Embedded - 92: Everybody Behave, Please

Episode Date: March 11, 2015

The Linker post for this episode: Make Anything James @Laen Neal from OSHPark spoke with us about starting a business, helping open source hardware, and throwing wild parties.  OSHPark got its start from DorkbotPDX. If you are in Portland, Oregon, check out their meetup (started out on Mondays, now first Tuesday of the month, look at the CymaSpace meetup calendar for the Maker Meetup). Open Source Hardware Association (OSHA) PCB Design School blog Bay Area Maker Faire 2015 is May 16-17, Bring a Hack dinner is usually Sunday.  This time we really did talk about the Maker Pro book.

Welcome to Embedded, the show for people who love building gadgets. I'm Alesia White, here with Christopher White, and we're joined by James Lane Neal. We're going to talk about building boards. But before we get started, I want to follow up on our chat with Sophie from last week. Check out my blog post on Element 14 about how to win the new Hackaday prize. The link will be in the show notes. Hi, Lane. Thanks for being on the show today.

Hi. Thanks a lot for having me. So, tell us a bit about yourself. Well, I'm Lane. I run the Osh Park Community Printed Circuit Board Order out of beautiful Portland, Oregon, an open hardware mecca. And so Osh Park makes boards.

What do you mean by community circuit boards? Well, the way that we get our prices down is by grouping together lots of orders from lots of different people. So to me, it feels very much like this is a community of people who are coming together to make this happen, to be able to do panelized orders and reduce the price for everyone. And so I had a board that was, oh, maybe an inch on each side. And I ordered them from Mosh Park. And I only wanted 10 because I didn't really want a whole bunch. If I had done that through someone else, I would have

had to get a whole panel. How big is a panel? FABs typically charge a very large setup fee, and then the materials cost is really a small part of that. So a whole panel ends up being about, a production panel is in the range of 18 by 24 square inches, or sorry, inches. So that's about, that ends up being usable space around 300 square inches of usable space and so about 300 more of my board or 290 more of my board than i wanted yeah the fabs will end up charging you the the same for a very small board or for a very large one because most of the most of their cost is in is in producing these photolithography films for the manufacturing process.

And so with your system, I might have gotten one of my board on ten separate panels. Correct. Because I'm on a panel with a whole bunch of other people. Yeah, exactly. So you'd, you'd get, uh, I, I take, um, lots of different designs from lots of different people and I, uh, I assemble them into, uh, into, uh, manufacturing panels, uh, and then send them all in as a group and then get them back and then break them up, ship them out, uh, to everyone who ordered. Okay. So this is all nitty gritty stuff and we've jumped right in here. I'm sorry. No, no, it's fine. I was curious about how

you, how it led you to, to form this organization company. Sure, sure. So, um, uh, about probably six years ago is when I started getting interested in, in electronics in general. Uh, uh, throughout my life, I'd been kind of curious and interested, but I had a really hard time getting a foothold before. So I went into software, which was a lot easier. I ended up just, there was a, in I think 2009, there was a Make Magazine video podcast about introduction to Arduino. And I hadn't heard of it before. And that was just transformative. It ended up being able to see how you can easily control electronics with software.

It was kind of a big change for me and very exciting. And I joined a local electronics group, Dorkbot PDX, which was pretty small at the time. And as it grew, we figured that we had quite a few of us who were learning how to design circuit boards around the same time. And all of us were placing these orders from China. And it was taking like a month to get back. And half the time, the boards weren't manufactured correctly. And we realized that we could combine our orders and order a full panel from a U.S. manufacturer. Get them back in a week and they mostly worked. So I started just doing that as a local service to our local electronics group. And we started getting emails from people in universities going, hey, could we get onto this?

We see what you guys are doing. Could we put a board on there? And yeah, we needed it because it takes, like I said, it takes a lot to fill a panel. And so we got, I took their stuff and put it on and it just started growing from there for really a couple of years before it became Osh Park. I realized that it was getting to the point that really this was a business thing and not a community thing because it was just getting that big. And so I spun it off into Osh Park. And it's your main thing now.

Yeah, for about a year and a half, it's been my main thing. And before that, you were a sysadmin? Yeah, I was a Unix geek in web operations mostly. And so now you're putting a whole bunch of the different boards onto a panel. Do you really just play Tetris all day? I write software to play Tetris. Does that count? That's not as much fun.

It is. It is. It's great. For really the first couple of years, it was working on improving that layout so that I could fit more boards in the space. Because when I first started, I was getting probably about 75% efficiency, and now it's in the range of like 96%. But you read other people's layouts. I don't even know how to do that. I mean, you read their Gerber files? Yeah, exactly. Since the manufacturing format for circuit boards is this Gerber format,

which is just a straight text format with coordinates in it, I was able to just kind of go to translate that. So if someone's board starts at one inch by one inch, I can just add five inches to that, and now it's five inches into the panel. Oh, that makes it sound easy. Yeah, really a lot more work than that. Why is it called OSH Park? Well, the user community that I feel the most connected with and really want to serve with this is the open source hardware community.

So that's the OSH. It's so obvious. It was staring us right in the face. For some reason, I was connecting it to Oshkosh or something. I figured Oregon or something. Would be nice. But open source hardware. Okay.

That actually makes a lot of sense. Yeah. Why did you, why open source hardware, okay. That actually makes a lot of sense. Yeah. Why open source hardware? Is it because that's how you've learned, or do you have other reasons? Yeah, so my whole life I've been a Unix geek, and that's really built on open source software. I really believe in open source as a way for the betterment of all people. Being able to learn from open designs, being able to make changes to it, to understand how it works, is, I think, really a wonderful thing about the open source community in general.

