2018, Chapter 1: Adventures in 3D Printing If I’ve seemed scarce in the early weeks of this year, it’s at least in part to digesting the impact of two important prototyping technologies we just brought on-line: 3D printing and 3-axis CNC. Today, I was finally able to spend some real time with our 3D printer…and immediately got results that surprised me. So, I figured I’d share my experience, and perhaps help some companies understand the pluses and minuses of this technology. “Hoo boy, another 3D printing fanboy here,” you may be saying, mentally tuning out. Well…in short, not so much. My experience with 3D printing ranges all the way back to the early days of 3D Systems (I was aware of the company in the late 80s, had friends working there from 1990 on, and my agency did their marketing pretty much exclusively from 1994-1998), as well as with SLS (I remember the early metal-printing units that sometimes had some very spectacular meltdowns), and, in addition, leading the marketing and commercialization of several different technologies, including 3D System’s MultiJet dot-matrix scheme to some really crazy stuff (like metal printing by sonically welding thin sheets of aluminum—and yes, that was just as insane as you might think.) Ironically, I have the least experience with today’s dominant 3D printing technology—FDM, or “them stinky printers that take reels of plastic filament.” The reason for this is multifold, ranging from some really spectacular breakdowns of early inexpensive FDM machines, to the relatively crude structure and surface finish that most of these machines produce. Disclaimer: perhaps there are some really high-res and fine-finish FDM printers out there—please feel free to correct me if I’m wrong. Disclaimer 2: this is gonna be a heavy engineering-and-business chapter, focused mainly on helping new or established companies decide if 3D printing is for them, so you may be mightily bored if you’re not all that into the technology. Disclaimer 3: as with most of our stuff, this is based on our experience—which means “NOT based on encyclopedic knowledge and extensive experience with every 3D printer on the market.” We could be wrong. Please correct us in the comments. This is supposed to be interactive (and not just a forum for cat pix, thought I do enjoy them.) Confused? Don’t know what all this 3D stuff is about? Perhaps best to start with a short TED talk from the developer of a new technology, CLIP (Continuous Liquid Interface Production), which can be thought of as an outgrowth of stereolithography (the original 3D Systems tech, and the technology used in the 3D printer we selected.) https://www.ted.com/talks/joe_desimone_what_if_3d_printing_was_25x_faster It’s a good talk, because it really goes through a lot of a benefits and pitfalls of modern 3D printing. To summarize: 1. 3D printing allows you to make stuff that isn’t produceable via machining or casting. 2. 3D printing isn’t really 3D printing; it’s 2D printing, over and over again. 3. It is (usually) slow as crap. 4. The amount of postprocessing might surprise you. Go back and read #3 again. SLOOOWWWWWWWWWW. As in, hours for a typical part. As in, use in production, no way, Jose, unless you’re talking low-quantity or high-dollar parts. The Carbon 3D printer appears to be a significant breakthrough in speed, and may pave the way to using 3D printing for actual production, but there are additional pitfalls there as well; I’ll get to those. Aside: we won’t be getting a Carbon 3D printer for some time, not at $50K/year lease, 3 years minimum. I can totally see that paying off for a larger company, especially if they are making actual production parts with it (it doesn’t take many avoided injection molds to pay that off.) “So if it’s so slow, why’d you bother getting a 3D printer?” you might ask. Well, it’s because we never planned to use it for production. We planned to use it to make prototype parts, iterate them, and then send them out for injection-molding or die-casting. “Oh hell, that sounds disappointing, you’re just gonna make a few gewgaws with it and that’s it?” Well, they’re far from gewgaws, but yeah, we’re not going to make 100,000 light pipes on it, or custom-printed transparent chassis or knobs shaped like chess-pieces or extensions to connect a front-mounted button to a rear switch (already discussed this—they don’t fit THROUGH the transformer on many products.) Nope. What we’re gonna use it to try out new designs, see if they fit, see if they work, and use them to show our vendors what we want. This is going to be especially important this year, as we put the finishing touches on the turntable and get it into production. The turntable doesn’t really use many plastic parts, but the ability to run off a 3D print to check fit and tweak in advance of the metal is invaluable. “Still sounds boring,” you