Embedded - 129: Atoms Not Bits

Episode Date: December 9, 2015

    We ask Matt Berggren (@technolomaniac) to envision an ideal world of product creation.  Matt's Spartan 6 FPGA shield is on the Hackaday Store. Its build process is documented on Hackaday.io. You can see Matt's other projects on Hackaday.io/matt. Matt works for SupplyFrame. Wishbone is the open source bus system that Elecia likened to ARM's AHB/APB.

Welcome to Embedded FM. I'm Elysia White, here with Christopher White. On the Supercon show, I said I didn't ask Matt Bergeron into the car of interviewing because he'd get his own show later. Well, later has come, and Matt is joining us this week. Hey, Matt. Welcome. Hi, hi. Could you tell us a bit about yourself?

Yeah, absolutely. So my background is predominantly as an engineer. I work for SupplyFrame now, which a number of people might know is actually the parent company which owns Hackaday, Hackaday.io, a bunch of other things. Spent, I don't know, 15 years or so working in the EDA space. Was director of product at Altium before I signed on to SupplyFrame. And run a bunch of workshops in the city, trying to teach people how to build electronics.

In addition to that, on the board of a company, started between myself and Nick Martin, who was the old CEO of Altium, doing IoT and connected devices, but really targeting the Chinese market. So it's a company actually based out of Shanghai. So I live a pretty full life. Well, I think that was a good introduction.

Although I have plenty more questions. I wanted to do an experimental thing. I hope you are amenable. We're doing this now? Yeah, I wanted to start off with the lightning round because I figured, you know, we'd get some more ideas for questions to ask. Okay. Not that I didn't prepare ahead or anything. Fantastic. Fantastic. So I'm looking for one word answers and we're gonna go as fast as possible okay okay the first one should be easy what is your favorite airport burbank which schematic capture should i learn for documenting home builds not for making products it hurts but eagle

what's your favorite programming language c what programming language should colleges teach first in their cs major curriculum these days i'm going to say javascript oh oh i know i know i know i i shiver when I say it. What is your favorite software keyword? Malik. Oh, man. We're going to have so many things to disagree about. Verilog or VHDL? VHDL.

Is a mobile phone an embedded system? No. Favorite city to visit for fun? Chicago. Domestic or international? Chicago, hands down. Oh, we agree on that one. No commentary.

Which schematic capture should I learn for making products? Altium. All-time favorite Hackaday IO project? I really like the iDrive-o-matic that the guys did this year. Yeah. Do you already have Star Wars tickets? No. What?

They weren't available. They sold out. Favorite word in a foreign language? Ooh. Kui. Kui, kui, kui. I think we need collaboration on that. Okay, so that's

a Mandarin word for yes, you can. But when I lived in China for a bunch of years and there are a lot of words which are repeated a bunch of times

in a sentence. And it's like kui, kui, kui, kui. So, hey guys, can you solder these 0201s by hand so ah yes it's like saying cool cool yeah it's like i can i can i can which is great because it means that something is about to get on okay hufflepuff or gryffindor oh uh gryffindor mac or pc i'm pc android or ios ios han or chewy sorry what was that han or chewy oh han all the way. What? No. Oh, yeah. Oh, yeah. Yeah. I mean, he's, come on, he's a scoundrel. Chewie's an engineer.

Yeah, but Han's a scoundrel. Like, he's what engineers would aspire. He's a consummate hacker, right? No parts were original on the Millennium Falcon. He's the guy that pieced together this thing that, yeah. Yeah, Chewie was the guy who pieced together the Millennium Falcon. He's the guy that pieced together this thing that, yeah. Yeah, Chewie was the guy who pieced together the Millennium Falcon. True, true, true. Han was the one who kept bashing it up. Yeah, well, I don't know.

I just, you know, there's just something so cool about Han. I just can't get over it. Okay, final bonus question. Beaches or mountains? uh mountains all the way all right cool i think we understand you completely now fantastic i'm fully vetted okay what questions should we add for the next person's lightning round oh um uh object oriented or procedural oh that's a good one that's a good one so i mean just you know uh like like that kind of stuff is just genius uh for me um because you know coming from predominantly c background that stuff is really interesting to me. I can't stand high-level languages.

I'm sorry, and I say this as a guy who spent his career in EDA tools, which are, of course, all big object-oriented projects. I'm sorry, I just love C. I really love C. So Altium's written in C++?

No. Altium's actually written in Delphi, which is an object-oriented version of Pascal, which is... I know, I know. So imagine Altium designer. And Nick and I have had Nick Martin, who's the guy that founded the company, since gone on to bigger and better things.

He and I have had this discussion a number of times. And at that time, Pascal, it was really Borland who was pushing Pascal. And they were successful in a couple of places, Eastern Europe, France, and Australia, to name kind of probably the places where they've been most successful. So they sort of won the war in the early days of these languages. And it was at that time really Pascal, not even object-oriented so much. It was all still using pointers very heavily, that sort of thing.

So he started writing autotracks actually in Pascal. And autotracks eventually became Protel. Protel, at that point, when they moved to Windows, they were doing a lot of object-oriented programming, obviously at that stage. And Delphi had sort of emerged as Borland's solution to object-oriented programming for Pascal developers. And so all of Altium Designer is actually written in Delphi.

Now, more recently, they added some support for C Sharp, certainly through the API, which signals to me the fact that they're going to do probably a lot more development in C Sharp and try to move away from the Delphi IDE and into Visual Studio. It's probably pretty hard to hire people with Delphi experience. You know, the core development team right now is largely coming from Ukraine.

And it's actually reasonably easy to find Ukrainian developers who are some of the savviest people I've ever worked with capable of writing Delphi. But there's a lot of downside risks to Delphi. But there's a lot of downside risks to Delphi anyway. You really have then to track with... You have one order of removal from what Microsoft is doing on the Windows platform with DirectX and all the other sorts of things,

which all team designer depends on pretty heavily. And what the developer of your IDE and your software development tools is actually tracking. So Visual Studio will always have those capabilities first. And then Embarcadero, which now is the company that owns Delphi, which to me kind of sniffs of a Smith Micro, like one of these companies that buys these

dead languages at fire sale prices. Gosh, I can't believe I'm saying that out loud. Those companies come in and snatch them up and then they've got to maintain and develop the software for that. 64-bit support took so long

for Embarcadero to add 64-bit support. Where Visual Studio, it was like, okay, we need to add 64-bit support for Windows, so we need a development environment because we're Microsoft, and yada, yada, yada. So I think that's the big downside risk. It's not so much, because it's a language,

and at some point you're going to rise above the language. But then you're really struggling with, hey, is this supported? Is that supported? There's this new capability in the operating system that I really want to be able to leverage in the application and does my IDE, do my development tools actually support that yet?

So, you know, if you're two years waiting for 64 bit support, which you were, that's pretty tragic. That's pretty risky. And even today, Altium Designers is still 32-bit. Well, that makes it difficult. I mean, I know that there are huge communities still seriously running Windows XP because it is what they have. Right. And so many things still, that is the

best operating system, the most stable operating system for them to run on right but some people are still designing embedded systems with pics that should be a question yes yeah i've got strong opinions i've got strong opinions on that one i'm not sure that them designing with the at mega is any better. No, I'm mostly trying to be inflammatory. We faced this at Altium. When we moved the corporate headquarters from Sydney to Shanghai, we had something like 90% or 95% market share, but almost no revenue.

Which signals the fact that you just had everybody using the software for free, right? Downloading a crack copy and it would propagate that way. But you're absolutely right. The vast majority of people there run Internet Explorer 7 or were for the longest time. So any sort of new features chrome-based features any of those things that would kind of be be really great and kind of valuable features for

people in in uh predominantly sort of developed markets um or more developed markets um you have to be very slow to add those. Adding capabilities to the software, one of the things that we constantly got from our Chinese partners and the resellers and the people on the ground there was, hey, is there a way to take out all of these other features? Because they slowed down my computer and I run Windows XP, or I run, even worse, like 98 SE. And you're like, whoa, okay. So if you want to move the software forward, you've got to take that into consideration. Not everybody has this $2,000 machine or $2,600 machine.