And open hardware is, of course, I mean, how I learned. So, I mean, when I got my very first Arduino and then started designing boards and then started wanting to incorporate an AppMega onto my own designs and looking at the schematic for the Arduino and the schematic for the bare bones board to figure out what I needed to actually include in my design

and really helped jumpstart that education. And so what do you do to get other people into open source hardware? Well, I don't know. Really, being active in our local electronics group is a big part of it. One of my employees does this Maker Monday class where we just kind of get together and people bring their stuff and they just make. So by default, things tend to be open unless people go, well, no, I'm going to hide it. Especially with electronics, you can look at a circuit board and see how it's made. The software might be hidden if there's microcontrollers, but things are open unless you try hard to close them, really. And so that Maker Meetup is at Osh Park?

No, it's at, actually, I forget the location now. Simma Space? I think it's at Simma Space, yeah. Simma Space. Simma Space, yeah. It's interesting because I've been active in software for a long, long time, and it's only been recently, or it feels like recently, that open source hardware has even been a thing.

Yeah. And I haven't followed it really that closely until it's getting started with the podcast and talking to lots of people who are involved in, in open source hardware. Do you see it following sort of a similar trajectory? Do you see it? It's only been around for a few years. It's kind of in that same phase that maybe open source software was in the

late nineties. Are there enabling technologies that are, that are kind of in that same phase that maybe open source software was in the late 90s. Are there enabling technologies that are kind of missing? It seems like you found one and you're trying to fill the gap. The manufacturing gap, yeah. I think that it's really benefited from a lot of the mistakes and experience of the open software community. There are, like in open source software, we've now had really decades of people trying to release open source stuff,

other people trying to use that open source stuff for closed source things, and this kind of arms battle of licensing and stuff like that that we're now kind of starting to see with hardware but we're able to avoid a lot of it thanks to being able to look back at the open software groups

and seeing what they did about different things. A lot of open software was licensing, coming up with the perfect license for open source. I haven't actually seen anything like that in open hardware. Well, there's several open hardware licenses now. I mean, Creative Commons covers a lot of this. There's the Open Source Hardware Association

has pointers to various licenses and sorts of things that you need to include in this in order to make it open source hardware. Since a lot of people like to pop onto the name open source hardware, but then they're like, here's the schematic, but they don't have board layouts or source code

or firmware for the devices on it. And so these are battles that are kind of being fought behind the scenes in the open hardware community. Or just have the Gerber's but not modifiable schematics. That is very common. I mean, there's lots. With software, it's like, sure, here's my code, but I'm not giving you the make file, so you're just going to have to rebuild all that part. Yeah, you've got to figure that out yourself.

In open hardware, there's a certain microcontroller platform that, I don't want to start a war, but they released the Gerber files. They released, I think, the Altium layout files, which is prohibitive to use that. I mean, that's heavy licensing fees just to get that. But then, of course, you can't get a hold of the main chip that's on the board. It's only available through NDA and through massive 100,000 quantity purchases. Yeah, that is, and it's sort of disappointing because it's something that everybody wants to say is open and yet you can't quite. But the Altium part, I have trouble saying, yes,

you have to use free tools because so many of the embedded compilers, the pay-for ones are in some way better. Absolutely. Sorry. That's true in the open software community. This is another place where you can look back and go, the software may be open, but you need to use these expensive compilers in order to produce good code or to even build it. That's something that happened that the open software movement struggled with decades ago. And really the result was to build out better open compilers. So like GCC and LLVM,

all extremely good compilers that have come from this. So I think that in order for hardware to be open, I think that we're going to have to improve our open tools. And thanks in great part to CERN contributing to the KiCad code base, I think that KiCad is well-positioned to become that. That was actually my next question. Okay, so which open-source tools do you think are going to win?

Yeah, I really think it's... I mean, KiCad, for its flaws right now, it's still a very new tool. I think that it's the best position for it. I mean, we have, there's, open source wise, there's really KiCad and the GNU EDA tools. And I think KiCad is clearly superior there. Me personally, I'm still stuck in Eagle. Is there a way to translate between Altium and KiCad is clearly superior there. Me personally, I'm still stuck in Eagle. Is there a way to translate between Altium and KiCad? I don't know.

I mean, that would be one of the things. I bet Altium would say, no, that's copyright or patented or something. You can't do that. But it's one of the things that with NumPy and MATLAB, being able to write code that looks remarkably similar to my MATLAB code in Python is really, really nice. Yeah, yeah. And transfer all the old scripts over. So I hope somebody's working on that.

Yeah, well, I mean, KiteCat has open format. And things go through, like exporting from Altium, I'm sure you can get positioning data. You can get the Gerber data, the ODF data. I've suddenly spaced on this other... There's another manufacturing-slash-file format that you can use to pull that data out. That's an open format that you could possibly parse into.

So it's not even proprietary necessarily where you'd have to reverse engineer anything. Yeah, it should be possible to write conversion tools. And yet none of us are jumping up and down to get started on that process today. No. Going back to OSH Park, what should software engineers know about Osh Park?

Because that's really what I am. I'm not really a hardware engineer. Yeah, well, Osh Park is a, in addition to being a place that you can upload your own designs and get inexpensive boards, it's also a place that you can um we have a large shared projects library where you can find other people's designs and and kind of work and order those designs too so if you're if you just need um a very simple breakout board for for an at mega or whatever then i mean that's that's a that's a place you can go search the shared projects, find one, order one, and be on your way. But you don't assemble the boards, right?

