Bản chép lời đầy đủ của S2 E5: Rapid Prototyping Deep Dive with Greg Paulsen

with? And I, how I guess, how can people avoid this, and how can they think about more about the design requirements rather than the manufacturing method and so on and so forth? If you have examples also of people doing this and not failing, but just kind of, narrowing their design space, I'd love to hear more. And so many examples come in my head with this, like, when I think about the when I think about a design where this. This person is designing something, and they had this idea, and with the power of accessible CAD, accessible 3D modeling, and 3D printing in house, so having a desktop 3D printer, usually something filament based, they're making this into fruition. From idea to reality. And sometimes these folks don't have formal training in other technologies, so they become really good at designing for desktop FDM, which is fantastic for low volume iterative design, especially if you have it in house. It's accessible. You're getting an idea and you're printing it and working with the next day. And I can already tell you that whenever I work with a client who already has A little bit of manufacturing, like just even a 3d printer, like they're better to work with because it can start understanding principles saying Hey, it turns out gravity does exist outside of a CAD space. Things can break, and they start to have, get this empathy for manufactured product, but at the same time, they may be designing with constraints for a process that won't necessarily scale to where they want to be. When I talk with a customer, I usually try to get some time, the timelines in place and where they want to be. So I may start talking to you and you may just be describing where you want to be in 6 years. I want this in every household. I want, I want it to be like this. And then, I may bring you back to reality check. Like, where do we need to be right now? Are you about to, debut this to your first 50 customers? wHat does your next six weeks look like? And so I usually think about like this, like I always say six weeks, six months, six years, because it kind of helps differentiate where you are in your product development cycle. So are you going outside this R& D concept phase and now you're trying to get into something that. Would have some trouble scaling and manufacturing, but I just need to get it out. And then within this budget in 6 months that may be I'm moving to more of a harder tool to method. So something like molding, injection molding, for example, and, your design will start to go that direction where you have to think more about uniformity of walls where undercuts that could cause side actions, which, increased tooling costs and drafting goals, which I think if you haven't Worked in a world where draft angles are required. All of a sudden become a headache. If you're not designing them into your cat or being conscious of them like in your previous design cycles and then that six years is sustained production where it's here's where I need to be figured out the manufacturing processes and work cells. In order to make that possible. So it's a little bit different than just like here's design make parts. There's entire fulfillment supply chains, delivery dates. There may be some turnkey packaging and other things involved with that. That's in a different lifespan of itself there. But yeah, figuring out where you are in your design cycle, what your goals are for right now. Am I doing a functional test? Do I have a connector that needs to hit a certain weathering IP68 or something like that? And. And I just do. I just need to test that connector or do I just need to test that whole part? What processes can help me get to that next stage? Do I need to go through? If I'm making a medical device, do I need to start thinking about FDA, what things are what materials, what processes are going to help me get through those initial approval stages and condense my lead time to where I can make a part that actually could go to my client without being, research or prototype. That's where we kind of can help consult. Okay so to kind of, center in around the rapid prototyping and kind of the transitionary phases of manufacturing, it would be, I think, really helpful for the listeners to know what methods exist out there because I think there are some very obvious ones and maybe not so obvious ones. If you were to kind of go in your mental catalog and say for these types of industries or these types of product applications, we see rabid prototyping methods of so and then kind of transition from there. I'll keep asking questions as you go. Oh, no, absolutely. And so I've done a lot of webinars on different topics and different, when to choose a topic. So choosing between additive or machining, but let's stick with additive, which is, 3D printing because we have a gamut of technologies there. One of my webinars that I have not done, but it's a topic that I want to touch on is like, when you shouldn't choose SLS or MGF. All right. So these are kind of bread and butter commodity industrial additive manufacturing technologies. Selective laser cinch. Yeah, for people that might not have heard the acronyms. Absolutely. So a lot of people who are used to 3d printing are used to the filament base. So you have a filament and it melts and extrudes and zigzags back and forth to create your part from bottom to top with powder bed fusion technologies. You and both SLS and MGF kind of fall in that line, they have different ways to fuse the plastic, but they're doing very similar behaviors where you have a layer of powder, almost feels like flour, and this powder is actually plastic, it's usually nylon. That layer by layer about the thickness of a sheet of paper goes across, and then with selective laser sintering, a laser will etch that cross section. Of your parts. If you imagine a part split into thousands of cross sections, it's going to etch that cross section and that laser is going to melt

that cross section as well as fuse it to the layer underneath. Give me that Z direction. That 3rd dimension has 3D printing. And then grow those parts from bottom to top as multi jet fusion fuses a little bit different. It has a fusion agent and a detailing agent. So it actually deposits a kind of this darkened agent across the build chamber on this cross sections. And then a single heat source goes across and creates a melt state. So they have a little bit different technologies, but they're both doing similar things layer by layer melting powder. What's important about this. Is that wherever there's not a part, there's still powder there. Go to And so I'm able to suspend parts in a three dimensional space like they're floating when you look at traditional 3D printing like filament, it needs to be, secured to a build plate. You always think about your X Y platform and you have supporting structures that are growing to take overhang features because it turns out again. Gravity does exist in manufacturing. And if you don't have something supporting an overhang, the overhang just flops down and you get it like a naughty mess with the print. With selective laser sentry and multi jet fusion, these powder beds, if you imagine me taking a golf ball and just kind of sticking into flower of deep down and letting go the part neither sinks nor floats. It's just stays there. If you imagine a 3 dimensional space an SLS about 13 by 13 by 23 inches, I can take different geometries, complex, housings next to rotors next to clips next to everything else and basically nest them computationally. About two millimeters away from each other, And that allows me to build at any given time about 30 to 300 parts per machine per night. And with that, it means that for you, the price range, even though this is a very expensive technology and SLS machine is about. 