MacroFab Engineering Podcast #180
This week, Riley Hall of Fictiv joins the podcast to discuss how Fictiv connects engineers and designers to job and machining shops.
The US Mint Denver produces 30 million coins a day. Denes, the tooling department manager, discusses with us how production at this scale functions.
Stephen is on the hunt for the next step in his electrical engineering career and shares the shifts in the industry and what employers are looking for.
Background/Career
Very High volume manufacturing
Working with Tier 1 Manufacturers
Visit our Public Slack Channel and join the conversation in between episodes!
Chrissy Meyer’s of Root Venture. At Root, she invests in deeply technical teams that are solving problems in robotics, hardware, industrial automation, AI, and machine learning.
Parker is an Electrical Engineer with backgrounds in Embedded System Design and Digital Signal Processing. He got his start in 2005 by hacking Nintendo consoles into portable gaming units. The following year he designed and produced an Atari 2600 video mod to allow the Atari to display a crisp, RF fuzz free picture on newer TVs. Over a thousand Atari video mods where produced by Parker from 2006 to 2011 and the mod is still made by other enthusiasts in the Atari community.
In 2006, Parker enrolled at The University of Texas at Austin as a Petroleum Engineer. After realizing electronics was his passion he switched majors in 2007 to Electrical and Computer Engineering. Following his previous background in making the Atari 2600 video mod, Parker decided to take more board layout classes and circuit design classes. Other areas of study include robotics, microcontroller theory and design, FPGA development with VHDL and Verilog, and image and signal processing with DSPs. In 2010, Parker won a Ti sponsored Launchpad programming and design contest that was held by the IEEE CS chapter at the University. Parker graduated with a BS in Electrical and Computer Engineering in the Spring of 2012.
In the Summer of 2012, Parker was hired on as an Electrical Engineer at Dynamic Perception to design and prototype new electronic products. Here, Parker learned about full product development cycles and honed his board layout skills. Seeing the difficulties in managing operations and FCC/CE compliance testing, Parker thought there had to be a better way for small electronic companies to get their product out in customer's hands.
Parker also runs the blog, longhornengineer.com, where he posts his personal projects, technical guides, and appnotes about board layout design and components.
Stephen Kraig began his electronics career by building musical oriented circuits in 2003. Stephen is an avid guitar player and, in his down time, manufactures audio electronics including guitar amplifiers, pedals, and pro audio gear. Stephen graduated with a BS in Electrical Engineering from Texas A&M University.
Special thanks to whixr over at Tymkrs for the intro and outro!
Are you a web that needs PCB assembly? Look no further than macro fab, we've got you covered whether you need a single prototype or high volume manufacturing including system integration McWrap is your solution. Use our easy online interface to upload your files. Get a quote in minutes and order your PCB assembly without speaking to anyone. A few days later, your high quality PCBs will arrive in the mail, visit Mac fab.com today and sign up and get started.
Welcome to the macro fab engineering podcast. I am your guest, Chrissy Meyer.
And we are your hosts Parker Dolan and Steven
Craig.
This is episode 180.
Christie Meyer has spent the past decade taking hardware projects from early concept through high volume production. After getting her MS in electrical engineering from Stanford. She led several projects at Apple and square and was a founding team member at Pearl automation, a vehicle technology startup. Recently Chrissy joined route ventures, a seed stage hard tech, venture capital firm at root, she invest in deeply technical teams that are solving problems in robotics, hardware, industrial automation, AI and machine learning.
So Christy, that's quite a background. How did you get started and all that stuff?
Yeah, well, first, guys, thanks. Thanks for having me. My background is actually a little bit crazy. I don't think it's the typical path for most venture capitalists. Most most investors and people in the VC world come from slightly different backgrounds than I do. But I, you know, at route we invest in startup startup company, so I know you mentioned this, but we do a lot of like dev tools and software and robotics investments. But like most of our investments are in hardware and manufacturing. And I've been like I've been a hardware not from the very very beginning like from very early on. So I worked on everything from like military satellites to credit card readers to iPods and that was that was back when iPods were still cool. Like they were still very much an N thing ever that
there? Were they own their own device besides being combined with your phone?
Which, which which generation iPod was it?
So? Well? Did I work on or did I Oh, yeah. Oh, well work on. Okay. So I did. Oh, gosh, I want to say like, four generations of iPod Touch. And one iPod Nano and the very first generation Apple Watch. But the first the first iPod that I actually owned was a pink iPod Mini.
Was that yeah, that was a little square one.
No, no, the square the square one was the Nano. This is like the big chunky one. Like if you look at it today, it's like I've still got I've still got it in my closet and miraculously like the things still still boots up and still works. You guys remember your first iPad?
I got had one of them. I don't remember exactly which one it was I remember it had a game on it where it would like play like 20 seconds of a song and give you like five choices. And you could guess what song it was? That's all I remember from my iPad. So
I haven't I the only iPod I ever owned was a shuffle.
Well shuffle is an awesome product. Like I feel like that's probably even though I never actually worked on a shuffle. That's probably my my all time favorite iPod. Like you could take that thing anywhere and you could be beat the crap out of it. And it survived like pretty much everything.
Okay, so I actually looked at the iPod Shuffle. The second generation is where I thought was a nano. Because the one I had was like the USB stick style.
Oh, yeah, that's way backpack. That's way back. Yeah, but like, let me Overmind back. So like the the story of like, how I got into hardware and how I got into engineering like it's gonna sound a little bit crazy. But um, I was nine years old when I decided I wanted to become an electrical engineer. So my my dad's old college buddies gave me one of those RadioShack build your own radio kits for Christmas. You guys know what I'm talking? Oh, yeah.
Oh, yeah. With little springs and you know, the 115 one kind of thing. Yeah.
Yeah. Yeah. I mean, I like where you follow just like the instruction buck and you're just like connect from spring a to spring B and spring B to spring C. And like when you're done, you've got this like barely functioning radio, like super super staticky. Right. Yeah, I loved I loved that thing. Like I literally spent hours locked in my basement like playing with it. So I like my nine year old self walked away thinking that that is what engineers do all day, like day in and day out.
Push components into springs.
That's pretty much all there is to it, right?
Yeah. Yeah, exactly. Just out of curiosity, do you still have that set?
