📍 Have you ever had an idea for a physical product that you just knew had the potential to change the world? Maybe it was something you dreamt up in your garage or a solution to a problem you encountered in your daily life. Whatever it was, you knew it was a winner. If only you can turn that idea into something... Well, you are not  📍 alone! 

Countless entrepreneurs and innovators have stood exactly where you stand filled with passion and drive, but unsure of where to begin,  and that's where "The Builder Circle" comes in.  

My name is Sera Evcimen and I'm a mechanical engineer, hardware enthusiast, and hardware mentor.  I've had the privilege of working with numerous hardware companies that are passionate about solving some of the biggest challenges in the world.  And I will be your host as we explore the exciting and complex world of physical product development.   Welcome to the Builder Circle season two. I have

Guest Introduction: Scott Miller

 Scott Miller here again after the widespread appreciation for your episode in season one. I thought I'd ask my friend Scott to come back. So thank you so much for coming and being on the show again, Scott. It's so nice to have you again.

Oh, I'm psyched to be here. And thank you so much, Sarah, for having me and for putting on this awesome podcast for everybody out there in the hardware world. 

Oh, it's my absolute pleasure. And I get to very selfishly indulge in these amazing conversations with incredible people like yourself.  I'm incredibly grateful. 

Discussion on Mass Production and Injection Molding

And today you and I are going to really dive into the intricacies of mass production, what that looks like in a sense from the angle of injection molding, which is A process that a lot of startups eventually venture off into after they go through the prototype and rapid agile development stage of hardware.

And I feel like this comes time and time again, where startups don't really know when that transition point is appropriate and how to  go through that process. 

Understanding the Transition Point to Injection Molding

So diving right in, in your opinion when product teams are going through this kind of design exercises and trying to figure out what their product is going to look like.

When should a product team know to convert to injection molding? And maybe talking through, it could be like technological readiness level or even just unit economics or even just supply chain strategy. What are your thoughts on that? 

Sure. Yeah. 

Insights into Injection Molding Process

Injection molding is fun just because it's such a deep body of knowledge. And I learned something every time I get involved in it. So I'm definitely not an expert, but but I've been fascinated by it over the last 25 years or so.  So when I think of injection molding at a high level, I think, what do we want to accomplish with it?

And generally, it's really good for high volume of plastic parts where you're trying to minimize the piece costs or your overall cost of goods sold, but have enough volume that you can justify the cost of the tooling, which can be, quite expensive. And when I think about  how this all works, I really look at it in terms of a progression.

The Progression of Manufacturing Techniques

So you wouldn't want to do injection molding if you're just building five parts and that's it. It would just be a lot of extra work and not worth it. I always go back to, if you're building one unit and save a dollar, you save a dollar. If you're building a million units and save a penny, you save 10, 000.

So you want to think through, when does it make sense? And it's certainly possible to do some pretty simple math to figure out what are the crossover points.  But when I consider injection molding as I mentioned, you don't usually jump right into it. And to show you'd be focused on just getting the thing to work.

And there's really nothing better than 3d printing at this stage because joy of 3d printing is you can print anything, doesn't need to be manufacturable in any way, shape, or form. And often it isn't so you can have undercuts. Thick walls, whatever it takes, but at least you can prove out to yourself that the, you can get at least one unit to do what it's supposed to do. 

And then when you start to have confidence in that, then you may jump into say CNC or machining a part. And the nice thing there is that you get access to more materials. Maybe you want to use stainless steel or titanium or. Delrin or something like that can be hard to 3D print, but you don't need to pay all of the upfront cost for tooling.

And also you don't need to spend all that time because when you do tooling, it's like a project within a project and it has to be debugged. Whereas with machining, you can do it, relatively quickly to lay out your tool paths get a lot of different materials you can pick in them as a designer rules are pretty straightforward that are in some ways kind of more logical than injection molding. So I might think about doing that. And then after I've built up 20, to, to 30 or so units, and I'm feeling really good that design works, I can handle my manufacturing tolerances and things are performing as they should, then I'll probably jump into rubber molds or silicone molds.

And basically the idea there is you'll use stereolithography  to 3D print. Part and then pour silicone around that, and that creates a rubber mold where you can typically get about 50 units out of it before the integrity of the mold is is gone.  So that's a great way. And there's a lot of excellent materials that actually mimic the injection molded materials through, there are thermostats.

A thermo set that you'd use for the rubber molds versus a thermoplastic that you'd use for injection molding.  And then after that point, when you're feeling good, then maybe you jump into soft tooling and then you go into hard tooling. But yeah, there's a different stages that you'd want to progress through.

 That's super insightful. I think a lot of the time when people think of injection molding, they're thinking about the hard mold right off the bat. And that's when you get tooling costs of maybe 20, 000 or something along those lines. And  having that progression makes a ton of sense.

How would you determine what unit count would make sense to go to that next stage?

Understanding the Economics of Injection Molding

With an injection molded part, there's basically two main pieces, the cost of the tooling and then the cost of the unit itself. And roughly for a three by three part, if you were to tool that in China with a mold base. I'd say that's probably a 5, 000 tool, assuming it's nothing crazy. There's no slides, just a pretty generic like a wheel or a housing.

And then for the cost of the part, what happens there is you look at the cost of the materials. Let's say you build it out of ABS. We can look up the cost of ABS in terms of dollars per pound. And then there's a transformation cost to take it from the pellets into that final shape. And that's basically driven by the cycle time.