470, 000, it's, these are, this is why it's industrial, right? An industrial machine is, they're not cheap. But when you look at it at scale, like when you're running fully nested builds for you as a consumer, you're essentially just renting that space, that volume that your part's taking up. And so you may find it's actually often more economical than industrial FDM sometimes, because you have that three dimensional nesting. And what I love about it too, is you have more isotropic results. So you have parts have similar mechanical properties, regardless of orientation. So if you think, if you compare that to your filament base, where things, where parts are kind of sandwiched on layer by layer, they're weak at the next layer. So between those layers, they have a weak spot. So if I build a pencil horizontally, FDM. It may have better strength, but it'll be look really course. But if I grew a vertical where it's a bunch of little circles building from the bottom to top, it'll probably break faster than a pencil would break, if you're putting pressure against it. So you don't have a mechanical advantage there. And again, and also nylon itself is kind of this commodity material that is in these powder bed processes, and they and it is flexible when you design it thinner. When you design ribs and stiffening features, it acts stiff. It's a very mechanically sound material and just naturally tough. So even though it is can be your lowest cost industrial application for 3D printing. It still doesn't it doesn't mean it's bad. It just means it's commoditized It's kind of like in cnc machine 6061 is like the bread and butter millable material in the united states. It doesn't mean that's bad It means actually it's pretty dang good and it's so good that it's used in a lot of industries And so it's cheaper, to do and a lot of people know how to mill it so 85 90 percent of the time Those process, the SLS and MJF processes actually work really well for your application. You can even do secondary smoothing like chemical vapor smoothing to move it from a mat to sugar cube into more of a semi gloss finish and a sealed surface to it. So you could do some extra things to it. And then we also do have FDM, because FDM, what, when is it good to have support structures? Larger, bulkier parts where they can thermally warp and stuff, picture frame style, designs. And FDM can build up to 36 inch parts, so we can, do larger parts with that, and that's that filament base. And then a lot of the other technologies are resin style, so they're, that's when your base material is actually a liquid. And that liquid gets cured with U. V. And that's stereography. We have next a three D. L. S. P. C. We have carbon digital light synthesis. D. L. S. And PolyJet, which is kind of, it's a resin base, it's almost like an inkjet printer, but with three dimensions, so most of the time it's used for multi color printing or multi material printing. In, so you can add digital rubbers next to kind of these acrylic place acrylic based rigid materials. But those, the liquid based materials, so there's resin 3D printing. Processes have acts like materials so polycarbonate light, which means stiffer or abs which is kind of a blend between stiffness and some flexibility. Or polypropylene which is just a little bit more flexible. They're never quite the same properties as that polypropylene or that. That that polycarbonate what we give with those other processes, but I do use those for. Smoother surface finishes from the initial print. As well as typically better mechanical tolerances. So if you do need a higher detail resolution, resin really fits the bill there. Yeah, definitely. And I've had experience, I mean, with resin, obviously, I've I've done prints on Formlabs printers which have been really good in terms of in terms of the tolerances that you're looking for. I've also worked with Carbon Printer the Carbon Printer. It's really fun. I mean, you can you can print lattices really well, specifically in rubbery material, which is super interesting. Yeah. I finally gave in and

bought the, some Adidas shoes with the carbon DLS, like mid sole. So I was like, I should do this. I'm in the industry. So yeah, I'll actually be going out some events later this year, sporting those shoes. So I'm kind of excited to do that. Oh, but yeah, for example, yeah. DLS is a resin based printer that. It kind of turns it upside down, right? You're building, you're growing on a build plate, but you're kind of pulling the part vertically out of a resin bath, and it's is so interesting because it, the design space changes so much because it's going backwards. We've had several failed prints where it just kind of ripped at the middle stage because it was either too tall or the support structure wasn't there. But yeah. Fascinating. No you're on the ball. So stereolithography is like is a resin bath where the parts are being lowered into the bath and a laser's hitting the top of the resin and curing it to the layer underneath. What is cool about that in stereography is I could typically do a higher mix of designs. So in one large build, I can do, 20 different designs and of different sizes. So SLA is it's also about, I don't know, 40 years old in the in the additive tech space. So it's very well researched and it is very consistent. I think build success rate on industrial SLA is like 97%. I mean, it's pretty much once it's programmed, you have a decent idea that it's going to be successful when it's done printing. success is great. I, personally, it's just so messy to deal with the resin, the curing, the post process but it, I agree that it has the least amount of just squiggly mess that happens with more filament based printing. But yeah, you, when you were mentioning DLS though, or those, the, even an XLSPC, those are upside down. And actually, when you're pulling a part out, you're absolutely right, by the way, gravity can Yeah, problem typically requires thicker support structure. Even on the part itself, you may see more noticeable like bumps from where it's supported and parts like lattices or designs that are cylindrical or designs that are kind of self supporting tend to be more favorable in that design, then essentially throwing any design at it because because, yeah, if with a high mix world, You may be running about a 60 percent success on first print. Now they have the advantage of actually being fast, like super fast. So even if you get a fail, it's just reset and start, go start going again. And then once you have the support and the strategy up, it it tends to print, very very robust, very reliable reliably but you have a. You do sometimes need to kind of reorient or adjust your supporting structures because you're pulling a part out and I keep on saying this, but it's so true in a cat environment. Your price is floating in the middle there. But once you apply gravity to anything it's a challenge and I, I actually, I'll give, I'll tell everybody this in design for 3D printing. First off, fillets are your best friend Fillets, reduce acute breakpoints across your part. And also when you are designing. Features on your parts. Think about what's going to happen if