You make the PCBs, which is just the board on its own, and you don't put any of the components on it. Right, right. That's up to the end user. I mean, we have plans to offer an assembly service, a PCBA service in the near future, but we currently don't have that now. It was okay with me. I soldered and it was a good excuse to use the soldering iron and the microscope, but when I think about larger boards that I want, they would all be fully assembled because I only want to solder so many things.

And yet I did one board where I got it made and I did the kitting and then I took it to a local house or a local design house manufacturing firm. And they put it together for me in their machines. And that seems like the sort of thing that's really hard to do for small runs. It really is. For every new part that you want to do, you have to get measurements of the part. You have to figure out how it will go into your picking place machine. You have to figure out orientation, how you're going to inspect it. It's a lot of work for every new part. And you have to get the kits from people, or you have to offer service to buy pieces.

Right, and if they're providing the kits, then that means you have to keep track of inventory, you have to very tightly control, match that with the actual job. There's a lot of opportunities for error. And then if you kit them, say you have a library of parts that you say, these are the ones we know how to place, and these are the ones we'll put on your board, everything else you have to do yourself,

well now you have a whole other layer of of things to worry about and everybody's going to want their resistor in your database yeah i don't know that's so much uh their resistor but certainly um uh people are very religious about the voltage regulator they use the um the diodes that they're that they want on there um it it it's it's a, it would be a real problem. But for your standard maker, that's true of engineers, for professional engineers. For home makers,

I don't think that's such a big deal. If they have a nice Eagle library that's full of the parts that they want, or of course, Kacad library, of the parts that we can offer them, then I think that most will use that. Does that mean you have to maintain inventory?

Yeah, I'd be maintaining inventory for those and I'd be able to buy in larger quantities and offer that price discount to everyone. Yeah, that makes sense. A lot of boards are a lot of people get started with the Arduino and so many of the boards they go to

then are the Arduino base. I mean, the AdMega base. Sure. And go from there. So, yeah, you probably could keep a database of maker parts. Yeah. But are people using Osh Park for only maker stuff,

or are people starting to use it for work stuff? No, as much as my target is makers and small engineers, it's really larger businesses have become, small, medium, large businesses have become most of my revenue now. Was that what you expected when you started the company? Not at all. I was expecting it to stay kind of a maker community thing. I figured the larger companies have the ways that they get their boards and usually would order their prototypes from the same place that was doing their manufacturing, or their full manufacturing. But as it turns out,

the engineers working at these companies really like to be able to just design a board, send it off to manufacture at a price that they can put on their expense report, and then without having to go through the whole company procedure for ordering a prototype through their main manufacturer with purchase orders and all that. For that reason, it's become pretty interesting to those engineers. I can definitely see that because small startups,

if you're doing prototypes, maybe you haven't set up the whole supplier relationship with a larger CM or something, and, oh, I need five boards to test out this idea, that's a lot easier to go to Osh Park than it is to try to come up with another solution. Yeah, exactly. I see a lot of larger companies

where it's just some employee somewhere who's putting down their own credit card, and it's just some breakout board for some piece. It's some test harness board. It's some minor piece of their larger project that is not going to go into their final product, but it is going to be very useful for them while they're designing. Yeah, debug boards and test boards and prototypes. And when I've started generation two and wanted to poke at new things,

I certainly don't want to go to my normal CM and distract them. They're finally building Generation 1. Don't mess with them. Right. So yeah, I could definitely see that. How big is OSsh Park now? We're now at three almost full-time employees, two additional part-time, and me full-time. That's processing about 5,000 orders a month. And was that what you expected when you were like, oh, it's okay.

I'll make a business out of collecting the dork bot PDX boards together. Yeah, not at all. I think I got really lucky to be in the right place at the right time, that there was this need. And I had done the math and I said, oh, well, if it grows this much, then that could be a sustainable business that I could quit the day job. But I certainly wasn't expecting it, no. Well, and now you've not only quit the day job,

you're employing other people, which as a small business owner, there is some barrier there. I'm way too terrified to be employing other people, which as a small business owner, there is some barrier there. I'm way too terrified to be responsible for other people. Absolutely. That was a huge mental block. It was, first off, all the things that you have to do legally before you can hire employees. There's, uh, employees like, and there's, there's, uh, getting workers comp insurance. If you have even one employee, you have to do that. If you, there's all, I mean, there's all these very minor things. And, uh, and like, uh, when you,

when you quit the day job, uh, well now you're responsible for your own health insurance. Um, when you hire, when you bring on employees, well now you're responsible for their health insurance. So all these things ended up being really worse in my head than they were when I finally did them. But those are all definite barriers to bringing on other people. Yeah, everybody always thinks, oh, I'm going to make a company out of this idea. They forget the make a company part. Right. Yeah, really. Well, and hiring someone to be on your team at a company is always tough.

Interviewing is one of those things that's just, how do you tell in 40 minutes if you're going to like this person in four years? Right. Whoever you hire, you can't expect them to like the company as much as you do, for example. Exactly. And when it's your own company and you have a passion for it, you can't really pay people to get that interested. Right, right. Luckily, I mean, this is another spot where I just got really, really lucky. When I quit my day job and when I started doing this and when it really started taking off, I hired my first three employees very quickly, all within the same month. And I got really lucky, and they're all fantastic. It could have gone wrong in so many ways, and I'm really lucky that it didn't

yeah, you are and I'm glad for you what things do you wish you could go back and tell yourself about Osh Park before you started it? good question I think that, uh, really, I, I, uh, I, I wish I had had a, uh, a co-founder, for example. Uh, it's, it's a, it's, it was a, it's a lot of work. It was a lot of work and it is a lot of work. Um, uh, being a single, a single, a single founder company and then to be self-bootstrapped, so you don't even have investors who could potentially help you out guiding things. I wish I had gone in an accountant earlier,

so I wouldn't have had to spend all that time doing accounting myself, because that is boring work. Not very interesting. What would you have looked for in a co-founder? Somebody more business-oriented or more hardware-oriented or you but cloned? Yeah, see, I'd probably just want either a me that's cloned or someone maybe more on the business side.