And that does limit what you can do as a software developer, certainly, and as a company trying to propagate sort of new features, especially in the EDA space where it's so much a box ticking exercise. Like at this point, I'm not sure that there's a huge amount of differentiation except at the user experience level, to the extent that we're kind of all converging on the same tool, right? Oh, they have differential pairs. Oh, we need differential pairs. Hey, they have length tuning for differential pairs. Oh, well, we need length tuning for differential pairs. Oh, and by the way way they want series terminations and oh okay well we'll add series termination so we

can compete with cadence or mentor or zookin or you know any one of these sort of things so um you know how you wrap up those features though and make those features available to somebody and then and then uh you know the sort of the converse to that argument is that in those markets, labor tends to be cheaper. So people manually doing link-tuned differential signals with series terminations in them, it may actually be worth the cost of hiring somebody rather than paying for the upgrade, for better or worse, unfortunately. So you went from LTM straight to SupplyFrame, or was there stuff in between?

So interestingly enough, I'd been talking to SupplyFrame while I was at Altium, largely because SupplyFrame had this extraordinary repository of data. Here's this company that owns fine chips and owns, well, now Parts.io. Parts.io didn't exist at that point but a lot of these great tools for like data sheets and pricing and availability information all these other things and we were looking for ways that we could partner um but at the point which the board

sort of took a hard line uh and without getting too distracted, suggested that our investment in connected devices and a number of other things, though probably great, didn't really maximize shareholder value. They had voted to remove Nick as CEO. And that was back in 2012, October of 2012. When they did that, I decided that, you know, at that point, if we're not really going to be chasing after the sort of other technologies, right, the sort of blue ocean technologies, these things which might be potentially transformative, but really focus or double down instead on core schematic and PCB capabilities that, you know, I kind of wasn't interested anymore. So I had left and Steve Flagg, who's the CEO of SupplyFrame, and I had left actually to start this company with Nick.

And Steve Flagg, who's the CEO at SupplyFrame, reached out to me and said, hey, you know, if you're in free agency, you know, we'd love to have you, we'd love to have you come in and just see how, you know, how we could collaborate in some way. And I did. And I have so much respect for Steve and what the SupplyFrame guys are doing that it was kind of a no-brainer for me because I think fundamentally so much of what's happening in the engineering space comes down to, well, so much of the next generation of technologies

for making electronic products depends so heavily on data that SupplyFrame, who is really at the center of the data made absolute sense for me it was like okay yeah you know look if if what you're telling me is you're tracking 100 million data sheets and 100 000 reference designs and you've got pricing and availability information from all these different distributors and things like that i can probably do something with that information that you're not currently doing today um so you know being the the you know core of my core being is all about trying to transform this space i want to try to make electronics accessible to other people and uh steve and the guys at SupplyFrame, Alec Brodich, just had come on very recently then as CTO.

He comes from Vertical Search. And fundamentally, the content play is about this combination of having really great content and then being able to search and present that content in new and kind of interesting ways. And that, to me, seems like it has the potential to be truly transformative in the electronic space. So that's when I signed on. What is your ideal view of somebody using what you're envisioning look like? I mean, I know what it looks like to design a board and pick parts today and going to various websites and trying to. So take me through what you want it to look like and what you want the experience to be.

Well, so I think my idealized view of this is really to look like what you want the experience to be well so i think my idealized view of this is really to be able to operate from the top down and it really has to do with what the distribution of the market is um right now you've got roughly 30 software developers for every hardware engineer in the world and so much of what this uh and it's kind of a throwaway term now, IoT. I actually don't like to use it because it harkens of mobile phones controlling toasters and things like that. And it has nothing to do with that. But so much of this next generation of products is really about building software applications that can reach out and touch the real world.

And the real world is fundamentally hardware. It's fundamentally analog right this is a real experience we're dealing with atoms not bits at that point so you know it's kind of crossed that uh that great divide there i think what you really have to do and you know i say this coming from schematic and pcb so it it actually sort of calls into question the whole early part of my career. I would love to be able to look at this from the top down. And that means, you know, keyword search that can associate a capability with a keyword and encapsulate that whole capability. So take as an example, temperature sensing, right? Yeah, of course, there's a chip at

the center of that, but that chip requires a couple of other components to be able to support it. It's got a bus interface on it, probably SPI, I squared C. It's got some parametric things that are sort of relevant in that context, right? Like, it's range. Like, are you measuring an oven or are you measuring the sun? Those are very different things. Exactly. So what's the low value?

What's the high value? What's the gradation in between? Because in biology, I may be concerned with a much more sort of logarithmic interpretation of that data. And what kind of power are you putting to it? That would be another of the parametrics.

Exactly. Exactly. And, you know, what I'd like to try to do, I mean, the ideal for me would be that when i implement a software api in javascript and i say i want to sense temperature or i need temperature as a requirement for my application that the system could actually infer what the low-level elements are um and interestingly enough this company meshavan which which nick and i had started in shanghai we've been experimenting with this concept of what we call generative circuits uh and the idea with the generative circuits comment uh concept was that we could

encapsulate all of the functionality that's required to deliver this capability right and when you think about electronics, and I've said this so many times before, you know, it's just not infinitely subdividable. As long as you sort of rise above the part number, there are really obvious categories. So if you take a category like communications and you say, well, okay, we've got Wi-Fi,

we've got Bluetooth, maybe we have Ethernet. Yes, we have serial, but most people have moved away from that. We've got USB. There's just like not a lot happening in that space. And interestingly enough, I mean, in the tenure that I had at Altium, I spent a lot of time going to companies because before moving to Sydney to kind of work as a part of the R&D team, I was actually kind of the engineer in residence, FAE, that would get sent off to all these major accounts. And I find that every organization was treating USB like it was a new problem, or Wi-Fi like they were solving it for the first time.

And it just seems to me like that's such an outmoded concept. And I think what's emerged, if anything, and so many things have emerged, I shouldn't trivialize it, from the Arduino space, is this idea that people are really just trying to stitch together a solution to a problem. Where that fails is when you want to then connect that with production and logistics and doing something at volume and doing something with a real return on investment. So it seems to me that kind of the ideal implementation approaches this whole process from the perspective of how do we move up two or three layers in abstraction and say these low-level elements are solved for you. And that could be something like a TI reference design.

I mean, why is it that every reference design that i get i end up having to recapture in my own schematic and pcb tool like it just doesn't make sense to me so this generative circuits concept it's kind of interesting we we started to play around with this where i could start writing code in javascript and it would actually generate a json file which uh was an equivalent of these little modular circuit elements. And inside of the modules, I could hide all of the complexity. So the module could be six layers with blind and buried vias and have all of these other things inside of it.

But it didn't really matter to the consumer of that at the software level because I would normalize the interfaces between these little modular elements. And what I would produce on the back end of that from that JSON file was a collection of modules and their connections, which, if you know anything about the EDA space, is kind of the equivalent of a netlist. So from that, I could then very quickly from the JSON file translate that into an Eagle netlist or an Altium netlist or a Pads netlist or whatever, and actually generate a PCB file, which had all of these modules in it. Now, what's interesting about that is then if you take those modules and you treat each module as though it encapsulates a single capability,

then you can put those on breakout boards and just chain them together on a breadboard. And what exists on the breadboard is exactly the same as what exists in software, but is also exactly the same as what will exist in production hardware. And that to me is a really interesting concept. So I think that's the holy grail. Now how we get there from here is actually still a reasonably complex problem, but it's a knowable problem. It's not like it's outside of the realm of possibility.

It's really a matter of can we do this and then get enough content to come on board to sort of backfill the requirements? And that means, you know, sort of taming the manufacturers and the distributors, which are so much the creators of a lot of this content. Because every chip that TI creates, somebody inside of TI created a board for that, brought it back, stuck it up on a scope, and figured out, is it working according to spec? So where are those designs? How do we organize those designs in a way in which we normalize the interfaces between those things so they really are compatible with each other? I mean, the power world, you sort of fall under one of two buckets, right?

