Greetings folks, and welcome to the Electromaker podcast. This is a very special episode, as you've probably seen by the title. We are interviewing Eben Upton, who is the creator of Raspberry Pi, and who still leads Raspberry Pi to this day. Regular listeners to the podcast will know that for a long time I've been wanting to branch out and start doing occasional interviews of people, and I cannot be happier with who we spoke to for this very first one. We started this interview actually with Eben in his car, and he was slightly concerned that the audio quality might not be up to scratch. So, he very generously said, 'Look, let's just start this interview again. I'll drive home, and we can do that somewhere a bit quieter.' Which is why this interview starts somewhat in medias res, as it were, because we need to go over a couple of questions that we've already talked about, and we were talking a little bit about Embedded World, which is a conference held in Nuremberg, in Germany, which we were both at, but of course he was there working, giving lots of interviews. I was walking around making videos for Electromaker. And then we get into the meat of the interview. He was very generous with his time, but that said, there was still only a limited amount of time. I had a list of questions as long as my arm I wanted to ask him, but I hope I cherry-picked ones that you would find interesting. There is also going to be a video version of this interview on the Electromaker YouTube channel. I'll probably also cut some clips out of it, too. But for now, without any further ado, the first Electromaker interview podcast with Eben Upton, creator of the Raspberry Pi.
Embedded World. Yeah, it was amazing. I mean, it was such a... It was I think people couldn't quite agree on whether it was a good turnout, whether it was a better than the usual turnout, worse than the usual turnout, but there's just an enormous amount of energy. I mean, I think people were just desperate to get back in the room with other human beings, and get back to life as it was before, you know? Absolutely. Of course we had a giveaway, you know, we gave away 25,000 Picos on the stand. You know, we really... our message... I think a lot of silicon vendors were taking to the show the message that you can't have our silicon. We don't have any silicon, and we were taking the message we have so much silicon that we're going to give away free silicon from our stand. So, we had this tiny little stand. And all very often kind of thronged, you know, people in front of it. Often people wanted to talk to us about the existing product, the Linux based product. But a lot of people also wanted to kind of find out about us as a semiconductor vendor, which was really gratifying.
Yeah, absolutely. Yeah. Well, I mean, I think that's one of the things that as we talked about just previously to recording, that the RP2040 came along at a very special time. It probably always would have been the success that it is, but yeah, you were able to kind of say at a point, 'Look, we have this.' And it put it into the hands of a lot of people that maybe might have hesitated.
Yeah, that's right. I mean, it is a product we believe in. Obviously, it's brought a lot of our experience of what makes a good microcontroller, what makes a bad microcontroller. We've used a lot of microcontrollers over the years. We have our ideas of what makes a good microcontroller, and really that's distilled down into RP2040. So, it's a product we believe in. It's got a very attractive price performance point. These things start at seven? These things go down as low as 70 cents in middle volume. And it's got a lot of performance. So, we do think it was going to be successful anyway, but I think what's happened is engineers are all conservative, it's good that engineers are conservative, we stick with what we know. And I think what's interesting is the shortage of other more established microcontroller products has given design engineers a license to experiment a little, try new things. And obviously RP2040, which came into the market in January of 2021, it kind of came along at exactly the right time to happen to that newfound willingness to experiment, and to be of service. One of the wonderful things for us with RP2040 is there are businesses, particularly in that maker pro kind of space, which would have gone to the wall. They come to us saying, 'Look, the alternative to buying a chip is I go out of business.' Being of service to people, particularly smaller customers, has been gratifying. It's been a morale booster for us.
I can imagine. And I mean, from my perspective, I studied music, and so I'm very involved in kind of synthesizer world. And it's amazing to see the catapulting of the amount of different complex Eurorack modules that are all using RP2040. A lot of people do. But as we talked just now, about Embedded World, which was very nice. I was there, too. And as someone who lives in Germany, it was wild to be in a big room full of people without a mask on, because that's the first time that's happened for me in 2 years. It was very strange. But we're also talking just after the Raspberry Pi Pico W and H have been released. And while as we're saying, the RP2040 is the exciting part of that, it is rare for a company to have such a wide range of people to react to things that they release, right from the side of the industry that follows what Raspberry Pi is doing closely, to hobbyists and even younger makers who are getting into it. So, from your perspective, how has the launch of the Pi Pico W gone?