So how many parts can you build an hour? Let's say the cycle time 60 seconds, which would be a little bit long. It's typically 30 to 45, but it'll make the math easy. You'll get 60 parts an hour. And if the press Cost 60 to run a 60 an hour. Then you're looking at 1 a unit to build one of those. And let's say within that part, it's probably 0.

50 of plastic. So that would be 1. 50 for the plastic part. And then there's some labor and markup on it. In round numbers, let's call it like 2. So we've got 2 for the part and 5, 000 for the tool. So what you could do is figure out how many parts you need to build to break even versus say 3d printing them or machining them.

And there's also, of course, the associated cost of just lead time getting them in your hands and are you, and this is why I think you, you mentioned this immediately, is that when you're in this rapid iterative  design process, this is just not the way to go especially if you haven't gotten any customer feedback or testing feedback on the design that you have in hand.

That's right. Yeah. Typical tooling, which is assuming it's not proto labs, which is its own beast. But if you're going to tool it in the U S or tool it in China, I typically expect for a part that big about six weeks to actually just make the tool. And again, the tool is its own project in itself. So you would give the tool maker your part, and then they would likely have to go and put draft and round on it, which we can talk about in a sec, look at uniform wall thickness, undercuts, all these different things.

And then from that, they'll go and design the tool with the ejector pins, the cooling jackets and all this different mechanism that will actually make the part. And then there's a fair amount of EDM or electro discharge machining and CNCing to build it. And then they have to do test shots to actually get your part.

So if you were to go and realize, oops, we actually wanted to change that a little bit, then that would require going back to the steel. And there's a concept in tool making called steel safe, where you can always add plastic because that's machining away steel. So that's easy to do. But if you actually had to take away plastic, then they have to open up the tool.

They have to fill it with weld and then go and remachine it. So there's all this extra work. And then if you do that, you're going to get cosmetic imperfections, witness lines and things like that. And the tool isn't going to be as good. So  it's definitely not to say that once it's tool that's cast in stone or cast in steel, you definitely can modify it, but it's just those modifications come at a greater cost that I'd say, unless you're trying to build a prototype tool that you're going to throw away, you should have pretty good confidence in your part.

That it's not going to change drastically.

And  that makes a ton of sense. And It every single, a change order does add to the lead time no matter what. And I guess my another question I would have in this instance, I can't go without talking about design for manufacturability in this case.  If a company is working with primarily plastic parts, or they have some type of plastic casing and that they're eventually going to have to do injection molding, because that's when you go into kind of the millions and I guess it depends on the projected amount of parts that are going to get created, is it a good idea to start having those considerations early on so that you're not 3D printing parts that could absolutely never be injection molded?

Design for Manufacturability in Injection Molding

Because I have this maybe irrational fear of assuming  that the manufacturing process can absolutely do everything and is just this unicorn manufacturing process. And then those assumptions make it to 3d prints that get tested and everyone's The test results are excellent, like we want, we really want this to exist in the world and then they go to injection molding and then injection molding is you can't have that lip there or you can't have this little groove there.

That's not going to work. Is that a fair fear to have? 

Yes, absolutely. And we would see that all the time at my last company that firms would work with these amazing industrial design shops and create an iconic a sculptor shape and then come to us and we'd be like, Oh my gosh, that's totally not manufacturable or at least at a, a cost it's reasonable.

So I think the more you can design with DFM, Design for Manufacturability, and DFA, which is Design for Assembly, in mind at the early stage, much, you'll be much, much better off. And I think this is something that comes with experience. After you've done it a few times, you  understand the design rules and criteria, and then can just apply those earlier.

But yeah, as you were touching on, injection molding has its unique set of rules that are not always so obvious and a few of them are generally you want to have a uniform wall thickness so that the part cools evenly. You can think of injection molding is basically squirting toothpaste. So not like a liquid, but a paste into a mold.

And then having it cool, which is how you can subsequently eject it. So if you've got massively different wall thicknesses, some parts are gonna take longer to cool than others, and that's gonna increase your cycle time. But you're also gonna get these huge sink marks which could be a big problem. So ideally you'll have a two millimeter wall thickness everywhere, which takes some special effort to be able to do that.

Whereas if you're machining something, you really don't care as long as the walls aren't too thin.  You also have to think about undercuts. So for your housing that you described, if you had holes in the side, that would take you'd have to have steel in there to make the hole, and that would trap the part, which would be a problem that you'd never get it out of the tool. 

So what you'd have to do there is have a side action that would pull the pins out of the way, or there's some cool stuff we can talk about later with shutoffs that you might be able to do, but these are some of the things to think about. And then one of the. Other ones all the time is this concept of draft or draft angle.

So to explain that, what we could do is think of a cylinder with straight walls versus a Dixie cup with tapered walls. And basically what happens is as the plastic or melt cools, it shrinks a little bit. And anything on the inside, it's going to grab onto really tightly. So if it was a straight wall, it would grab on so tight.

You'd never be able to get it to release. Whereas if we can taper the walls a little bit with some draft, then it will release more easily. And typically it depends on how much texture you have and the material. But if you just had a lightly textured wall, it's probably a half degree of draft, which doesn't sound like it's that much.