might say. “I thought 3D printing was gonna usher in an age where I would just download a product and print it on my desk, saving all this old-fashioned ‘factory’ nonsense.’” Yeah. Uh huh. So, with that, let’s rewind and do a little primer on what 3D printing is…and isn’t. (Not) the Key To Printing An iPhone on Your Desktop When 3D Systems was young, there was a run of irrational exuberance. You could make all sorts of crazy stuff that you couldn’t make with machining! It would replace conventional manufacturing! It was going to allow people to print their own cellphones right on the desktop! It would dramatically shrink the time it took to develop new products! Yeah. Let’s break that down: You can make stuff you can’t with machining and casting! Yep. True. The problem is, this is only of partial utility, especially when the surface finish on the parts isn’t exactly consumer-grade. It would replace conventional manufacturing! LOL. No. Especially with early-90s processing power. Creating a sliced 3D file (STL) was no joke in the early days. Machines would choke on it. And generating supports and orienting parts was a real art. Printing glitches (read, bad data or a brownout) killed multi-day builds. Plus, you were dealing with messy, toxic liquid resin. It was in no way ready to replace conventional production. You could print (any consumer good) on your desktop. Even more laughable. The best you could do was print a plastic shell. You weren’t going to print a screen, a PCB, a microprocessor, or anything else. Oh yeah, and those early plastic shells…it was a real accomplishment when they got resins strong enough to withstand snapping together. It would shrink the time to develop new products! Now this has shreds of truth. With rapid iteration of prototypes in-house, development could frequently be accelerated. That is, if you were using heavy-duty 3D CAD already, and had the space for a machine twice the size of a phone booth, and a budget that started at $250K and went up for materials, maintenance, and support. And even that was heavily caveated. Faster development didn’t always translate to the manufacturing Holy Grail: faster time to market. There always irrational exuberance with new technologies. See Gartner’s Technology Hype Cycle. Within a few years of launch, 3D Systems had refocused on the “rapid prototyping” aspect of their technology, and started seeing some real traction. They also developed adjuncts of their technology that allowed fast investment-casting, and developed new technologies intended to reduce the cost and up the speed of 3D printing (anyone remember Actua?) And, as with any technology, alternates came online. Selective Laser Sintering brought metal parts into the realm of 3D printing (and also, more practically, fast wax masters for investment casting). Then, Fused Deposition Modeling brought the price of 3D printing down within reach of a much wider range of users—hell, there are FDM machines selling for $200 these days! And yeah, some companies started experimenting with 3D printing for production. And some had successes, like MrSpeakers with his 3D-printed headphone cups. Aside: this was a really interesting application, since it played to the strengths of 3D printing to create parts that could not be made any other way, and also used acetone vapor to improve the surface finish. And some surprising applications emerged. Many jewelry companies now do every single design they make as a 3D-printed wax model, then investment-cast the result in precious metal. For such a low-volume, high-dollar, semi-custom application with small parts, 3D printing is ideal. And just-in-time 3D printing for medical and dental applications is now becoming mainstream. And you’d be shocked at how many specialty costume parts now come straight out of a 3D printer—from Loki’s helmet to Wonder Woman’s armor. None of these uses were mentioned in the early days; hell, 3D Systems’ former head of marketing was shocked to see a high-end ring manufacturer printing all of their wax masters. Still, with all this change, we at Schiit never invested in a 3D printer. The reason was simple: the more experience you have with 3D printing, the better you understand their shortcomings. Seeing only a handful of messed-up builds coming out of various 3D printers is enough to make you start wondering how reliable it is. Knowing you’re going to have to be intimately engaged with 3D modeling before doing a single part is also a barrier. And, with limited engineering resources, even a reliable, easy-to-use 3D printer could end up being a rabbit-hole of time-wasting and frustration. And, in fact, that should serve as a template for the first question any company considering 3D printing should ask themselves: 1. Am I ready for 3D printing? Am I already doing 3D modeling? If not, remember you