I mean, really what I want is someone who's technical so that they can take over all these things that needed to happen in the first couple of years. And so you can take a vacation. Yeah. Thankfully now with employees, I mean, that's the other option instead of co-founder employees. And thankfully I'm now to the point that I can take time off if I need it.

Yes. For the first three years of Osh Park, it was getting to the point where I was working the day job for eight hours, driving home, and then working on Osh Park for eight hours. And if I, for whatever reason, couldn't do that eight-hour shift, then the next day there was 16 hours of Osh Park to do, because the work just didn't go away. Sounds like a parable. Yeah, it was rough. It was, that was really brutal. So, and then when I quit my day job, I was like, all right, I can, I finally have all this time to do that, but then I, I'll finally

have free time. But, uh, then I, when you're working eight hours at the day job, eight hours at the night job, then that means that when you quit one job, you still have your full day. Yeah. And you still have all of those things that let fall all of the the your family and friends are sort of understanding when you're working really hard but when you now they all want to have dinner with you and and you haven't seen them for a long time either so right yeah so why Right, yeah. So why purple boards? I like the color. I was, when I, the very first panels of Osh Park, of the Dorkbot PDX order were green.

And that's ugly. The traditional color. The traditional green color. And that's boring and that's ugly. And when I, so I went to my fab to see what other colors they could do. And they said, well, we can do blue, red, yellow, black, whatever. And so I did blue panels for a few times. And then I was like, well. And then somebody brought to one of our Dorkbot meetings this little development board from Japan.

And it was this beautiful, brilliant purple color. And I was like and it was this beautiful brilliant purple color and I was amazed I had no idea boards could come in that color and I said well I want that with green boards or blue boards when I'm going through my Flickr feeds and RSS feeds

looking at people's projects I couldn't tell if I was a part of that and that's very important to me to see the projects that I helped build in my own small way. And so having the purple boards let me very clearly figure out that, aha, that board is very likely one that I made. And it was a big source of pride. And now they're pretty iconic. I didn't know that I would use purple in a system now.

It used to be on my list because I always did a versioning thing where we'd usually start out at green and then either go to yellow or blue. And so you could see if you were on board version six because you just counted through the rainbow. Yeah, yeah. But now I would skip purple because I would just assume that that was a prototype board that I got first from Osh Park. Yes, yes. Certainly, there's a mix of people who are like, well, you know, purple's done such a good job of prototypes that when I see a purple board, I think this is a prototype.

There are other people who go, well, with the purple board, this allowed me to really make the circuit board the centerpiece of my project because it's gorgeous with the purple and the gold. So I get a mix of both of those things. Does anybody know why green is the default color? I think it was, I'd kind of look this up and there isn't a clear consensus on why it was. Nowadays, it is mostly because of the history, but also because it's one of the quickest curing times when they're photo imaging it. It absorbs light quicker than other colors to cure faster. But is that a chicken or an egg problem? I mean, is it because they've sort of tuned the curing process for the green? Yeah, I think it is a chicken and egg problem.

I don't see any reason. Well, actually, let's see. So it's a UV curing thing, so I don't know if purple would absorb more of that or not. But certainly light needs to be able to go through it, and the green is kind of a thinner color because it needs to cure all the way to the board. It is definitely more translucent, the green color. Like there's a color of solder mask that's just clear. It has no pigment in it, which I think is also very pretty.

It cures very quickly. It shines right through. The UV just cures it very quickly. Excuse me, I'm typing clear PCB. Yeah, when we add a second color, it's probably going to be clear off the bat. Oh, man, that's beautiful. Isn't it? Because you get to see the like this gorgeous copper color

and uh like i worked at at intel like 20 years ago and and they they put out one of these boards that was just this beautiful copper color and i and i didn't know anything about pcb manufacturing back then or the solder mask or any of that but i um uh someone said yeah that's clear and it's kind of stuck with me ever since because it is really nice. We can see those beautiful copper traces all shining underneath the clear solder mask. But it is a little visually busy. It is. It's good for, say, one of the complaints. Two layers would be okay.

Four would be mind-boggling. Well, I guess that would probably have planes, so it wouldn't be that bad. But yeah, one of the complaints people have about the purple is that it's hard to see the traces underneath of it in order to debug your design. So with clear, you can easily see where your traces are going. Oh, I routed that one.

All right, so plaid is probably not going to happen it definitely can it actually can one of the things we've worked out with our fabs is doing multiple mask colors per panel so the idea being that I'm probably not

going to get a lot of requests for blue or green or whatever certainly not enough to fill a panel but perhaps enough that I could run part of the panel green, part of the panel purple. Here you're finding out what software engineers think is important about PCBs. Well, yeah, can I have a board that's on the line between them? I want the one that's sort of ombre faded. So I've worked out some code for my panelizer that can

mix it. You'd have to do kind of a dot pattern, a dithering pattern in order to do that. But you can certainly do both of them on there. And so with that dithering pattern of some dots of green and some dots of purple

or whatever. This will totally make my hardware run faster. Speed stripes, got it. Exactly. Osh Park now with speed stripes. That's 200 megahertz. The board's 200 megahertz? What?

