Drop it in the river: The surprisingly simple philosophy behind manufacturing 2,000+ circuit boards with a tiny team

Stephen V. Chavez (Siemens) (00:02.222): Hello and welcome to the Printed Circuit Podcast where we discuss trends, challenges and opportunities across the printed circuit engineering industry. Joining me today on the podcast is Are Halvorsen, Principal Design Engineer at Microchip. Welcome to the podcast Are.

Are Halvorsen (00:16.065): Thanks. Thanks for having me, Stephen.

Stephen V. Chavez (Siemens) (00:18.09): Awesome. Yeah, it’s great to have you today. So, you know, could you start by telling the audience a bit about yourself and your role at microchip?

Are Halvorsen (00:25.504): Yeah, so I work out of the Trondheim office in Microchip. We’re about a hundred people. I lead a team of four other engineers and we create eval boards, development boards, EVKs for Microchip’s new silicon products, mainly microprocessors and microcontrollers, but also other types of silicon.

Stephen V. Chavez (Siemens) (00:51.406): That’s awesome, awesome. Can you tell us about your background? know, basically how you started in the industry, you know, how you evolved and how you ended up at Microchip at this role on Microchip? Because I know you guys are doing some fascinating architecture there. So can you elaborate a little bit more of your background?

Are Halvorsen (01:07.988): Yeah, I it depends how far I should go back, but you know, I, I was a 15 year old kid at some point and I said, you know what? I’m bored of school. I don’t want to do this anymore. so I started training to become an electrician. so true, true equivalent of high school. And, I got to a point where I actually was offered a job, as a, you know, someone that

Are Halvorsen (01:35.443): installs telecommunication equipment, an apprenticeship. And I had a teacher back then that pulled me aside and said, Are, you know what, you don’t want to take this apprenticeship, you should become an engineer. And I told him, so what’s an engineer? I didn’t even know I had no idea. Yeah, so I was bored with with the framing of framing of high school. So I went to the university. I got my

Are Halvorsen (02:24.534): did all the maths and chemistry, physics. I met two very engaging teachers that really got me interested in science again and maths. I applied to become an oil drilling engineer. It’s a master’s degree and there are 10 spots in the country every year. I got in and I figured out, you know what?

Are Halvorsen (02:34.154): I don’t want to be an oil drilling engineer. after all, so I started electronics, started a bachelor’s degree. I found an open spot and no requirements, you know, just sign up and get going. So I did, I did my bachelor’s. I wanted to start automation, you know, working with PLS systems and things like that. So I applied to do that, but I hit the wrong button in the web interface.

Are Halvorsen (03:08.918): and I ended up with electronics. So, and that’s where I learned about microcontrollers. That was my first introduction to microcontrollers, was during my bachelor’s. And we were lucky enough to have a company called Atmo locally in our city. And this stuff fascinated me. I thought it was really great. So I contacted them and said, you know what? I want to work for you guys. And I was lucky. I got a bachelor thesis with Atmo.

Are Halvorsen (03:40.403): I got an internship where I got to meet a lot of great people, applications engineers and others that did hardware. I started to study digital chip design, a master’s degree in that, but I’ve never worked with that at all. So I started full-time in Atmo in 2012 doing eval kits. Yeah, and we were acquired by Microchip in 2016. And I’ve been here ever since.

Stephen V. Chavez (Siemens) (04:07.566): Wow, what a story. My goodness. I love the fact that, you you started in one direction and then you pivoted a few times. I think many of us in the industry, you know, we kind of navigate, we’re not, you know, uncharted waters and you just your career just unfolds. And, you know, we look where we are today. And it’s like, you know, if we had look back at that young kid that started 15, 16 years old.

Are Halvorsen (04:25.762): Mm-hmm. Mm-hmm.

Stephen V. Chavez (Siemens) (04:35.406): We would have never guessed this. is where we were ending up, you know, or this is what we landed. So that’s I love it. I love it. You know, I want to jump in right into, you know, some some topics. So the first thing I want to get your thoughts on is, you know, can you share with me the basics of how your mixed-signal or your high- I don’t want to say this year. High mix, low volume turnkey operations works at a high level there, you know, at Microchip.

