One Laptop Per Child Talks

OLPC Laptop Physical Design

In October 2006, Mark J. Foster, Vice President, Engineering and Chief Architect of One Laptop Per Child and a leading computer portables expert, presented OLPC technology at the Stanford EE Computer Systems Colloquium. Specifically, EE380. The Electrical Engineering Department's, Computer Systems Curriculum

Due to its length, the transcript of his speech was divided into three parts. Below is Part 2 of Mr. Foster's speech. Please continue to Part 1 and Part 3 for the full transcript.

Mark: Mechanical design. Again, I warn you. This is not just computer architecture in this case. If you really want to solve a real world problem, you can't just restrict yourself to the pc board, because dropping a pc board in the middle of Guiana, is not going to help. You need to give them a product. Something you can hold, something you can carry, something that's strong, You've got to solve real world problems, and that's why I have to talk about mechanical design if you really want to understand the product that we're trying to accomplish.

So first, absolutely safety first, of course reduction of hazardous substances, yes yes yes, even though that's not a requirement, of course we're going to do it, it's obvious. We tried to make the thing, it's rounded, friendly, easy for kids to hold, it's got a very convenient handle on it. But, moisture, dust, dirt resistant is an interesting challenge, it's really hard. Thermals get interesting too. Even if you have a low power CPU, when you can't pump air through the box, tricky, lots of work required to solve that one, even with low power levels. But again it's important, because you need to see the kids. A lot of them live in huts, on dirt floors, and this is not a clean environment, and it's not a matter of teaching them to take care of it, the environment itself doesn't support them, so you've got to do something about that, and we are.

[question from audience] So there's no airflow at all, whatsoever?

Mark: There may be a tiny amount, but there is almost none at the moment, let me put it that way. We're working on thermal solutions as we speak. I can't tell you what the final solution will be, but it will certainly be nothing like traditional systems. Grabbing this thing, and this is blowing hot air all over the place [laptop on table].

A very extra rigid shell that we call mainframe internally. It's an internal structure that helps reinforce the outer chassis unlike a normal notebook. There's actually in fact, a plastic wall that sits between the LCD and the motherboard. The way the system is sitting right now in that top right here, the motherboard actually sits on the top and the LCD is on the bottom [sectional picture of the machine]. They're sandwiched together, so it's unlike other notebooks in that way as well. There's almost nothing in the base except the keyboard, touchpad, and battery pack. And why do we do that, because it increases reliability.

When things move and twitch, and you've got to run wires through hinges and connectors, reliability goes down. And so, by doing this one single board with the LCD right underneath it, we substantially improve MTBF (mean time between failure). We improve the reliability of the box, that's what it's for. Another little trick is we are doing 3D connector reinforcements. What we're doing there is actually internally on the plastics, one of the most common failure mechanisms on portables is your connectors. All these things that are sticking out that you are plugging stuff in and out of. And they break.

They start getting loose, and they get looser, and they eventually break and you got to get the thing repaired. So again, if you're sitting in the middle of the Sahara, you can just bump down to the nearest computer repair store and ask someone to fix it. And we don't want them to break to begin with. So we're actually building plastic around the connectors, to constrain them, to hold them in position, to provide much more strength, because again, what we're trying to do is deliver a system that will actually work in the real world for a long time. And then bumpers, these are an interesting story, I better keep it short for the day.

Basically those bumpers are sacrificial elements of the chassis. They can break, we can snap them off and put new ones on. And in fact, this is something I can't commit to yet, but we're even looking into new chemistries here, and I mean really new chemistries. We've looked at some polythyrene blends for the bumpers. Interesting tricks, why would we do that? If we can provide even more shock resistance, we will. I can't commit to the bumpers, this is literally a science project, that portion of it, but in fact we've got some of the top chemists on it, seeing if they can come up with a way to make this thing even more robust and provide further protection for the system.

It's the kind of thing that we're doing all over, is that we're trying to build a machine that's appropriate for kids. And there's lots and lots of stuff as you'll see as we continue on this that it's focused on kids. And another one is certainly the LCD. Change the subject here real quick. Nicholas did this before on a small scale, and he gave a bunch of kids in Cambodia, I believe, some IBM notebooks. Big expensive IBM notebooks, little kids.