Yeah. I guess I was the one that started the board's going to run faster without saying what was on it. So what about Other Mill and the desktop CNC machines that people are making PCBs just on their desk? I think that's awesome. Anything that makes boards easier to make is great. Of course, they have the main limitation

that makes them not a scary thing for me, but instead just a very cool thing, is doing plated holes is not possible with those. So you have to put your own vias through there to connect the tops and bottoms of the boards because they're just two completely separate sides. They're very cool, and it's great to be able to reduce the manufacturing time. Being able to reduce turn time is huge for any industry

in order to reduce your development time. And the other mill does a great job on it. They can't do the plated through holes. And it's only two-sided, and you do two-sided and four-layered, right? Yeah. Most people can do their designs in two layers for our service. Like our service, the two layers is vastly more popular than the four-layer. Well, I guess the Eagle usually does two layers if you're using one of the free versions.

Yeah. Let's see. I went to your prices. It's $5 per square inch. Which includes three copies. So if you're a one-square-inch board, you'd get three copies of that for $5. Okay. Which is actually pretty cheap.

Yeah, that's with free shipping and everything. You're getting your one square inch board in the mail in under two weeks for $5. How much does it cost to get like a 300 square inch panel done? A lot more. But yeah, I mean, because I'm sharing with everybody else. Yeah. Right. sharing with everybody else. Yeah, if you're doing that, getting it yourself, it's

really a big cost, which was a problem during the first days of the circuit board order where I'd fill up maybe a third of a panel. And that wasn't anywhere near enough to cover the cost of the whole panel. That's sort of scary.

But you still have to ship it out because people are still waiting for it. Are still expecting it. And one of the things I was doing back then, and still now, I thought was very important, was having a known time that you're going to get your boards back. Nobody wants to submit their boards and then to order boards and then go, well, I guess it'll be back sometime in the next month. Could be a week, could be a month. People want a firm schedule. Like, hey, if I order my board by this date, it will definitely ship by this date.

And so back then I was like, okay, well, this is the date I committed to. It was this date, it'll go to fab, and this date it'll be back. So if it's a third fold and that means that i'm i'm filling up the panel with whatever stuff i want built and then uh and hoping i can do something good with it do you still do stuff um for yourself i mean it's so hard to go from having this hobby to making your hobby into a business and now are you back to having a hobby where you get to use the board that you're making well for um for when i first started this i was just getting into the electronics hobby i was i was i'm very excited about doing my own boards and this was this was my way to do it um i really hate breadboarding so i'm really not good at it i

mess it up um a lot and same with like a lot. And same with doing wire wrap style stuff. So this was my way. I could just design it on the computer. I knew that it would be electrically correct. I saw it on the parts and do it. But it very quickly became too busy for me to do this. I was spending my time doing the order and didn't really have time to do it now.

Now with my great employees, I'm starting to build again. So yeah, I can finally start using it again. Are you excited about it, or have you just found a new problem that needs solving? No, I'm very excited about it. Because this is, like for me, I can always, like I get them a day faster than everyone else because they show up here.

So I really love to be able to just kind of go, like spit out a design. Like yesterday I was like, oh wow, the Teensy microcontroller has these really cool capacitive sensing um features so i just i just spun up a board with a bunch of uh a bunch of like touch pads on it and and plopped it onto a panel and and i'll have it back uh on wednesday they're funny that i mean that it used to be i still part of me thinks well part of me also thinks that cell phones are new, but part of me thinks that it should take three weeks to get a board back.

Yeah. And then you should have to send it out for assembly. part. So, I mean, hand assembly is not such a hard problem. But yeah, I mean, it still takes, like our average time from ordering a board to shipping it is about nine days, nine or 10 days. So you're still looking at a couple of weeks. Do you work with multiple board fabs? Or are you kind of, sort of, are you going to go for the vertical integration and buy your own board fab soon? Well, I mean, eventually, it probably going to have to go that way. As it is, I work with three different fabs all around the U.S.

Since before I would allow myself to quit my day job, I wanted to make sure that I wasn't locked into one fab who, if that place had a fire, for example, that I would be out of business. So yes, I added another fab and in the process added, and since then I've added yet another fab to kind of distribute the work between all three. And do you know them? Are you just one of a million customers for them or are you starting to be a significant part of their work too? Yeah, I worked out that for the three fabs I use, I'm their single largest customer for all three of them. So that does give me some leverage there and it lets me

really get to know the fabs. And it does start making you think about having a personal fab. Yeah. I mean, that means you probably do sum up to a whole fab or will soon. Yeah. I mean, running a fab would be a really rough thing. The process of producing circuit boards is a messy process. It's a very harsh chemical process. It's a process that if you're not continuously upgrading your equipment, you're going to

fall behind. It's one that I'm perfectly happy to let the fabs do for now. And with my increased volume, I can negotiate them down on price. So that's not a huge deal. But yeah, it's in my two-year plan that I'm going to have to start looking for an acquisition target there, I guess. Wow. That's pretty cool.

It is. It's really scary. I am a homemaker slash geek. To look at things at this scale seemed totally weird. And now we're back to this business wasn't really what you expected, was it? No, I certainly was prepared for it. I mean, I had done the numbers and I'd looked at my growth rate and I still do this on about a monthly basis just to kind of extrapolate, okay, well, if this is where I'm at now and this is the current growth rate,

then in one year, two years, and five years, where will this be? And what will I need to do to deal with that? Wow, I want to read your business plan. What missteps did you have? I mean, what things went wrong? You've made it sound pretty, wow, I just tried it and it all worked. No, it was a lot of pain. At the very beginning, putting together a circuit board panel with a bunch of different designs and to the point that it can actually be fabbed properly ended up being a tough process. Like when my first panels, I had no idea.

