It encourages people to be interested in science, technology, engineering and maths. It’s a step education thing, I defy somebody not to be excited by that, I just defy a kid not to be excited by taking pictures from space.
Enthusiasm radiates from Eben Upton. By day he’s the Technical Director and ASIC architect for Broadcom. By night, and on weekends, he’s the driving force behind the Raspberry Pi, that small computer that has been revolutionising hobbyist computing and the future of technology itself since its launch in 2012.
Tall, dressed casually and with a giveaway clue as to how he travelled to work that morning – cycling glasses clipped to his t-shirt – the founder and one of the trustees of the Raspberry Pi Foundation is explaining to me the background to the amazing minicomputer, the three-by-two inch media centre, web server, retro gaming hub, educational tool and even, as we’ve been discussing, space-bound photo snapping machine.
But Upton isn’t your usual computer designer. He and his team didn’t build the Raspberry Pi to retire on. What they planned was something audacious, something fantastic.
They realised that with the right hardware, the Raspberry Pi could change the world.
The Origins of the Raspberry Pi
We didn’t originally see ourselves could building a general purpose computer.
Devised in 2006 and inspired by Acorn’s BBC Micro (a 1980s computer that was built for the BBC Computer Literacy Project and sold largely to schools in the United Kingdom), the Raspberry Pi began life as a much larger device, the aims of which were established when Upton collaborated with teachers, academics and enthusiasts with the aim of devising a computer that would inspire children.
“We didn’t originally see ourselves could building a general purpose computer. We didn’t see ourselves building a PC. We saw ourselves buildings a programmable device. We started off with microcontrollers, I’m sure you’ve seen them on a big orange board [as seen on the Raspberry Pi website, this was a microcontroller-based platform that could drive a telly, plug it into the component
of a television and yes, it was a really interesting device. It had about the performance, I’d say of a 1980’s microcomputer.
“You could build it yourself, you could solder it together by hand.”
Instantly I’m reminded of 1970s breadboards and very basic computers constructed by hand, of Steve Wozniak assembling boards in Steve Jobs’ parents’ garage back in the mid-1970s. Of course, computing has moved on considerably since then, but Eben is clear that this idea of a computer that can be reconfigured by hand is important.
“That’s been a key to the success of the project. People want something that they understand.
“It was really only sort of post 2009 that we could start to build something along the lines of a general purpose computer, [when] the right chip came along, a Broadcom chip with exactly the right featureset for Raspberry Pi became available.”
The Raspberry Pi that saw release in early 2012 doesn’t have any removable parts, of course, but it can be reconfigured based on the particular purpose you have in mind for it. This is why the Raspberry Pi ships without a keyboard and mouse. As Eben puts it: “we had a particular vision that the thing would be useful and we built it and it was, and it’s pretty much that simple.”
The British Computer, Made in Britain
I don’t mean in a jingoistic way, but using a chip that was made in Britain, using an IP that was made in Britain … it’s British.
Computer manufacturing in Britain since the end of the 1980s has largely been non-existent, with international 16-bit wonders like the Amiga and Atari ST, PCs, Apple Macs and games consoles replacing the successful homegrown 8-bits. Other than a range of Acorn computers in the 1990s aimed at educational establishments, these have all been largely based on designs from the USA and Japan.
The days of 8-bit classics like the Sinclair Spectrum ZX80 and Amstrad CPC 464 are long gone – but in the Raspberry Pi, British computing has something to be proud of once again.
I mentioned this to Eben.
“I don’t mean in a jingoistic way, but using a chip that was made in Britain, using an IP that was made in Britain, including the ARM obviously [ARM processors originate in Cambridge]… it’s British. One of the things that changed for us is we have a global dimension that we didn’t probably see when we were thinking about educational application which was our original thought.”
But what of those greats that went before, Sinclair, Amstrad and Acorn? Could Eben one day be compared to giants of British 8-bit computing, such as the idealistic Sir Clive Sinclair (designer of the Sinclair ZX80 and ZX Spectrum) or the shrewd businessman Baron Sugar (the founder of Amstrad, responsible for home computers, word processors and IBM compatible PCs across Europe in the 1980s and 1990s), or even Christopher Curry, founder of Acorn Computers (whose legacy lives on in ARM)?