have to purchase and learn 3D CAD as well. Or staff to it. And those costs can be much higher than the printer. If 3D CAD isn’t already a part of your design process, you may not benefit from 3D printing. Hell, it might tie your engineering resources in knots. Am I prepared for screw-ups? You’re gonna print a lot of junk. You’re gonna screw up files. Your power will glitch and you’ll lose a build. If you expect laser-printer perfection every time, run. Away. Fast. Okay. Have I scared you away? No? Cool. Let’s look a little closer at the process, then move on to the story of Schiit’s First 3D Printed Part (with photos.) I Come From the Future (Where I Work In An Auto Body Shop) The above subtitle aptly captures the gist of 3D printing. It’s a really bizarre combination of super-high-tech hardware and downright antique finishing methods. It’s actually no wonder it’s caught on as the technology of choice for Hollywood’s specialty costume shops—they’re used to molding parts out of resin, then sanding, polishing, and painting to get the various finishes they need. No matter what 3D printing technology you choose, you’ll be confronted with this dichotomy. Stereolithography? Get used to washing, clipping, UV post-cure, and maybe even sanding. FDM? Get used to acetone vapor polishing, sanding, and painting. SLS? Get used to post-curing, tumbling, filling, additional oven sintering, plating, and more. As an example, consider the 3D printer we chose: the FormLabs Form2. This is a stereolithography printer that’s designed to be as user-friendly as possible. And it does a pretty damn good job of it. But it still comes with: A futuristic-looking desktop machine with orange transparent shielding, allowing you glimpses of the laser as it traces out each layer of the part, and a fancy touch-screen interface for running it all. A workbench full of tools that include scrapers, clippers, gloves, bowls (for holding 99% isopropyl alcohol) and instructions about how to use these medieval parts. #1 you might expect. #2 engendered a “Huh?” moment from every engineer and technician at Schiit. Except for me. When asked, I sighed. “That’s how 3D printing is.” Mike looked confused. “What do you mean?” “I mean, you get this super-futuristic looking machine, but to get usable parts, you usually end up with a setup that looks pretty much like an auto-body shop.” “A…body shop?” Mike asked. “Like, sanding and painting?” I nodded. “Like sanding and painting.” I remembered my first experience to the back of the Scicon shop (a 3D printing service bureau). I was shocked how much it looked like our old Odeon shop, where we sanded, painted, and finished loudspeaker cabinets. Mike looked horrified. “Are we…” “We’re gonna be doing as little of it as possible,” I said. “This is supposed to be easy to use. But we still have resin to deal with.” “Resin?” “As in uncured epoxy.” Mike’s look of horror deepened. “Oh, crap.” “Yeah. It can get everywhere. Hence the gloves. Hence the isolated workbench. Hence my paranoia.” “I’m glad you’ve done this before,” Mike told me. I shrugged. Because in reality I hadn’t. All my experience with 3D printing had been on the outside—in marketing. I saw enough of it to know its pitfalls, but we’d never had a 3D printer ourselves. I really didn’t have much idea what to expect. I didn’t know how hard it would be to get the printer running, or how long it would take to get usable parts. I did, however, figure the learning curve would be long, and we’d have a ton of screwed-up parts before we got anything that resembled something we could use. And the inclusion of the workbench with a bunch of clipping, cutting, prying, and cleaning devices didn’t make me feel all that confident. Aside: why did we choose the FormLabs Form2? Mainly because I figured stereolithography would give us the best chance at making finely-finished, accurate, and usable parts. I also knew at least some of the pitfalls: the messy liquid resin, supports, etc. It also seemed to be produced by a company that had worked hard to make the printer as user-friendly as possible—most of the bitching about their product centers around the cost of the resin and build tanks, which I don’t care about that much. We don’t need to minimize cost of parts; we need to maximize our chance for good parts. We also purchased two accessories from FormLabs: the first was an automated cleaner that looks a bit like a futuristic deep-fryer, which automates the process of cleaning the uncured resin off the parts. It came with its own alarming set of accessories, including a siphon pump straight out of a 1970s aquarium setup, and a float for measuring specific gravity of the alcohol (necessary, since it becomes contaminated with resin over time.) The second we still haven’t received yet as of this writing: a UV post-cure oven that automates the post-curing