Oh my gosh, I don't like I feel like my mom got rid of it when I was in high school. And I'll never forgive her for that.
I can imagine by now like, just like the cardboard just all falling apart. Rested springs and
Yeah, I bet it would still work. I bet it would still work today. That's fast forward a few years. So I did. I did study doubly in school. And I had like, I had a pretty fun run of internships while I was in school. So like, my first one was actually working on a steel pickling line. Have you guys heard of steel pickling?
I don't think so. I mean, I can't imagine putting IBM's in a jar and then putting them on sandwiches.
It's kind of close, kind of like almost almost no, it's it's um, it's like a process for you take like raw rolls of steel and like run it through super concentrated hydrochloric acid baths. And it just like cleans off rust and debris. So it's actually some pretty nasty stuff like that those acid baths are not. Not nice. But, but I mean, it was a freshman in college. So I was like, thrilled to have any job. And they, they hired me to build user interfaces for their their Siemens PLC, so that ran the pickling line. But at least I mean, that's what they told me. But I think I think the reality of the situation is that they hired me because I was like, the only person that they could find that was small enough to like climb under the equipment and repair it when things broke down.
That's how it always is. I did radio stuff up in Oklahoma, and like, they picked me probably cuz I was the youngest person to climb the towers.
When I, when I interned for the power company, and they opened up one of the large power transformers they sent me down inside the transformer. They didn't actually have me do anything. They just sent me down in there. And they're like, how is it? Like, Oh, thanks. I'm covered in oil. And it's like 120 degrees inside this. Just close the lid. Yeah, a little bit of hazing of the internet. Suppose
that you learned so much about transformers? Right?
They're hot noise.
That's what I learned. You never want to do that again? No, no, yeah. No. I mean, I loved it. I really did. It's like, I would come home, like covered in oil and grease. But like, it was, it was probably my first exposure to like real world manufacturing, like, you know how things actually get made and processed. So like, that was kind of like my first step into manufacturing. And I like I never wanted to turn back after that.
Just out of curiosity, what about it really stuck with you,
like building things physical and seeing things? Seeing things like happen in the physical world is just, I don't know, there's something really magical and there's something really, really, really special about it. And I actually like the next summer, I guess, I'm moved up in the world. I got a a desk job building like MATLAB and Simulink models for massive diesel engines. Okay, and I'm like, I hated it. Like, I hated it, like Simulink models was not my idea of fun. But like the highlight from that summer was the week that I actually got to spend working as an operator on the engine assembly line. So that's cool. Yeah, they opened it up to like any intern who wanted to take a stab at it could just like go spend a week working on the line. And I mean, they obviously like they put us on the stupidly simple stations like the ones that were like pretty much impossible to like, mess up in any way. But I mean, again, like I just I thought it was the most amazing thing ever, like being able to see real physical products come together and like being part of it was so much better than sitting at a desk building Simulink models.
Yeah, I'd say you would have rather been like testing the machines and that kind of stuff instead of simulating them. Yeah,
it I think it's I think being on like being on the manufacturing side and and the testing side is is like and being able to see things come together it it really takes like, it takes things to a whole nother level and like if you're isolated and just sitting down and focusing on design, and I don't know, like design without manufacturing is nothing right? Like what's the point?
It's software, right?
architecture, software, you mean just typing?
Oh,
I wasn't gonna be that many, but architectures very similar to that.
So you but you have worked on a handful of design projects, right? Yeah.
Yeah. Well, so my, my first like real job out of school has was like, writing test scripts for military satellites. So like, I was just, I was a lab rat. I sat in a lab just like day in and day out and wrote Perl scripts that would run on these like, giant test racks full of like, top of the line. Xilinx FPGA is like, it sounded cool. But honestly, like, it got pretty, pretty boring after a while, like I learned pretty quickly that like government work can be slow at times.
So who picked pearl? Was, was that something that you picked? Or was it someone,
it was all the rage at the time? Like, I'm trying to think this was like, Really, before Python started taking off. And it was it was pretty, pretty fun to like, try to see if I could write like the entire script. And like one one single line.
Would it be done in Python? Now?
I would assume so I don't know. I don't know what their what they would be using. I don't know what they'd be using now. But it's like, the thing that killed me is like, after I got bored with writing Perl scripts, I moved to California and went back to school. And right after I went back to school, I found out that my satellite project had been cancelled. Like, completely scrapped. So that was like one and a half years of my life, like effectively down the drain.
Were there any signs on the wall about that?
Not when I was there, not when I was there. I like I think that there were some significant like Congress budget cuts right after I left. But yeah, it's unfortunate how that works out.
So what kind of what kind of tests you run on satellites?
Like, top secret? Like?
Yeah, I guess so. No, I
mean, the idea was like to stress the hardware as much as we possibly could. And I think that's like, it was basic validation to make sure that like the designs were working like as, as expected, and things were meeting spec, and then also just like running as many stress tests, as you could
write the old rule that if there's a spec written, then there's a test to test that.
Yeah, pretty, pretty much, pretty much.
Well, so then you went back to school? And what did you focus on for your masters?
Yeah, so I actually worked on wireless sensor networks. So I, like took these little notes and connected them up in like a network around my, around my dorm room around my apartment, and then worked on that different protocols to try to make, like, try to increase increased throughput and reduce latency. Which was fun, which was, which is fine. I mean, I loved I loved the stuff I was doing on the wireless side. But it's also really funny that like, around that time, I got an internship at Apple, and this was like, this is about the same time the iPhone came out. So I interned for the iPod team. And like, pretty, it was pretty clear from like, day one, what I was going to be doing was like, nothing related to what I was working on at Stanford, like no wireless, nothing related to like protocols for like, wireless mesh networks, you know?
Well, okay, so what did they have you working on? So
so when I joined, it was actually a it was an iPad, new tech team. So we were looking at a bunch of new technologies, things that were probably a little far out that they weren't going to be like built into products anytime soon, but things that like if they worked out, it would be like it would be incredible and I think at the time like iPhone was really starting to take off so iPod was kind of seen as like, iPhones little brother and as a result because like volumes were lower. It was the testing ground for a lot of new technologies. So like before something got booked into On iPhone, we would try it out on an iPad program where volumes were a little bit lower, and you could handhold the suppliers a little bit more closely. So I did that for a summer. And we also did like, we also did a lot of like competitive tear downs. Like I remember taking a part of zoom. Back go and zooms were a thing.