“You’ve got to remember these guys, sold millions of computers and they created a revolution, they created an industry, they created my job, they created Acorn which created ARM, they created the Cambridge tech scene which created the core that we use in our chip. There’s this continuous lineage starting with these fellas.
“I think if in 10 years’ time we transform the UK tech industry skills pipeline then I think we’ll deserve some credit. Right now we don’t deserve that.”
You’ll be intrigued to learn just how Eben intends to transform tech skills both in the UK and further afield. But first, how exactly is a Raspberry Pi constructed?
The Raspberry Pi Construction Line
In order to understand more about the Raspberry Pi – including how the costs are kept so low – I visited the production line, where I was shown around by Gareth Jones, Senior Manager of the New Business Division at Sony. The background to the plant and the process involved in building a Raspberry Pi are both very interesting.
Construction of the Raspberry Pi first began in China back in 2011, but the results weren’t considered satisfactory. As a result, the Raspberry Pi Foundation moved production to the UK.
Situated in a former colour TV manufacturing plant in south Wales (a peninsula in the United Kingdom) are four production lines (soon to be six in order to keep up with the Raspberry Pi’s remarkable demand) producing 12,000 of these amazing little computers every day. This is one of several collaborative partnerships for Sony, and the end product – a working Raspberry Pi device – is constructed here for Premier Farnell, under licence from the Raspberry Pi Foundation (they are one of two licensed manufacturers, the other is RS Components. Both are British companies).
Owned by Sony Corporation and with a large portion of the factory producing broadcast television cameras, the factory – part of a larger complex of offices – is a hive of focused activity.
The process of building a Raspberry Pi is quite remarkable. Utilising component suppliers throughout Europe and the Far East, the process begins with solder paste screen-printed onto a board. Most of the manufacture is automated, controlled by cameras and positioning controls to correctly align the board and apertures in order to get the necessary 100 micron thick paste deposition.
Following this, components as small as 0.5 mm are mounted on the board, loaded into a machine on a reel and picked up using a robot arm with a vacuum nozzle (a method known as double-sided SMT, in which the components are mounted on both sides of the printed circuit board). As you can see from the photo above, this stage of the process resembles an early computer spewing data on tickertape – the truth is quite different, of course! Components arrive from the suppliers on these reels of tape, each holding 10,000 transistors, diodes and other discrete components. A system called TIMMS manages the process, monitoring components and acting as stock controller, alerting operators when a reel of components is about to run out. Again, a camera is used to check the parts on the nozzle for orientation, and when placed they’re held in place by the solder paste.
Due to the volume of production around 400,000 components are mounted each day. As the process continues, larger components are added, again with cameras checking correct rotational and X-Y positioning. If there are any problems, the board’s orientation is adjusted to suit.
The secret of the Raspberry Pi’s success is unknown, but it probably has something to do with the main chip and its space-saving, package-on-package arrangement, in which logic and memory chips are stacked – the same approach as used in the manufacture of smartphones. This begins with the CPU component picked up and placed onto the board, once more held in place by the viscosity of the solder paste. Once in place, the RAM chip is picked up, dipped into solder and placed onto the CPU, with closely controlled accuracy to get the right depth – too much or too little solder would be disastrous.
Following this, the boards are passed through an oven at 238 degrees Celsius, the temperature at which the solder will evaporate. The Raspberry Pi boards are almost ready, save for testing, which is done in several stages.
First comes the visual inspection, again performed by camera and computer. Following this the hand insertion part of the process adds the Ethernet, audio jack and other components that are too large for the automated line. These components are passed through a flow solder machine (a fast way of soldering components in bulk), with joints then checked visually. Testing continues at eight test stations, with four people manning two stations each.