process, necessary for some resins. Alternately, we can simply put the parts in the sun for a while. But yeah, the fact these accessories exist should give you a hint that 3D printing ain’t like buying a new color laser printer. FormLabs also directs you to download their own file preparation software, which you use to orient the part (important to get good mechanical properties) and create the supports (necessary to keep the part from sagging and deforming during the build, and which have to be cut off the part after it’s finished.) The software estimates the amount of resin you’ll use to build the part, allows you to choose a layer thickness (100, 50, or 25 microns), and calculates the time to build the model. It takes what used to be a very tedious job and automates it pretty much completely (it even has a “one click print” function.) But, even with this automation, it assumes that you’re going to feed it a prepped .STL file, which is a specialized 3D CAD format that has been prepared to slice into layers. Don’t know what an STL file is? Can’t create one? Don’t pass go, don’t collect $200. And that brings us to the next question you should probably ask, before embarking on 3D printing: 2. Am I ready for the realities of an imperfect process? Do I have the tolerance—and space—for the mess? Again, this ain’t clean, easy, silent, or smell-free. If you expect getting usable parts to be as easy as pressing a button, you’re in for a big surprise. Do I have the engineering time? This can eat up a lot of time. Are you prepared to give up a week of your time to get set up and running—or do you have another engineer that can do it? Surprise and Shock: A Tale of Schiit’s First 3D Print “Okay, okay, I get it,” you say. “3D printing isn’t for everyone. It’s not production. You’ll probably screw the pooch. So how’d it go for you?” In one word, surprisingly. When we received the machine, I set it up myself in only a couple of hours, being super-slow and super-careful about it. They did a good job of engineering out the worst parts of stereolithography. You never touch the resin (it slides in like a cassette), the build platform and tank are disposable (again, so you don’t have to deal with resin), the build platform is snap-fit, the software takes you through leveling the machine and getting on your wifi. Which is where I hit the first glitch. None of our wifi zones seemed to be in range. We never figured out why; we just moved one nearer the machine and it worked. But that set us back a bit. When that was resolved, I still wasn’t ready to start printing. We needed 99% isopropyl, buckets of it. “Why?” Tony asked. “To clean off the resin,” I told him. “Do you really have to?” Images of a dripping stereolithography platform came unbidden. More images of sticky resin fingerprints on the machine’s case, on the desk, on the automatic washer, on the light switch, all over the lab…shudder. “Yes. We really have to.” I emphasized that we really needed to keep everything clean, this could get really ugly in a big hurry. “Like Ten Thumbs,” Tony said, recalling the contractor who had put epoxy on the floors of Schiit’s first office…without adding hardener first. “Yeah. Like that.” Now Tony shivered. “Gotcha.” Once the isopropyl was in place and I’d filled the washer, I had no excuse not to get started. 3D printing companies always give you sample files you can use to print, but I didn’t want to do that. Oh no. In fact, I had a specific application in mind: Light pipes. Light pipes were why I’d purchased clear resin to start. Light pipes, if they worked, would solve one of the biggest pains of Schiit production: the need to bend and form LEDs through the front chassis holes. Aside: seems like a small problem? Well yeah, maybe if we’re talking one LED. Three? Oh hell. Ten? Oh hell no. It’s actually a huge time sump, and that’s even before we start talking about the LEDs we accidentally break. Eliminating all of that would make everyone very happy. We’d looked into designing light pipes before, but I’d never been thrilled with any of the solutions, most of which required additional machining, and could still result in the light pipes breaking. We’d quoted an injection mold for a light pipe, but the cost was such that I wanted to me sure that the damn thing worked—and worked well—before we sunk the money. However, recently I’d figured out a new approach, one where we didn’t attach the light pipe to the board at all. Instead, the light pipe would attach to the top chassis, and simply slide over surface-mount LEDs on the board. It wouldn’t even touch the LEDs at all. Sounds great, right? Again, yeah. If it worked. Which I didn’t know if it would. Which is one reason I wanted to use a 3D printer to find out if it worked, and, if it didn’t, tweak it until it did. At that point, I could be confident in investing