I just threw all the boards down and sent them to be manufactured and I'd get them back and there'd be all this evidence of rework on the panels. Like the fab would go through and be fixing all these places where there were otherwise be shorts or breaks. And at the time I was like,

I didn't think much of it. I was like, wow, good thing that they, good thing they have inspection. But the fab was getting really upset with me and I had no idea. And until I submitted my third panel and the sales guy calls me on the phone and goes, look, our production manager, our floor manager says your boards are awful. Your panels are awful. And they're a pain in the rear to manufacture. We'll continue doing them, but it's going to cost you, I believe it was 10 times as much. Oh, my God.

And so I was like, wow, my business is dead. There it goes. I mean, that's it. But I took one of my panels, one of the files from my panels, and I shopped it around to pretty much every fab in the US. And there's about 300 of them. And I found one who, and finally one of them said, yeah, you know, we'd do it, but these things are really bad about it. And I was like, oh, of course.

Well, it doesn't have to be like this. I can do this any other way. And really the main thing is I was filling up this empty space. Since the panels had a bunch of empty space, I'd put protoboard, perfboard on it. Perfboard is a circuit board that has just a bunch of plated through holes at 0.1-inch increments to provide a very easy prototyping platform. It's a really useful little piece of board,

but it meant that my panels would take eight hours to drill, and that shuts down. Most fabs only have a couple drill presses, so that piece of equipment was completely in use for eight hours. I'm respecting all of the perf board on my desk a lot more right now. Yeah, it's like average drills per inch on a circuit board is about 20 drills per inch. On a perf board, it's 100 drills an inch. So if you're in a whole panel like that, that's five times more than the fab is used to dealing with. No wonder they hated you. Yeah. And that right there was the main thing that if the first fab had just told me that,

I could have adjusted it and it would have been just fine. But there were other things. Like if you take one board that has a very high copper coverage on it. It maybe has ground planes. It has lots of traces. And you put it right next to a board that has maybe just a few traces and no ground planes and not much copper on it. The first one gets seriously under-etched, and the second one gets completely over-etched, which causes shorts and it causes breaks.

So that was the evidence of rework from my first fab. And so the guy says, hey, you know, if you fix this, it would be a lot easier to manufacture. And so that really threw my panelizing algorithm for a loop because I was just trying, I was optimizing for space filled, not copper coverage. Tetris, yeah. Yeah. So, I mean, I needed, and so like my first few panels after that. You needed another axis.

Right, it's another depth of the bin. It's this type of problem is called bin packing. And this kind of turned it into a 3D bin packing or a weighted 2D bin packing algorithm. And so that made my first few panels after that were very inefficient, but fabs were more willing to do them. And so I found a fab that would do them, even though it was about twice the price of the fab that I was using before. It let it continue. It let the company continue. Sounds like they need to get their act together.

I mean mean come on if you drill a lot of holes then learn to drill a lot of holes why didn't the first fab tell him this that's true well the first fab um is is one of the largest in the u.s it's uh it is um they didn't need my business they all they knew is that i was throwing these crazy panels at them and my panels are completely unlike any other job this fab would ever get. Right, because there are a whole bunch of different things all over the board instead of one of, or maybe three boards combined. Right. Most places order one design.

Yeah. I was ordering a hundred designs on a single panel. And they were like, this is, this is messed up right here. We're not doing, we don't want to do it. But thankfully I found a fab that, that was willing to.

And, uh, and I, over the years have, have now become their, their largest customer. So I'm, I hope they're glad.

I'm certainly, certainly thankful that they, uh, that they were willing to. So the first guy's just set to forget it. Yeah. They just didn't want to talk about it. What about the ones who said, here's what you're doing wrong? It was still out of what they were comfortable making,

even when I was fixing these problems. Just due to the size of the panels, the complexity of the panels, they said they were very kind about giving me that piece of advice, but even with that, they didn't want to do it. But you said, like, thank you and sent them a brownie basket, because not everyone will tell you that you messed up. It was incredibly nice of them to do that. Because they didn't have to.

They could have just said no like the other 299 did right and there was there was a uh and and right around this time there was a uh a blog being run by a a uh a fab engineer um i think it was called pcb design school and uh pcb design school uh shoot i forget the name but yeah he did this this this wonderful blog that had all these, like each entry would highlight a step of the manufacturing process and kind of talk about what the things are. And I emailed this guy and said, look, I'm having a really hard time finding someone who will manufacture my panels.

Can you give me an idea why? And he gave me the most in-depth critique of any of them, which was fantastic, because it enabled me to fix the process and eventually get it to where it is now. That's cool. I mean, it goes back to people are what help you make a business absolutely work yeah

and having somebody who is willing to look at this odd customer and and think well maybe there's something here you know

if we can work this out maybe maybe this will grow into into a you know an important part of our business so yeah

having some vision there is finding somebody who's willing to do something a little outside their comfort zones yeah any other catastrophes that happen that you want to tell us about every now and then there's a there's a there's a catastrophe and and it's like and it's always very shocking when it happens like like i've i once got a panel back and it was, uh, uh, it was completely wrong. Like the, the, like it was somebody else's. It was, it was one of my panels, but they had manufactured it completely wrong. And, um, that is thankfully, thankfully not my fault. So the fab stepped up and fixed it very quickly, but, um, I, I pride myself on, uh, on fast panels and reliable panels.

Like, you know, your board's going to work. And we've had one situation where it just didn't catch it in time. People are getting their boards, and that is just so embarrassing when that happens. And that's, of course, a catastrophe with all those. Because it's like, whoa, this is totally my fault. I'm refunding the whole panel. And that's a big financial hit. But it's one that I really felt I had to do since it was such an obvious bad thing.