you hold your part by that feature. I one of my, one of the things I've seen, and it breaks my heart every time, is a person will design something pretty big. Like a large part, which usually can be expensive to, to build. And, hundreds of dollars, or even low thousands of dollars sometimes for larger parts. And then they'll build a pin in it. And. They'll design it like a little pin and that pin depending on where it is, can be really fragile or even, get crushed by the gravity of the part. I've a hundred percent seen that before, but I've also seen where they get this part, the part's perfect. They take it out of the box, instantly break that feature, that thin feature and their parts scrapped. Or it requires like drilling and reworking those parts, and you have this, very expensive piece that now you are now you're having to kind of scrap or remake or glue and epoxy, and I call those god pins, those features. I've seen rocket little model rocket ships where they have antenna on top, and when fall in the antenna breaks off and like the cosmetics of the part is gone. I'm like, that's where you sometimes when you're designing for additive, you may want to look at those features and say, can I turn that into a whole with a press fit pin, like use off the shelf components instead because I could buy, 25 steel pens for four bucks, and, commodities. oh, please. it. I mean, it's the same reason why we don't design in threads. We usually recommend for to use press fit inserts into, yeah because you can tap like different materials like SLS and MJF tap like butter, and we could do those services for you press like brass screw to expand inserts are very good for certain technologies in the put it, heat it, insert. It's easy, right? Yeah, thermoplastic. He's a heat stake insert, and we do that service like on our on zombies website. You could just click inserts. Put how many you're in and you need and attach a drawing that easy. 📍 Very interesting. This podcast is presented to you by Pratik, a startup advising and coaching company that is geared to help hardware entrepreneurs get their ideas from a napkin sketch into a lab and out into the world. Um, okay, with that. Podcast Break So this is, I think, a really natural point in the podcast to do a podcast break and talk about hardware horror stories. I'm sure that you've seen a decent amount of those. So yeah, if you could talk to kind of, when either a hardware startup got it wrong and that set them up for issues. Whether it's within their, I guess, manufacturing strategy and maybe they either understand undersized their strategy where they're doing too much of rapid prototyping and not transitioning over quick enough or something along those lines. I think listeners definitely benefit from hearing the horror stories so that they don't

have to experience it themselves. Yeah, absolutely. I'll draw from actually some of my early years in product development, even pre Xometry where we were serving clients of all different types, and even including, very professional large entities where certain things where they transition to production. They, knowing what I know now I wish I knew that when I was, like, in my second year of career, and raised my hand and would have said stop right now. I have a speaking earlier, but on a much larger scale, designing for a 3D printing process, and then moving. To a injection molding. I remember working on an optics device where the initial goal of the optics device was to build using fused deposition modeling Ultem, which is Ultem's a very high strength great durable material. Their original design, because it's designed for 3D printing, didn't have draft angles, didn't have any of these things that you need for molding, , it's kind of bulky, and again they had the Ultima as their spec, but they realized that this housing needed to hold atmosphere pressure. A lot of times in in kind of, military projects or different devices, they actually have, are slightly pressurized, not overpressure or anything, but slightly pressurized in an environment inside to keep the electronics safe in a bay. With FDM, FDM is inherently porous. Like those layer by layers, you have micro gaps, so you can never really seal it, or if you're if you are trying to seal it, it's not really production viable at that point. Let's move. So they're like, let's move the molding, but they didn't think now that we have the freedom of molding, what materials can be available for me? Because in 3D printing, you may have a dozen materials. In injection molding, you have, I don't know, 30, 000, you have a lot of choice, going through, and that's when you usually could, even if you're new to this, you could benchmark saying, hey, I like that camera housing, what's it made out of, and it could be, PCABS or a glass film nylon, and this part probably could have been glass film nylon, and they moved to Ultim. So comparing Ulti pricing is about $55 kilogram or higher, compared to most materials are about a dollar, a dollar pound, or sorry, pound a dollar pound or $55 pound for these premium materials. So the parts became kind of prohibitively expensive. And also the design was not quite optimized because it was just basically a drag and drop into an industrial manufacturing process. So the design itself, created expensive tooling. It created a, it created a challenge on the material costs. And even these materials sometimes have requires have more special requirements. So it just became. A much more difficult project than it really needed to be because the because this 1 to 1 transition going from like this 3D space. And now that's our changing processes, and this is my why I asked, 6 months, 6 years, like, where do you want to be? Like, I think we could have stepped back and

said raise your hand and said, here's some design changes we want to do now. And here's the material substitutions and take that extra, 2, 3 weeks of rapid prototyping and iterative design. Working on that in order to create a much, much more successful outcome. And then so I've definitely seen that transition happen. And this is this was not a cheap project. This was, high budget. Requirements, but even, veterans in the industry can make these mistakes. I think the other thing that I've found too, and this comes down to when you are transitioning to a, different technologies is understanding the difference between checking off a project, benchmark and, successful outcomes when you're building an integrated device, you may be creating enclosure or housings around a PCB design, and if something changes there. And you've already released tooling this is where Emmys and ease really need to be best friends with each other and be over communicating, but if you have a design requirement change. And the mechanical components are released to production. Like the CEE finds they need extra capacitors, or something within their design, or a different junction, or they need to move a feature, you may either need to scrap that tool, Or jerry rig internally, like lots of Kapton tape possibly doing a 3D print a 3D print little frame to help move the PCB around, or add extra components to it. Or even post machine your production unit and that's what just my word of caution, but I've seen it way too many times, actually, where it's oh, whoops. And then you have to go and bring all your, all your low cost injection mold components or your CNC milled parts and bring them back to the mill and, modify at best. For that, or you do a tooling rework, which can