Are they still ran out? Just now? I'm
on? Unfortunately, they're not actually like I had a zoom. Oh, no
way. Did you really? Yeah, I
actually liked it.
I'll try not to hold that against you.
So on doing the tear downs, I, I imagine that you got a lot of insight into like manufacturing and that kind of stuff.
Yeah, it was. It was fun, like trying to figure out why things were designed the way that they were. And like, you know, you come to realize that, that so many of the decisions that people make when like laying out a board or like architecting a product or doing mechanical design, like it's it's purely for manufacturability. Right, like, so many decisions are made along the way to make things easily manufacturable and like consistently manufacturable at really, really high volumes,
which that's something I'd really like to get into. Because you have plenty of experience with large scale manufacturing, right?
Yeah, yeah.
So I mean, I think that would be really awesome to kind of go down that route and sort of crack that egg on, what does it take to make a million or 10 million things?
Yeah, I like this is something that I love to geek out about, it's just, I think that there aren't too many places that you can go in the world where you get, you know, a bootcamp or a crash course on, like real design for scale, and like design for like high volume manufacturing. And I think that like, a lot of people just don't realize that the way that you the way that you approach design is is totally different, like totally, totally different. If you're going to build 10 of something, or 1000 of something or like 10 million of something. And I don't just mean like the decision to like 3d print something versus machine something versus injection molded, something like that. I mean, that's important, too. But like, I mean, done, the way you lay out your PCB, and like the types of issues that you care about, and the issues that you don't care about. I mean, the biggest shift and thought process is like, when you're building 10 million of something, you actually really care about the point 1% Failure Modes, like really care. And I'm not I'm not trying to shit all over Samsung, but like, if your battery blows up, point 1% of the time, like that is definitely not okay.
Well, yeah, it's like 10,000 plus people, right, and now have their pants on fire.
Or worse, or worse.
Yeah. Right. And it only takes one to get in the news, right?
Yeah. Yeah. I like the other. So the other crazy thing is like, when you're shipping 10 million or something, it's it's not super uncommon to hit silicone bombs that like nobody has ever seen before.
They make an irata data sheet just for you. Well, you're always
like, wait, wait, like, this isn't listed anywhere on the Rada like No, it's not because the chip maker doesn't know about it. Like those are not fun problems.
Right? You're hitting their point. 1% Exactly.
Yeah, exactly. Which I which and those, those types of things are really hard to reproduce. Like I've literally had times where I spent the night overnight in the office staying awake, sitting up next to racks of like hundreds and hundreds of devices that were running stress tests, just trying to catch one silicon bug. And we had like we had everyone on the team taking shifts staying awake through the night because if a device had the bug, like you'd have to grab the logs off before they were overwritten and you'd have to like stop the test to try to preserve it in that that state like before the whole the whole device rebooted. So yeah, it's like you hit these these really tiny problems and and it's just like that silicone suppliers are not are not even aware of. Anyway, I'm going off on a tangent here but like when you manufacture in like high volume you should expect to run into like issues that no one has ever ever seen before.
So where if you if you're a hardware design team, designing their first couple products So, when do you start thinking about this kind of scale? Like, when do they need to start going, Okay, I need to start designing for this high volume kind of stuff,
oh as day one, right as early as possible as early early as possible. So it's like, you've got to keep it in mind. Like, while you're actually architecting the product, like when you're selecting your components, like when you're figuring out how you're gonna, like, piece everything together. But then there's also like, you know, you people talk about design for manufacturing, like, you can also, you can also do a little too crazy. And you can do, you can do a little bit too much of it too soon. Like, I think there's, there's a balancing act there, people like to say, oh, like, there's no such thing as doing design for manufacturing too early, like, you know, start DFM right away. But the reality of the situation is like, DFM is often a process that, that you can't do, like alone in a vacuum, like you're having to go back and forth, and back and forth with suppliers, and you're having to like, figure out what their technical capabilities are and what they can do. So it's like, if your design is changing, and like a big way, and it's, it's like, doing so frequently, then you don't want to waste everybody's time, right? Like you don't, you don't want to waste the time of like, you know, your entire supply chain and get them upset and angry at you before you even get started on the project.
So actually, with the design for manufacturing, when does the manufacturer actually come into that? Do they? Are they part of the design for manufacturing from the get go? Or do you kind of do the first couple of rounds and then present to them what you have?
Yeah, usually, usually, like you want them involved from from step one, like the moment you start DFM, like, the first thing you should do is talk to the manufacturer. And that's, that's both like the manufacturer who's going to be SM ting your boards, or like assembling your final product. But also like all up and down the supply chain, do like whoever's injection, molding your parts, like you want to be talking to them as well, or whoever's, like die casting your frame, like, it's important to, like, involve them from the first from the first step in the process. Because ultimately, like, you know, you can design whatever you want, but they know their capabilities, they know their their equipment, they know what they can do, like far better than, than you ever will. So you have to lean on them early to be able to tell you like what's okay to do and what's not okay to do.
They will they will also know their point 1% Failure Modes and assembly and stuff and help you work around those. Right.
And like I hate, like, I hate to admit this, but like, there's been like, way too often. I've gotten the DFM feedback from like, you know, a flex supplier or somebody that like suggests, like 20 changes, and I like look at it, and I like roll my eyes and I'm just like, ah, like we're on a schedule. Like we don't have time to like make all these changes, just like it'll be fine fab it, just go ahead fab it, it'll be fine. And then like, sure enough, like, it all blows up in my face. And I'm like, Oh, I really wish I would have listened to that.
Oh, yeah, I've certainly experienced that before, where there's a customer who they're ready to go, they provide all the files to you. And they're like, Okay, we've got a, you know, our money burning a hole in our pocket. And all of our design flaws are here, we just give it to you, and you make it happen. And I look at him, like, none of this is ready. Is even slightly ready. And then they're like, well just push forward. And I'm like, okay, hold on, this is gonna be fun. So your
money? Your money? Yeah, yeah. How do you deal with that? Like, how do you talk? How do you like, convince customers? That No, really, you should go back back and make these changes to your design?