It's either switching or it's linear. And then in there, you've got some other gradations in there. You've got USB, you've got USB power, battery power, the DC, AC, whatever. And then you've got battery backup for your USB connection or battery backup for your DC power supply or whatever it is. But that's all kind of knowable.

And what's really interesting to me is that that sort of commoditizes so much of this. It does create probably, it comes with its own set of risks, certainly for the semiconductor manufacturers. Because at some point they could be optimized out if their chips are too expensive. Or they don't have the right sort of partners for some of the discrete devices that they choose or things like that. So their costs become too high. And so much of this is a race to the bottom anyway.

But I think the real differentiated position, certainly when it comes to production hardware, and I think we know this as people who have actually built production stuff, so much of their differentiation is in the area in which you're actually moving the science forward right so more precise sensors uh more efficient power supplies all of those other sorts of things that's where their true differentiated position is nobody has a differentiated position on the

lm317 like it just doesn't happen so um you know i think there is uh there is market for that and the more that i talk to software developers, the more that I've worked with software developers extensively over the course of my career, building hardware for them, the more that I realize that I am the bottleneck in a lot of cases for them being creative and reaching out and touching the real world. And I think once we accept that we can build sort of the API to the real world, then at that point, what emerges from that is kind of the App Store model for electronics.

If I can stick this temperature sensor over here and it gives me temperature information and I can interact with that over an API and I can discretize temperature sensing and I can discretize motor controller, HVAC controller, lighting controller, any of those other sorts of things into these smaller modular elements, it's sort of like taking the mobile phone and taking kind of an exploded view of it. Like if you could pull apart all of the constituent elements of a mobile phone and distribute it across a space, well, you would have a baby monitor and a hospital monitor and all these other sorts of things. So at that point, a couple of different markets emerge. One is for these small sort of OEM-style off-the-shelf devices,

which plug into really the notable platforms. And let's face it, there's a lot of people building IoT platforms today, but I can almost bet today who the winners are going to be. It's going to be Amazon, Samsung, and Apple. And Microsoft will have some small stake in that. So if we accept that that's going to be the platform, then we can start building these modules which plug into those platforms.

And then on the front end of that, software developers, what emerges is an App Store model, which works for the Android Store and works for the app store from Apple and that sort of thing, where people can start to build really cool applications on top of those things. That makes a lot of sense,

but I want to go back to the original question. Sorry. Some of what you were saying, it sounded like you wanted to build, I want to say 20 questions. There used to be an online 20 questions game that had everything in it. And of course, it's just a binary tree.

With 20 questions, you get a pretty huge space. And so I feel like to some extent, it sounded like you wanted to build 20 questions for electronics. So when I say I want a temperature sensor, you say, well, do you want it to be analog or do you want it to be digital? If you want it to be digital, do you want it to be Wi-Fi? Do you want it to be USB? Is that where you're headed? And I like the power part because that's probably the hardest part for me with my software background.

No, absolutely. Absolutely. And I think that's exactly it. You know, interestingly enough, we had this model and we sort of all dismissed it as oh that's so old thinking but visual basic introduced this model right where we would sort of reduce these things down to uh events and properties and we would create these small component elements which were some series of events which were tied to those things and some properties that those things had and you didn't really have to think about what was behind those controls. But you could really operate from this sort of controls-driven development environment. And this is, wait a minute,

Visual Basic applied to... Kind of applied to, well, applied to electronics design, right? So what are the components of a system? You know, not a lot of people really care about what processor is in their mobile phone but they absolutely care that they can take photos and that they can interact with a touchscreen or that it has some accelerometer magnetometer these other sorts of things but in terms of is it telling me what direction i need to go in um can i tilt it while I'm playing a game? These other sorts of things. So it's kind

of, you know, reducing that to reducing the electronics problem to a series of controls. And then, you know, I think power is interesting because it's one of these sort of fundamental elements, but it could actually even be derived from those other elements to the extent that once you understand what your power budget is um then you can actually roll backwards up to well what power supply do i need the system can actually infer that instead what you're defining is well what's the paradigm for powering this thing can it be battery powered no it can't be battery powered because now you're drawing you know i don't know 600 milliamps or something like that oh okay i mean theoretically yes it

could be battery powered but it's not going to have a great battery life you know what what are the conditions under which you're going to power this thing i have a lot of of clients who come to me having built a system and it does what they want but it usually, its power situation is terrible. It lasts for an hour on battery and they meant for it to last for a month. Or they put on NeoPixels on there and they didn't ground it right and it went badly. And it mostly works except sometimes there's weird flickers they don't know how to get rid of. Things like that. And I worry that this model you're building, these corner cases and these edges, they still seem hard. Well, I think they are hard. But I don't know that we need to make that...

Look, these are all hard problems. I think it's a question of whether or not somebody who has the real domain expertise is going to go in there and solve those problems once and for all. So, you know, like TI, I love TI as an example because probably more than a lot of other people have really tried to push that space forward and try to reduce power to a relatively knowable problem. I mean, WebBench is a great example of that. I can go into WebBench and say, hey, this is my input. You know, WebBench is a great example of that. I can go into WebBench and say, hey, this is my input. This is my input. This is my output. This is how much current that I need. And they can design you 60 different power supplies that all meet whatever that requirement is. And it's an extraordinary amount of energy and effort that's gone into that. Now, TI are kind of the resident experts. They are the domain

experts in that space. So I think what I'm proposing with that, and again, this is a very idealized world. I'm not saying that I'm running out and building that today, though I've gone really far down that path and actually building that to the extent that I can actually already generate a netlist and I can deal with a lot of these things in a modular way but the um what what's interesting to me is that we have these domain experts in this space and we just haven't really found a way to sort of conform their content um to a way by which I can just go in there and parameterize my input. And from that, I don't know that that actually means that I plug in those parameters or that the system can't evaluate the component elements. So imagine if Adafruit, for every module that I can buy from Adafruit,

told me what its power consumption was. And I think they largely do. But imagine that every module vendor... Well, yeah. So for Adafruit, some of their modules take power, and I would like them to add, here's the ways you can get it to reduce the power. You take off this resistor, you clip that LED.

So, yeah. Right, right. So having some understanding of what that program logic is in there is obviously you know that's a deep engineering uh problem uh or probably deeper than than what you would expect for the average enthusiast to kind of come in there and be able to to make inroads and and try to try to address those things but i think the the fact is is that you know all of it's knowable you know the physics hasn't changed and this is the you know this is a great thing about it is that it's all kind of

knowable and and if you can build a system which actually allows for that uh sort of input you know whether that's your keyword whether that's your parameterization um then you then I liken it to the way that we always use software as the example for abstraction. But I think there's another great example of abstraction which comes actually from the mechanical world. And if you've ever spent any time around really good mechanical engineers, you'll find that really good mechanical engineers, especially those who have spent a long time in the industry,

will have parameterized models for the things that they typically create. with thread count, thread spacing, all of these other things, screw head type, all available to them within their MCAT tool. Plug in the parametric information and out pops a model. And it's a beautiful parasolid model. It's got all of the detail in it. And then they can export that to STEP or they can send that to their PCB guy who can bring that into their PCB tools. And it doesn't really matter.

But the fact is that over time, we develop this sort of ability in the mechanical world. And we've developed this ability in the software world also to parameterize things. And I think what's kind of missing there in the electronics world is enough time or enough energy on the part of any one person or group of people or sort of tribe or whatever to go back in and sort of refactor the way in which we're doing those things to try to make that stuff just a lot more accessible. Because you're right.