be, costly and time consuming to do. But that's something where, especially in hardware. Everything we're designing is usually an enclosure or some sort of packaging around a, electronics and with supply chain shortages, hardware going out of stock or obsolete constantly, you may find that that's like your design changes much more than you expect your tool life cycle to last and just being aware of that, talking with your team, identifying the risks in your bill of material, And even compensating with CAD kind of understanding where you're going to be like, is this design going to last through multiple reps, multiple, multiple generations of product, or is this going to be one off, it'll change your manufacturing approach or tooling strategy, you may have other. You may find that it may be best to move forward with 3D printing, for example, like until like things solidify and you kind of are building work cells around it. Because you're just going to be changing tools way too often, or you may be adding extra space like you're talking your ME, buying them the favorite six pack of beer, make sure they give you some extra room as a EE, so you're designing so you're designing things that can work together for future generations. I mean, that's really sound advice. I think that happens all the time, specifically with PCB designs nowadays due to the supply chain shortages. And just generally, there's also a lot of internal discussion. Sometimes that happens later on where it's is it a firmware problem? Is it a electrical EE problem? Or is it a mechie problem? And I think those conversations happening before tooling happens is probably the best approach there. I guess I actually had a question and kind of I guess like prying a horror story if you've ever seen any, you were talking about this kind of powder bed printing process and you said that the thing about it that's really cool is that it can accommodate overhangs and stuff that doesn't have support but it can print it because of its inherent structure. Have you ever seen and this is again referring back to the manufacturing method bias that we talked about. Have you ever seen people design it to that powder bed, but then it can't scale? Or is there, are there methods from that powder bed design that can scale? Because that's not really obvious to me, so I'm curious. Yeah, and actually, yeah, just kind of abstracting 3D printing scale, scalability has to do with how many can I fit in that build, and then how many machines can run that build parallel. So it's very different than, formative production where it's closed cycle time, open part falls out, closed cycle time, open part falls out. With additive manufacturing, you're kind of doing smart batches. So usually the larger the part is, the less scalable it may be moving to a production stage. But you're to your point to the more complexity that you add within that design, SLS is really fun. I could do organic shapes. I could do off angle features. I could add a grid and lattice inside here, but it may lock me down into that process. And this is where, in your design stage, sometimes we have this idea of CAIV, which is cost as an independent variable. What are you trying to sell it and sell this to your customers for? And, if you're finding out that, Hey, this SLS part cost me this much, and you hope that it could scale down. To this level, we can help work with you and actually design what a work cell looks like, production work cell, but ultimately we have a cost for these builds and you can only fit so many in. So unlike injection morning where the tooling cost just keeps on monetizing down every single time we do a cycle, like some of that tooling cost is absorbed in that part. And when I move to thousands, 10, 000, 100, 000, all of a sudden tooling cost is negligible. And it, you kind of cap out in additive manufacturing, and it just depends on how much value are you designing into that part. It's, because it definitely, you do, it definitely do have break evens, but if you find that I'm trying to get to this cost, you may need to take a few steps backwards, look at that single part, and say, doesn't need to be two parts.

Doesn't need to be three parts, and each of those three parts can be injection molded and figure out what that next stage is. That makes a lot of sense. Yeah, that's that I guess that's what I was trying to get at because I do think that oftentimes I can see engineers getting really excited about rapid prototyping process and using it to the utmost degree. And then now they have a design that absolutely can't be manufactured in a different setting. So it's it can be manufactured in low quantities, but cannot scale. So having, I guess the takeaway here is that if you're using rapid prototype manufacturing method, it's really important to understand what it's unlocking and what potential future higher volume ones won't unlock. So really understanding the design constraints, what it unlocks, what materials you can use and all of that, like just general. Futures of what a rapid prototyping Applicate rapid prototyping manufacturing technique provides you and then understand that is scalable to a degree And maybe talking to experts to say how much how many? Maximum can I build with this and then scoping your design according to that. And then once you move to that kind of higher scale thing, okay we need to reevaluate our design, maybe get rid of some features, maybe split it up. What what materials am I going to use? Kind of ask those questions again, because I feel like sometimes there's this narrow vision. We need to ask what materials we're going to have, what design features we're going to have early on, and then you just commit to it, but that isn't quite how that works, because in the early stages, you're doing technology development, so you have to answer these fundamental questions, and you need to be able to iterate quick, and that's why you use these rapid prototypes. Mechanisms, but then the question changes now. It's okay. The technology inherently works. How can I make this exist in the world? And so you need to ask yourself the question again of what is the material? What technique do I use? What features can I build out so that it's cheaper? And as you said, cost as a variable, because now cost is a variable. Whereas in the very early stages, it might have not been. Yeah, and you're hitting the nail on the head, for sure on this it is, and by the way, I am probably one of the most excited people out there for additive manufacturing and rapid prototyping technologies, because it does just bring access, and I've watched this industry, even, so because rapid prototyping additive manufacturing, 3D printing exists. Yeah. CAD got cheaper. Yeah. CAD got more accessible because now you have this, people are like, hey, I got accessible ways to manufacture, but STL, like they, they're they don't have CAD access. So then you're seeing these programs like Onshape, Autodesk Fusion 360 these lower costs programs free CAD coming in. And being accessible and giving professional ways to design and execute. And now you're getting access to manufacturability. And by the way, bringing it back to Xometry, that's kind of why we exist, too, is we're democratizing access to manufacturing. And we're using AI and machine learning along the way to, for pricing and scaling supply chain management. But it is, it's, it's 100 percent understanding kind of where you are and understanding that sometimes your choices are change. Going back to