But if for our box that we're building, it was six inches tall, that half a degree is going to add up a lot. So we might be three inches at the bottom and I can't do the math in my head, but maybe two and a half inches at the top. And that might substantially change, the function of the thing.

So we'd have to think about the draft. We'd have to think about the parting plane. So where does the two pieces of steel come together? And then that's where your draft would originate from.  So that would change the shape. And then we also think about rounds, because if we imagine this melt flowing in and through the cavity,  we don't want to have any sharp corners because those will create a lot of shear, which will create a stress concentration or generally weaken the part.

So we have to have everything flow. And where this gets tricky is if you design all this stuff in too early, then imagine that we have some draft on our walls. It gets really tricky to measure the distance between the walls because it depends on the height or the Z because you'll get different measurements and that creates a lot of challenges and really increases the complexity of your model.

So you want to be thinking about draft from the beginning, but you don't necessarily want to put it in. So there's a balancing act here. 

That makes a ton of sense, and it's probably easier when the parts are smaller so it doesn't become this explosion of draft angle.

Yeah. It just adds all sorts of complexity. And especially when you're just trying to get that thing to work, but to go back to the failure mode that you described, it's really important to have this stuff in mind and understand how it works when you design your parts to be 3d printed so that you can design them with straight walls, but just make it a lot easier when you do transition to injection molding.

Engaging with Injection Molding Shops

And this might be irrelevant question, but I am curious from the point of a founder that potentially has not done very specific design for injection molding experience and potentially doesn't have resources like yourself to be able to ask these questions to or in the room.  If they were to go to an inject say they have a design in mind that is not stable yet but they want to see if they're headed in the right direction.

Do you feel like it is reasonable to go to an injection molding shop and just show them the design and say, is this something that could be injection molded? Eventually, we're not there yet, but we really want to get your thoughts on it. Is that a reasonable request to make, or would that be too, would that put the risk on the relationship if they were to end up to work with them but are just not ready yet? 

Yeah, I love that idea. For the tool makers, they do this day in and day out and probably forget more every day than I'm ever going to know. And this is really complex stuff. So I think getting them involved early is great and they'll have phenomenal feedback. Typically they'll do a detailed assessment of looking at the draft angle and there's a lot of automated tools like Moldflow to help out with this.

But just making sure there's enough draft angle, undercuts, where would the parting line go?  There's CFD modeling, so as the milk comes in, you can see if there's a pressure wave, or if there's areas that might not pack and would be a short shot. And they have the expertise to run all this, and I often will see maybe a 10 page PowerPoint presentation come back, which is just eye opening because Every little corner, everything matters in injection molding.

But I think the trick is to reach out to them. It's probably okay to share one part and just say, can you give us feedback on this to build the relationship? But what one wouldn't want to do is how is to waste their time doing a full FM on everything. If you didn't have an intent to work with them.

I think that would damage the relationship and just not be the right thing to do as a. Fall back. I know for AutoCAD, if I remember correctly, the students get free access to their whole portfolio. So if somebody is a student, they could try to run mold flow and they may not fully understand what it's saying, but it could give.

I think it's a valuable experience. The other option would be potentially uploading something to proto labs, which uses automated software and I think gives Pretty good feedback on the problem areas. And that way it wouldn't be wasting somebody's time. And, maybe you want to go forward with proto labs anyhow to get some some parts quickly.

So that might be a little more automated way to to approach it.

I really like that. And maybe as you as you go through those, there will be small design changes that happen. And then once you get to a point where you've implemented those changes and you feel like the design is stabilizing a little bit, maybe that's when you start that relationship, even if it's not going to be an immediate purchase from an injection molder, but just to get their thoughts and ideas early on, but not so early that you damage the relationship.

Yes. Yep. And I think if you just limit it to one part that's. That's fine. And what it does is it lets you understand how they work, how they think, what sort of feedback they're able to give. So I think that's a very fair thing to do, or it's very double sided.

That makes a ton of sense. Okay so that is a perfect I feel like, download of how to deal with injection molding. Obviously, there are a lot of other manufacturing techniques that work. At scale could you speak to the ones that you've personally experienced and either had really good experiences with or really bad went in too early, too late, just I'll leave you with that.

Other Manufacturing Techniques for Scaling

I know it's a little bit broad and I just wanna get your general thoughts and we can dig into the details.

Sure. Yeah. So the ones that we bump into is injection molding is ubiquitous. Most consumer products have some form of plastic and you can do a lot with it. We'll also typically see a lot of compression molding, And a good example of this would be the buttons on the remote for your TV,  typically soft rubber.

It uses a thermostat  process instead of a thermoplastic, so you have different materials, typically soft materials where you want to change the durometer. And it's cool the way it works. Instead of a high pressure application, think of more like a waffle iron, where you'll lay down the unvulcanized rubber, and then over a course of eight minutes, the waffle iron will open and close a few times to get the gases out, and then basically create all that crosslinking. 

Which gives you your final product. And the cool thing here is you can do different colors that are just beautifully welded together and you can do different wall thicknesses.  So if you want to have really thick buttons, but very thin connective parts that give you the springiness compression molding doesn't care at all about wall thickness.

So that's a nice, thing you can do with that. We talked a bit about CNC ing or machining, which is great, and you can mill or turn, or sometimes with these five axes, it's not clear if it's a mill or a lathe because everything's spinning. But that lets you typically use metals that might be hard to process otherwise. 

