.05³ mm³/voxel → 3,400 mm³ per second.
    430 mm x 320 mm ÷ 0.05² → 55,000,000 dots/layer
    55,000,000 ÷ 28,000,000 → 2 second a layer.

    200 mm thick ÷ 0.05 → 4,000 layers
    4,000 × 2 sec/layer → 8,000 sec/full scan. → 2.2 hours a block.

    That’d pretty good, if it were true!
    If you note the graphic of the printers, the closest (left) one shows 4+ hours.
    Seems like they’d not choose the worst-case to show…
    So the printing is probably slower than that, for a “full block”.

    So my nozzle-count estimate, or the 5× performance-over-my-inkjet guesses are off by a bit. Maybe down (all in) a factor of 3 or 4?

    Could be.

    2.2 hr • mean(3, 4) → 7.8 hr, max res, all in.

    Yah, I’d bet on that. OVERNIGHT runs. Full-shift runs. That kind of thing.

    GoatGuy

    Reply
  102. ⊕1: brief and to the point. Sophisticated musing…

    I think THE ‘problem’ with Von Neumann casted machine actors (so far) is that none have come even remotely close to building “the brains”, the “nervous system” and the critically important sensing metrology to (ideally) build replicas that have such significant productive capacity that ‘building more’ is only a wee fraction of their hourly enterprise.

    For instance, chipmaking. Here we are, 60+ years since the invention of the transistor (74 years, 1947, Bell Labs, Murray Hill NJ… and 59 years for Integrated Circuit // Noyce) and today it takes dozens of different purposed, fabulously precise equipment, scores of exotic materials, and an invisible “feeder supply chain” that is just as remarkable, if in many ways more-so. My point?

    When I think of Von Neumann self-replicators, I tend to think of them as “whole cities”: like the Silicon Valley, hundreds of key manufacturing centers effecting the supply chain of “stuff” that allows chips to pop out the far, far end. E.g. enterprise specialized in making wire, for them others specialized in making wire-drawing dies, yet others in solvent borne shellac insulation; yet another industry of copper mining, another of its smelting, yet another in its electrolytic purification, and yet-yet another in turning blocks of it into giant spools of wire-drawer’s feedstock.

    And that’s to make the magnet wire in a hundred grades, to make just the WINDINGS for the motors which power just about every aspect of the Von Neumann collective. Yet others making the iron, more reälloying the pig into magnetically hard steel, roll stock, lamination punchers, motor assembly fabbers. Very likely independent specialists in fabbing all the itty-bits: bearings (oh, the ball bearing specialists!), housing casters, … and the METROLOGY behind keeping these plants accurately working.

    Never mind… I could go on endlessly in this chain. But the metrology behind it is what I feel is the vital thing that no Von Neumann approximate has achieved. Those itty-bitty (or huge) sensors of position, temperature, acceleration, velocity, mass, pH, angle, concentration, specific gravity, ambient conditions, viscosity, surface roughness, hardness, tensile strength, elasticity, porosity, resistance, capacitance, inductance, and ALL THE REST.

    I can’t even imagine a Von Neumann collect that’d be notably deficient in all these, and at LEAST N² more contributing technologies, needed to make, part by part, out of substantially more specialized materials than “inkjet printing” seems remotely able to muster.

    So yah… ⊕1 especially to “locally sourceable material MAGICALLY on tap”.
    Especially for that.

    GoatGuy

    Reply
  103. Here’s the information that should have been in the article (from h20195(dot)www2(dot)hp(dot)com/V2/getpdf.aspx/4AA7-3333ENW.pdf ).

    • Multiple parts produced at the same time, or large parts, in a powder bed 430 x 320 x 200 mm (16.9 x 12.6 x 7.9 in).
    • Parts can be arranged freely in multiple levels in the powder bed to optimize packing density, productivity, and cost.
    • No build plate required, compared with selective laser melting (SLM).
    • Low-cost, high-quality final parts for serial production up to 100,000 parts.
    4
    • Best-in-class price-productivity.
    3,4
    • 1200 x 1200 dpi addressability in a layer 50 to 100 microns thick.
    • Finished parts with isotropic properties that meet or exceed ASTM and MPIF Standards.5
    • High reusability of materials can reduce materials cost and waste without compromising part quality.
    6
    • Density after sintering > 93%, similar to MIM.