those hardware nightmares, sometimes it's, I've seen CDRs, like critical design reviews phases, where it's just basically project managers looking for them to show a PowerPoint and saying, okay, please proceed. But really a very important part of a critical design review is saying stop. Stop, assess, rethink, what are what it's treated like a pre Yeah, How can what are the five things that could go terrible if we move forward, and how can we mitigate that in the next few days in order to have a more successful outcome in the future? such a good call out because I do think that there is this over emphasis that hardware startups because of honestly, I maybe it's the investment culture. I mean, I could get into many reasons of why this is the case, but there's this over emphasis on speed, which I totally understand. And there are parts of it that need to be that you need to be speedy. But at the same time being able to have these natural stoppers because at the end of the day, if you go down the path of the wrong design, oh boy, is that going to cost you so much Absolutely. So I completely agree with that. I want to quickly talk about, because we didn't get into it in the earlier stages of the conversation, but I really also want to get your thoughts on metal printing, because I know that's kind of up and coming. I haven't seen a lot of startups use it, honestly. There are a few. It's just the barrier of entry seems a little bit high for a lot of people. One there isn't as much common knowledge around metal 3D printing. People would rather go to mills and the old kind of techniques. So I'd love to kind of get your thoughts on different types of metal printing and what they can be what they can enable in designs. Absolutely. This will go back to that theme of how much value am I putting in my design, and that's going to help you understand when and where to choose metal 3D printing. Metal 3D printing. So Xometry, we offer two of the most popular ways of achieving metal 3D print. And metal 3D printing is, it's metal. It is it is, it's not metal like it's real metal. We're making parts out of stainless steel, aluminum alloy and these these parts two major methods are direct metal laser sintering. Similar to that plastic powder bed you're putting a fine layer. Of metal powder down in an inert chamber. A laser is hitting that and it's essentially creating like a little microweld there, unlike that plastic powder bed metal misbehaves. Yeah. will stress and want to flex up and move all around. And in plastic powder beds, you keep the whole build

chamber hot in order for parts, not the plastic parts, not to kind of flex and warp up in metal, you're essentially centering them. You're melting them to a thick build plate and the entire build. Which requires support structures. Just think your FDM print, only now my support structures are not, let me use my fingers and remove. It's let me use an angle grinder and remove, right? It's, it is a, they're metal supports but they're, your parts is fighting against that build plate the entire run. Design rules for metal 3D printing, Are different than design rules for CNC machining, the less work my milling does. So the more bulk I keep in my part on CNC milling is the cheaper part is. Cause I have I have a bar of material and I'm removing features from them, removing material to make the features of my part. With additive, the more your parts on a diet, the less work it has to do in order to produce those features. So why would I choose additive manufactured metal products? I will, I'm usually having a design that will have similar, looks to what a metal injection molded product or a die cast product may have with the addition of me being able to add more organic geometry if I'd like. So generative designed features work really well with additive manufactured metal components or what I'm building features that have. Either inaccessible areas. So lattices is kind of like the typical example, but that could even be, hollows. It could be things where as long as I have a place to empty the uncentered powder I could have, some internal channels and chambers to lightweight a design. Or I could build these off angle features that would typically in CNC machining. Require 5th axis or multiple geometries and just add overhead time complexity to the build, which increases the cost of those parts. So there definitely is a break even curve with additive manufacture components. And typically, though, you are finding that most of the time in when you see industry examples, it's in high value, lightweighted designs. So that's why you see it in aerospace. It's because they have the budget to work and design out and build out these multi thousand dollar parts or 10, when you see the big stuff, I mean they're tens of thousands of dollars, if not more, to produce those parts. But the savings, the ROI on them are just insane because every pound saved on this is this many gallons of fuel per year and you and your weight is a true metric Yeah. with metals. Binder jet metals, which is another technology, has more scalability, and that's that is, again, a thin layer of powder, but it's an ink binder, I kind of think like a ink jet deposited glue is going down and holding the metal pieces together, layer by layer, so when I get my parts out, they're in kind of a green phase, and then we de bind those, which removes that polymerized binder, and that's a kind of a brown phase. And then you then you go to a sintering oven where that material goes to the very high heat. And essentially those, that powder beads will melt together. And if I caught sintering and it does have, it's about 97, 98 percent dense. So it is not, it's not like there's a little bubble in the middle of it. It just means that if you look at a cross section under microscope, you'll see tiny little channels of porosity on it. So it's not good for gas type applications, but it can make very durable Very strong and regularly complex metal parts, in typically a steel alloy Binder. That's nice because it doesn't require that support structure. So you have more volume with it, but it's usually is still catering to parts that are smaller than a golf ball. And when you think about production, so I could do larger parts and binders that I've done very cool, big auger screws and stuff with it. But when I'm thinking about single alloys typically part smaller than 4 inches are preferred. But in production, parts smaller than an inch cube is more production viable for lots of reasons, and it does require more tuning than direct metal sintering. So Biterjet processes because they have a furnace state there's some warping, deflection, shrinkage that happens. There's better software out there for that but something to understand in your product development life cycle is you may need to iterate a few times. Using that design, once you find a design that's tweaked in a way where it's successful, it's usually very repeatable, but you may be working on several weeks timeline. So it's very different than RP rapid prototype method this is my son being one of the rigs being right, like one does not simply metal 3D print. It just, it does require a little bit more, uh, unfortunately, a little bit more tribal knowledge. At this point, I think software will augment it the next decade. It'll just be part of our CAD program saying this part is going to be metal 3D print, optimize, click, and then all the CAD data is there and I pray for that day, but for right now, it is it still does have a lot of learned experience once you adopt it. Thank you. It may substitute from your need to die cast metal injection mold and working at working in those scalable phases because you can do a little bit more, with these technologies and get rid of that tooling and that lead time up the tooling. But there is there's definitely some it's definitely a higher barrier to entry higher cost option when you're jumping into it. That makes a lot of sense and It's super interesting to think through. It seems we talked about how traditional kind of plastic rapid prototyping methods have still kind of, a vast difference from what it means to scale in a different manufacturing method with potentially the same material. It seems like there's even a larger gap of difference between metal 3D printing and other methods to, Do higher quantity manufacturing for metal. And actually, as you were talking, I was thinking about one time I visited aerospace company there. They built rockets and