Another one that comes up all the time is die casting, which in some ways is like injection molding, but for plastic,

yeah. for

injection molding, but instead of for plastic, but for metal, and it's often magnesium or aluminum or zinc for the alloys that you can use there and with die casting, you do typically get a little bit more of the injection molding rules that you have to follow.

Frequently for the die casting too, there'll be a lot of post processing. So as the parts come out, you may have to grind away the sprue or drill some holes or tap it, which would have been hard to do. So that's it gives you a little bit more freedom, but all those extra secondary ops are going to add time and money, and also introduce a potential quality flaw.

So your scrap rate might go up. 

In terms of other cool ones, blow molding is great. If you want to make something hollow. So we've all seen those red gas cans around the Jerry cans. Those are most likely bull molded and basically you take a tube, which is called a Persian. It drops down between a clamshell on the clamshell shot seals it and then a needle injects air and it just blows it up to take the shape. 

Another example, this would. Be like a soda bottle, but there's a little unique twist to the soda bottle is that it's both injection molded and blow molded. So they'll, because it's a massive undercut, so they'll mold what looks like a test tube and then blow mold the shape to get the really thin wall thickness and the part that holds.

But if you look at the threads, that precision requires injection molding molding is pretty cool.  And we also see roto molding, which. It can be used a few ways. You can do like massive kayaks where you just put in the the pellets,  basically it's a shake and bake or more like a rotisserie where it just gets hot.

And

insane how large they are. If listeners should definitely Google rotomolding machines it, if you look at it, you would all, my first thought when I saw a rotomolding machine was how is this economically scalable because it's just, it's this massive equipment that goes into a massive oven.

And the one that I saw at least, and it's just, it's fascinating. 

It is, yeah, I think the cycle time is maybe 15 minutes at least and I guess similar to compression molding. The trick there is having a lot of cavities so that you can paralyze, whereas injection molding is very serial.

And that this is also very much for hollow

That's right. It's a great technique. Whereas you can't easily injection mold a hollow part, you'd probably have to make it like a clamshell.

Exactly. Yeah.

They also use rotomolding for doll heads in hands and feet, which is really freaky, but they'll basically print the sculpt of whatever face you want, and then electrolytically grow a shell around it, melt the wax of the sculpt, and then they inject a little PVC.

Shake and bake it. And then when they pull it out, it's like this alien giving birth. The whole face pops out and it's pretty horrifying to watch but nonetheless, very cool.

I'm definitely okay skipping that video. 

Yes. Yeah. That's a, that one's something to see. But yeah, that's used in doll faces all the time. Then we have like standard sheet meddling, which is often just done with a strip of metal coming in off a reel. And then a progressive dye that hits it progressively get whatever shape you need.

So a good example would. That would be like the back of a Tonka truck. The bucket of the dump truck is a sheet metal. There's centering, which is basically squishing a lot of basically metal dust into a final shape and then also forging, which is basically banging metal into the shape that you want. 

And then another one, which I haven't done, but I've certainly watched a lot of YouTube videos is how they make pots and pans, which is deep drawing. And basically, the idea there is you take a circle of whatever material you want. And then there's the outside shape of the pan or pot and then a plunger that grabs the material on the outside so it can't go in and just pushes it right in and the metal kind of cold flows around and that's magical and yeah, that, that's a really cool YouTube video to watch. 

I, yes, this is, I think going into why I started mechanical engineering. It was just watching these videos and how it's made and seeing how fascinating all of these processes were. And yeah, these, I think these techniques are important to be aware of because a lot of the times when Hardware, hardware startups begin.

The scale part is feel so far away. But at the same time it's looming over you because you need to eventually get to the point where you can engage in these steps. And it's really specifically when you are not a manufacturing expert, you might not even know that certain processes exist.

And then once you do know they exist, you don't know how to prepare for them and how to engage in them. So I guess like in terms of technological readiness level, so diving into a little bit more on the product development life cycle, right? Where there's there's the first kind of  understanding your requirements, understanding what problem you're solving, and then making designs for that, having conceptual designs for that, and then turning that into a prototype and testing and so on and so forth.

How many in your mind, like, Where within that cycle is the best time to one, not only consider these kind of more mass production manufacturing techniques but also how can you implement them in each of those stages in a practical way without. Constraining your team because usually bandwidth is incredibly limited as well.

So I think that would be a useful tool for founders to say, Okay, we're in the preliminary design phase.  Eventually, we're going to get to this mass production scale. So what can we do right now in this preliminary design phase to think about that and incorporate it into our thought?

So I think there's a couple ways to to be able to do that. One is just based on experience after one has been down this path a few times, that knowing that early decisions cast long shadows you'll know how important it is to try to get some of this stuff right. And just be looking at it through a lens, even in the prototyping of geez, I think we're going to mold this part and stamp this part.

And things like that. But the trick is that takes experience and it's hard to do that if you haven't been down the path before. So to mitigate that risk, another option would be to bring in an expert for a design review. And she could just look at it and be like, Oh, have you thought about where the parting line is going to go here?

Or shoot, you've got some really tall walls. If we injection mold that. Draft is going to become a factor so that you don't need to solve everything there, but at least it's on your radar screen. The thing that always worries me is the unknown unknowns that to go down a path and then have a sort of an eye opening moment when it's too late.