Well, you know, with with anything, I think there's, there's, I think you have to weigh the options, pick your battles, maybe you know, that list of 20 things, I guarantee you not all 20 of those things were 100% critical, maybe three of them were go and rank them and say like, look, we have to do these three, we would really like the other 17. But we'll be happy if you give us these three, you know, work it out, work it out that way and and build your case and explain it well to the customer and explain why it needs to be that way. And what they can expect if they don't do that.
Yeah. Do you ever make the changes for them?
I try very hard not to ever modify a customer's file unless they have given me explicit ability to do that. You know, unless it's something very minor, like moving silkscreen on a PCB or something like that. Yeah. Yeah,
I've found that when we come to customers with with issues like that, they are happy. And like, they usually always take our advice. It's very rare that they say, you know, don't care, though the money on the fire. Let's do it. That's very rare.
Yeah. Well, and everyone's under the gun, right? There is no such thing as like, just an open loose timeline. Everyone's always like, under the gun.
Yeah, yeah. But like, the thing that people that people have to realize is that if you don't fix it up front, like, it's only gonna get harder to fix, like, the further along in product development you go, right?
Always.
And it's just like, So, point point. 1% problems matter when you're shipping 10s of millions of things. So like, even even if it's one of those like, slower priority manufacturing issues, that that's flagged by the like factory, you really have to go through those with a fine tooth comb, and you really have to, like, make sure that you're addressing their concerns, and you have to trust them, right. Like, you have to trust them that they understand their process, because if it turns into a point 1% failure, like, like, it's not okay, and like, heaven forbid, like it, you know, returns and recalls of products are like really expensive. Like, really, really expensive. I feel like I feel like you know, expensive from a monetary perspective of like replacing a customer's unit and processing an RMA is like, it's expensive in the moment, but like, when you're a company like Apple or Google or, or Amazon like, failures in the field are, are expensive from like the perspective of like, public perception and PR. Because all it takes is like one person's blog post to go viral about some issue they found with their phone and like your stock price can drop instantly. Like you can quickly wipe 10 billion off your company's market cap so so like an issue that causes a product to fail like point 1% of the time, like really isn't really isn't acceptable and like even more importantly, if it's like a Layton failure like meaning it's not caught before it leaves the factory like that's really not okay.
Yeah, armies and stuff like that always remind me of the airbag that was the khaki Takata I think airbag issue with all the Army's well not arm is recalls. It's like, I mean, you got to think of every single vehicle that has one of these airbags in it, you're paying a mechanic 300 $400 to swap it out, that doesn't include the parts. That's like you're talking millions, like, I think it was something like, like 20 or 40 million vehicles were affected. And they're still repairing vehicles that day. It's like, a decade later or something like that.
Yeah, that's insane. Like, that's the last thing you want to happen. Right? And like, I mean, sometimes, I mean, product development is tricky. Like, I don't want to, you know, I don't want to make it sound like they did. They, they, you know, knowingly did a million things wrong, but at the same time, like, you've got to be super diligent, like as early in the process as you can, and you got to test the test the heck out of everything to try to avoid problems like that.
Well, okay, so So what kind of checks and balances do you put in place to prevent these kinds of things?
Yeah, I mean, it's so I, I think that like, you know, you can test on the line as much as you want. Like, you can test like, you can do like functional testing on the line, you can, you know, you can do like, oh, man, I'm forgetting like the name for it, but it's like, effectively, like, tested as a customer what at the end of the line, but like, really what it comes down to is is like, even more importantly than that, and even more importantly, that like benchtop validation, like electrical validation, like on the bench, like you've really got to beat these products up, like really got to beat these products up. Like you've got to drop it from like a meter on the granite over and over again until it breaks and like you've got to send it to like the highest you know, reasonable temperature and the lowest reasonable temperature and like do that like hundreds of times like I think that like my all time favorite is playing with the ESD gun.
That's about the actually mentioned ESD. Oh, yeah. Unfortunately, yes.
Yeah, like that's my favorite like that. That is fun. But like, the idea is that you don't just like zap it to see if you'd like pass a spec. You test it. You failure, like, yeah, you abuse it. Yeah, you keep testing, and you keep testing and like, you keep increasing the voltage or like, you keep increasing the temperature or the number of like, times you're dropping it onto granite or whatever it is, until it fails. And then you figure out what your weakest link is. And then like, make it better next time, like, make it better for the next iteration. It's like the I feel like the secret to shipping and like really high volume really lies in the amount of testing that you do to make sure that your product is going to survive, and that it's going to be robust. And like, the more you can like front load that testing, like early in the development schedule, like the better off you'll, you'll be.
So I got a funny story about that is because I work with a couple badge like conference badge makers, and one of the groups they test by giving it to their kids, because it's, it's they think it's a very good analog to give it to a bunch of really drunk people at a conference. So if it can survive the kids, it can survive the conference. For testing, yeah, so it seems to work pretty well.
That's awesome. I'm gonna start doing that with every every product I touch from here on out.
How about designing for assembly when we're talking about very high manufacturing?
So is there a difference between manufacturing DFM, then would that be DFA?
Yeah, I mean, typically, DFM refers to like, the parts like the individual components. So like, you know, def, I'll do DFM on my like injection molded piece of plastic or like my stamped metal part or whatever, whatever it may be. But like DFA refers to, like putting all the pieces together and making sure that they work, work well together. But it's it's weird. The terminology is weird, like I've always used DFM for effectively, all of that.
Yeah, that's what I would say, I would be like DFA DFM is all of that. Because at the end, you're still manufacturing the entire thing. Yeah.
Well, well, okay. So the distinction I might put on that is DFM might be something that your engineering team in house they do to the design that they created. And DFA might be something where you rope in your manufacturer, and you're discussing it together how it's actually accomplished. Yeah, I
don't know. I've always thought about like DFM, even even DFM I've thought about as like, you're working with the supplier directly. But it's not like your final it's not whoever's doing your final assembly, it's like, you're working directly with the injection molding house or, you know, the diecast better or whatever. So it's not like a purely internal thing. But I also like, the weird piece of this is that like, when you talk about PCBs, right, like people talk about people say DFM Right? Like, it's really an assembly, like you're placing things on a circuit board, but people still call it DFM and not DFA.
So we should just flip them.
Pretty much.