they showed me their industrial 3D printer. They said that they did a lot of nozzle printing. Because the geometry that they were trying to achieve were almost impossible before 3D metal 3D printing existed. And the machine was as big as my living room. It was, I think they said it was the largest in the world. So I guess metal 3D printing is At this point in time, more so reserved to potentially industrial application startups where they're building one of something or two of something and it needs to, the functionality is so important that you need to be able to extend your design space so much but you are very adamant on the metal material that you're using and traditional methods just don't get you there Absolutely. I think this is where we're seeing the initial adoption. That being said, I do, I strongly believe that metals are going to really take a lot more form and a lot more precedent with with accessibility in the next several years, I've been seeing more improvements, especially on metal binder jetting that where it's going to be more cost effective, as well as software, that software is going to make this much more accessible and less of a tedious experience. Here's the thing about plastics versus metal printing. If I have a plastic 3d print and I need to modify it, I have a drill. I have files. I have sandpaper. I can make it work. I could add proxy and then sand it down. Yeah, I could do a lot to modify my part and I do that. And I actually, part of my earlier career. I had a whole toolbox by me just to modify a part before we actually received it into receiving inspection in order to bring it up to our tolerances and spec. That's okay. Use reamers, whatever you need with metal. If you wait a couple weeks and get that metal part, and say you have a square slot for, a keyway, and the part doesn't fit, you're just out of luck. I mean, you need to machine, or you need to bring it to especially if it's a non round feature that can't just be drilled on a machine, like if, say, you have a star, a star spline or something like that. You have to either scrap the design or figure out broaching or something really clever to augment that. So those headaches when your design doesn't come out just the way you're hoping for with metal, it can be much more difficult to work with, especially if you can't have a good predictive result on what the outcome is. But I really have to emphasize that. You could still like Xometry, we instant quote metal 3D prints and you're able to see this pricing and say you do are working on design that does require the robustness of stainless steel and you're working on lower quantities of products. It may just make sense for you to choose this over traditional machining. I've seen even traditional companies that have used that are typically working in investment casting. And they've done studies where they're like, Hey, our annual need for this investment cast part. Once we've have our initial implementation. So think MRO. So maintenance repair work there. They're finding that replacing that production with additive manufactured metal parts for maintenance and repair. Is a cost saver cost and time saver because they could get that part, just as functional actually often higher, better properties and then investment cast in a, a sub two week timeline. And that allows them to very quickly respond when they need maintenance when their parts are down. So there's places where it is filling in, even in just the typical manufacturing world, right? The traditional, which is still the major part portion of manufacturing out there. aNd I'm just seeing more and more applications. But you're absolutely right. The adoption's a little bit slower. The costs are a little bit higher. And and meeting and bridging that, here's my design. What am I gonna get? Really requires software in the loop to bring us to the most success. yeah I agree. And I think that specifically DFX when it comes to design and what manufacturing method needs to be used is super important. And I guess on that side of things, what are some common pitfalls you've seen product developers or startups, if you have specifics specific examples kind of the common pitfalls when it comes to DFX. Yeah. A lot of times I just start with really basic analogies. Let's move the plastics really quickly. But if I'm taking a CNC design or design that would be typically machined and moving into a plastic or if I had that CNC design mindset, actually, FDM, fused deposition modeling is that filament based 3D printing is very analogous in its design rules. Because I am, I, that, that layer of material, that thickness, that bead and strand of plastic has a width to it. It's almost like drawing with a crayon, right? Or at least a one millimeter colored pencil or something like that. And When you think about wall thicknesses, you typically are working on that, 1. 5 millimeter or 60, 000 safe zone, if you will, like for wall thicknesses there. And it's very analogous. To how I have a design for a CNC machine part. So if you're comfortable with designing for machining or you kind of know those rules, usually designing for FDM follows very similar rules. And you could cheat a little bit, but you know, go off angle features and things like that with additive manufacturing. But it's still like that accessibility, everything else about that. You do want to think about that in the design because it does require supporting structures. Designing for molding, putting parts on that diet. Instead of adding thickness to a part using ribs for strength, right? That type of design, it works really well for those resin based technologies, as well as for a plastic powder bed technologies. Those analogies work, very well in that phase, and you could go a little bit thinner on walls. I still will say if you are always designing to the minimum wall thickness, it's like going to an auto store and saying, I want the