So I think, yeah, just getting that voice of the table for DFM, which is the manufacturing, and then something I think we might touch on is design for assembly. How do you actually put the parts together is critical. And again, not everything needs to be incorporated upfront, but but it is good to have that voice at the table. 

So I think I think what I'm hearing is basically there, obviously there isn't really a solution to not having expertise and a lot of these teams especially in early stage startups could be pretty young teams that don't have the experience or have not worked in these. And I think you. Pointing out something super important, whereas within the kind of design cycle, sometimes because startups are going so fast and they really want to learn so quickly that there isn't really a stop stopping moment  where they're trying to figure out, okay, what, at what point do we want to bring people in to react to this?

And oftentimes there's a little bit of wait, let me make it perfect. Before, before I get people involved, but I think that thought. should be put on the side of the table. And the reality is the moment that you have something that you are prototyping and going through a few tests, I think is a good moment to have some type of design review, whether it's a conceptual design review or preliminary design review, and then having a a pocket of subject matter experts, whether it is a potential customer that you feel like you have a pretty good relationship with and knows that you're in an early stage and wants to get feedback or someone that could also be someone that advocates for the customer or someone that can at least bring the perspective of a customer.

I think that's important. And then secondly, the perspective of.  Manufacturability and someone that's done this type of manufacturability and then if there is someone that has done similar systems that don't exactly it's not an exact kind of one on one, but something that's maybe have a, has a different application, but very similar, I think bringing those people in to react to your design early on will be a really good risk mitigation effort in your design.

Do you feel like that's a good,

Oh, absolutely. Yeah, I think if you can just make your unknown unknowns known unknowns, they're like, all right, these are the things we don't understand yet, then you can plan for them.

Yeah, it all comes down to cost quality and schedule, and you don't have a good plan, you're going to suffer on some of those, and maybe your competitors are not.

So it's really important to, to have a plan. Of course, no plan survives contact with the enemy, but I see so many companies that build a beautiful product and stretch what they can do with the low volume techniques further than they should. And then they've got demand, they're building them, but they're paying through the nose for the product that should be a quarter of the price.

And then it's difficult because all their engineering resources are supporting that one. How do they break away and do the new one and then have a smooth transition? Whereas if they had planned a little bit better up front, or at least knew how to plan for it, then they could have a much better outcome.

And I see that happen all the time, especially for teams that haven't done this before. 

Absolutely. You bring up an excellent point. I think it's also not only that does the demand increase and the amount that they're paying per per unit increases, but there is a limit, there's a capacity limit that certain manufacturers hit when they're trying to get to  the volumes. But sometimes I've definitely heard of Of startups that were like this one client potentially wants 20, 000 of these and currently we are building 20 of them a week.

And so it's it's just, you get to that point where you are almost at odds with your reputation of being able to deliver certain things. So having this strategy of kind of a handshake from potentially like manual in house to manual outsourced to semi automated to automated and then also from not low quantity manufacturing techniques, medium quantity and manufacturing techniques, if they even exist.

And then high quantity manufacturing. It's just, there needs to be, as you go through your technological readiness level or market readiness level. Whatever you are measuring your ability to succeed in the physical world they, that, those kind of phase gates need to unlock these next steps in, in your manufacturing strategy and your assembly strategy. 

Yes. Yep. And I think the key thing that you touched on there is that you've got a plan or that there's a progression, as opposed to going and then figuring it out when you get there. Yeah,  because that's what's so tricky. 

Yep. Yeah. And it's it, there's that whole saying of you plan. God laughs. And it really is. You can plan all day long and then you'll learn a bunch of stuff through your testing and it'll change. But plan to replant is fine. But you do have to have some type of strategy going into it.

And when you try to when you just become a firefighter in your job, that becomes, It's. That starts to be very risky and also could lead you down the wrong path of a decision because it's driven by reaction rather than proactivity.  📍 I very much agree with you.   

Hi there. It's your host Sera Evcimen. This episode is sponsored by my company. Pratik Development LLC "Pratik".. I started. Pratik to help her to restart up like yours, where I personally partner with you to get your hardware idea out into the world. If you're interested in advising coaching or. Consulting feel free to reach out at hello@pratikdev.com you can find the email. email. in the show notes.  

 okay, so I guess, okay, we've talked about a lot of manufacturing techniques and not being able to scale  at the right time and so on and so forth. 

So I think this is a perfect and natural podcast break to corner you again  to ask you about either.  A single hardware horror story or some hardware, stor horror stories that you can share that are either related to this or not. But I'm I know that I always get a kick out of the ones that you share with us.



Hardware Horror Stories: Lessons from Roomba

So if you don't mind,

sure. Yeah I have actually two quick ones. One, as it relates to injection molding and they're both.  I'll just tell the stories because it was so long ago.

Uhhuh. 

And most people have heard of the Roomba. So the first one, when we were doing it ourselves, and this was back in early, like late 90s, early 2000s, we had absolutely no idea what we were doing.

And there were so many unknown unknowns.  We just worked really hard and were in China in the factory all the time. But we had some issue on a Friday night and we used to work till I think 10 o'clock every night and later on Thursday for unknown or for known reasons, but I won't get into them.  So we had a Friday night around 10 o'clock and we realized that there's some problem with the tool that like a piece of plastic wasn't fitting as it should.