Are Halvorsen (04:41.676): Yeah.

Are Halvorsen (05:00.854): Hmm. Yeah. Yeah. I can talk a bit about that. So we can start with kind of how things are operating today. And I can talk a little bit about approaches we tried before that didn’t really work or scale at all. So we have a team of let’s say 15 people in Asia. They are our operations team and they ensure that we manufacture more than 2000 circuit board products.

Are Halvorsen (05:49.059): using contracted manufacturers. So these guys are the real heroes of our operation. So when I do a design, I make sure it works, we get the prototypes back. When we release it, I don’t hear about that design ever again, unless there’s a problem. And that’s kind of the philosophy we wanna go for. As long as we do our jobs right.

Stephen V. Chavez (Siemens) (05:49.059): Yeah.

Are Halvorsen (05:56.129): then the operations team can make sure that it manufactures and that it’s stocked in the shelves and we can focus on whatever’s new. So how do you do that? It sounds great, but how do you even do that? We have a common component library used across the company. People do their hardware design.

Are Halvorsen (06:40.403): with that component library. If they’re missing parts, they obviously have to add the parts into the library. They complete their design and they generate a production package that is, let’s say standardized. They upload the BOM,they provide it to the operations team. They give a review and say, okay, everything’s in order. You know, they ship it off to the contract manufacturer. They do their thing. They handle procurement of components, they handle assembly. We also work with a dedicated PCB broker, so not an individual PCB manufacturing site. We do that because we are able to scale, we can manufacture at different locations. can, you know, if a factory stops working for whatever reason, we can just move a design to another factory. That’s also a key.

Stephen V. Chavez (Siemens) (06:55.502): Hmm.

Stephen V. Chavez (Siemens) (07:12.366): Mm.

Are Halvorsen (07:22.914): key thing in this. Yeah, we get the prototypes back. We validate that the design works. We develop automated manufacturing tests using Python and you know, we talk to our hardware, we do some self tests, we do that in test fixtures that we, you know, I draw up electrical connections and the contract manufacturer takes care of the physical part of it.

Are Halvorsen (08:05.326): Yeah, and then we do a first article or a series zero build where we qualify everything. I get a piece of hardware back and I’m asked to say, you know what, this is good. Once I approve that, it’s in the hands of the operations team. That’s kind of the, let’s say the short version of how it works. Yeah.

Stephen V. Chavez (Siemens) (08:14.062): No, no, no, you know, I would tell you, know, I know I can tell you, I get it. It can go really deep when you think about the overall process of designing, you know, a circuit board, especially when you’re dealing with multiple teams in multiple locations and you got the whole gamut from your supply chain, external.

Are Halvorsen (08:33.004): Mm-hmm.

Stephen V. Chavez (Siemens) (08:39.114): stakeholders, which is your fabrication and your assembly team. And it takes a lot. It’s a lot of communication has to go on a lot of of collaboration and integration of how you’re handing off and how you’re exchanging. And it, you know, if you’re not following the industry best practices, it can definitely be very expensive, meaning time and money. then, you know, you don’t communicate correctly or you don’t get all the information. You know what? You’re forced to do a respin because you miss something.

Are Halvorsen (08:59.906): Hmm.

Stephen V. Chavez (Siemens) (09:08.494): So yeah, you definitely have to follow industry best practices and it sounds like what you guys are doing there is state-of-the-art and you guys are on top of it, especially with the success you’re having there at microchip, know, you talked about manufacturing. So, you know as an engineer, know, let’s go to our next topic I want to talk about which is design for manufacturing and turnkey manufacturing tasks. Can you elaborate a little bit more of your process and how you guys are addressing DFM or design for manufacturing and then the turkey manufacturing regarding testability.

Are Halvorsen (09:41.931): Yeah, so I think the first thing about DFM is you have to think about it from the start. I think that’s the key mistake that people do. You have to know the capabilities of wherever you’re assembling your board. Let’s say for us, that’s a contract manufacturer. We know that they can provide automatic image inspections. We know what kind of problems you can catch with AOI. If we have a really big, complex design, can ask them to do manufacturer defect analysis, so MDA. You basically build a huge bed of nails fixture and you have a piece of electrical measurement equipment that does a matrix of measurements between basically all the needles, resistance, inductance, short circuits, capacitance, that type of thing.