But but the thing is, like when it comes down to, you know, your your team of engineers make a decision to put, I don't know, an extra screw on something for 100 units. Okay, that's something but for 10 million units, that's a lot of extra work for whoever's doing the manufacturing, and that has a lot of implications on things. If the decision comes about, that you were adding an extra fastener or whatever, how does that get relayed to the manufacturer or the assembler? And what kind of how do they gonna roll back and say we like this or we don't?
Yeah, like, that's, that's a really tough part. And like, the truth is, it varies like, from organization to organization, like, I wish that there was like, one set way that that it worked for, like most, you know, contract manufacturers, but like, you know, the, the reality of the situation is, like, you know, when you're dealing with high volume, and when you're dealing with like top tier manufacturers, like you've got to know when to put your pencils down, like when you should not add that screw or when you should not add that fastener. And like you really need to like stop making design changes and it becomes really sensitive, like really, really sensitive so like, I always found that it works really well to have like one point person whether that's a project manager or a systems engineer or like I you know, and an engineering manager or like some factory lead, I don't care what you call it, but like one point person that is like, ultimately the decision maker on at any given time, like whether or not a change gets in and if so, how it gets communicated. hated how it gets communicated to like the factory because, like, keep in mind that these like top tier axioms they're, they're, um, they're like optimized for continuous production. Like once you get production started, they run like a well oiled machine but like, you know, getting to that continuous production state can be really, really hard. And like one mistake that I see a lot of startups make. And like, especially now that like I'm on the investing side, like at root, we have a lot of a lot of hardware startups in our portfolio. And like, one mistake that I see people make is that they're moving as fast as they can. So instead of going like pencils down on a design, like you're still making pretty major design changes, you're adding screws and fasteners, like as you enter mass production. And I mean, I'm sure you guys have seen it, like, I'm sure you guys have lots of war stories, for sure.
Get when you get a new sets of Gerber's or component lists, and you're like, it's already halfway through production, guys. What are we what are we doing here?
So we did did Apple have like a single point man, for entire, like, the i Whatever iPod number. Yeah, there's a single pointment on it.
Yeah. And that's what, that's what engineering project managers were. And that's why like, I mean, I was an EPM, and engineering project manager at Apple and, and I loved it. Like, it's, it's fun, and still technical. But at the same time, like, you are the person that is like, in charge of all communications with the factory, and all decisions related to the factory, up until production. And then once you get into production, there's this shift, and then this, the operations team moves in. And they they take over and it's their responsibility from there on out because, you know, as an engineering team, you like move off to go work on on new, new projects. But yeah, like, it was pretty clear, especially in like the early I've had days, like, there was one point person, and if you wanted anything done a design change made like something had like something to be done an experiment run or something happening in the factory, like it was it was one point of communication. Which was great. But it also meant, like as a PM, I was on the phone at like three in the morning. times that I can count, right?
I wouldn't imagine you'd have much time to do any design yourself. You're basically putting out everyone else's fires.
Exactly. Exactly. All the time, both metaphorical fires, and occasionally a literal fire. Though, no, I mean, it was it. That's That's exactly how it worked. And it was firefighting, like it wasn't a design role. Like I was not sitting down and doing like iPod schematics. And part of me missed that, like, all of my education was like, super, super technical. So like, part of me missed that. But at the same time, like the thing that I really liked about the role was that, like, I could be building a product development budget one minute, and then like, negotiating with a supplier the next minute and then like in the lab, taking scope shots, trying to like debug some, like crazy issue that we found, like, the next minute, you know, it was like, never a dull moment. And like, you bounced from thing to thing to thing, and you got exposure to so many different areas. And like I really wouldn't trade that for the world.
So I got a question for you. That's Apple related. But but kind of a little bit off. The I actually have a customer that has asked for Apple's level of quality for their surface finish on like MacBooks and things like they'd like that that bead blasted process.
Pretty, pretty expensive.
Yeah, yeah. Well, and that's what I that's what I told them, I was like, you want that level of quality in the, you know, 100 to 200 piece. I'm amazed that Apple is capable of doing that in the million range, you know,
yeah, but the the thing that you have to like remember is that when you're working for and this is like any big company that was this is like not just apple but like any big company with deep pockets, like they can afford to start production with 50% yields on something right sure. Like they can they can they can afford to like throw away, you know, to in every 10 iPhones bill like it's it's, it's not ideal, but like they can sustain the For a while, because they have like an infinite pile of cash that they can that they can tap into. And like if you're a small startup, or you're a smaller, smaller midsize customer like, like, I'm sorry, you can't afford to do that, like you can't have Apple Apple level, cosmetic finish on your on your parts, because it's prohibitively expensive. And you don't have leverage, like you don't have leverage with your supply chain to even push for that, like, you have to go in under the assumption that like you're nobody to them, right. Sure, sure. Beg beg for for, you know, for somebody to even like crank out parts for your product.
For sure, and I wouldn't, I mean, you can probably speak to this, but I wouldn't be surprised if Apple had massive quality documents, right? Oh, yeah. Well, I mean, they how things are supposed to look?
Oh, yeah. And they have they have armies, like armies of quality engineers, that are like, that live full time in Asia. And like, they're, they have, you know, very specialized quality engineers, too. So like, you know, there's, there's, you know, dozens of quality engineers that just focus on batteries are just focused on displays, or just focus on silicone or connectors, or, you know, speakers or whatever it may be. So, like, there's super specialized quality engineers, and, you know, hundreds of them for each particular part and component. So, so those people live on site at the suppliers, like they're there day in and day out, but like, no startup has, that level of resource is like no startup can, can can go can go make that happen as much as they'd like to.
So to speak in a quality control, testing and validation. I guess it depends on your product, but high volume, do you do you still do every unit? Or do you do batch? Or do you just kind of spot check it or combination?
Yeah, all of the above, like all of the above, like, when you're talking about incoming quality control, there's some like pretty clear sampling guidelines. So like AQL levels, and like, you know, tell you if a batch is like, so big, you should inspect, you know, X number of parts, and this many, you know, failures are allowable. But when you're talking about like outgoing quality, it's like, there's, of course, there's always going to be like, things that are tested 100% of the time on every single unit that goes down the line, because you need to verify, verify that the functionality is there that it works that like all your signals are connected, you don't have any cold solder joints, like you know, you've got to your your connectors are seated, like there's, there's tests that you have to run to exercise, like every single piece of the system, like every single subsystem, but then there's also going to be like, much, much more in depth quality testing that happens at a sampling level that you pull X number of units at random off every every everything mine and you go through a huge test plan, like, you know, exercise, exercise things a lot in a lot more depth.