It's not just a power problem. It's not just power supplies. If I look at, hey, I've got this transceiver, but I mean well it's not just power supplies you know if i look at hey i've got this transceiver but i can put it in a low power mode what does it mean to put that into a low power mode what's that going to do for my power consumption exactly how long is it going to take to wake up all the other sort of stuff that goes along with that and um interestingly enough you know we all read the data sheet like it's the first experience reading the data sheet. And I guess the holy grail, if I could have it,

would be to take that relevant information and distill that down to something which makes sense without having to read all of that content. I mean, anybody who's grabbed the IMX28 data sheet and gone through all 2,200 pages knows that you're going to get to page 936 and there's going to be that one gotcha. But I think the fact is

that probably 95% of the systems that are built today could actually be built from small modular commodity pieces. The other 5% is kind of reserved for those people who are genuine electrical engineers who are trying to push the science forward and exploring new areas and building new technologies. But I don't know that those two things are so tightly coupled

that you can't have one without being totally dependent on the other one. I think those two things can operate independently. And you can say, great, just build a system which has temperature sensing and a camera on it and some touch control and some other things and distribute that system because they're all designed to be wirelessly connected and all you need to do is select the bits that you need and you know this is kind of the phone blocks thing right yeah this is making

electronics into lego blocks which as far as trends go i totally agree with you 10 years ago there was no building things up like there is now. I mean, you can just go take everything you want from breakout boards and just smash them together. And usually it works. Sometimes you end up with three volts mixed with five volts, and that doesn't work so much. I know for sure. Don't ask me how. But there are still so many gotchas although no there are there are i mean well but he said 95 problem maybe maybe it's really the 80 problem but you know there's some right yeah you're solving some large number of the design problems especially in prototyping or hobby and like you said that five percent okay that includes people

who need to do something truly novel or or who are making a whole mess of them right it's cheaper high volume low cost we've got to customize this further uh or size you know size is always difficult sure or i would include rf stuff in that because as soon as you include rf your your whole life is ended that's not as true as it used to be. I can't tell you the story that I wanted to use an example. But let's just say something didn't work when a particular material that had nothing to do with the electronics was changed. Right, right. Is this like the Apple iPhone where their outside edge is a... Worse.

Yeah, yeah, yeah. It was like, it doesn't work on Tuesday. Well, you shouldn't wear flannel shirts on Tuesdays. Yes, exactly. Yeah, exactly, exactly. You know, but coming at this from the perspective of, you know, my background is sort of interesting because it's sort of diametric in that, you know, I've got a deep knowledge in this FPGA space, but then also having spent a long time in the RF space, so many of the problems actually could be solved in the RF space. And as long as we just sort of demarcate these other parameters

under which this thing will operate, I think there are rules of thumb that we can enforce that will produce good, meaningful designs for the RF that is used most commonly. So Wi-Fi, BLE, ZigBee, the 802s, right? Those things, I think, it's reasonably straightforward to be able to say, okay, great, use these bits. And by the way, because they're modular and because they're all, you know, they've all independently passed FCC and whatnot, you'll be able to use those.

And as long as you don't disturb that part of the circuit, then your cost for certification will be substantially lower. I think that is achievable. I think when you start to get into other areas of RF, certainly there are challenges there.

And then I think, you know, like RFID and things like that, you've got some finickiness that happens in there, but even that can be parameterized. You know, we tried to address a lot of these problems like with touch interfaces at Altium, where we would do these auto-generated

capacitive touch geometries on the board. And all of that stuff was totally knowable. Like, we could do that programmatically. So it's a matter of how do you organize yourself. And I won't dispute the fact that they're hard problems. But I think at this stage in the industry, you're either solving hard problems or you're not really moving the industry forward. You're solving hard problems or you're just making it prettier. Yeah, exactly. Sometimes that's hard. industry forward you're solving hard problems or you're just making it prettier yeah yeah exactly

exactly and that you know that's yeah yeah i mean that yeah that that absolutely i mean for me it's extraordinarily hard extraordinarily hard because if if i had my druthers my house would just have you know a din rail running through the living room um which you know may not be the most aesthetically pleasing thing in the world but but it is a nice source of heat for those late autumn months where you don't really want to turn on heat and cause your hair to fray up. So, I mean, along these same lines,

interestingly enough, this borrows a page out of a playbook of the FPGA world. Which I did want to ask you about. The FPGA Arduino, I'm not sure I want to call it Shield, that's on Hackaday. They have their little store now, and I shouldn't have said it that way. They have their magnificent store now. And what is it? Tell me about this Arduino-compatible Spartan F6 FPGA board. Okay, so short answer is that we were launching Hackaday.io, and we were launching this project's page.

And somebody had mistakenly asked the extreme electrical engineer in the company, hey, will you build a project and put it in Hackaday.io? And I was like, yeah, I'll build a project. So I decided that I was going to build this Spartan 6 FPGA board. And part of the angle on the Arduino side of things was that if I connect the Spartan 6 to the SPI flash, and then I connect that SPI flash into the SPI bus on the Arduino, then as long as you can generate a bit file, then you can use the Arduino

to program the SPI flash, trigger the reboot on the FPGA, and the FPGA will come up with the configuration. So the Arduino could actually become the programmer for the FPGA. And I've done a little bit of work with Node Zero Port, making it possible to actually work through Node.js. But there's kind of more value in probably then also exploring all of the other shields that were out there. So if I wanted to build a reconfigurable hardware platform,

I could actually plug in a shield which had 3G or had Wi-Fi or something like that, download the bit file, pass the bit stream through the ATmega into the SPI flash, trigger the reboot of the FPGA and the FPGA would come up configured. So I could actually do this sort of cloud configured hardware. Now that was kind of, you know, that was part of the premise. But you

have to understand like there was a backstory that was happening at the same time. I was working on all these IoT products or what I call the sea of connected devices back in Shanghai. And we were trying to figure out how we could do this highly configurable hardware platform. And FPGAs are interesting because I can drop in any sort of peripheral core inside of the FPGA

and use that to control some downstream thing. And they're inherently reconfigurable. So today it could have an SPI bus, tomorrow it could have I2C, or it could have an LCD controller, and the next day it could be a touchscreen controller, or it could be an audio interface, or a whole variety of different cool and interesting things that I can do with the FPGA. And I started to kind of explore some of those concepts. And one of the things that I was playing around with a lot, we did this at Altium, and then I'd sort of taken this to the extreme after I left the company, was this idea of SPI to Wishbone.

And if you're not familiar with what Wishbone is... Yeah, that was the question. What's Wishbone? Sure. So IP cores, for some guarantee of compatibility on IP cores, all these bus interface standards emerged. Wishbone, Avalon, a number of things. Because we knew that we were dealing with what's really a chip design problem,

whether that's for ASIC or it's for FPGA or it's something else, we need some standard interface for the IP cores to plug into one another. Now you see how all this story fits together. But we needed some standard interface for IP cores to actually plug into one another. And Wishbone emerged as this open standard for IP to say, if you're going to build an FPGA IP block, if you have this standard signal set between block A and block B, these two things would be compatible with one another.

So if you think about it in the FPGA space, it's a little bit like POSIX is intended to be in the software space or in the embedded software space. What we're looking for is compatibility. So the idea of FPGA-based processing, we started inside of Altium. I mean, we didn't launch the concept. The concept was much bigger than that. But we had done a MIPS core inside of Altium. And when you think about FPGA-based processing,

processing on an FPGA, you've got a processor core. And then it's really up to you as a system designer to determine what peripherals you actually want to pack around the outside of that core. Now, to make sure that those peripherals are then memory mapped into the processor's peripheral memory space and that you've got some other memory interface on the processor, we use Wishbone as a bus standard.