I think the launch has gone extremely well. We... I guess a lot of organizations, ourselves included, have had to refocus their attention over the last year on sustaining existing products. So, there's maybe been an adjustment in allocation of engineering resource away from making new things, and towards doing the work required to keep existing things in production in this very strange supply chain environment that we're living in. So, there's a hunger, particularly in the hobbyist space, the hobbyist and education space, a hunger for new things. Not many people are releasing anything new. Pico W has come along at a good time for that. Obviously, it adds this new layer of capability in Pico. We produced the first board, unsurprisingly, with our own silicon on it. Since then, there's been this explosion of both end user products using RP2040, but also of other people's take on the RP2040 development board concept. And so, it's kind of nice to come back to that a year and a half in, and add a really important new capability, a lot of people have been asking for in the platform. We're glad we've been able to do it at only a modest cost increase. So, we've gone from $4 to $6 for Pico to Pico Wireless. And yeah, it's been well received. It's been great to see people getting them out of the box. When we sent sample units out to people before launch, it was good to see how quickly people were able to get online with it. And since units over the last couple of days have started arriving in the hands of actual civilians, it's been great to see that experience has been replicated in the wild.
Absolutely. Yeah. And I think as you say, there's a great worth in not only getting such a small price increase onto the wireless board, but to be able to get it out there, and to have good documentation straight out of the bat. From reading the documentation, I agree, one of the things about it that's quite lovely is just a few lines in the REPL, and you can search Wi-Fi in the REPL. And for people that have fought with various other boards in the past, I'm sure that's something that will mean a lot to them. And to just briefly touch on the other release, there's the Pico H, and of course there's going to be the Pico WH. The idea of actually putting pins onto the Pico, being able to put it into a breadboard, is one of the things that I know in the maker sphere people did mention about the Pico early on. Was there always an intention to make a more breadboard friendly version of it, or was that more based on feedback from the original Pico?
I think we probably would have got there in the end. I mean, if you see, we have our Zero products, and the Zero products are the smallest big Raspberry Pi at $5 and $10. And they have headed versions. There is a Zero H, and I'm not sure I've ever seen a Zero H in the wild, but there is a Zero WH. So, that convention follows on from the Zero products. We did a little bit more work this time. The H products are not just the non-H products with .1 inch two rows of .1 inch headers soldered on. We've removed the castellations, so the non-headed boards have castellations to make them easier to use as an embedded module. Taking the castellations off the side, which is nice, just makes them a little bit slimmer. And then rather than using a pair of rows of headers, we've had a custom frame designed. Well, when you look at it on its own, it fairly closely resembles a DIP socket. It's actually got two rows with crossbars, with four plastic crossbars between them. We've created a new standard for debug header. Previously, 3.1 inch space pads for debug on Pico, Pico W. On this, we have a little JST three-pin JST connector, which will be our future standard for zero wire debug on Raspberry Pi microcontroller products.
Cool. Yeah. And as you say, just even breaking those things out a little bit opens it up to a much wider range of people, because there's a lot of people who are still doing their first breadboarding tasks. And it's interesting to me that there are a lot of younger people who will be getting their very first experience of any kind of hands-on embedded work with the Pi Pico, and of course the new Pi Pico HW. Moving slightly away from the Pico, but still staying within the Pi products that have had their day in the last couple of years. So, every major Raspberry Pi release board has always been very popular since day one, and the main line boards, and of course the Pi Zeros, because there's nothing cooler than a very tiny Linux computer. I've got several Pi Zeros, and I love them. The compute module, however, I might be wrong about this, but I feel like that's one of the ones that had a slightly slower uptake. As in, as more recently as the Compute Module 3 Plus, it maybe wasn't something that was a massive seller. Then the Compute Module 4 came out, and it seems like the entire range is one of the most coveted things that Raspberry Pi now makes. And why do you think they took off in such a massive way?