Treat it like a customer, that kind of stuff.
Yeah, I assume you guys do a lot of you know, follow a similar process. Yeah,
it always surprises me, like the difference between electronic manufacturing and I was digging into automotive stuff because I'm a big gearhead but hearing because I talked to a lot of mechanics and and dealerships where they will get vehicles that are they have like 50 problems with them, like switches and stuff that's like you would think because, you know, electronics, will you test that kind of stuff? Yeah, I mean, factories, automotive factories, just don't test that science stuff.
That's crazy to me. And I think they just assume if it's together,
it's good.
That's scary. It's terrifying.
It might be that manufacturers automotive world treat dealerships as like the Final QC. But that's still like, man, it just feels weird. Well, it could also
be the opposite to it could be that they're like pushing as much as the QC, like, you know, upstream to their supply chain, as well. So that they're they're like relying on you know, whatever subsystem supplier to do, you know, a thorough level of testing so that they don't have to repeat it all. Final assembly level.
You would think that but there was a really popular posts online a couple weeks ago where like, a truck showed up at a dealership and it was a white truck and had a red door. And just like, Man, how, how does that get through quality control out of the factory? Yeah, not my job.
Have you read all the Tesla stories like people are seeing quite a few horror stories and even some of my friends that that, you know, ended up buying model threes have like been horrified at some of the quality issues that their their cars shipped with.
Yeah, I think that's it's the same thing, except the problem with Tesla is they don't have dealers to do their Final QC for them. Yeah.
Well, the other problem was with Tesla is that they're like, manufacturing final assembly line is like a tent in their parking lot, right? Oh, yeah.
I completely forgot about that. Are they still doing that?
I hope not.
I hope not either.
Alright, so quick question about getting involved with a tier one. This comes from just the idea of like, if you don't know a tier one manufacturer, how do you even reach out and like begin that conversation?
That's a good question. So all of the big tier ones these days have have dedicated like business units or dedicated divisions that are working that are like, dedicated to working with startups and small, smaller companies. So those are often a really, really good starting point. I mean, even even like Foxconn has one. But I like the thing that I would caution. People like that. I really think that that like, the most important thing to know about working with top tier factories is that you probably shouldn't be working with top tier factories. Like, I'm really, like, I'm joking, not joking. Like, I really, I really do think that like, hardware nerds, like myself tend to get like over excited by name recognition, like you say, like Foxconn, or quanta, and people just fall over themselves. It's like a, like a celebrity crush, you know,
for electrical engineers. Yeah,
it is, seriously is, but like 99.9% of the time, like, it's not a good idea to work with these factories, like a very, very rarely is, is a good fit. Like, I've got to be careful what I say I have close friends at all of the big top tier contract manufacturers, so I really hope they don't listen to your podcast. Getting myself in trouble.
So why aren't they a great fit for your new widget?
Yeah, well, I mean, don't get me wrong. So like, there are really great things about working with Fox cons of the role, like they have amazing process control, like I guess, I mean, they pretty much have to. And they like you also get like full access to their supply chain. So like as a small company, like you don't have the ability to work with like best in class injection molding to houses, but like, if if Foxconn is doing your final assembly, they will handle that relationship, like you can get best in class suppliers for like, pretty much any part and like whatever you dream of, they have it.
Or they can build it, or they can or they
can build it. And it's always, it always amazes me that like, you need like 1000 new human operators overnight, and they're just like, okay, no big deal.
They make those two
vertical integration. Yeah, but but like, the reason that I caution people against it is because it's really risky. Like, you have to realize that if you if you're not Apple, like if you're if you're working with somebody like Foxconn, and you're not Apple, you have to assume that you are that you're literally nothing to them, like they will drop you so fast, it will make your head spin. And like I've seen it happen to so many hardware startups like poor poor startups and like scrambling to find a new factory partner like right before you enter production is pretty much the worst thing that you can have happen to you. So like getting the A team at a smaller factory is oftentimes like, way better than getting the C team at a top tier cm. Like you definitely lose access to some of the resources they have, but like you get real support, like real, real support through like all the ups and downs of product development. So I'm giving you guys an opening here. This is where you're supposed to like insert shameless plug for macro. I'm
actually actually one thing we don't do on this podcast is the shameless plugs for microphones. Well, okay, so I think Chrissy, yeah,
I've heard something before. And I want to get your opinion on this, I've heard that you don't want to be, well, obviously, you don't want to be low on the list for your contract manufacturer, because they can just drop you at, for whatever reason, the kind of rule of thumb that I've heard is you want to be number two, or number three on their list, and you want the other people around that list to be about your size. So in other words, you still want it where they pay attention to you if you asked for something. But if number one leaves, they don't tank,
they don't go out of business. The whole the whole company that your CM doesn't tank, right. Yeah, right. Yeah,
no, I think that's I think that's spot on accurate. Like, I definitely, I definitely agree with that. Like, the one thing I would add to it, though, is like, make sure that there is like, especially if you're working with like one of one of the bigger names like it makes sure there's some strategic alignment. So like, what I mean by this is like, they have to have a reason to want to work with you. And the reason should not be like, they believe that I'm gonna sell a bajillion products, and they're gonna make a ton of money off of me like
that is, that shouldn't be their reason,
that shouldn't be their reasoning, like that's unlikely the case and like they know it. So there's got to be some other reason they want to work with you like, an example of it would be like, maybe you are the first time they're building a product that's going to be used underwater. And if they like knock your project out of the park, it will help them land other customers and like potentially bigger customers who are also making underwater products like that. That's what I call like strategic alignment. And I think it's like really important, regardless of who your contract manufacturer is.
So one thing on the macro crab thing is, because how Mac Feb works with the our partner network is we don't have that typical, like getting fired by your CM problem.