Stephen V. Chavez (Siemens) (10:20.92): Mm-hmm.

Are Halvorsen (10:39.586): So you take a, what you call a golden sample, you profile the board. So they don’t even know how, need to know how the design works. They just profile a board you know works and they compare those results to the other, to new boards that you make. So you’re making an electrical profile of the design. So you don’t need that for really simple stuff, but if you have a board with lots and lots of connections and nets, then that can kind of make sense.

Are Halvorsen (11:12.755): What we do differently, I think, is what we call the fixture test. We don’t have pre-made solutions with million-dollar equipment and flying probes or anything like that. But we make sure that in the design phase we put test points in critical areas of the design. For MCU board designs that’s typically you need access to your programming pins. That’s a given. You need access to measure voltages around the board if you have any power supplies or anything like that. Maybe you want to measure inputs, outputs. You want to measure, you know, since I work with eval boards, there’s always USB, right? So I have a luxury of always having USB on my product, which means that I can measure the current that goes through the USB cable. I can look for short circuits before I connect it to a PC. So I have a luxury there.

Stephen V. Chavez (Siemens) (11:55.203): Mm.

Are Halvorsen (12:12.866): So we put this board in a custom made test fixture that connects everything you need. And we have a Python framework that is a GUI that the contract manufacturer uses. From their point of view, that GUI is the same for every product. It’s very simple. You open the GUI, you have your barcode scanner, you scan your serial number.

Stephen V. Chavez (Siemens) (12:34.092): Hmm.

Are Halvorsen (12:41.74): The GUI takes care of downloading the appropriate test software if it’s not already on the computer. You hit the big play button, no configuration, you know, those guys, they don’t know the product, we do. There’s one big play button. Once you hit the button, you wait, you see the test progress. When it’s done, it says passed or failed. That’s it. So we own the production test, not the CM.

Stephen V. Chavez (Siemens) (13:07.118): Mm.

Are Halvorsen (13:11.072): And in that way, you know, we, if things fail, they say, you know what, we’re having a lot of failures. Please go look at the test logs. So we can go look at the test logs, look at the parameters, you know, and do, and worst case, we can also do remote debugging. So I can log into the test station remotely and I can kind of work with the technician and figure out, or their test engineers to figure out what’s actually wrong.

Stephen V. Chavez (Siemens) (13:33.838): Mm-hmm.

Are Halvorsen (13:36.035): So I’ve been able through that setup without having access to any measurement equipment or oscilloscopes or anything, I can’t touch the hardware, I’ve been able to pinpoint things like LDLs being mounted wrong. I’ve been able to pinpoint things like, you know, we made a design error, we swapped the ground and the power pin on an IC, things like that. So the test logs and the collaboration with the test engineers can…

Stephen V. Chavez (Siemens) (13:58.873): Mm.

Are Halvorsen (14:05.394): give you lot of insight. It’s really hard to debug without equipment.

Stephen V. Chavez (Siemens) (14:11.436): Yeah, I I will tell you, know, yeah, I totally understand because as a technician, you know, early in my career, you know, being down in the lab, you know, troubleshooting and I refer to this like chasing the dragon. You’re trying to find a needle in the haystack. You’re trying to figure out why it’s not working. sometimes it’s straightforward. It’s easy. Other times.

Are Halvorsen (14:23.147): Hmm.

Stephen V. Chavez (Siemens) (14:32.728): You can be there for weeks and you can’t figure out. It’s very difficult to figure out what’s working. And that really excites me when I hear you talk about the testability regarding DFM in regards to your MDA, is your analysis that you’re capturing, your bed of nails. mean, those fixtures, those bed of nails fixtures, especially if they’re clam shell, they’re not cheap. Those are very expensive and your custom fixtures you have.

Are Halvorsen (14:56.962): Hmm.

Stephen V. Chavez (Siemens) (15:01.931): It’s good. you know, I’m curious your your test points, you mentioned test points on your boards. You put them on strategic locations. Is that correct? Is that what you said? Strategic locations? You don’t do 100 % on every net, just specific locations. Is that what it is?