You do have to balance that. I mean, there's the embarrassment of it and the financial hit that, I mean, that that's hard to eat. Yeah. On the other hand, if you don't do these things, people may not be loyal.

And most Osh Park people are pretty fanatically loyal. Yeah. It's, it's, it's, it's amazing. I don't know that I deserve it, but,

um, really, uh, like one time I bought something off eBay, and I got a good deal on it, but it turned out that the shipping price was like an LED matrix when I first bought it. It's a $1 LED matrix, and I was like, wow, that's a great deal. And it had $4 shipping, and I said, okay, well, I'll just buy 10 of them. And then that's $10 with $4 shipping, no problem. And so the guy says, no, that's actually going to be $40 for shipping. And I was like, okay, well, then I'm sorry.

I didn't know that I want to cancel. And he got really irate. And it occurred to me that the reason was that he wasn't making enough money to provide good customer service. And it was really from that experience that I decided that really the main way that I can compete with any other fab is by just providing fanatically good service. Is there a seasonality to the boards? I mean, do you have months where things are heavy and months where things are light? It tends to be pretty consistent.

There's, of course, Lunar New Year. Chinese Lunar New Year is a big burst. Because everybody's building boards here because we can't get them anywhere else. Yeah, because China is closed for really a period of, it ends up being like four weeks when you include the shipping times and things like that. So yeah, there's of course that period. And strangely, right after the beginning of the year, November and December are kind of low. And the home hobbyists have more time, but the companies are ordering a lot less.

But then in January, it just spikes. And I imagine part of that is Lunar New Year stuff. And then prior to every Maker Faire, there's a little burst of traffic. I want to ask you about Maker Faire, but before that, is there any other data associated with building the boards that is strangely interesting. Hmm. I mean, like the seasonality, or you get a whole bunch of people wanting this sort in that time,

or at megas only come in bursts of 10. Yeah, there's, let's see, so definitely like ordering patterns from around the world I think are very interesting, because like we'll get Definitely ordering patterns from around the world, I think, are very interesting. Because you'll see a burst of like, okay, in this, for example, this Eastern European country,

all of a sudden there'll be just this big cluster of orders. And then that country will just kind of stop for a while. And so you know that there's just a bunch of makers in that country who are all starting on something. And you see this little burst of activity, and then it calms down. And you'll see these very localized bursts as hobbyists, I think, are getting together, learning how to design a circuit board, or working on a project together and doing these things.

So Maker Faire, you were at the Bay Area Maker Faire last spring. I love Maker Faire. It's my favorite conference. Conference. I guess it is a conference now. It really is. Yeah, I don't know what else to call it it's I mean it's a show and tell right and it's it's really awesome to be able to see uh like because this is my time to see what what my customers are making and um like at Maker Faire it's not a requirement that things have utility it's not a requirement that things have a commercial use. A lot of these are just art projects

and that they've done something really interesting with their boards. I'm thinking of somebody did this pendulum that is just precisely motor-controlled such that when it swings this pendulum down and then it just intricately stops it at the other end. And it's like, well, that's really cool. I think that's really cool math that they put into it. I think that's really cool projects.

And so that's what I really love about Maker Faire. Are you speaking this year or having a booth? Probably not having a booth. Really, I like to be mobile. I like to be able to see all the... I want to see everyone's projects. Yeah, if you have a fixed thing, you don't get to see much. Exactly.

I don't want to be sitting there. Really, my customers are the people who are showing off at Maker Faire, not necessarily the ones who are walking around Maker Faire. So those are the people I want to visit and those are the people I want to see. That makes sense. I mean, having a booth is a way of selling things, but it depends on whether you want to go. It sounds like you go for inspiration almost. Yeah. I mean, like I said, the reason that I made it purple was so that I

could see what people were making with my projects and with my boards. And so that's very important to me, to see what people are making and see what I'm a part of in some little way. And are you speaking this year? I haven't submitted a talk. I don't know if that's too late to do so, but I'd like to. All right. I can't help you there. I have no idea.

What should I talk about? That's always my problem, is coming up with the topic. Well, there are so many things. I mean, building a business is completely not trivial. Or turning a maker-focused business into something larger. Maker Pro came out, the book. I know Sophie was in it and a few others. It was an O'Reilly book.

Maybe it was a Maker book. I don't know. Were you in it? No, not Maker Pro. Okay, so it's a book about people who are professional makers, and it's all essays about how they got into it and what they do. And the thing is, it's sort of cool because people who you might not expect and how they do it and what they mean i mean there there are the makers who grow their own food and have no impact on the

infrastructure and sure they get the title and then there are people like sophie who are professional engineers who in their spare time like to build things art projects and and interesting things for themselves and those are makers too yeah and you can make a living at either end of the spectrum yeah and i find that amazing but you are also making a living on the maker spectrum because you're making it possible for everybody else i've always kind of been a support person so like the uh when i was a unix admin that's really a a support position other people are doing are they're writing their apps to be just deployed on your platform you're you're uh they're coming to you with uh with problems and and that that's

working on other people's problems is very attractive to me. So I kind of consider this kind of the same deal. I am a maker of support as much as I am a maker myself. And it would be hard to talk there about being in the support role without it being an advertisement. I have some problem with that sometimes, is that I don't really know how to give a talk without having at least a little bit an advertisement. I have some problem with that sometimes. I don't really know how to give a talk without having at least a little bit of advertisement.