Which is amazing. Like, it's always been, like, it's always one of those things that like you are constantly on edge through like a whole product development cycle of like, you definitely want to push your factory like you want. You want to build a great product, you want to go this fast as you possibly can. And you're kind of like pushing them to the extremes, but you don't want to push too hard. Because you know that if you push too hard, like they've got other customers then like, they could just walk away and like it's a very real risk. So like, having not about that risk is Yeah, it's like pretty valuable.
Have you have you ever been dumped by a cm?
i Okay, so I have never been dumped by SEM. But I have come close. So many times. Like I like I've actually I've had them dumped me. And then like, this was when I was consulting, but like I've had them like dump me and the team I was working with. And then like, two days later, like, change their minds. So
you were on your like number four or five customer for them? No, I
think we were at the bottom of the list. But like, that's an emotional rollercoaster. Like, let me tell you, like, I spent that those like, the, like 48 hours between between between when they dumped us and when they decided to like, take us back. I spent like most of those 48 hours, like calling every single person up their management chain and like spending time on the phone with like the president of the company to like, convince them like why we're a valuable partner and why why were strategically aligned. You know, it's like one of those like crazy emotional roller coasters that you're like literally awake all night on the phone trying to convince the president of a company to take you back.
Sounds awful.
Just just like just like high school dating, you know?
It's it's funny to think about that way because like if you go to the grocery store, you're gonna be able to buy groceries in the cashiers like now you're not going to be able to buy groceries here.
Also Also if you're dating in high school, you kind of want to be number one on the list. You don't want to be Yeah, I suppose. So. So in the in the notes you had written that you wanted us to ask you about some pet peeves of manufacturing. You want to go ahead and share some with us.
Like oh, they're they're definitely there's some things about manufacturing that I love to hate and like you probably shouldn't have asked me this because cuz you're gonna get me started on on ranting and I don't know if you're gonna be able to get me to stop.
Oh, this is perfect. Please go ahead. You sir. Yo. Yeah.
So so first thing on the like top top few things that I hate about manufacturing is, is glue like glue is the bane of my existence yet somehow, like every freaking product I've ever worked on has used glue at some point along the assembly line like have you guys have
to guess is it because it's unpredictable?
Well, it's unpredictable and like you spend hours you spent hours getting materials kit and like people operators brought over the line and like test stations up and running. And like, finally everything is like ready to go. And then the damn glue machine gets clogged like again for like, the 10th time in two days. So you have to sit there and like wait for another hour for them to disassemble the whole thing and like, find a new nozzle. And then for some reason like that nozzle is like only stored in a warehouse that's a mile down the road. And the one person who has the key to that warehouse is on his afternoon smoke break
like this is this is oddly really specific here.
This has happened to me so many times. Like it's crazy, like I can't even count. So like when it comes to glue. I just I don't care how you dispense it. Like I don't care if it's by hand, I don't care if it's like automated dispense on some fancy XY table. And I don't care if it's straight up Loctite 495 or some like super fancy two part epoxy. Glue always, always turns into a disaster. And I feel like I have spent more time in factories waiting on glue than anything else.
I just don't like it from like this assembly aspect either. Like if you need a fix, like especially because you know, you know, I've never built a million things, but you know, a couple 1000 It's like, oh, I need to repair some of these things. Oh, it's glued together. Nevermind.
Yeah. Are you see the those? You're the customers were constructions come over and they say apply superglue to an area is just like, Oh God, this is going to be awful. Especially
when it's right next to her screen. Oh, yeah, well, that's gonna you're gonna have a your point 1% Failure is going to be someone's gonna put a thumbprint on that screen with glue.
And I feel like I'm like, by extension, I also hate underfill. Like, have you guys ever made any PCBs with underfill?
I have not? No, no, I actually do not know what underfill is. Oh,
so underfill is this black, goopy substance that you deposit around the edges of ICs. So it basically protects ICs and helps prevent balls from cracking or chips from popping off the board in the event that the product gets dropped or see some sort of hard impact or vibration.
Okay, I've seen this stuff before on PCBs have disassembled.
Yeah, but the way that it works is that you you have to run the board seven SMT line just like you normally would. And when they're done, you apply underfill around the outside edges of a chip. So you rely on capillary action for the whole thing to work like the underfill is supposed to wick under the chip and fill up all the empty spaces between balls. So it's great in theory, and theory, it sounds fantastic. But in practice, it can be a nightmare. Because if you if you apply it inconsistently, it's worse than having no underfill at all, any voice or any air pockets in underfill can be very bad. And you can't tell that they exist just from looking at the chip. There's no way of telling like what the underfill looks like underneath the chip. So there's no there's no fast or easy way to inspect it. And when you get voids in the underfill it might still be great protection in the event that you drop the product, but it's almost guaranteed to fail in thermal cycling. Because when you have like different sections under an IC that do and do not have underfill they have different very different coefficients of thermal expansion. So if you leave your leave your iPod and a hot car in Houston, Texas in the middle of the summer mismatched coefficients of thermal expansion means that the ice is going to see some pretty intense stresses as the thing heats up. And I've actually seen it caused chips to completely D laminate like as in the layers of silicone start to peel apart. So
yeah, it kind of looks like basically it's like a conformal coating that goes underneath your BGA Yeah, yeah.
Also seems like X Ray would be some x ray sampling would be a way to detect that. But that's also a giant pain in the ass.
Yeah, I mean underfill can be a useful tool but like you have to be really sure that you're applying it well otherwise it can be very, very bad news.
Got it? All right. So glue and underfill
your way keep going. I've got I can keep I can keep going. Yeah.
Oh, yeah. No, that's that's just that's just the first thing.
Like antenna is how do you guys feel about antennas?
Well, what kind of antenna? Are we talking? Oh,
good question. Good question. So chip antenna or like a trace antenna on your main PCB? I have no problem with that's totally fine.
But you don't have to do anything with it. Yeah, yeah,
exactly. But the moment you tried to do antenna on a flax or it's a PCB trace antenna that's separate from your main board, and there's like some form of interconnect, the whole thing becomes super fragile and super dependent on precise assembly. Because like, all of a sudden, you'll tighten a screw an extra quarter turn. And it can detune your whole antenna.
Oh, cuz yeah, flexes the board, right.
Or, or you can have some like plastic piece that sits next to the antenna that comes in a little bit thicker than normal. Still, like still on spec for the drawing. But maybe it's on one opposite end of the distribution compared to previous samples, and that can completely detune your antenna. Or I swear I'm not making this up, like, you change the color of the ink printed on your glass faceplate. And it determines the antenna.