And then Wishbone has these higher level structures. Like there's this concept of what's called a Wishbone interconnect. And the idea with the Wishbone interconnect block is that it's got a standard master interface on one side and then a series of slaves can connect into however many master interfaces you want to extend it with but then all of those slaves are peripherals in the processor's peripheral memory space so they'll have a base address and there'll be some address bits which

are allocated based on the number of registers that each one of those different peripherals requires to be able to configure the peripheral right so if it's an fpi controller or it's an i squared c controller that sort of thing i've got to have some mechanism to say okay and it lives at this base address and the address space that this consumes is this much and then that gets uh resolved to the processor's peripheral interface um and then the processor can address those peripherals just by starting at the peripheral memory space and saying okay great at peripheral memory space at this address that's the spi controller and it requires this many bits to configure the the spi so now bringing this down to the arduino world so so talk about beginning with great enthusiasm

uh and then realizing kind of the practical realities of things um you know i'd done this you know all of that works great inside of an fpga but uh i developed a core uh and we we'd iterated on this a few times which does spi to wishbone and the idea was is that if i put an spi interface on the fpga just using an spi controller uh core inside of the fpga then i could extend a discrete processor a fixed processors peripheral memory space by way of the spi bus so imagine now that i've got the 18 mega 328 and i want to add a bunch of peripherals well the, the ATmega328 doesn't have a lot of peripherals. Maybe I want to add a touchscreen controller, or maybe I want to add some LED controller, or maybe I want to add some of these things, which FPGAs are really good for extensibility.

Just by way of the SPI bus, which is running at whatever, 200 megahertz or something like that. I've got the means to interface to the FPGA over SPI. And then on the other side of that is this Wishbone interconnect. Then I write a little protocol which sits on top of that, which says, okay, now I'm sending this data over the SPI bus. This is the peripheral that I want to talk to, which is at this base address, which requires this many address bits that I want to talk to, which is at this base address, which requires this many address bits to be able to configure the registers

for whatever that peripheral is. And that actually allows me to create this really extensible system. So if you think about that in context of Arduino, it was kind of like, you know, it was taking the idea of the Arduino and saying, okay, you know, cool, use the Arduino, but why don't you build something which has like real configurability in terms of its interfaces and

that sort of thing. And if you just packed a collection of Phoenix connectors around the outside of it, you know, screw terminal connectors, or, you know, you put in a right angle header or something like that, you decided that you wanted little plugin boards, then you could actually extend that to be able to do all manner of different things. The realization with that is that it's probably way over the top for the enthusiast to think about, because it starts to get into a lot of murky water about programming FPGAs. The prerequisite for that is that you understand how to build an FPGA design

and that you can actually then understand the IDEs that go along with that. I was about to say, that's a pretty big hurdle. There's all manner of different challenges involved in that. I added the JTAG interface on there just with the hope that at some point when I get some free time and, you know, that's probably coming reasonably soon, at least at this stage. It's kind of long in the tooth, so I've got to come back to it or I'll feel really terrible about it. I've got to come back to it and design some simple tutorials and examples

and that sort of thing for how you do some basic FPGA stuff, then how you do some IP-based FPGA stuff, then kind of introduce the concepts of Wishbone, and then introduce the concepts of two streams, one of which is cloud configurable systems where the FPGA provides the interface, but the interface is inherently configurable.

And then on the flip side of that is these just desktop, highly reconfigurable systems where you could play around with all manner of controller and all manner of sensor. And the FPGA just works sort of as a proxy, right? Just, you know, it's an arbitrage thing, handling the interface between whatever those different peripherals are. So, you know, it was one of these things that you know i spun up the design it was like a week and a half to spin up the design i've done a lot of fpga boards in my time um but i i i realize now kind of reflecting on it that um it it's probably more useful to teach basic logic than it is to teach uh some of these higher level concepts or at least i have to go through that

sort of first step so i'm in the process of trying to configure some content for that i'd actually created recorded a couple of videos and things like that which i haven't released yet that'll take people through some basic logic and that sort of thing very similar to uh like what uh clive maxfield did with like bebop to the boolean boogie and you know how computers do math and those sorts of things uh taking some of those concepts um rather than going to the other route which is to you know grab uh you know patterson and hennessy's computer organization and design and do a bunch of examples out of that book and try to implement all of those on the FPGA. So it's a fun design.

It's a fun board. It's got a simple JTAG interface. You can configure the FPGA over SPI, which is kind of fun. All of that stuff is easy to do, and you can kind of do all of that. But again, easy to do. Big quotes around that one. Easy for the experienced user. Yeah, exactly.

Easy for FPGA developers. Those skilled in the arts. God, why don't you know how to do this? I mean, that's a curse of knowledge, right? I wish that I could simplify some of those things. But unfortunately, it's just one of those things that you have to get through to really understand. Yeah.

So Wishbone, just to make sure I understand it or to check that I don't, Cortex has a bus system that's sort of similar. And I use a lot of ARM Cortexes. Yep. Is Wishbone the open source version of the AHB, the high performance bus, and the APB? Yeah, exactly. So it's a very similar concept. Absolutely similar concept. So, I mean, and the APB? Yeah, exactly. So it's a very similar concept, absolutely similar concept.

So, I mean, they're total analogs for each other. Well, then let's go on to teaching because we've done that. Oh, yeah, fantastic. You said that you're thinking about how to do for future instructions for the FPGA board, but you've been doing a lot of workshops over the last year for lots of things. Yeah, quite a few, quite a few. Which was your favorite workshop? Ooh, you know, I really liked the, I did, I did one which was on soldering, but it was, it was more than soldering. It was just sort of

soldering techniques and things like that. But actually took um a room of 30 people who had never soldered anything before in their lives some of which had soldered some things you know mostly through hole and by the end of the day we uh had reflowed boards with bgas on them um and used hot air to do 805s and And they were doing 805s by hand. We did a USB interface, which the micro USB has a pretty fine pitch. I don't know. It's like 0.8 millimeter or something like that or 0.65 millimeter for the pins on the micro USB connector. And everybody was successful.

Nobody failed. Everybody managed to do it. Were there a lot of microscopes involved? Not at all. Not at all. You know, it was mostly, the only person that had a microscope really was me. And that was only so, you know, I used one of those little Andon Star cameras, which are fantastic and super cheap. I used that little Andon Star camera so I could actually just show them a close-up of what

i was doing with the soldering iron tip um but you know i really that that was one of those that was just like total instant gratification because everybody was building something and everybody's worked and it was such a confidence well you know i'd like to hope that it was a confidence builder for everybody who who came and i've run that now twice uh twice or three times i can't remember um but i've run that now a few times and people have been really enthusiastic about it but that's actually part three in what's shaping up to be a four-part workshop series. And how long does it take to prepare a workshop of, say, a day's length? I mean, that one has a lot of hands-on, but you still have to do the instruction. No, no, no, absolutely.

So, you know, when I did the first workshop, which is sort of an introduction to a PCB, I mean, we start with what is a circuit board? Like, what is a PCB? And I pass around pre-clad boards or pre-clad board panel and a whole variety of different things. We go through everything, right? What's solder mask? What's paste mask?

What's, you know, what's good silkscreen? What's bad silkscreen? How do you create a PCB footprint? Like, what should your pad geometries be? Which is one of those how long is a piece of string problems for a lot of people. That was

probably the longest to prepare. That took me the better part of probably two weeks to create the content for it. For the soldering workshop, it also took me two weeks, but that was largely because the first batch of boards came back and they ignored my v-scoring notes uh on the pcb files so my boards didn't come back v-scored

so in my office i had set up three other mills uh all spinning in this amazing orchestra of 16,500 hertz uh well i mean 16,500 rpms not really hertz but this amazing orchestra of cnc as i was mailing out these boards which came back from the board manufacturer after they ignored my my notes on that and i took about a week to mail them all out but it was about a week to prepare for that content um other mills say don't don't use this on boards yeah because yeah yeah well so so i mean technically uh so so it's not really cancerous polyglass material isn't cancerous if you read the msds on polyglass it's not actually cancerous but it is an irritant but it's really bad um yeah i mean it will get in your nose it will get in your lungs it's going to take a while to get out all these other things note the glass part of this right

right right so any fiberglass substrate like that is going to create dust which you don't want to breathe so the the this my solution to it was you know both a a lot of fans which is not a great solution please don't ever do that and then the other part of it was actually taping shut the uh the machines while they were running um and by taping the sort of outer perimeters and everything then i kept really the dust to a minimum all the dust actually settled in the machine and then i would go back in there with a with a vacuum and suck all of that information or suck all of that nastiness out and into the vacuum cleaner. But like anything, I mean, the vacuum cleaner probably did a better job of dispersing that around the office than anything else. So lesson learned,

but failure was not an option. I had a bunch of people booked into a workshop. So I said, no, this has to run. And yeah yeah i kind of went over the top with it um but yeah we built six boards during that workshop it was actually really cool uh and they hand soldered every single one of them it was really cool i had a really good time that is worth making your office a super fun site yeah yeah it really is it really is christopher's giving me this look like, no. No, whatever you do. Please don't do it. No, you know, I came back from,

I lived in China for a number of years and I came back to the city and- Oh, the air quality with all that stuff is going to be better than China. Yeah, exactly. Exactly. Of course. Air quality, food quality.