Here the MAC address will be downloaded to the Pi, and the DC power, audio and video output will be checked, taking 60-90 seconds. After this, the boards are placed in ESD bags and prepared for shipping. The testing process hasn’t ended, however. An independent QA team tries the boards, running them as customers would, in order to test the Raspberry Pi as it would be used in schools or at home. QA defects are apparently rare, with only 115 returned from the 450,000 produced in this plant (I’m told that of this figure, only 18 were actual defects – the others were likely faults due to poor SD cards).
Gareth Jones tells me that the Raspberry Pi is being considered as a processor for controlling industrial processes, with a division of Sony in San Jose interested in using the computer for a project they’re developing.
Most crucially, however, is the news that the plant has had many school visits. In a world where manufacturing is generally thought to have been largely outsourced to China, this is vital as it shows children that not only is a popular British home computer being built but that manufacturing jobs exist, and they’re not about being covered in grease and dirt.
The Raspberry Pi Ethos
We’re not just doing this because we think Raspberry Pi is great, but because we think a very small number of easily fixable things are broken.
Figures indicate that the Raspberry Pi has sold around 1 million devices, with orders still to be fulfilled at the time of writing. Of this amount, the Foundation estimates 2-300,000 have ended up in the hands of children, far exceeding the 10,000 devices a year they expected to be sold to schools. “One way or another it’s in the hands of kids whether [via] parents, schools, teachers, grandparents or [bought by] the kids themselves,” says Eben, “particularly with stuff like Minecraft on there.
“The thing about that is it is really meaningful to kids, you can make a program to build a house rather than having to build a house brick by brick. Once you have built a program that can build a house you can build 2 houses and that kind of automation is something that is really valuable.”
But getting this computer into the hands of children is only the first step in Eben Upton’s plan to combat the tech skills shortage.
“We have a country where we have vast numbers of people who need jobs, we have vast numbers of companies crying desperately for people… not even PhD mathematicians, just people who can program. We could rely on immigration for a little bit, that would work. The problem is that turns the country into a parasite on India, we and America will suck India dry of all the best programmers. That’s an appalling piece of behaviour.
“We can’t keep doing that, that’s a short term fix.”
It’s unusual to meet someone who appears to have an answer to such an important problem. Eben continues, arms animated with enthusiasm for what he sees to be a simple, long-term solution, as I observe that this isn’t the sort of response we hear from government.
“I don’t know why. We all know it and it’s just a realisation borne of just spending hours reading CV’s. Just finding enough people to staff an engineering team. We’re not just doing this because we think Raspberry Pi is great, but because we think a very small number of easily fixable things are broken.
To combat these problems, the Raspberry Pi Foundation was set up with six trustees (among them David Braben who is best known to many gamers as the co-writer of classic space trading game Elite) to manage the running of the organization. To assist the trustees, Eben and co have hired six employees. “We had no employees for the first 700,000 Raspberry Pis sold, so we have really staffed up having six employees. We have a much larger community of volunteers who we spend their evenings and weekends working on Raspberry Pi stuff, some at uni who spend their days, evening and weekends working on Raspberry Pi.”
“We’re a small group of guys, we think we can make a difference and the fact that we think we can make a difference means we think the problems are fairly straightforward. Like the thing with kids, if we thought we were forcing kids to learn to program, we wouldn’t get anywhere. What I think we have realised with Raspberry Pi is if you give people the tools, they’ll do it. We don’t need to push the program, give them tools a chance to build a house in Minecraft, give them a chance to make the cat run round in Scratch. Giving people the chance to do physical computing, that’s been a real surprise to me.”
Where Next for Raspberry Pi?
With a particular constellation of maths and science skills, what happens there is we fight over that pool of people what they tend to be is middle class white boys and we fight over this pool, we, and the physicists, mathematicians and investment bankers just fight over this pool of people and it’s a zero sum game.
With 1 million units sold and a growing reputation among hobbyists, home theatre enthusiasts and its aim of being a child-inspiring computer fulfilled, is Eben’s work done? Has the Raspberry Pi done what it set out to do?