Yes, and I'm very uncomfortable with self-promotion, so that would be tricky for me. Yeah, I understand. Yeah, so just email me your ideas for a talk and I'll see what I can do. Write it up. He'll read it. Write it up, I do write it up yeah we should submit it too shouldn't we i guess i don't know i think that there are lots of things you could tell people even if it's here are the ways to learn how to do a board and let me help you build it or don't let me help you build it. Just go out and build a board. I think you could easily make a talk about all the stuff that got you interested in building boards as well as the scale of building a business.

I think, I don't know if they're doing it this year, but sometimes Maker Faire has like a pre-conference for the more business oriented things, more along the lines of what you do, which is to support the Maker community. That's true. So we should look and see if that's there. I should go to that. That'd be kind of fun.

I went there the first year. It was cool. And you also, last year, sponsored the Bring a Hack dinner. Yeah, that was fun. Which was a Maker Faire after party. Yeah, no, that's one of my favorite parts of Maker Faire is the Bring a Heck dinner. Jerry Ellsworth's been running it for years. And it's so cool because it's everyone that you want to meet at Maker Faire all crammed into one room with all their coolest projects that they've been working on for the previous year.

And it's fantastic. I love it so much. I look forward to that every year. And last year is the first year that the company was really successful enough that I'd quit the day job and had employees and all that. I thought, well, I guess I can buy a few pitchers of beer. I can buy a few pizzas and and uh and kind of give back to it and you did i mean i i was sort of shocked i was happy to pitch in for beer and pizza and and you were just like no wander off yeah i mean why not it's uh

it's uh since it's it's it qualifies as an advertising budget. It's mostly written off on taxes. And it's such a great group, and I get so much out of it that it's totally worth it to me. Well, and for me, I have trouble going to such parties. I've been trying to be better about it. But there's too many people all in one space, but having the bring a hack part of it, I brought Maxwell, my little, are you okay widget. And that made it so much easier to say, Oh, what, what did you bring? And it was a lot easier to talk to strangers.

Well, that's the great thing about Maker Faire general is uh is you you always have the uh the the question what do you make what do you like to make and um and that's you always get an interesting answer to that even if someone is is not currently making anything they go oh well i'm interested in like doing audio stuff and uh and and especially the bring a hack dinner where you actually have your project with you it's like yeah check out this this weird synth I made, this weird effect that listening to the capacitive touch of this touchpad or whatever. It really breaks the ice. Everybody is very interested and friendly,

so it's really a fun part of the conference. And so the pizza place where you had it, did they just not believe you about the number of people expected? I mean, I heard you tell them 50 and then you moved rooms a little while later because it was clear that there were going to be way more people than the little place, the patio they put us in.

Yeah. Plus the patio was getting cold. Yeah. So like I had, I had called it up previously because Jerry had said, or the old place, which was Harry's Hofbrau or something like that, closed down. So we couldn't go to that place anymore.

And this seemed to be the only place large enough to handle this. I really thought it was only going to be 50. I think it was more like 250. That would explain some things. Yeah. And, of course, now that i've talked it up there's probably going to be even more this year i'm sure it'll be fine but uh yeah i haven't said where it will be yet definitely right definitely going to be reserving that that that place in a

much in a much larger capacity because that whole backroom area was great but also i'm not sure they're going to want us back some people like, like it had the 12-volt track lighting. And somebody didn't have, did you see this? No, I did not. They didn't have their power supply. Oh, no. And so they jump up on the table with alligator clips, the clip to the track lighting.

And, oh, geez. Remember, people, leave things better than you found them one of the waitresses like like slap them down it's like no no like what are you doing but it was uh yeah so we'll we'll see what they say everybody behave please they're probably gonna have my credit card this time you what they're probably gonna have my credit card this time as some sort of deposit so be nice oh i would love to come again so if you do do it yeah i look forward to it i think we're about out of time christopher uh we still have questions so we this is like checks we can't be out of time because we still have questions exactly excellent uh okay what's up with lane are you trying to be a palindrome what's up with you

so usually your name when it's a cap when it's an l it looks like i aN, which for a while I decided must be E-N. Yeah. But. So Lane is my last name backwards. Yeah. My real name is James Neal, which is a pretty common name. And I like to be distinctive. I like to have.

I mean, if you say Lane, you pretty much know who you're talking about. There aren't many lanes and lots of James Neals. So that kind of happened. My name, Palindromic, just doesn't work in any fashion at all. So you got lucky. Yeah, you could do just Topher backwards. I don't think that comes out in a pleasing way. Red Pot?

Exactly. Red Pot Search? Okay, maybe not. And mine already is the same backwards and forwards. What? What? You totally blew my mind there for a second

I really believed you come on it's just a bunch of vowels only if you spell your nickname with one letter that's true alright

well even though we have more questions maybe we'll have Lane back on again yeah awesome I'd love to any last thoughts

you'd like to leave us with this time? No. Thanks a lot for having me on the show. This was a lot of fun. Good. It was good to have you. My guest has been Lane Neal, founder of Osh Park. Thank you to Lane for being on the show

and thank you to Christopher White for co-hosting and producing. It's always nice when your producer loves you. Takes out those words you say improperly. And if you'd like to say hello, hit the contact link on embedded.fm or email us, show at embedded.fm. If you'd like a sticker, I still have some.

Send us a note. Tell us you told somebody about the show and include your address, and I will send out a sticker eventually. Thank you for listening. And now I do have a final thought to leave you with. And this one's from Amanda Palmer. And I don't know if you remember Chris Veck, he was on the show a while ago talking about empathy driven development. He suggested reading The Art of Asking, How I Learned to Stop Worrying and Let People Help.

And it was a good book. It was weird, but good. And so the quote I'm stealing from there is, in both the art and the business worlds, the difference between the amateurs and the professionals is simple. The professionals know they're winging it. The amateurs pretend they're not.