MachinePix Weekly #40: Alfred Jones, Senior Director of mechanical engineering, Peloton
MachinePix Weekly #40: Alfred Jones, Senior Director of mechanical engineering, Peloton
Alfred Jones talks about the challenges of mechanical engineering for the masses. This week's most popular post was a chain manufacturing machine ⛓
This week’s interview features Alfred Jones, currently Senior Director of mechanical engineering at Peloton. In my last interview Shyam gave an electrical engineer’s perspective on the challenges of shipping consumer hardware—this week Alfred brings a mechanical engineer’s perspective.
(PS: want to work with Alfred on the Peloton team? They’re hiring Hardware Engineers 🚴🚴)
The most popular post this week was chain manufacturing machine. This machine cold-forms the chain links from steel stock, but depending on the size and application chains can also be hot-formed.
I’m always looking for interesting people to interview, have anyone in mind?
—Kane
Interview with Alfred Jones
What got you interested in mechanical engineering in the first place?
You’re going all the way back there! So I grew up in an era where toys were “assembly required.” They don’t even say that any more. My dad would buy me a GI Joe in the early 90s, and you’d have to assemble the vehicle. You had to put the top housing on and the bottom housing. Maybe some side structures. You had to decal it. You even had to pop plastic parts off the runner!
Anyway, so the most enjoyment I got out of my toys was putting them together. After a while they would break, and I would lose interest. So I would make new toys, turn them into something else. I’d take two different broken toys and tape them together, or find a new way to assemble them. In addition to all that, I loved Legos, K’NEX—I just like making things. I like the physical and mechanical aspect. I was also a very analog engineer. I did a lot of electrical circuits before I was trained, but it was all analog. No coding or anything.
When I went to college I thought I would do computer science, but it turns out I wasn’t great at typing. I’m better now. So I went for mechanical engineering and I loved it. I loved crafting, building, even differential equations, believe it or not! Structural analysis. All of it.
What’s something people may not appreciate about mechanical engineering consumer products?
Mechanical engineering and product design is often taken for granted. The reason is that Amazon, Google, a lot of the big companies, glorify tech as software. But it all rides on hardware! Mechanical engineering is a spoke of the wheel that pulls together everything.
The hard part about customer experience is that a lot of it is invisible if done well. A lot of what we think about goes into how it will feel, how it will be used, what it’ll feel like in your hand or on your skin. If it's a large piece of machinery, what’s the safety, the reliability? What I think customers take for granted is quality. You expect that when you buy something it’ll just work. But not all products are engineered or created equal. With the good products I’ve worked on, quality is engineered from the beginning. You think about the different failure modes, how you design for them, how you design them out, how you optimize materials and assembly. All these things are very challenging, and you have to juggle all these variables without losing product identity, industrial design, and the product usability and experience and look and feel.
I can’t stress enough how much thinking goes into making sure something doesn't fail. A lot of my life I’ve worked on mobile devices. We have to design all those for dropping. If you just slapped a screen into a case, it would probably break instantly if you drop it. Furthermore, when you design for high volume, you have to specifically engineer for that, too. You have millions of units that will be made with different vendors using different tools. The first one and millionth one have to have the same experience and reliability. All of that makes mechanical engineering so special and unique. All of that is taken for granted, because users, the first and the millionth, kind of just expect it to work. I used to laugh that blood, sweat, and tears, and years of our lives, went into these projects, and people would just throw them to children and expect them to work. If you mess up, there’s a flood of bad Amazon reviews. It’s brutal.
The last interview was with Shyam, an electrical engineer on a lot of consumer hardware. He mentioned it was really challenging to collect true usage data from consumer hardware.
You have to do a lot of extrapolation from things you know. When things start shipping, you’ve committed to a product and a manufacturing system, but you don't have a ton of data! Statistics, statistics, statistics: it’s funny, with those classes I took in school, I didn’t realize how important it would be to understand distributions, curves, and regressions as a mechanical engineer. But it’s so important. You can’t rely on huge data sets here. I can’t make and test a million of something just to show it works.