Yeah. No, I came back to the city and all of the meetups that i was going to were very sort of um and i'm gonna say this it's gonna sound kind of kind of negative but i don't mean it that way they were very hand wavy um you know i would go and it was a lot of like two minute lightning talks this is what i'm building i really need funding for this or i need somebody who knows how to do this but i wasn wasn't actually learning anything. And so, and I'd seen like what tech shop was doing, which I think is brilliant. And it's, you know, it's very, very cool. Uh, but I wanted to do something which was just total grassroots, totally altruistic. Listen, I've been doing this now for 25 years. Let me just take you through,

let me give you the information that other people have given me. And so I decided that I would kind of take people through that process. And it was, you know, what's a circuit board? How do I design a PCB from scratch? You know, actually having somebody there who can answer the questions

while you're fumbling around trying to figure out why eagle is so stupid and allows you to run tracks over the top of each other um you know like how do i solve these problems how do i re-corner something how do i drop a via and switch to a different layer versus dropping a vn staying on the same layer how do i drop a via without having to hit the stupid middle mouse button like all these other sorts of things so you know i i really wanted to take people through that process um and i guess it's you know for me it's it's really a labor of love um so it doesn't feel like work to me even if it takes two weeks to prepare something it's like

you know that's cool uh that's easy um because the the trade-off for me on the you know what comes back to me on the back end is just extraordinary to watch a bunch of people now able to design things. Is this part of your supply frame job or is this totally extraneous to it? So it started off totally extraneous to it. And supply frame has been very, very gracious in saying, listen, we'll support you in doing this. I've been very careful not to make a hard connection between those things, but nobody in supply frame said,

we should do this, except for me who said, I should do this. And then supply frame came on board afterwards and said, look, this is really cool. What you're doing is potentially really transformative. And so they've become really supportive. Yes,'re doing is potentially really transformative. And so they've become really supportive. Yes, you can use the office on Saturday to do this.

Yes, we'll buy tables and chairs. You shouldn't be paying for bagels and pizza out of your own pocket. I really and truly, this whole thing, I haven't charged for any of these workshops, but I've been really careful not to try to make it just a marketing stunt. Because I think that's the other risk, is then people feel like they're going to uh design con or something like that and like a workshop in in that environment which you know i've i've for what

it's worth i you know i've i've learned less and less over the years either that means that i'm getting smarter and smarter, which is probably not the case, or just the quality of information that's coming back through a lot of those things. It tends to be very marketing-driven. It's like, yeah, I really wanted to see how you use a spectrum analyzer.

I didn't really want to learn the specs of the spectrum analyzer that you want to sell me. You helped put on the Supercon conference recently. the specs of the spectrum analyzer that you want to sell me right well you just i mean you helped put on the super con conference recently and well you got an amazing set of people to talk there was always some well no we're not really here to sell our stuff but of course of course spark fun is going to use the SparkFun thing in their workshop. They're not going to use Raspberry Pi. Well, maybe they would. Maybe they do have other workshops.

But they're going to use something they sell, and they're going to mention that they sell it. Because that's how they pay for their people's time. Right. Right. And I don't think there's anything wrong with that. I think it's when it becomes, you know, I'm an engineer and I've been marketed to my entire career. Yes. And it's just overwhelming the amount of really bad marketing that I get that seems to be written for somebody else.

I mean, it just, like, it doesn't connect with me at any level. And I don't, you know, I think there is a right way to do that there's a wrong way to do that and i think the right way to do that is to make sure that what's core to your message is is is all about how you improve somebody's ability to do something um and Um, and they'll get that you're valuable as a, as a, as a, as a by-product of that. I don't think that, you know, parading yourself around in, you know, with, with your products and having, you know, the, the, uh, end cap from an aisle standing next to you on stage with a bunch of boards on it and these other things, you whatever uh somehow you know advances your position i think you've really got to give people substantive content and then they will absolutely listen to you you know i think

dave jones has done more for selling hardware equipment than probably many of the marketers inside of those same companies and And I think Chris does this. I think you're successful in this. I think many of the people who are really trying to advance the understanding have, you know, they've become the trusted advisor more than anything else. Which is to say, if you're a company and you're listening to this right now and you would like to spend a god-awful lot of money to make us... Where are you headed with this? I was going to say that they could pay us for advertising, but the truth is we always say no to that, so why bother?

But I think you're right. I think it's like talking to somebody you work with and getting advice on tools it's somebody you trust and you know it's a it's an opinion based on experience not based on well i work for this company so i have to come up with a story for why this is a good thing well and even if we were selling things i wouldn't want to sell things that i didn't want to buy i mean sort of like glow forage yeah we we did say we have a link and use our link if you want to buy one, and it was very cool. But that was something I really wanted.

Right, right. And we won't be afraid to say it's terrible if it's terrible. I know. Although Dan knows where we live, so I'm not sure that's good. He's got your shipping address now. But, you know. No, but I agree with that. I agree with that.

But I think that is the new marketing, right? Because we all have banner blindness, right? We don't see the content that's sort of bombarding us on pages in the way that we used to. And there's sort of one of two directions that you can go you can try to make those things more prominent and flashy and like you can take up the whole page and make them click before they could see anything exactly exactly and and and there are places you know i lived in china and a lot of the chinese websites have a lot of that you know with the flashing and the blinking and

the you know the gold trim and all those sorts of, oh yeah, absolutely. And I just look at him like, oh my God, I'm so repelled. All I want is this one piece of information. Yeah, when they make you do that, when you already want to buy their product. Exactly. That's the worst part. Exactly. Look, I just wanted one spec because I wanted everything else about your product. And if you just let me have this piece of information, I will push buy.

Right, right. I will still, you know, I'm going to give you money, right? So, you know, I enter a lot of those engagements, like surgical strikes, right? I've done all the research, I've homeworked everything, and then I go to the site, and then I buy the thing. I don't need you to advertise all of the latest and greatest of these sort of unrelated things to me. I think there's one dimension of that is relevant marketing or relevant content is useful.

I see you're buying a part we're about to end of life. Have you considered this other part? Oh, absolutely. That's cool. Yeah, absolutely. Well, DigiKey has actually done a great job of that. I was just on the DigiKey site the other day and i'm buying these nrf 8001 chips which i'm going to use for the bluetooth component of this workshop series and as i'm buying the chip

the johansson ballon pops up and uh the oscillator that i need pops up and all of them are not you know it's not advertised to me except in this one small little line underneath the the various purchasing options which was uh the people who purchase this also purchase this right i mean amazon figured that out oh yeah and i use that so often absolutely well when i looked at it i realized that i could actually construct the entire system from these recommendations that DigiKey was making to me because they were exactly the parts that you need that go along with the NRF 8001 reference design and the same is true like if you go to DigiKey now you plug in the ATmega328 when you get the 328p and then you descend into the page where you're going to choose the quantity that you

want to buy you'll see that they have the oscillator listed there they've got some headers listed there they've kind of got all of the bits and pieces that are core to designing with the ATmega328p. Those are features that absolutely I love and I want to have. And those are the things that we as SupplyFrame are actually trying to implement, right? But then I think there's the other part of that,

which is the sort of hand-wavy thing. I want to know, if I'm enthusiastic about electronics and I don't know anything about it, I want to know how to use an oscilloscope. If you show me that on your oscilloscope, just trust the fact that I'm smart enough to then decide, yeah, look, you know, I mean, the price point is right. It's a great scope. And they showed me how to use it. I'll just buy theirs. I mean, why am I going to go run out

and buy this other thing? Given the commoditization of so many of these parts, then yeah, if A and B are functionally equivalent, price equivalent, then I'm going to choose the one that gave me something else.