Are Halvorsen (15:07.202): Mm.

Are Halvorsen (15:11.841): Yeah.

Are Halvorsen (15:16.29): It depends on the design. So I can actually show you a couple of designs right now where I can show you the difference. Okay, so let me just set that up.

Stephen V. Chavez (Siemens) (15:22.446): Sure.

Stephen V. Chavez (Siemens) (15:27.778): Yeah, know for our audience, you can’t see because this is a podcast, you know, Are’s got a unique setup where he can show us some circuit boards. So we’ll kind of talk through this and hopefully you can visualize and he’ll try to explain in more visual effect of what he’s showing.

Are Halvorsen (15:40.374): Mmm.

Are Halvorsen (15:44.035): Yeah, so let me just zoom in a bit here. So this is a really, really small design. You know, there’s not a lot of room on this design. You can’t have a hundred percent test points here. On the other hand, the design complexity isn’t that huge either. So on this board, I don’t do any MDA testing. There’s no electrical profiling on the board. There’s AOI going on, so you can…

Stephen V. Chavez (Siemens) (15:55.054): Exactly.

Stephen V. Chavez (Siemens) (16:08.301): Mm.

Are Halvorsen (16:13.11): you can catch a lot of soldering defects. Obviously you’re not able to see defects in BGA packages or QFM packages without wettable flanks, things like that. So our approach to testing basically said that when we made this board, we figured out, okay, so if I know I can’t see under those packages,

Stephen V. Chavez (Siemens) (16:27.662): Mm-hmm.

Are Halvorsen (16:39.212): Can I do some functional testing to infer the fact that it works? And that’s exactly what we’ve done. So I can’t zoom in anymore, but hold on. Yeah. So if you see the larger QFN next to the USB cable, those small DFN packages next to it, and all the circuitry over there, that’s actually an embedded debugger on this board. So this part of the board,

Stephen V. Chavez (Siemens) (16:51.008): No, that’s good. I could see it.

Stephen V. Chavez (Siemens) (16:57.326): Mm-hmm.

Are Halvorsen (17:06.69): connects to USB and it lets you program and debug the actual chip we want you to evaluate and that’s the small one on the other side. So by connecting to the outputs of the power supply, by connecting to…

Are Halvorsen (17:28.188): Yeah, by connecting there, we know that all of those small DFN packages actually are soldered correctly. We measure outputs from the LDOs on the board to verify. Yeah. So part of that functional testing also involves having the embedded debugger on the board program the target, being the PIC MCU in this case. We load special firmware in the PIC MCU that exercises all the lines, the connections between the PIC MCU and the onboard debugger. And the debugger reports back to USB. So we’ve taken a design where we have a lot of small parts, like there are electrical level shifters here that are six pin miniature packages. I can’t do that with AOI, but we infer it from actually exercising those pins in our test setup.

Are Halvorsen (18:39.694): We mentioned that this is low volume and high mix, right? 2000 plus products. So test time isn’t that critical, right? I don’t need to test this board in a millisecond. I don’t need to test this board in one second. I build them in batches of 500 to a thousand at a time. So, okay, the test takes 20 seconds. So what? Doesn’t matter. So that’s kind of, you know, if I was doing something that was really high volume and I needed to get my test town done.

Stephen V. Chavez (Siemens) (18:39.694): Mm.

Are Halvorsen (18:51.564): test time down because it was important for the margin of the product, I wouldn’t have that luxury. So when you test silicon, you can’t do it that way. You can’t spend 20 seconds testing a piece of silicon. If you do that, you can’t sell the product. Nobody wants to buy it. It’s too expensive. On the flip side, you can see the test points really. So the test points that I’m pointing at here, that’s the programming lines for the debugger chip.

Stephen V. Chavez (Siemens) (18:54.722): Mm-hmm.

Stephen V. Chavez (Siemens) (19:14.243): Mm.