in an injection mold and making tons of them. So I knew exactly what I was going to do on the 3D printer, first thing: a light pipe. And that’s where I started the morning of January 22, 2018. I had an old light pipe design (one that was supposed to be attached to the PCB) in 3D CAD, but it wasn’t right. I wanted one that attached to the front panel, had at least 5 light pipes in line, and was snappable down to smaller sizes. Not a big deal. I’m familiar enough with Alibre 3D that creating a part like that takes only about 15 minutes, after a couple of measurements to confirm the geometry of our current front panels and LEDs. A few notes on this design: 1. It’s a guess. Will a 45 degree chamfer reflect the LED light, or should it be a radius? No big deal, I could try both. I just decided to start with the simple version. 2. It’s designed for the geometry of our current LEDs. I’ll need to do a board with final geometry to see if that changes anything. 3. It’ll have light bleed. We’ll have to see how much we have to mitigate that later. 4. It’s not really made out of “light pipe material.” Which is fine for for a first prototype. I exported the 3D design as an STL file, opened it in Formlabs software, and had it automatically orient the part and add the supports. This literally took less than a minute. Aside: I really can’t emphasize how important this streamlined process is—it’s completely different than our CNC, which I’ll get to in another, future chapter. Probably after another chapter where I take you through an engineering design exercise—the design of a tube amp similar to the Vali 1. Did I mention that 2018 would have a lot of engineering chapters? It estimated using a tiny amount of resin, and gave me a build time of 1 hour 9 minutes, using 50 micron layers. The build time would be halved with 100 micron layers, and doubled with 25 micron layers. Could it really be that easy? I wondered. Nah. Probably not. I’d probably see some spectacular failure. But we had to start somewhere. “**** it, print it,” I thought, and pressed the button to start the build. The machine started its gyrations—first by filling the build tank with the clear resin, then heating it to operating temperature. Tony and Naomi came by to look. “Is it printing?” Tony asked. “Soon,” I told him. “It has to heat up, it looks like.” “Oh.” He seemed disappointed. And for a while, there really wasn’t much to watch. The machine ran the wiper over the build window a few times, probably just to evenly distribute the resin. We waited while it slowly came up to temp. Finally the build platform whirred down into the clear resin. I could see the laser tracing the first layer. Pretty exciting! (Really, I thought it would fail. I still thought it would fail, or give us an unusable part.) Then the platform came back up, the wiper swiped across, and down it went into the resin again. “It has to do that every time?” Tony asked. “Yep. Every layer.” “How many layers?” I squinted at the screen, which was counting down the time. “406.” Tony shook his head. “Oh boy that’s slow.” I laughed. Yep, it was. And that’s the huge advantage of the Carbon 3D printer that I linked to previously: it never wipes. It never stops. It just builds. And it doesn’t use a laser to trace the part, it uses a DLP engine (ironically, something I suggested to 3D Systems, literally 10 minutes after being introduced to the technology. Not to say I have any special insight, it’s just nice to see I wasn’t completely full of it.) Everyone came in to see the 3D printer running. They all pronounced it a bit of a snooze. Which is good. Excitement is the last thing you want in 3D printing. Especially if you’re talking early metal SLS excitement, which included some very big fires. And they were right. The 3D printer doesn’t make much noise, just some understated whirrings. Which is good, since it’s in the engineering lab. It also doesn’t emit any terrible odors, since the build envelope is completely enclosed. The alcohol wash machine is similar—not much stink, which I was worried about. All in all, boring. Good. I went upstairs to work on some other engineering stuff. When I came down, the print was done. The platform had raised above the resin. It was still dripping, but I could see something that looked very much like the part I designed, stuck to the upside-down build platform. No way. There was no way it just friggin worked. Well, there was only one way to find out. I unclipped the platform and put it in the washer, and let it start its 20-minute cycle. Again, everyone came by to see the finished part, but there wasn’t much to look at, with a clear part submerged in clear liquid. Finally, the washer was done. I unclipped the build platform and flipped it over. And there it was: a completely clear light pipe, exactly what I’d designed. I was a little