I hated electromagnetics it's
my least favorite class. Like I've worked with some amazing antenna designers over my career, and I've got utmost respect for them. But it really is black magic to me, especially when you're getting into millimeter wave bands. That stuff is insane. But antennas always have a way of making final assembly of a product so much harder than it should be.
You weren't involved on the iPhone that you had to hold correctly. For to make a call.
I was not I was not those those. Those are my projects. I was
operator error, right? That wasn't Apple's fault.
I had no part and the antenna gate scandal.
You know, okay, so one thing one antenna pet peeve that I've run into and it's just awful.
And this should have been caught way early on in InDesign. But I had a customer that wanted an antenna and it was an external you know, camera SMA style connector, with a pigtail that went through an enclosure. But the pigtail wasn't long enough to a to assemble the thing on a bench you had to like clamshell it part way and connect it to a board, then close everything and screw it together. It was terrible. Oh,
service leaves. Yeah, service leaves are important. Yeah, I've worked on too many products where you the first time you make a wire or like a flex in her cable interconnect. You make it way too short, because you forget that you actually have to assemble it like a human's hands have to go in there and be able to physically plug it in, right and the second iteration is got this crazy s bend and it's super super long, just so that you can get the get the thing assembled.
So I had we had a similar issue, except that it was the enclosures already had that, that antenna on it. So I had the pigtail inside the enclosure right. But you couldn't put the PCB in it, because they would interfere with the wire Oh came out of the side of the enclosure. So you had to remove the antenna, put the PCB and then put the antenna back in.
So I worked as a as a repair technician for years.
Doing PA systems and mixing boards and keyboards and things like that. And you sort of get an appreciation for the assembly when you have to be the guy that takes it apart to fix it. And so many times so many times you're trying to rip something apart and there's 18 ribbon cables. That's like how the hell did they put this together in the first place? Like it's hard enough to take apart?
Nobody know what it is designed for repair. Everybody does talks about design for manufacturing and design for assembly but nobody ever talks about design for repair
designed for poor schmucks.
We're talking about we're in a capitalist society. It's thrown away, Stephen.
Oh the ivy My bad yeah.
consumerism society,
I should say, all right, and anything else for pet peeves?
Yeah, well, I could keep going. But the list is long. I've also got, I don't know, some of these, some of these start to get slightly more specific to manufacturing in Asia and even more specific to southern China. But there's a couple of tried and true rules that we live by, just because, you know, we've been bitten too many times. And one of one of them, one of the things we like to always tell rookies before they went over to China for the first time was, if you care about something, don't let it out of your sight. Don't ever let it out of your sight. Because the the way that I mean, I get it, like manufacturing environments are super, super chaotic. But things have a tendency to disappear. So you've got this PCB that hit some sort of bug that your firmware person is like trying to debug and figure out what's going on. And it's the only one that's ever hit this and it's like, super, super valuable. And then one of the factory engineers says that they're going to go take it to the FAA lab to look at it under a microscope, and then they walk away, and then you never see that board ever again. So it's just this like, repeated thing that happens again, and again, and again, you find something that's important, or, you know, something that really is of high value, and then it has a tendency to disappear. And I don't mean, I don't mean things like phones or laptops, I mean, like boards that desperately are in need of some sort of failure analysis. Yeah, you gotta
babysit.
So I got a pet peeve go for designs, is when you have when you're assembling a product and the part of the build is soldering the wires onto the board. So instead of having a connector there that plugs into the board, because I've actually seen some products that just hate soldering, having
connectors or connectors take up space, like they're big, if you're trying to make a product as thin as possible. Like you can't use board to board molex connectors, like that's not gonna, it's not gonna work.
Now, one of the I've taken some stuff apart and like the antennas, where we started to so the only way to take it apart is actually desoldering everything
yeah, that's, that's, that's no fun. Like, that's not fun. I just I hate I hate anything that's hand to solder because it's so it's so inconsistent people make mistakes, but I'm totally fine with hotbar soldering.
Oh, like LCD screens and stuff like that.
Yeah, yeah. Yeah. How about how about soldering is often done in high volumes is you know, as long as you dial in the process, it's then like, super consistent moving forward.
Actually, I talked to a customer off the ledge today, they wanted to watch well, they wanted to install these connectors that went through the front plate on a product and then put a PCB on to that and then solder it directly on there. So it is, it would be impossible to remove the PCB after soldering. And it's not like one soldering point. It's 16. Large soldering points for this. And I was like, if you if you have one RMA, you will regret this. And you will have one RMA.
Yeah. Cuz what do you do with a failed unit at that point? Because you can't take it apart? It just goes right into the E waste bin.
Yeah, it doesn't get taken apart. And
maybe maybe you can salvage the screws.
Yeah, that's. So we're not doing that with that customer. We're doing something
else. But you are doing it on the macro
amp. The macro amp is different. Because it's one one off. It's a personal project. And I don't I can serve the board from one side. So you can still service the board. Okay, I can serve as the Yeah, i Everything is one side stuff. So I can I can serve it at all from the side that I can look at.
That's not bad.
Yep. Well, thank you, Chrissy for coming on to the podcast.
Yeah. Appreciate it.
My pleasure.
It was a lot of fun talking about high volume manufacturing, because something that you know, not a lot of people have experience with.
Yeah, yeah. No, it's it's a it's fun, and it's definitely it's definitely something that I'm super passionate about. So I enjoy talking about it. But I also apologize for ranting on all of my manufacturing pet peeves and I guess glue isn't always terrible.
Well great with that. Would you like to sign us out?
That was the macro fab engineering podcast. I was your guest Chrissy Meyer.
And we were your host spark Dolman and Steven Craig. Later everyone take it easy Thank you. Yes, you are listener for downloading our show. If you have a cool idea, project or topic or your hatred of glue is just as strong as ours. Let's save it and I know Tweet us at Mac fab at Longhorn engineer are at analog E and G or email us at podcast at macro live.com. Also, check out our Slack channel. If you're not subscribed to the podcast yet, click that subscribe button. That way you get the latest episode right when it releases and please review us wherever you listen, as it helps the show stay visible and helps new listeners find us