Are Halvorsen (19:19.072): So that’s our entry point into the whole board. You program the debugger chip with a bootloader and firmware. Then you go into USB and you do some voltage measurements over on the edge. And that’s enough to, you know, I can confidently say that this board works a hundred percent when I bring it out of the fab without doing any electrical profiling. this…

Stephen V. Chavez (Siemens) (19:28.664): Mm-hmm.

Stephen V. Chavez (Siemens) (19:45.667): Yeah, you know what? I really, I really like that board. The look of it, especially, you know, me being a, you know, 35 years of designing, I’ve seen several different type of test boards and I really like the castellation on the edges. So you can mount that on another board if you had to or solder it down and then just the way it’s condensed. mean, we talk about today’s complex packaging being smaller and tighter. And I mean, that’s a perfect example of, you know,

Stephen V. Chavez (Siemens) (20:21.058): very small test board that you’re able to do, you know, it’s a really good functional test. Especially through the USB, what you’re doing there. I’m curious, you know, when you generate these boards, are you guys utilizing, you know, ODB++, are you guys, you know, for your manufacturing outputs, or are you guys using 2581, or are you still using standard Gerber with Netlist? What do you guys…

Are Halvorsen (20:38.05): We’re still on the Gerber train. That’s something.

Stephen V. Chavez (Siemens) (20:42.262): No, no, it’s nothing wrong with that. mean, it’s still the highest demanded outputs is Gerber data still in the industry todayaccording to I-connect007 and PCEA, these two different entities have their respective magazines, in the research they do Gerber data with the netlist is still the number one requested output data. But, you know, with the intelligent formats, we can do a lot more, you know.

Are Halvorsen (20:48.482): Hmm.

Stephen V. Chavez (Siemens) (21:11.054): Regarding you know testability and so forth like that. So I thought it would at least ask because I was just curious Discussion came up the other day So, you know with testability, you know, let’s let’s talk about you know problems I suppose they wouldn’t talk about scaling, you know these kind of things to processes To different teams and locations because you mentioned, you know, you have 15 designers in Asia or 15 people in Asia So you’re all in different locations. So, you know, can you elaborate our you know?

Are Halvorsen (21:14.114): Mm.

Are Halvorsen (21:18.53): Yeah.

Are Halvorsen (21:27.33): Hmm.

Stephen V. Chavez (Siemens) (21:39.81): Talk about this, you know, the scaling in these kind of processes.

Are Halvorsen (21:42.773): Yeah, sure. So back when I started, you can kind of bring it back to when I started my career in Atmo. I was sitting in a room with three other applications engineers, hardware designers, or not hardware designers, they were application engineers. They knew how to do hardware design, they knew how to do high speed design, they did firmware development, code.

Are Halvorsen (22:31.905): I’m losing my words. Code frameworks to let customers code their own applications. They released examples, they wrote user guides, everything. It’s everything from A to Z. So one thing is creating the actual piece of hardware, but if you want our customers to use it and learn something from it, you need to provide actual firmware and documentation.

Are Halvorsen (22:52.547): that a big part of the job we still do in the team I run is exactly that. It’s not just hardware design. I don’t want to say just, but it’s more than that. the team I joined was very cross-functional. We had people that knew a lot of different disciplines. We had programmers. We had people that made traditional

Stephen V. Chavez (Siemens) (22:52.547): Mm-hmm.

Are Halvorsen (23:08.918): traditional debuggers, right? Tools that you buy and you plug into your own hardware and you program and debug your own hardware. Those guys were combined into the team of applications engineers and designers. And that’s when we invented or created these onboard debuggers that I was referencing. You you need a lot of different disciplines. need deep knowledge of the silicon and the debug systems. You deep, firmware knowledge,.

Are Halvorsen (23:41.005): you need and obviously also how to design the hardware. So that cross-functional team developed these kind of complex boards. Hold on, I can show you some examples again. So we started with something like this. This is a credit card sized board that let you evaluate a microcontroller. It had some sensors on it, some capacitive touch and some different things.

Stephen V. Chavez (Siemens) (23:41.005): Mm-hmm.

Are Halvorsen (24:06.498): problem with them is you always have to bring a debugger. So if you went to a training session, you know this board is quite cheap, you can manufacture it, but you still need a debugger. This is a hundred dollar tool or something like that. So, okay, what can we do about that? So this cross-functional team just said, okay, let’s just put the debugger on the Easy, easy, easy enough. We did that. We created some.