surprised at how clear it was. I was used to stereolithography resins being yellowish (the old days strike again), and I wasn’t really prepared for something that looked nearly optically clear. Of course, it was also stuck to the bottom of the build platform, and there were a couple of dozen supports that needed to be cut away. One of the tools Formlabs provided was a specially-designed lever that fit in a release slot that the software built into the part base. I pried gingerly, and it popped right off. This resin didn’t absolutely need a post-cure, so I didn’t have to worry about UVing it. I went straight to the next step: cutting away the supports. And I held in my hand a very credible-looking light pipe design. No. Friggin. Way. It really couldn’t be that easy, could it? Of course, it probably wouldn’t fit. Either the LED towers wouldn’t match the chassis, or I’d boned the measurement to align the part with SMD LEDs. Naomi was working on a Freya, so I borrowed the front panel. The light pipe prototype dropped right in. The LED towers fit perfectly. I shook my head and laughed. “It works,” I said. “Does it fit?” she asked. “In the chassis? I still need to hack up a board and find out.” “I have a parts board you could use,” she offered. Cool. I took the parts board (these are usually old prototype boards, or stuff that was too damaged in production to be sellable), pulled out a bank of 5 LEDs, and replaced them with the only surface-mount parts I had handy: the amber LEDs we use for voltage references. Aside: before you get too excited, amber looks like butt. Scroll down. We’ll be sticking with white, thank you. There was no way the light pipe would work, would it? I must have screwed up some measurement. It would hit the LEDs, or be too far away, or something… No. It fit fine. The light pipe proto sat right over the LEDs without touching them. Did it reflect light? Did it bleed too much? I fired up the Freya board…and sat back, disappointed. No light shone through the light pipe. Hmm, back to the drawing board. Maybe it didn’t reflect enough light… No. Wait. The amber LED was really, really dim. I mean, barely on. So dim the room would have to be totally dark to see it. This was with the same 10K current-setting resistor that give us the bright white LEDs that some people like to complain about. Hmm. I replaced it with a 1K resistor. It was brighter, but still nothing to write home about. Wow. I went to 150 ohms. Now it was bright… …and the light pipe worked! Yes. You read that right. In one day, we designed, printed, tested, and confirmed a working light pipe design. Is it perfect? No. We still need to try a curved profile. We need to get some white LEDs. We need to see how bad the bleed is in normal operation, and see if we need to do something to mitigate it. But we are far, far down the road to solving one of the most irritating problems we have in production—in one day—and that’s pretty amazing. Custom Knobs, Transparent Chassis, Design-Your-Own Products! What’s Next? In short, none of that. Like I said, we aren’t going to be using 3D printing for production. Will we use it to prototype new knob designs, or especially inserts that allow for easy push-on assembly? Oh hell yes. Will we make complete chassis from it? No, not unless we’re prototyping new approaches that are intended for die-casting. Will we use it for tweaking about a dozen parts on the turntable? Absolutely. Will we use it to allow mass customization? Nope, sorry. Will we use it to solve a myriad of mechanical irritations, by designing clips and brackets that can be injection-molded? You bet. In short, we have a couple of dozen applications already in mind. Which brings me to the last question that any company considering 3D printing should ask themselves: 3. Am I ready with enough applications to make it worthwhile? Do I have enough parts that can be prototyped in plastic to make having a 3D printer—and its attendant cost, upkeep, and mess—to make it worth having? If not, then it may be best to take a pass. Should I use a service bureau instead? There are tons of companies out there who can run 3D printed parts for you, leaving the finishing and mess in their shop, rather than yours. We used such a service to produce the prototype castings for our turntable. If you don’t think you’ll need to iterate a bunch of parts, it might be best to use a service bureau. So yeah. I’m not a fanboy. I got into this eyes-open. Yesterday I produced the first part on our 3D printer. Today I’ll be making some alternate designs to test how well they work at reflecting the LED light. By the end of the week, we’ll have something worked out that can go to injection molding. And then, we’re well on our way to solving one of our biggest production pains. Next up: turntable parts. And a couple of things I can’t talk about. Not yet.