And as a testament to what can get done, if you work really hard, we identified this at the time we had a great factory. They picked up the tool at 10 PM, drove it to the mold house. They worked all night to either machine or EDM.  To fix whatever the problem was. And then by eight o'clock that morning, we're making new plastic.

And thank goodness it worked perfectly. Now the challenge with that is we had no idea what they did and none of it was recorded in the mechanical database. So if we ever had to move factories or build another set of tools, it totally would have been impossible. Like the only bomb  that existed was in the tools.

There's Yeah, and that just happened time and time again.  So if you have a great partner, it is possible to do things that quickly.  Luckily, our volume was low and we we're able to do some quality testing. It's not that I would recommend that technique, but it still blows me away today that within 12 hours, we were able to make a change.

And luckily for us, it worked out. Okay. There's many times where you make a change here and it creates another problem over there. So we. Thank you. I wouldn't recommend it but we did good luck that time.  But yeah, the other Roomba story, which sort of sticks in my mind is we're had built, I don't know, maybe two or 300, 000 Roombas.

We're feeling pretty good about ourselves and wanted to launch a new version of the Roomba that was minimally improved. There wasn't anything like substantial, but different colors and a few different SKUs.  And for reasons I still don't understand, we changed the processor and it all, like everything worked great, but we decided to let's change the processor because it seems like a good thing to do.

And we did that and things were looking good,  but as one of the final tests, about three weeks before we were able to ship the Roomba, we put it in its playpen, which is basically a four by four piece of plywood with some two by fours around the side. And half of it was carpeted, half of it was a linoleum, and in the middle was a metal rail.

And we noticed almost instantly all of the Roombas were like flies on flypaper and just stuck on that metal rail and dead, like dead as dead can be.

Oh

And you're like, Oh, this is not good. This is not good at all. And especially because A, we had no idea it was coming. And B, it was three weeks before we were going to ship a lot of them.

So I'm like, Oh my gosh, what's going on?  And what we did is in investigating a little bit, we realized that the Roomba is like a giant Van de Graaff generator. So it's not only picking up dirt, it's picking up every electron it can from the carpet and storing it. And it's nice beautifully shaped.

top housing and at times like they would get up to 25, 000 volts that we could store in these things and whenever it drove over metal, it would discharge. In fact, it would, the spark would jump about an inch from the robot at all. And that which was spectacular. In fact, it didn't just jump, but I think it had seven.

It went from the main board down the wheel spring to another board and then to the metal like it bounced everywhere  and the old processor we had was very immune to static. The new one, not so much. So we're like, all right, now what do we do? And these guys were dead. There is no bringing them back.

There's no sort of they just totally got fried. So if you look on that Roomba, you'll see now everything is mechanically recessed, as far as it can be, to increase the gap so that we couldn't get a spark. So the DC jack used to be flush with the end, and somehow we were able to pull it in about an inch.

We did put some transorbs and do some other stuff on the board to try to protect it a little bit. But that's one of those where if we hadn't done that testing, that might have been the end. So test early and often is good. Build it in from the beginning. But that was just so striking with literally like flies on flypaper.

They're all just stuck,  stuck there.  So

that must have been a fascinating day.

yeah, that was a long three weeks. There's many of those that were long, but but we learned a lot. And also if it's not broken, don't fix it.

Yep. Definitely. Yeah. And also, I guess testing all the environmental conditions that potentially could exist is another one. 

Yes. Yeah. There's so many lessons that we can learn from that. But yeah, that was horrifying. 

Oh, wow.

And the only good thing there was they all broke. So the things that are so much harder is if it takes a hundred hours to break and only 5 percent of them break, those are much, much more challenging to catch, but it's much easier, when everything just breaks at once.



It's basically a system design problem at that point at the, once it's five units out of a hundred, it's just identifying why that happened and what the effects on yield and do you actually fix anything because it's,  I feel like that would be a harder call.

It is, and I think there's a human nature to try to look past it, be like it wasn't that bad. But but at the volumes we would ship five out of a hundred as a non starter.

Yeah 

yeah, and we did have plenty of those too, but at least this one was like a good practice as we were trying to, get our legs under us.

We knew we had to fix it.

That's a really good story. I love that.  Thank you so much for sharing. Okay. So going back into assembly. 

Design for Assembly: Reducing Part Count and Assembly Time

I want to I, we touched upon it a few times and I want to close off the podcast episode thinking about assembly. So assembly can happen in many ways. One in my mind is that one could be almost like I, I call it permanent assembly where it's welding , and then there's impermanent assembly where things are clicking into each other and you can actually disassemble it.

I guess you can disassemble welding too, but it'll be a little bit more chaotic of a disassembly than just clicking something off. I'd love to get your thoughts on.  Different types of permanent assembly that you can think of and then and we can also talk a little bit about design for assembly when it's when it comes to actually deploying something in a contract manufacturer and having technicians work on putting it together. 

Yeah. Yeah. So one that it's not not that sexy, but I try to use all the time is just glue and screw. So the nice thing is if you screw it together, you can typically unscrew it to work on it. And it's not, it's semi permanent but it's not as permanent as something being welded. And this is really good as you're ramping up a product because the rework is good or as we're getting today into better sustainable design.

Yeah. Potentially there's a replaceable module that you could do that with the downside, like screws cost nothing. So that isn't really going to drive the cost and it's an unskilled operation. If you design it the screw housing will guide the screw in there. It's yeah, it's an unskilled operation.