“I think we have been fairly candid that there is a lot of software work to do. Focussing on software rather than running of after the next faster processor – there is no better chip we could be using at the moment certainly in the price bracket. The good thing about software is you spend money on software once and you get it back on every device. It’s quite expensive, optimising the software for the Raspberry Pi but then everyone benefits. All those million people who bought a Pi last year, they get the benefit on their Pi, it’s not like they’ll have to buy another one. Getting mileage out of the software, is really what we’re doing here. I’m sure some point in the future we’ll do another Pi, I can’t obviously see what we’d change today but if a better chip came along we might consider it.”
Might he consider a larger device, mounted on a bigger board?
“I think the size is important – the size is important for cost. I think one of the things that allows you to do that, to hit that price point, is the stacked memory configuration [the package-on-package described above]. If you want to move away from that, I think much bigger than it is at the moment wouldn’t be cool anymore.”
Eben isn’t too concerned about the competition either – at least, not the current competition.
“There is nothing else in the price bracket. I think people are managing now to get devices $50 range, although ones that claim to be $50 don’t tend to be available so in practice people who are managing to get into the $70 range with things that have comparable levels of performance.
“Some of these are based on a variety of Chinese devices often with very low end Cortex processors and very low in multimedia. So people are managing to get a device with a lot more CPU and a lot less GPU in that price bracket. Yes, it’s interesting to see those emerge. I have yet to see one that is particularly threatening to us.” Would it even be a threat at all?
“The issue of threats is an interesting one, I don’t really acknowledge anything as being threatening but our aim is for there to be lots of small programmable computers. If someone builds a lot of small programmable computers that’s good.”
Small programmable computers, in the hands of enthusiasts and children. Not an instant solution to the technology and engineering skills crisis, but a positive step forward.
“Occasionally you hear someone say ‘there’s no skills shortage, there’s no engineering skills shortage’ – my ass! The point where I am besieged by qualified interview applicants for the jobs we have here in this building I will believe there is no engineering skills shortage.”
But aren’t maths, science and technology subjects that only the academically gifted can excel at?
“There are skills someone can bring to their computer programming, there are tasks which are basically best done by people who are mathematicians but I think there is an enormous potential skills base of people who if you get them at the right age and you give them access to the right tools they will learn to be great computer programmers. You see guys going for interview in the games industry in particular come through the non-traditional route. I came in via several courses at Cambridge, but I also had a hacking background as a kid so I was programming when I was 10.
“I think that there is limitless potential. The thing with programming is it’s not very hard, not academically hard. I know a lot of people who are very extremely gifted computer programmers but don’t have traditional academically ability. It seems to be completely decoupled in the same way as being a good carpenter is decoupled.
“I know people who started work at 16 and they are great computer programmers so I think the impact is potentially enormous. The risk is that we might position it at the most academically able kids, the people who already have maths. With a particular constellation of maths and science skills, what happens there is we fight over that pool of people what they tend to be is middle class white boys and we fight over this pool, we, and the physicists, mathematicians and investment bankers just fight over this pool of people and it’s a zero sum game.”
So what is the answer? How does Eben propose to fight against this constellation of skills, a self-perpetuating skills crisis that ignores potential?
“What you want to do is chip away at the three things I said, white, middle class, male. Double your pool, get girls involved, ethnic minorities get involved.”
The point where I am besieged by qualified interview applicants for the jobs we have here in this building I will believe there is no engineering skills shortage.
Eben Upton is seemingly not a man whose aim is to build a business out of nothing and retire on his yacht. Rather, he is almost certainly a man who noticed a real-world problem and realised what the solution is. Most people might try to enter politics or activism in order to change the world – Eben Upton chose to build a computer as a long-term solution.
That computer is British is a source of pride for those of us that know. We usually look at computers as devices devoid of geographical constraints and concerns. Endless streams of computers, tablets and smartphones built in the Far East drop into our hands on a daily basis, their birthplace understood but its relevance ignored. It has become largely unimportant, a fact of life, of gloablisation.
Does the Raspberry Pi represent a sea-change, a move towards computers becoming more parochial? I doubt it. Instead, the Pi is an enabling computer, a powerful and potent piece of hardware that can genuinely revolutionise the teaching and learning of computer sciences not just in the UK, but worldwide.