The class. Exactly. So I guess tying all of that sort of back into the workshop thing and then also to the super conference, I think everybody who spoke at the super conference who presented something that was of value and their product was demonstrated in that context. And I think everybody respects the fact that they were demonstrating their product in that context. And nobody was put out by that.

I think the risk is that you get up there and it's this sort of old world marketing. The kinds of people that know what a SWOT analysis is and all of this other stuff. Positioning something to me like a mailer or something like that. Just give us your email address and then we'll send you 18 emails every week. I mean, I have 5,200 unread emails in my inbox and probably 60, maybe 70% of that is marketing so well a few of

those were mine but but actually you know what's great about what's great about google is is that you know it consolidates all of those things so when i look at it in gmail it's like oh yeah so all of those threads are that one item um but then i look at the list and it's just you know it's relentless and so i guess as an engineer i've you know i've grown really distasteful of that you know i just are allowed to push the unsubscribe button it even works for most of these companies no and for for many of them it does but now i've got this new uh because you know my my title is technically a director title. Oh, you're toast then. You've got somebody to read your email for you?

Oh, no, I wish. I'm working on that one. Something to ask for for Christmas. Exactly. Or just somebody who deletes my email for me so I don't have to read them. No, but instead I get inundated with all these offers to uh to hire somebody on podcasts no no no no to actually hire employees right and it's all these recruiters but they don't have an unsubscribe link in their email so then i send them this formatted thing

which is like i'm not sure if you understand kind of the rules of the whole can spam uh legislation which was supposed to require that unsolicited emails you know with the purpose of blah blah blah blah well see i used to send them i would be happy to get your referrals but my company does not allow me to pay recruiters oh and they would go away yeah maybe that's the approach yeah see that's a much more delicate approach uh where my you know my approach is very instructive. What you're doing is illegal, quit it. Exactly, exactly. Oh, goodness gracious.

Yeah, so, yeah, I think the Supercon, you know, the content at the Supercon was really cool. And I think it was, you know, it was a really, it was a great mix of people. They were the kinds of people that we really wanted to have talk at some event. It was like a who's who of everybody that we really really really wanted to have give a talk it was pretty fun i i almost didn't go it really wasn't quite up to it and uh with the new house you know there's been a lot of stuff every weekend there's more stuff that needs to get done oh sure and san San Francisco now is almost two hours away. Oh, yeah. Yeah, it just, I was like, oh, I don't know.

And I was so glad I went. I mean, it was all the audience members were, yeah. You guys should do that some more. Although, I mean, I sort of brought that up because organizing and planning that and getting people to talk who weren't going to talk about their companies, that's hard. I mean, I know why ESC and some of the big embedded conferences have gone in that direction. It's just, you want somebody to talk at your conference, of course all of the Oscope vendors are happy to have somebody present because that's their job. And you're pretty lucky

that SupplyFrame has started to support you in this thing that you wanted to do anyway. But man, it is hard to find those people who want to do it anyway. Right, right. You know, I think they are out there though and i think there is uh the ones that do are so passionate about it um and i you know i guess i can only speak from from the experience that i've had with the people that i've that i've been fortunate enough to have met

um that we would do it anyway right i like ifly Frame didn't support me in doing it, I would bring people to my house and teach them how to do stuff. Or I would find another place to do it. I would find another venue. And the company has been really supportive of that. But through that process, I've met all sorts of amazing people

who all have some value to contribute back. And so many of them have said, yeah, look, I'm running this other workshop, which is about how to develop for Android. Would you like to come to this? And just amazing amount of content. And I think without those people that just blindly do it, it's hard to create the inspiration for those people who wouldn't probably do it otherwise. But then I think if we continue to drive that agenda, it actually creates more contributors in that space. Because people say,

look, this is a great opportunity for me to get out there and not market my company, but market myself by sharing all of this incredible information that I've been fortunate enough to learn from other people. I won't pretend that any of the information in my head that I didn't learn from somebody else. Oh, I made some of this stuff up. Yeah, it's probably something. Sort of accidentally. Yeah, exactly.

I mean, there was a lot that I learned definitely through pain and shock. Grab opposite ends of a dipole antenna on a 100-watt radio or something. But the opportunity to actually go out there and sort of continue to share that information is just extraordinary. You know, it's such a rare thing. I think we're one of the few, you know, and this is the only hand-wavy bit that I'll probably throw out there. We're one of the few companies right now that's actively doing that and actually trying to hold a hard line. And I think part of that really came from the acquisition of Hackaday, where we really had to learn what it means to have that very clear separation between church and state sort of thing. Don't make this purely about profit.

Make this about creating the best content, and the profit will flow from that. And Hackaday has been extraordinary in holding that line Mike Stish and Hackaday has done a great job as an editor to make sure that the quality of information continued after Caleb and Phil Terone and whatnot quality of stuff that's actually being shared

is really really good but also doing that in a way in which it didn't just become one of these outlets for advertising yeah exactly let us show you the next product by company B, C, D, whatever It didn't just become one of these outlets for... Advertising? Yeah, exactly. Let us show you the next product by company B, C, D, whatever.

Actually making sure that we had the right balance of content in there. I think some of that stuff is news and it is informative, but it's got to be balanced and it's got to operate with the right tone. I think they've done a really good job of maintaining that. Supply Frame has been really supportive of that, which a lot of other companies may not operate that way. But we've managed to sort of strike a balance there. Yeah, I think a lot of people were a little worried because there was the possibility of Hackaday

becoming an advertising outlet for SupplyFrame. And that would be hard because hackaday had it had a definite personality and it wasn't really corporate right right it was kind of the it was almost exactly the opposite exactly exactly it was like you know i would rather build this myself than pay you money for it um which is kind of fundamental that sort of anti-corporate sort of attitude towards things. I mean, it's almost, you know, it's almost Stallman-esque, right? But it, you know, it's managed to work out.

And I think even when we did Hackaday.io, we had all of these fears about, hey, we're going to build this project site and people are going to be able to upload their content and share it with each other. And like, even today, we haven't run ads on Hackaday.io, but what's happening on Hackaday.io is really positive. People are actually sharing projects with each other and other people are learning from it and other people are borrowing from them. We've got now 100,000 people in that space

and it blows me away. It completely blows me away. Sure, there are probably a lot more open source designs on GitHub or something like that, but there's no single entry point for that. So unless you can write a web crawler, you're going to have a difficult time finding that stuff or navigating those things.

And the code or the schematics on GitHub only get you so far. You really need to build instructions. Right, right, right. And the opportunity to actually have a relationship with the person that built it. Yeah. So we created this chat channel on there,

which is kind of like IRC, because a number of us are old enough to remember IRC. And everything that was virtuous about it, right? The idea that I could sit there and have this constant conversation going on sort of on one side of my monitor while I was writing code or I was developing designs or something like that. Having this sort of constant conversation going on

so I could ask questions because, you know, Kevin Morris, who's the guy that writes FPGA Journal, and I say this a lot, so it's a quote that I reuse all the time. The guy that writes FPGA Journal, Kevin Morris, who's an absolutely brilliant guy, absolutely love his writing, because I think he's really well-meaning as a media guy, has this great quote where he says that

engineers are in constant fear of being found out. And it's true. Where software developers are totally comfortable having 10 different stack exchange windows open on their computer, it's like electronics people, we're supposed to somehow be omniscient, right? Like, oh, yeah, didn't you learn that in school?