The trick is that these screw bosses take up a lot of space and they don't look very nice and they leave holes. So if you want to go beyond that, one of the ones that you and I were recently talking about is ultrasonic welding.

I'm a big fan of this. There's no consumable. You basically just use the plastic to weld itself to another piece of plastic.

It's very strong. So there's a lot of things to like about it. The challenge that sometimes doesn't become apparent until you've designed it in is one, you need to do a special design using something called an energy director, which is basically like a little tooth that melts to be able to stick the two parts together.

And often it might have a tongue and groove or some  sort of self aligning feature.  So that's one that you have to modify the design and know how to do that. And the second is you actually need to have the ultrasonic welder itself. And this can be quite a challenge. Typically building overseas, you would just specify ultrasonically weld this and it gets done and you're happy and you may not need to know more about it.

But as I found recently doing it domestically we don't seem to have as much knowledge, or at least widespread knowledge on ultrasonic welding. So you really need to find experts. And the machine itself can cost anywhere from 5, 000 to 40, 000. So that's another cost that, potentially have to incur.

And if you're not building that high of volume, it will be difficult to amortize that and make financial sense. So you'd want to think through that.  And then you also have to make sure the design is conducive to ultrasonic welding, where there's basically like two parts. There's a nest that holds the bottom and a horn that provides the energy to the top.

And you want to make sure that you can get the energy really close to the weld area.  So some shapes are more conducive than that than others for that.  And then once you have that, and it might be a, another podcast we could go into, but all the mechanics of how do you generate that energy? What's the right frequency?

And then how do you transfer it to the plastic and get it to weld? So there are quite a lot of different moving parts that you, if you're definitely doing it in the U S you'd want to understand, as opposed to just blindly specking out ultrasonic welding

I guess I have a question on that. Is, are there any constraints of what could exist under the weld? Because I know that I don't know specifically if you're, say, welding a housing to a, another plastic sheet or something and it has to have a PCB in it is there any constraints around that? 

Let's see. Sorry, there might have I lose my networks glitched out. 

I'll just, I'll ask it again if you want. 

yeah, cool. Just so we have a clean copy.

Yes. My, I guess my question is, are there any constraints around what could exist under the weld? I'm more so thinking around say you have a PCB board PCB is board say you have a PCB under there and you do a ultrasonic weld around it. Is it going to, is there, are there constraints of what you can put in or While you're doing it, or should it be because I feel like welding would be really useful if you can just close something off and it's permanently closed in there but if it can't, if some electronics or something else can't exist inside of it, I'm just curious if there's a constraint with that.

So you can definitely build waterproof enclosures like a clamshell that capture a PCBA inside. The trick is you don't want to cook the board. So you want to make sure all of that energy is going into the plastic and not into the PCBA.  And there are some tricks in terms of how you position and locate the board, because the, as you do the ultrasonic welding the two halves are going to collapse onto each other.

There's a hydraulic press or pneumatic press that you know squishes it. And you don't want the board to be the thing that's stopping it but you also don't want the board rattling around just because it's not a great customer experience. So what we've done is added a little bit of an elastomeric foam that basically will allow the plastic to squish and serve as its own end stop, but prevent any sort of rattling.

And with that, we've been able to get I think about 150  PSI of waterproofing pressure. It'd make a good scuba watch. 

cool. That is fascinating. And then, okay, so that, that's a really good method to be able to weld plastic to each other in a relatively permanent fashion. And then going into more on the design for assembly, are there any kind of watchouts or common pitfalls that you've seen that people do when it comes to designing so that something can be assembled in a contract manufacturer floor, even in the in house? 

Yeah, I'm a big fan of trying to reduce part count as much as possible within reason. And the reasons to do this are one, if you don't need a part, you don't need to order it. You can't forget it. It doesn't cost anything. So from a supply chain standpoint, it just makes life a lot simpler. And then in the US, our labor rate say is anywhere from  60 an hour to 100 an hour.

And because of that, you want to minimize the overall assembly time. Often fewer parts results in shorter assembly time. So as we're doing something, we look, within reason that, can we combine these two parts into one?  Or what assembly technique might be a little bit quicker. So like the glue and screw, that's going to take longer than ultrasonic welding.

So if you had a huge part within the bounds of what you can ultrasonically weld, that might lend you towards ultrasonically welding. But evaluating these things, there's a couple of methods that you can use to as well. One unofficially, I call the Justin Chan method, who is a person in  the Roomba team that taught me a lot of what I know about DFM and DFA.

And basically, with that, you would take a prototype  that you'd built, working prototype, put it on the table with your manufacturing partner, and they would basically take it apart with no input from you, piece by piece, and then put it together. And as they're putting it together, they're looking to see if one part can go in two different ways.

And if it can go in two different ways, are both ways correct, or does it need to be polarized? And effectively, can the product put itself together is the goal on what are the pitfalls. So I think that's really handy. The other more formal one is called Boothroyd Dewhurst, which is a great firm down in Rhode Island, and they've developed a methodology looking at reducing part count and assembly time, and probably more than we can go into now, but for example, If you're having a human assemble it, you try to design parts that are easy for the human to grasp and orient.