thinnest brakes possible. You're designing risk into your design into your design. You go to minimums where you need to go to minimums, but work with nominals or slightly larger than that in order to have a more consistent print because. Things happen. And again, mechanic stuff exists. People will squeeze your part or twist your part or a little drop on the floor and you don't want your part to break. You're the first time it feels stress. sO it's very different than resolution is not where you should be designing, resolution shows where you can put text and other images in on top of your robust design, but in general for additive manufacturing, like I really do think bigger bulky designs. FDM works really well for that. If your designs have more organic features moving to SLS, MJF or resin base technologies are going to work much, much better for you. I see this all the time where someone doesn't want to spend the extra 4 to move to SLS and they get like a very small part in FDM. And when you're talking about 10, 000 layers and, this coarse feature, your part just doesn't look great. I mean, no other way to say it. The machine did its darndest to make your design. The machine worked as hard as to make that design, but it just has naturally coarser features where other, other processes just will dance all day with organic features and smaller, wispier. dEsigns and also the material itself tends to be more forgiving. Yeah, I think you know for me, that's one of the biggest things is like usually where Like what am I choosing? What's my goals? Am I looking for mechanical robustness? Am I looking for? Detail resolution. Am I looking for clear? That's a very common thing that someone looks for. In that case, resins print clear parts. So we have several ways of printing a part that is translucent, and we have some coatings that we've had to bring up to, I'll never say transparent, I'll say IceCube, bring it up to IceCube clarity getting to optical is a very different conversation, and usually adds a couple, a zero or two to your order because that's manual work, but that's like my general guidelines is usually I look at the design and I can start instantly seeing what processes are good with that. mhm And then from there, it's what are your goals? Mechanical, is it stiff? Is it rubber? Is a metal, is it, there's a really great down select doesn't need to be sterilized. Does it like what temperatures is going to live in? Is it, does it need to have some sort of water or air tightness to it? That's is it gonna be dropped? It's gonna be hit. It's gonna be kicked. What, chemical exposure. All that will help downslide very quickly into what processes you'll be working towards. That's perfect. I think the parameters that you listed out are great for just people to think about and also know going into these conversations when it comes to manufacturing, because oftentimes I see a lot of startups jumping the gun of working with suppliers or manufacturers or exometry manufacturing. And they don't have their requirements squared up. Knowing your requirements and kind of going through that list that Greg just mentioned to be able to have a successful engagement, but also pick the right manufacturing, because if you don't have those requirements, or you find that out later, then you might be going down the path of the wrong manufacturing process. Are there any other parameters that might be helpful for people to mention? I would say just in general the, when you're looking at additive manufacturing, it doesn't have the vastness of accessibility in materials that machining or molding may have. You have under each technology vertical. You have a certain set of materials. That's a good set, but it's not the set. It's not the total set. And and so those attributes, just as I mentioned, impact resistant thinking about attributes like flexible, tends to be more useful than thinking about materials. I can print Nylon 12 in SLS, MJF, and FDM. FDM is the filament base. So all of a sudden you have a, you have that Z strength difference, right? You have MJF may have a certain difference to SLS. And thinking just straight on materials will not always get you exactly where you want to be with your design. Even I've had people that are working, saying I need a part to be ABS. But their design resolution was much better fitting for a resin based process. So it's more about what do you need your design to do? They're like, I needed to I wanted ABS because I have this pin here that needs to kind of slightly deflect to move in. And then on that case, we're like, okay let's look at this, this semi flexible resin instead. Because what they really wanted was a snap. They, eventually it'll be APS, but right now, I just need it to snap in place, and so it brings you to a very different conversation, and usually gets you in the right technology and material vertical out of manufacturing when you're thinking about attributes. Yeah, that's a good call out to system interfaces are important. Usually what the part is going to interact with could sometimes dictate how the part is designed or manufactured or what material you even use. So I think having that kind of system level thought and going into conversations where you communicate that is. A super important parameter to mention. Okay, great. This has been a great conversation. I really want to kind of wrap it up with your thoughts around general just development and innovation in the hardware development space regarding AI or other developments that you'd be interested in mentioning. Oh yeah, what a time to be alive. I mean, just honestly, that's, I, we use AI as a core, a Xometry's process. So what I mentioned instant pricing, what's actually, kind of happening behind the scenes is we have computational geometry, interpreting a 3d model that you're uploading. It's a, and it's looking at it's driving different attributes of this, but it's different

than a cost plus where we're like, Oh, how long will it take to build here? Let me add an overhead and do this instead of saying. Parts like this on a competitive marketplace and this process at this quantity with these features cost this much. So it's that machine learning black box where you just give it a bunch of information and over iterations and iterations, a decade of information, millions of parts quoted and. We're, and also we are on dog food, right? We're sourcing directly to our manufacturer paying for this. So we actually know the cost because we've done it before. So getting this market based pricing, uh, AI derived is how we move at speed that Xometry is being able to do that. And we use AI for our AI driven matchmaking as well. Not just what you say you do, but do you actually do it? Suppliers, based on their behavior in the work. We worked to build, we worked to actually match relevant work with the suppliers. How if you're on Netflix and you see that little match score saying Hey, there's a 96 percent chance you're going to like the show. They kind of see that the supplier see a percent match score saying Hey, there's like a 84 percent chance that you're going to take this job right now. Like we, we predict that. And we let them know, we're like, Hey, we think this is a good fit for you. And so it gives more relevance for the suppliers because suppliers, they like doing work that's in their sweet spot and actually makes it cheaper for you, the customer, because it's their sweet spot. So it's not they're not having to stretch to in order to produce that. So it's, they're not, adding expenses. And so for my side, the quoting procurement supply chain side is AI driven. I'm seeing, with chat GPT and other prompt based services. Like I'm seeing AI deriving so many interesting things. If you look at I've seen some prompt based design tools recently. So saying I want to make a enclosure, it's going to be a square, it needs a three millimeter wall thickness, it's going to be an inch and a half deep, I could say millimeters and inches because AI is just going to convert it, right? And I need four holes to put in with a, a half inch boss and a quarter twenty thread or something. And then all of a