    Reply
  104. OK… so… Not ONE single specification related to actual throughput or performance. Nothing about powdered-metal “inks” costs. Nothing about “voxel size / volume”. Nothing technical beyond the ‘printable’ volume itself. 400 mm (long) x 350 mm (wide) x 200 mm (high/thick?) But WITHOUT specifics … then things like 4× printheads, and 2× this and 50× greater throughput… Are kind of mendacioius. Disingenuous. I would imagine that the tech is quite speedy though. After all, HP has made quite a reputation for fast inkjet process. And a low-gum sinterable metal is a good thing. Just would be nice to know how quick it it might be. … I followed the linkie to the HP website. It was no different… nothing specific. And — pet peeve — the only photo of a part on the main website page? An 8 speed manual transmission shifter knob. Its like advertising motor oil… With pictures of sexy-curvy Mazerati cars. Or tooth whitening toothpaste… With perfectly photoshopped greco-asian female headshots… Just saying. A $400,000 metal printer for … shifter knobs? … There’s another part of me that likes the idea. Shifter knobs. Small decorative parts that are hard to mold. From old’ fashioned pot metal or newer magnesium-aluminum alloys. Maybe if the [i]“per knob cost”[/i] is low enough, once the machines are going 24 hours a day and demonstrate near-zero operational shutdown problems in production, well … setting up a line of 20, 30 … 100 machines, working in parallel, one might be able to make quite a bit of “output stuff”. Good for HP. GoatGuy

    Reply
  105. OK… so…Not ONE single specification related to actual throughput or performance. Nothing about powdered-metal inks”” costs.Nothing about “”””voxel size / volume””””. Nothing technical beyond the ‘printable’ volume itself. 400 mm (long) x 350 mm (wide) x 200 mm (high/thick?)But WITHOUT specifics … then things like4× printheads”” and 2× this and 50× greater throughput…Are kind of mendacioius. Disingenuous. I would imagine that the tech is quite speedy though.After all HP has made quite a reputation for fast inkjet process.And a low-gum sinterable metal is a good thing. Just would be nice to know how quick it it might be.…I followed the linkie to the HP website.It was no different… nothing specific. And — pet peeve — the only photo of a part on the main website page?An 8 speed manual transmission shifter knob. Its like advertising motor oil…With pictures of sexy-curvy Mazerati cars.Or tooth whitening toothpaste…With perfectly photoshopped greco-asian female headshots…Just saying.A $400000 metal printer for … shifter knobs?…There’s another part of me that likes the idea.Shifter knobs. Small decorative parts that are hard to mold. From old’ fashioned pot metal or newer magnesium-aluminum alloys.Maybe if the [i]“per knob cost”[/i] is low enough once the machines are going 24 hours a day and demonstrate near-zero operational shutdown problems in production well … setting up a line of 20 30 … 100 machines working in parallel”” one might be able to make quite a bit of “”””output stuff””””. Good for HP.GoatGuy”””””””

    Reply
  106. Of course if you are laying it down atom by atom, then there is a slight chance you can build up the various carbide inclusions and other intermetalic microstructures that give the high performance to any metal stronger than a simple pure metal.

    Good luck trying to fabricate the strained and distorted microstructures like work hardening or martensite steel.

    Reply
  107. Yes, and this “main limitation” is so profound that it severely curtails the usefulness of the entire field of manufacture. In 2005 I saw jet engine combustor “swirl cups” printed in stainless steel/binder which were wicked solid with braze material in a second operation. They were good enough to test the geometry and make relative comparisons between designs.

    The next step might be micro-peening of the layers as they are “laid-down”, but that product could only hope to match cast material properties.

    Now, physical vapor deposition of metal atoms, analogous to how pyrolytic graphite is made, might be the alien technology everybody wants to use to build their O’Neill cylinders. Nickel laid down by physical vapor deposition with zero porosity and extremely fine grain size – now that might be something we can work with.

    Reply
  108. You point out the main limitation with this process. It’s not a 3D “printing” process. It uses a laser to build up a part by fusing thin layers from a supply of metal powder. It might be suitable for a few limited applications. But as noted, the metallurgical quality of the finished part is not optimum.

    Reply
  109. OK… so…

    Not ONE single specification related to actual throughput or performance.
    Nothing about powdered-metal “inks” costs.
    Nothing about “voxel size / volume”.
    Nothing technical beyond the ‘printable’ volume itself.
    400 mm (long) x 350 mm (wide) x 200 mm (high/thick?)

    But WITHOUT specifics … then things like
    4× printheads, and 2× this and 50× greater throughput…
    Are kind of mendacioius.
    Disingenuous.

    I would imagine that the tech is quite speedy though.
    After all, HP has made quite a reputation for fast inkjet process.
    And a low-gum sinterable metal is a good thing.

    Just would be nice to know how quick it it might be.

    I followed the linkie to the HP website.
    It was no different… nothing specific.

    And — pet peeve — the only photo of a part on the main website page?
    An 8 speed manual transmission shifter knob.

    Its like advertising motor oil…
    With pictures of sexy-curvy Mazerati cars.

    Or tooth whitening toothpaste…
    With perfectly photoshopped greco-asian female headshots…

    Just saying.
    A $400,000 metal printer for … shifter knobs?

    There’s another part of me that likes the idea.
    Shifter knobs. Small decorative parts that are hard to mold.
    From old’ fashioned pot metal or newer magnesium-aluminum alloys.

    Maybe if the [i]“per knob cost”[/i] is low enough, once the machines are going 24 hours a day and demonstrate near-zero operational shutdown problems in production, well … setting up a line of 20, 30 … 100 machines, working in parallel, one might be able to make quite a bit of “output stuff”.

    Good for HP.
    GoatGuy

    Reply

Leave a Comment