Stephen V. Chavez (Siemens) (24:12.206): Mm.

Stephen V. Chavez (Siemens) (24:21.453): Mm-hmm.

Are Halvorsen (24:35.2): More scalable boards like this. Here we hid all the debugger circuitry on the bottom. Still needed a lot of tinkering from different people to get it working. Yeah, we progressed to more advanced stuff, more peripherals. Here we have ethernet and other things. Still the same thing. There’s a debugger on the bottom.

Stephen V. Chavez (Siemens) (24:38.222): Mm.

Are Halvorsen (25:01.378): This thing you could also at the time get for around $50 instead of buying, you know, expensive development kit and the hundred dollar divider. So that was kind of our focus back then. But in order to get all this to work.

Are Halvorsen (25:27.106): You needed to master a lot of disciplines. needed to master hardware design, high speed hardware design. You need to master firmware development. You needed to understand the manufacturing process, DFM. We needed to comply to our in-house product life cycle management systems. The architect of that system happened to sit in the same room as well, right? So.

Stephen V. Chavez (Siemens) (25:27.106): Mm-hmm.

Stephen V. Chavez (Siemens) (25:39.832): Mm-hmm.

Stephen V. Chavez (Siemens) (25:44.949): Mm-hmm.

Are Halvorsen (25:45.175): We were in this bubble where we could do everything locally and have a good connection with one team in Malaysia, in Asia, that was the operations team. That’s a luxury you don’t have all the time. That’s where the scaling problems start. So we were able to scale this to a design team in India. They made the same type of products. They became highly efficient.

Stephen V. Chavez (Siemens) (25:59.535): correct.

Are Halvorsen (26:13.654): The only way that happened is we talked to them every day. That’s what we needed to do because we were doing so many different things. You can’t just give this to a hardware designer and say, make this. There’s a lot of things to do. That’s when we kind of realized back then we had in our design flow or production flow, we had a lot of gating steps.

Stephen V. Chavez (Siemens) (26:17.368): Mm.

Are Halvorsen (26:41.41): So there were formal reviews, formal gating reviews. So I ended up being one of the people doing those formal gating reviews. So once a week, I sat down, myself and one other engineer, and we looked at submissions from other teams. And our job was to figure out, did they follow the process or not? It’s quite time consuming. It’s frustrating for the people on the other end, you know. If they did something wrong or, you know, they forgot something.

Stephen V. Chavez (Siemens) (26:49.998): Mm.

Are Halvorsen (27:10.892): Quite a tedious process and not very efficient. And to be quite frank, Atmo had about 6,000 employees, a lot of different business units. We were able to scale this to a team in India and two other teams within the company. Everyone else just said, you know what, this is too hard. We don’t want to do this. We’ll set up our own CM. We’ll deal with it ourselves.

Stephen V. Chavez (Siemens) (27:30.894): Wow.

Are Halvorsen (27:39.203): We’ll get you the hardware, we’ll get it to the warehouse, but you don’t get a say in how we do it. That’s the drawback of creating these complex processes is that if you have someone with hardware experience that really knows how to make a good piece of hardware, they don’t necessarily know the rest. And that was the Achilles heel of how we did things back then.

Stephen V. Chavez (Siemens) (27:45.486): Mm.

Stephen V. Chavez (Siemens) (28:01.102): Sure.

Are Halvorsen (28:05.314): So the difference today is that we don’t really have a lot of gating criteria anymore.

Are Halvorsen (28:16.138): Okay. So whenever you start a design in our system today, no matter if it’s a prototype on your desk or if you intend to mass manufacture it and sell it, you start the same way. You go into a self-service system, you generate part numbers for your PCB, for your product. And you can do that yourself. There’s no one that needs to approve it. Then you start working on your design. That’s in a component library.

Stephen V. Chavez (Siemens) (28:29.038): Mm-hmm.