What’s been the most unexpected learning after shipping millions of products?
Two things come to mind when you ask me that question, which may contradict each other.
First, the importance of the core engineering fundamentals. I think that, you know, the average student learns those in school. I did well, I was top of my class, but as soon as I went into the workforce, I thought I didn't need that theory anymore. But to this day, the importance of those fundamentals—beam equations, force analysis, material stress strain understanding—is so, so core to every single thing I work on, because we have to think about how things will fail all the time. That’s one thing I took for granted.
The other thing you don’t realize, and I see a lot of young engineers take this for granted in mechanical engineering, is that there are some things in product development you just have to learn in the field. No school or classes or books or projects will teach you these things: root causing failure, triaging costs and business decisions against mechanical designs—things like BOM costs or engineering costs. Things that could affect quality at scale.
What’s the wildest story that you’re able to share from your time shipping hardware?
I have one; you can figure out how to make it sexy. I was working on this product that was actually a flop. The Fitbit Flex 2, when it was incepted, was forecasted to be the most insane sales-driving Fitbit ever. The first Flex, which was the second Fitbit, was a small little pebble device. Back in the day, before Fitbits were wearables, it was this little device you would throw in your pocket to track your steps. Fitbit expanded into wearables where that was probably not important anymore. We incepted this product called the Flex 2, it was going to be, and still is, the smallest fitness tracker. At the time, it would have been the first swim-proof fitness tracker.
This was projected to sell two million at launch, and then eight million over the next year. We designed this thing in such a way that it could be assembled using a completely automated line. Essentially, the way this tiny device was held together was with two types of glue. When we talk about statistics and all the different variables that come into manufacturing, especially at scale, this is the epitome of it. If you know anything about adhesives, you know where this is going.
In product design we have EVT, DVT, PVT (engineering validation testing, development validation testing, and production validation testing): so we go to our EVT to validate the engineering aspect of it. It was an easy test, right? You test every part of the line, test for air leaks. But the scale at which we did our EVT was probably thousands. We go to DVT, and we do something like a fifteen thousand unit run at a thousand a day. And we went from a high success rate in EVT, to a ridiculously large build loss in DVT: something like over 50% build loss when we picked up the volume.
What made it worse was batch to batch, it would swing wildly 50% to 90% to 40%. It was all over the place. This was like 2015; it was supposed to be Fitbit’s big product. Big launches are so important to consumer hardware companies; so much revenue is driven by the on time delivery of these products. This product was supposed to be available for the holidays, but now we have these crazy fluctuations. I had to stop everything to root cause this issue, at the risk of losing millions and millions of dollars and missing launch, because every day we’re late is a huge hit to Q4 revenue, the biggest quarter for consumer hardware.
I had to run massive amounts of DOE (design of experiments), looking at every process step, doing trial runs, checking every possible part and vendor. Long story short, the issue was adhesives and the ways they were stored. The first batch was great because we had just gotten the adhesives. But batch to batch, depending on how the adhesives were stored, we saw wild fluctuation. We had to invest a ton into the facilities to build out humidity and temperature-controlled storage for all the adhesives.
We were able to make the launch, but not at the volumes we intended. But in the end it didn't matter! No one wanted the Flex 2; they wanted the wearable trackers with screens. It’s funny because you can find insane failures, pull off amazing fixes, and in the end, it doesn’t really matter. Mechanical engineering is hard!
What’s been your favorite product to work on in your experience of leading HW teams?
They’re all near and dear to my heart. Every single product I’ve worked on, either designing a portion or the entire thing or managing the team, is like a child. I don’t have a favorite child, but I tell each one they’re my favorite. They’re all problems that I’ve solved. I took an idea and turned it into reality, and along the way there were issues and challenges and failures, but in the end they were mostly successful.