sudden it'll be like. Is this the kind of shape you're looking for? We may see a design paradigm shift, almost the same shift that I saw where I learned on AutoCAD programs when I was in high school in the early 2000s, like 2000 2001. Where I would have to know what I was doing right text space in like this line is going to be four inches at this angle. This line is going to be here. This line is going to be here. And there was no real undo or Oh, let me change that parametric design later. Parametric started coming out around 2002 ish, and all of a sudden I thought it was how crazy is that? I just make a square and then I tell it what the dimension should be later. That paradigm shift, which is just common CAD design and the future. Okay. You may have a chat box in your design stage, basically, and this is where I, again, like these things, I hope and pray for, right? It's just being like, hey hey, I think I want to diecast this. And it almost acts those squiggly lines in a word processor where it's like, Here's some changes, and just, brings you up to manufacturability and allows you kind of manipulate and modify within your instance. That's what I'm hoping for in the near future, but that's where I really see AI driving better design. Less downtime with mistakes and even catching challenges with integrated product design before they happen, before they're released into the manufacturing life cycle. Totally. And do you know what the prompt based design is called? Is it through a company or? I need to look that up. I was just speaking with a couple of folks recently on this. So this, most of this is in beta stage. Right now. But yeah, if we could share that in the podcast notes, but I am very, I am super curious on this now. My, my bias is on manufacturability feedback. So I'm like, design is cool and all, but a lot of people have designed software, but a lot of people don't necessarily design for manufacturing. So that's my bias is like, how can I tell AI prompt to be like, I'm about to make this injection bolting, squiggly lines. Great. Just like a word processor, but even better auto correct. Think about like when you're typing with your thumbs, right? You could put completely misspelled word and it gives the right word out of it. Like, how can you put that design paradigm into your into your software as you're in the design phase? Fabulous. Okay. We'll definitely do some research and put that in the show notes so that people can check that out if it's available for public, that is so I'm sure that there's a lot of stealth people working on it too because it is really an open space. And I feel like the intersection between AI and hardware development is. Definitely going to lag from just AI's implementation into software, because I mean, it's kind of, it's home base. So I'm really fascinated with where the world takes AI and super fascinated with how we navigate the kind of IP challenges of it too. And since it's trained on data but this was such a helpful, we already have some names here. Tanso 3d and Catify AI. Cool. I'll definitely make sure to check them out myself. And I'm sure that our listeners will too. Awesome. But thank you so much, Greg. This has been such a helpful conversation. I think considering different manufacturing capabilities and the prototype stage for a router prototyping. They'll get so many useful chunks out of this. And do you have any last piece of advice for people working in the hardware space and trying to build ventures? Have a conversation. That's my last advice is when a manufacturer gives you feedback in your design, it is not a critique of you. We want your success. We, my success is your success. And everybody at

Xometry, we feel the same way. So when we're giving DFM or manufacturability feedback, our goal is to help guide you in a direction in which your results will be predictable and repeatable. And having that conversation is free. So let's have it. Let's have a let's have that talk. Get some cat up even if you're not directly ready for pressing go upload the file. It gives us something to talk against and we can help guide you in the right direction. And, we have so many tools through our resources page online that can help you get that right direction as well. But I just really recommend that because, yeah, when you work with your manufacturer, you do get better products out of it. That's a great way to end the episode. Thank you so much, Greg. Thank you so much. If you've made it this far, welcome to the too long didn't listen section. Before heading into this section, as a disclaimer, this episode was not sponsored. But Xometry did provide a. Little perk for builder's circle listeners, which, uh, which is $25 off any order that is above a hundred dollars. You can find the code BLD C I R P O D 2024 in the show notes for your use so with that, we're going to transition to the too long. Didn't listen. Section where I will quickly walk through some key takeaways. So in this episode with Greg, we started off talking about how. Existing manufacturing method. Could create a bias. And so making sure that you have awareness around design decisions, you're making to ensure that existing infrastructure doesn't inherently constrain your design longterm. An example of this would be you having access to a filament based 3d printer. But then your part once it scales. Requires potentially injection molding. But it doesn't have the correct. Designs for that. So making sure that is. Bias that you are aware of and that it doesn't make it to early stage design decisions. We then went into the details of the difference between filament based versus power bed. Fusion 3d printing talks about SLS MJF SLS, a selective laser centering M J F is multi-jet jet fusion. These are powder bed, fusion techniques, and the powder is plastic. Basically the big differences that it makes the parts is essentially float instead of being built on a bed. And then talked a little bit about resume based 3d printing with stereolithography, carbon, digital light synthesis and PolyJet. Biggest learnings from that is that different 3d printing methods require different designs and support structures. Important considerations would be to acknowledge the 3d print. We'll process. So is it upside down? Is it being built on a plate or being cured in a weird way? What materials are you using and what mechanical properties are you trying to achieve? What is the end use of the prototype and is it representative of what it can be with other manufacturing methods? So these are the questions that you should ask. Resin and powder bed fusion processes could lock your design in due to unscalable ability to handle overhangs. Keep that in mind. Another big point of advice from Greg, having worked with a lot of companies, he says that you need to really have an idea on target price per system or per part. Other than that, there were really useful nuggets throughout the entire episode. I highly encourage you to go back and listen, but these are the key takeaways that I was able to pick out. From the episode. So I hope that they were helpful. And as you go into your prototype phase, This episode should really help you decide on what type of prototype to do and how to scale from there. The opinions and information shared on this podcast are for informational purposes only. We always recommend that you seek professional advice before taking any action related to your business or personal ventures. Thank you for listening, and I hope that you enjoyed the episode