Yeah, it's as you say, Compute Module products for us they're always a slow burn. I think we're still in the volume up ramp for CM3, which is the 2017 product. So these have very long design-in cycles. They're very sticky once designed in, which is nice because it gives us some predictability in terms of demand. But they do take a long time to design in. CM4 has been very strange though, and we asked ourselves a lot what's happened with CM4. I guess I can point to maybe four things. One, I think people do genuinely prefer the form factor. We are very big believers in the SO-DIMM form factor, but this kind of perpendicular Hirose connector form factor. It is more area efficient on the baseboard. It does let you build much smaller baseboards because you don't have the metal frame of the SO-DIMM connector. So I think that's resonated with people. The inbuilt wireless, the option to have inbuilt wireless has been really helpful. We had some very candid feedback from the Android Things team at Google back a few years ago about the challenges that not having Wi-Fi on the compute module products had. They basically said, 'Okay, fine, you develop something on a Raspberry Pi 3 and okay, you can migrate to CM3, but then you've got to do a bunch of wireless conformance testing yourself.' So unloading that from customers and providing antennaed and non-antennaed onboard antenna and external antenna options was good. Power chain's a lot easier as well. Earlier few modules we'd push the obligation to generate a bunch of rails and to sequence those rails appropriately on the designer of the baseboard. That's not an onerous task, but it's not no effort. And so that's a factor. Just putting 5 V in, it makes all the rails. It gives you 3.3 V, it regulates you some 3.3 V to even use yourself. So that's been good. And I guess possibly also the supply chain thing. It's genuinely hard if you want to go to scale after you've developed something on Raspberry Pi is to go buy chips in the market and go chip on board. That path has become a lot harder in the current situation. And so it's all of those things together. We just blew the doors off with CM4, and of course the challenge right now is building enough of them to keep people fed.
The Electromaker podcast is sponsored entirely by the Electromaker store. In short, we have a store at electromaker.io. We are not immune from the chip shortages, but we do have a fairly good stock. If you are wanting to support this podcast financially, the best thing you can do is consider us when you are starting your next project. Thanks so much.
Um, and I mean the chip shortage is something that is almost inevitable that it gets talked about, but I think it is something that people are a little tired of hearing about, but for perhaps a slightly different question about the chip shortage, in general, apart from the obvious lack of parts, has the chip shortage actually changed any design priorities for the Raspberry Pi team going forward or is it more a case of kind of waiting out the storm and trying to get back to normal operation?
Um, there's an element of waiting out the storm, certainly. And there's an element of managing our customers through the storm. We do a lot of very intensive work with our customers to understand their demand, OEM customers, our industrial customers, to understand their demand profiles and make sure that we can... every vision is that while we're in a shortage situation, every object we make goes to a home immediately. We're not interested in shipping somebody a thousand CM4s if 900 of them are going to go on a shelf for three months. We'd rather ship you a hundred now and then ship you another hundred next month and keep you going. And that's labor intensive. I mean, that's a really good... we have the best part of a thousand direct OEM relationships being maintained by the commercial team at the moment. It is a punishing experience for the people involved, and it's not something I would want to subject people to for years on end, but it is getting people through, just getting our OEM customers through. So there's been a sort of a change of priority probably from the commercial team, the commercial organization within Raspberry Pi from, I guess, sales generation to demand management, to supply and demand management. Actually the same skills, right? The same people. The people who are good at one are good at the other. The same people, but it is strange not to be going out and creating demand. It's strange to be just servicing demand which is coming in. So change of focus there. On the engineering side, it's as I said, it's really about deprioritizing new stuff. Why Pico W and friends are so exciting, right? Because they are new stuff in a time when doing new stuff is challenging. It's about deprioritizing new stuff and focusing engineering resources on things like qualifying alternate parts. You know, one of the ways through this crisis is if you have a chip on the board and you can get it from three vendors, you make sure that you've done the work. You don't just throw the chips on and go, yeah, they should be good. Throw it on, ship it out the door. Actually building the prototype runs, doing the qualification, getting comfortable. Sometimes there are PCB modifications associated with it. So that's again labor intensive. Feels a little sad sometimes. They're victories, you know, survival is victory. Like keeping the product shipping is victory, but they feel like little victories where new genuinely new products like Pico W feel like big victories. But we're doing what we have to do to get through. We're doing what we have to do to support the customers, and particularly the OEM customers because you've got to remember these are businesses, often small businesses. If they don't get supply, they die.