I’ve also worked on products the world never saw, but they all improved my ability to make the next thing better. So I don’t know what my favorite is! I’ve worked on cell phones, wearables, and autonomous vehicles. AV by far is the craziest thing I’ve worked on, when you talk about underestimating mechanical engineering; I’m designing something that’s deadly, and I have to think about the safety of the pedestrians and passengers. But how do I make it look aesthetically pleasing? Comfortable? Affordable?
I think connected fitness devices, like wearables and Peloton, are probably the most satisfying, because they directly improve people's wellness and health and connectedness. I think those are the most satisfying, and I want to keep working on things that can positively change people’s lives. I got a lot of enjoyment working on cell phones, but they’re not as satisfying because they get thrown away. We design these things for 2–3 years but the average life span is 18 months.
Kyle, the iFixit founder, lamented about the short lifespans of phones when I interviewed him.
I have a love-hate relationship with iFixit. Every engineer loves it when they want to learn about something. But when you make something and they tear it down and rate you, you don’t always get to tell your side of your story in architectural decisions!
But really I love iFixit. I have a ton of their tools here in my office.
What brought you to Peloton?
I realized I just love working on things that positively, directly impact people. We’re not just a bike company, we’re innovating like hell in fitness. The whole company is really into fitness.
My job is to expand on the awesome product we have and build a more awesome team than we have. The culture and energy they’ve created with the community of Pelotons is amazing. Really great people: everyone thinks about our members and wants to improve people’s lives.
What is your personal favorite piece of hardware? From a mechanical engineer’s perspective, what do you think represents the craft well?
This right here: a DJI drone, my favorite piece of consumer hardware by far. It’s the best of mechanical engineering. It has motors, cameras, aerodynamics, structures—and it flies! From a system engineering standpoint, all the sensors it uses, the motors and batteries it uses, are all amazingly integrated. They have jam-packed a lot of technology in these little things that other companies are trying to do bits and pieces of, but DJI is doing it all. They make very elegant-looking drones that by far are the best drones, at scale. I have one where the arms fold down.
From a product engineering perspective, it’s just elegant. From a systems perspective, the amount of technical depth and integration you need to make this work reliably at scale. It’s pretty dope.
Any side projects you’re working on right now?
Currently no because my side project is my four kids and getting them through COVID, but typically I love doing a lot of different side projects. I love Halloween; last Halloween I projection-mapped our house. I have a whole woodshop in my house; the last project my daughter and I made was a trebuchet.
Any favorite books or books you’re reading now?
The book I last read was Extreme Ownership. This is the one I had my managers read as well; I like the philosophy on ownership and responsibility.
What’s your favorite simple (or not so simple) tool or hack that you think is under-appreciated?
Post-it notes! I have a whole bunch on my desk; that’s how I organize my life. As soon as an idea comes I can write it down and put it somewhere. I color-code them.
I love shoes, but I keep one pair of casual shoes where I don’t care what happens with them.
The Week in Review
I had no idea these existed before I saw the video. It’s always a pleasure to discover a new machine.
I would have bet this would be the most popular post this week because of all the lewd jokes, but the chain machine pulled ahead.
This caused a bit of commenting on why they were changing what looked like nice shiny wheels for rusted wheels. The new wheels are rusted because they are made of weathering steel, which forms a stable, protective, rust-like appearance. The shiny wheels are shiny because they were worn down from usage (and likely had a flat section—even trains get flats, but they’re a lot more metal. Literally.)
Also fascinating: everything is held together by gravity and sheer mass.
Notice the patterns painted on the fins and the colored nose to help the chase pilot and observers orient the missile in flight.
Postscript
After posting the Nirbhay cruise missile, I started thinking about how cruise missiles fit into modern doctrine with the development of UAVs…
If you enjoyed this newsletter, forward it to friends (or interesting enemies). I am always looking to connect with interesting people and learn about interesting machines—reach out.
—Kane