If it's something really small or hot or sharp, then that would be penalized.  And you'll just add up the amount of time it takes. And the goal is to try to minimize that within reason. So typically you'll look at your existing design and it's a spreadsheet exercise to figure out how long it takes and then using some of their techniques figure out oh I could combine these two operations or  if I made the part self locate that would give me an advantage and then you can end up with a better better design. 

I really love that. And I specifically really appreciate the have a technician from your contract manufacturer build it in front of you for the first time so that you can see what stumps them and what could go wrong because I often I've done this myself. And when I work with startups, I also say that initially, you should really be an expert in how you build this, where you have done it in house a few times, at least,  not even a few, I'd say probably Depending on how complex the assembly is, but like 50 to 100 times is probably a good idea, especially if it's mid complexity so that you understand what processes take the most amount of time and know that you are probably getting better at it.

And CMs, that might not be the case because they might switch technicians on you. And it might be starting from scratch. So the process times are going to be longer. And. I really appreciate having that insight, but also being able to, once the process starts being able to see, okay our design is pretty stable now but what we can do is you can deploy fixtures on the manufacturing line.

Say there's a tiny PCBA that needs to go into a little slot and the.  The technician is really struggling with either holding the tiny PCBA or the slot is it's a pretty press fit, fitted slot, so it's not super easily sliding in, you could have a stamper or something of that source, the kind of Precedent or something like that.

So I feel like you're definitely thinking more on the upstream, which is super important. And I feel like I've had to deal with designs that were not thought of in upstream and now downstream, there are issues that I have to solve. So I've created fixtures and simplifying the process for the technician, but I really love that perspective and there's a few things you can do along the stream of assembly, I think.

And it's a good call.

Yeah, and I think you're exactly right of trying to build 50 to 100 yourself, because then as the engineer, you realize where the pain points are before inflicting on somebody else. But the downside of that is you have all these assumptions and skills you've built up. And when you have somebody that's never seen it before, that's an intelligent, trained person, and they go out of their way to make the wrong assumption.

It's really insightful, because we often do see about a 20 percent year on year turnover of the workers. Of course, there's both like skilled and unskilled. I do mostly unskilled stuff, so you need to have the intelligence in the product. So it basically knows how to put itself together, but that takes a lot more work.

Whereas if you have a skilled worker, they're like a great technician that they can debug it and make up for a lot of the design sins that you might not be able to do in a higher volume product. 

Absolutely. That is excellent. And I think that's a perfect way to end this episode. And I really appreciate your insight, Scott. Obviously when it comes to mass production I trust no one else but you. And so I really appreciate it. I've had my little experience with mass production.

Mostly I've worked in industrial systems. So this is always very enlightening for me as well. Every single time we talk. So I greatly appreciate your time. 

Oh thank you for having me. This is a huge amount of fun. 

Thank you.  

Welcome to the end of the episode where we do the Too Long, Didn't Listen section so that we can get into key takeaways right away. All right, I'm gonna get started. So I already used this for promo, but I loved what Scott said, so I'll start with my favorite quote from the episode, which is, if you're building one unit and save a dollar, you save a dollar.

If you're building a million units and you save a penny, you save ten thousand dollars. So that's just something to consider when doing optimization in your design when you're in the early stages versus late stages.  Also talked about the majority of the conversation really centered around injection molding and the transition to injection molding.

So we kind of mapped out a really good transition where you start with inching. 3D printing, and then you transition to CNC for a little bit more stable designs where you want more accuracy. And then you go into a rubber or silicon mold for about 50 units. And then in injection molding, you transition from soft tooling into hard tooling depending on the maturation of your design.

Tooling itself is a project within a project, so it's really important to note that it requires design of its own, whether it's in house or external. It also requires strategy in thinking about cost of tooling versus cost of part and if that return of investment will be made. A rule of thumb is how many parts you need times the part cost out of injection molding.

So that, that can give you an understanding of if the tooling costs might be , a worthwhile investment for you.  And then  obviously design considerations are a focus on uniform wall thickness, avoiding undercuts, and including draft angles in your designs to make them injection molded friendly.

We go into details of the episode, you should definitely go back and listen to that. And then integrating manufacturing considerations early in the product development cycle to avoid future roadblocks. This is advice that keeps coming back again. Please do this. When you're developing your product, just think about manufacturing early on. 

Scott obviously suggests that startups should consult with injection molding experts early in the design process to ensure the feasibility and manufacturability with injection molding. And then also recommends different types of manufacturing techniques like compression molding, die casting, blow molding, rotomolding.

Rotomolding's a really interesting one. Highly recommend looking into YouTube on that. They're just these gigantic machines that get into, get put into an even larger oven. Highly recommend. And then he highlights the applications and considerations for scaling up. And then  talks about the steel safe mentality where you can always add plastic, but but to take away the steel needs to be welded and then remachined.

So that's a pretty big design change. And then finally we talked a little bit about ultrasonic welding. Talked about how it's a great way to weld plastic to plastic and it's great for permanent assembly but requires very specific design features and could be costly, so consider that going into that.

This episode was really about diving deep into the details of how you can scale certain types of products. We very much focused on plastics and that, those types of materials, but  listeners have different types of products they're doing, different materials they're using, and want dedicated episodes.

To uncover those types of manufacturing techniques, please drop us a line on LinkedIn or reach out to me personally on LinkedIn and I will make sure to have another episode where we talk about that part. So thank you so much for listening and I hope you enjoyed the episode.

End

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