Are Halvorsen (28:46.316): You can put things inside the component library yourself. There are no blockers there. There are, you know, people need training in being a library manager to formally approve components in the library, but everyone is allowed to put a component in the library and put it into their design, even if it’s unqualified, no blockers. Yeah. So that’s the approach, right? Self-service all the way, no blockers. That’s how we do it. So the, the analogy I would say,

Stephen V. Chavez (Siemens) (28:59.064): Mm-hmm.

Stephen V. Chavez (Siemens) (29:06.411): Interesting.

Are Halvorsen (29:16.418): for our processes, it kind of flows like a river. You start at the beginning, you do what you need to do. You create your part numbers, you create your library components, you do your design. And when you’re done, you have a package in your hand. You drop that package in the river. And if you’ve done your job right, it flows through the river and it ends up in the warehouse. That’s it.

Are Halvorsen (29:50.048): That’s the truth with modifications, right? There are some checks and balances. So we do have a review of the bill of materials before we actually let the package be dropped in the river. That’s one. We also have a gaining criteria where you need to justify why you’re making this product. That’s number two. And you also have to, you know, once you get your prototypes back, you have to put your stamp of approval on it. And that’s basically it.

Stephen V. Chavez (Siemens) (30:09.614): Mm.

Are Halvorsen (30:11.042): That’s how we’ve been able to scale it to more teams without enforcing how you do it. But we enforce how the files are generated, stored and managed so that that small team of 15 people can deal with all the products in the same way. Even if it’s like a small board that I showed you first or a huge complex $20,000 FPGA product. It’s the same process.

Stephen V. Chavez (Siemens) (30:36.492): Mm-hmm.

Stephen V. Chavez (Siemens) (30:40.094): That’s awesome. You know, it’s interesting because, you when you think about, you mentioned, you know, the multiple domains collaborating and in your tools and especially with, you know, library management. I’m curious, you know, what are you looking forward to? What are the things you’re looking forward to to get easier, you know, as you implement or you’re running with the Siemens tools or the Siemens solutions, you know, specifically like EDM and the collaboration doing? Yeah, I’d like to know your thoughts on that.

Are Halvorsen (31:09.398): Yeah, so right now, my team, we store our design files in Git repositories. We do a lot of software development as well, so we’re comfortable using Git. That’s not true for everyone. Everyone stores their files differently. The consequence of that is if someone comes to me and says,

Are Halvorsen (31:40.065): Can you tell me which design files were used to generate this output? I can point them to a specific git hash and say, this is the place it got from. But again, software development. More traditional hardware development engineers may be using Windows shared drives or, God forbid, just put it on their local PC. That happens too. You could put it in one drive, wherever, but…

Stephen V. Chavez (Siemens) (32:07.512): Mm.

Stephen V. Chavez (Siemens) (32:07.512): Mm-hmm.

Are Halvorsen (32:31.702): they might not necessarily be able to point you to the exact design files that were used to generate the output. That’s a problem, right? So when we do design review, for instance, we have to generate PDFs. We send the PDFs to a bunch of people. It’s hard to comment in PDFs. Maybe you need meetings to actually sit down and start circling with pen and paper. And that process is quite cumbersome.

Stephen V. Chavez (Siemens) (32:42.456): Mm-hmm.

Are Halvorsen (33:02.199): When we release design files, I mentioned we have standardized production packages that you need to upload. The bill of materials and all the Gerber data, the drill data, PCB specification, all of that. We can’t control how everyone stores those files. you know, people need to follow templates and maybe it’s not exactly how they want to do it. They need to do it outside the software, outside the tool chain. They have to manage files and that’s a complication.

Stephen V. Chavez (Siemens) (33:02.199): Mm-hmm.

Are Halvorsen (33:03.476): Even the component library. The component library actually, lives in a Git repository now to a lot of designers aren’t happy about that because Git is complicated. I’m looking forward to baking all that into one system, into one tool chain. know, once you have it in one tool chain, you know, that’s how it works. It’s all in there. You open your application and you work on your design and it’s just built in.

Stephen V. Chavez (Siemens) (33:21.656): Mm-hmm.