It's like, no, I didn't learn anything. I learned how to do matrices and differential equations. And I haven't used one of them. Really? those are useful yeah yeah yeah no i'm sure there's you know they're probably totally useful to somebody right yeah um so you know i think the the uh there is a bit of that and this is like one of those places where we can all sort of adopt an anonymous identity um or or an identity or even multiples, I guess. You know, there's probably some risk of that.

But then we can all go in there. We can talk about stuff. And IRC was kind of that, right? I could cower behind my ignorance and, you know, ask questions and, you know, purport to be a 13-year-old boy, you know, in a Cub Scout troop or something just so I could get my total idiot question answered for me. But, you know, this is actually created. Everybody suddenly creates a new account.

You know, I've got a great idea. You mean that works? Yeah. So if, if tomorrow we have 200,000 people on Hackadio, I'll know exactly why. Yeah, it's been really fun. It's been a great ride. It's been an absolutely great ride there. I like the story that you tell about you have this workshop,

you want to do this workshop, and then somebody, SupplyFrame in your case, helps support it. And I think that that happens more often than people expect. And that that is the reason why you should go ahead and take that extra hour this weekend and build something or document it or talk about it or prepare a workshop. Because right now it's your free time but what if what if somebody comes along and says oh well we can make that easier for you and then it's your free time and so much better oh absolutely you know interestingly altium when i worked there didn't have any we had training

classes and we taught training classes and they were expensive they were like two thousand dollars for a week and people would come and they would attend a four or five day training class and they'd learn how to use Altium Designer. And at that time I was an FAE and I dealt with a lot of the sort of top tier support cases really around simulation and signal integrity, which just happened to be kind of my area of usefulness, I guess, to the company. Maybe because I was like a real engineer, I guess. And I say that somewhat jokingly, somewhat tongue-in-cheek,

because most of the support people were technical support people. And they could answer questions in the software, but they couldn't answer engineering questions. So I got frustrated at one point, and I went home. And this is like 2002, so it's a long time ago. Um, and I created a video on my computer about, uh, spice simulation and fundamentals of spice simulation. Like, how do you get the spice simulator working? How do you set up models? How do you map pins? Um, how do you create models from, uh, multiple models? So you can black box these sort of functions and then simulate them as a unit.

And I put it on our FTP server and I started pointing people to it. I created an email signature in Outlook at the time because that's what everybody used. An email signature in Outlook and I would point all of the customers that had simulation questions to that. And when some people in the company got wind of it, it was like a revelation for them.

I mean, for me, it was like, yes, we should just do this because it's easy. And it's not because I'm particularly smart or anything like that. It was just that it was just easy. I was, if anything, I'm profoundly lazy and I just got tired of typing the same responses.

And mind you, this is an AVI. It's a function call. Exactly. This was an AVI file because that's all the video compression that was available to me. And so 128 meg AVI file sitting on an f2p server somewhere this is like a pre-youtube kind of thing and uh and the company

really embraced it they were like yes absolutely we should do this we should do this for every simple discrete kind of thing so over the course of the next several years, we started to do more and more videos. We got some screen sharing stuff. We started to share screens with the customers, that sort of thing. And it totally transformed the way that we did business. And then we really took a very serious approach to that. I mean, I totally agree.

But they were making money from those courses. They were making a lot of money from those courses. You know what? And we didn't lose any money right what people really gained from from that piece of content was insight into kind of how we operate as a company look you're not alone we're going to help you to figure out how to use these things we're going to teach you exactly how this stuff works. We're not going to presume that you know anything. Because oftentimes, again, going back to that quip from Kevin Morris, oftentimes we have this expectation that the customer is somehow totally aware of all of this capability. And we're somehow beneath the customer because we're the people on the front

line handling the support side of things or whatever. But the truth is, is that we all go through that exact same process. I mean, it's really the curse of knowledge, right? You can't know what it is to not know after you already know. So, you know, it truly is a curse of knowledge, and you kind of suffer through that for a bit. But then you realize that that's actually how you sell your product is by giving people the information that they need to be successful. And, you know, so many people have helped me to learn things that, uh, uh, I feel like it's my responsibility to teach other people how to do stuff and share that information. You know, I grew up in a house with a dad who was an engineer.

He's, he's very insightful, very back of the envelope, designs to 80%. The last 20% says it's just a waste of time. You're just going to iterate, iterate, iterate, stop counting Fento, Fair Ads, all these other sorts of rules and wisdom that come from being older and doing these things. So now that I'm a bit older,

I have this opportunity to actually turn around and kind of share the same stuff that was shared with me. And I think it improves the supply frame brand. I think it elevates the supply frame brand and supply frame says, yeah, you know what, look, we'll totally support that. But, you know, I don't know that they're unique in that space. I think everybody's trying to figure this out. And even as a company, you know, our customers are customers like Texas Instruments and Microchip and all these other folks. And it's our responsibility as a company also to then be advisors to these guys. How do you transform your marketing department

so you can reach into this audience and actually talk to these people on their terms and in terms that don't seem totally slimy or outdated, right? Because you're going to look really embarrassed if what you do is another mailer that you send to this audience of people. Yes, please. Either send me spam or send me paper mail.

Yeah, exactly. Wow. Exactly. We have kept you for far too long, but we still have a couple more questions. Okay, please, please. Can you say the Mandarin word for you can again? 可以. 可以.

可以. 可以. 可以. 可以. 可以. Yeah, they say it very quickly sometimes. I'm not going to try. All you have to say is like, here's another one.

You just, 对,对,对,对,对,对,对. So the word is 对, like yes, right? But when you're confirming something, you know, like sometimes we'll say yes, yes, yes, yes, yes. They say dui, dui, dui, dui, dui, dui, dui. Dui, dui, dui, dui, dui, dui, dui, dui, dui, dui. That is uniformly the response from my technicians back in China.

Kui, kui, ah, dui, dui, dui, dui, dui. So anyway. Thank you. Yeah, no problem. Do you have any last thoughts you'd like to share uh is it my final thought or is it my last thought this is the one you were worried about at the beginning fair enough fair enough no you know my final thought really is all about trying to build stuff which which can transform this this market right which can can really

transform people's ability to do and to do interesting and innovative things that's it so you know like if i left anybody something to meditate on it would be okay what can i actually do that has the potential to transform this space because it's so ready for transformation right now and i don't know that anybody is doing a great job of doing that right now i think there are a lot of people who who are well-meaning um but get quickly pulled back into the traditional approach to doing these things and i think we can we can probably uh we can probably operate differently. And I say this as somebody who's got 15 years invested in schematic and PCB. I think there are new and interesting approaches to this.

And it would be great to see other people actually riff on that. And if they need access to the APIs, to the data, and those other things, that's what we do as a business and that's the kind of thing that that we're in the business of trying to support with all of the content that we have access to so so that's my final thought is that is that too long a final thought no no that was good thank you trying to leave it something inspiring thank you for speaking with us. Oh, thank you guys so much for having me. Thank you, thank you.

My guest has been Matt Bergeron, Senior Director of Product at SupplyFrame, which is also EE Focus, Hackaday, and Tindy. Wow, I forgot to ask Tindy questions. Thank you for listening. Thank you to Christopher for producing and co-hosting and for letting me try out Lightning Round. I think we're going to do that one again.

Hit the contact link on Embedded FM if you'd like to suggest a lightning question or email us show at Embedded FM or if you'd just like to say hi. I also have a final thought for you this week. Tied you over till next week. From Leonardo da Vinci. Principles for the development of a complete mind. Study the science of art. Study the art of science. Develop your

senses, especially learn how to see. Realize that everything connects to everything else. Embedded FM is an independently produced radio show that focuses on the many aspects of engineering. It is a production of Logical Elegance, an embedded software consulting company in California. If there are advertisements in the show, we did not put them there and do not receive any revenue from them. At this time, our sole sponsor remains Logical Elegance.