Are Halvorsen (33:34.167): When you’re done with your design, or even along the design, you save your thing. And the tool chain takes care of tagging, revisions. It takes care of saving all the intermittent steps. Whenever you’re done and it’s, you know, when you go to review, you have an integrated solution where you just point someone to a website and they can start opening the design in the browser. can comment. You can see statuses on different topics that you had to deal with. You can…

Stephen V. Chavez (Siemens) (33:43.822): you

Are Halvorsen (34:03.33): All that traceability is in there. It’s in the toolchain. When you’re finally done and you kind of hit the approve button, we’re now able to have the system generate the output the way it should be and automatically upload it into the lifecycle system without having to manually manage files or teach people how to do it. It just works. That’s really lowering the bar to…

Stephen V. Chavez (Siemens) (34:06.989): Mm-hmm.

Stephen V. Chavez (Siemens) (34:17.87): Mm-hmm.

Are Halvorsen (34:32.898): for the engineers to be able to focus on what they really know, and that is hardware design. They know their products, they know how to do it, but they don’t have to be bogged down by all those extra things that you need to do today to get a design in production. That’s what I’m really looking forward to.

Stephen V. Chavez (Siemens) (34:37.272): Sure.

Stephen V. Chavez (Siemens) (34:49.422): Yeah, you when you think about a well-oiled machine, mean, that’s, yeah, I mean, that’s what Siemens offers is a fully self-sufficient ecosystem that allows you to focus on what you do best rather than having to, you know, be forced to do a lot of manual steps, which anytime you have manual steps, you increase your potential for errors to exist or to be created. you know, especially if you’re dealing with just PDFs and…

Are Halvorsen (35:12.642): Mm-hmm.

Are Halvorsen (35:17.804): Mm-hmm.

Stephen V. Chavez (Siemens) (35:18.062): You know, people commenting in PDFs getting red lines and you know, I have many experiences where stuff just gets missed. You know where you don’t have any gatekeepers or any phase gates to know, you know, did everyone collaborate or comment on it and are all the inputs of you know, did all the feedback come in and you end up missing something and like I said. You end up, you know, causing a forced rev or.

Are Halvorsen (35:25.44): Mm-hmm.

Are Halvorsen (35:36.354): Mm-hmm.

Stephen V. Chavez (Siemens) (35:44.042): even potentially you create the wrong version of the board because you miss communication. And yeah, it happens. It happens to the best engineering team today. So, well, you know what? We’re running up on time. let, let, you know, let me get your final thoughts. What are the final thoughts you’d like to share with our audience? You because I mean, I can keep talking to you all day, especially when I see all that equipment behind you and the circuit boards you showed us. mean, it’s yeah, I wish the audience could see what I’m seeing, but yeah, you know, what are your final thoughts that you’d like to share with the audience?

Are Halvorsen (35:49.249): You

Are Halvorsen (35:56.482): You

Are Halvorsen (36:15.106): I think we’re in an interesting industry. Our industry has consolidated over the last few years. You’re seeing fewer and fewer companies, but there’s still so much innovation going on with customers. The possibilities are almost endless. Today’s buzzword is data centers and AI, but there’s always going to be a new thing.

Stephen V. Chavez (Siemens) (36:35.966): yes.

Are Halvorsen (36:43.83): That’s kind of what’s kept me in this industry and this job in microchip. It’s like the constant change. The constant change really fuels me. It makes me motivated and it really wants me to get up in the morning and go to work and, you know, solve problems. That’s essentially what engineers do, right? We solve problems. Sometimes we find too many problems, but hey, let’s try and, you know, if we focus on the problems and we solve the problems, then…

Stephen V. Chavez (Siemens) (36:44.173): Mm-hmm.

Stephen V. Chavez (Siemens) (37:00.237): Haha.

Stephen V. Chavez (Siemens) (37:03.928): Mm-hmm.

Stephen V. Chavez (Siemens) (37:08.451): Ha

Are Halvorsen (37:14.595): endless possibilities of what I can enable other engineers and companies in the future to achieve.

Stephen V. Chavez (Siemens) (37:25.186): Yeah, that’s awesome. That’s awesome. Well, you know what? Like I said, we’re up against the time and I want to thank you for joining me today on the podcast and thanks everyone for tuning in as well, and to our audience – please continue to tune in for more trends, challenges and opportunities across the printed circuit industry.