And the question was, how are they going to work, are the kids going to take care of them? What's going to happen? And these kids had never seen a notebook before. Living in a village with basically no electricity, until a generator was brought in, they not only took care of them, you would not believe what happened. School attendance went way up, they were excited, they were enthused, and the things they learned, because they were given a tool that enabled them. Well, even with that though, things are still going to go wrong, kids are still going to make mistakes and trip, so we want to shock mount the LCD.

There's something else we did that's a little cool, is this little transformer hinge. You probably noticed that the system has kind of a skew on the front page. well, the reason, is it can go all the way around and do that. [picture of the notebook with family portrait on it] Now the basic pitch, one of the basic easy things to explain, when we're talking to folks that are interested in this machine is it's an e-book, and what we're telling them is, anywhere in the world, you're going to spend $20 to $30 a year on paper.

Not talking about royalties or anything like that, paper for the textbooks that you're handing out to kids. Even in the poorest areas of the world. Well, instead, over a five year span, you're going to have spent as much money as you would on this machine. And so our basic way of explaining it is buy this machine, download the textbooks, and give kids the chance to do far much more than they could do with those paper textbooks. That's a really important thing and that's one of the reasons why this ability to flip the machine around is so important, and it can obviously under portrait or landscape orientation in order to do that. But you need to hold it, and I wish I had the machine.

We're going to have them in a month folks. Our first system build starts November 7th. And we're actually going to have our first 1,000 machines, and I wish I had it here for you today, but it's coming very soon.

[question from audience]: What's the current weight?

Mark: Ah, that's a good question. much like the thermal question. We're currently sitting at less than 3 pounds. In fact, at the moment it's actually 2.8. But the number that I talk about is still 3.0. We're cautious, we want to make sure we deliver what we promise.

[question from audience]: Where's the keyboard? It's not on the other side of that handle.

Mark: No, it's in the bottom, so here, that actually shows you the stack. So here is the system base, the system battery pack, the keyboard rests in here, then up front you can just barely see the corner of the touchpad. Then this is the brick, that is, it is the motherboard up against the top wall and then below it the LCD. And then the snap on bumpers that surrounds the system and help protect it.

OK, in this mode, you can literally flip the lcd up, rotate it around, and bring it back down, and it's very easy to go back and forth. And in fact, I mentioned that it has this monocolor capability in the LCD. That button right there is the monocolor switch. Let's you very conveniently go between the two modes. Why be able to get to mono less quickly, because it takes less power and not run the backlight at all. Although even with our backlight on, our power consumption is mind boggling. Our panel consumes less than a third of the power consumption of just the backlight of any conventional notebook that I know of and that's with our backlight turned on. So our entire panel is just a third of just the backlight of other systems. very very power efficient.

[question from audience]: Do you think LEDs will come down over time?

Mark: We are using LED backlights. Good question. Absolutely. And it's wonderful, LED backlights are great, you can get rid of a lot of things that are a pain in the rear. Inverters, the guy that takes the battery voltage and pumps it up to 3-15,000 volts depending on the panel. I don't that in kid's hands. Especially if they're sitting in a monsoon. It's nice, they're reliable, and we've saved the space of an inverter as well. And it's much more efficient. That's certainly a factor in our efficiency.

And I mentioned before, some of the goodies that are in the machine, you can see the stereo speakers here, you can see the camera, we'll talk more of that. A little microphone input, and down on the sides, here, you probably notice that these guys flip up. Those are our wireless antennas. And I'm probably going to really screw up the video, but I want to show you that again.

Let's go to the first page, and see these little antennas here, that's another interesting thing. Those really do double our range. We get 3DB more gain with putting them up. These are not a gimmick, they make a lot of folks think they are, we've gotten some funny e-mails about them, but they're not. In fact, our wireless guy, was quite insistent on doing everything we possibly could to improve the range on the antennas. And that's what's actually happening. They work when they're down, but they get better when they're up.

Everything in the machine is certainly dual purpose, you can barely see on the side of the machine here, what we've got up top is these are the audio connectors. And then you've got a USB port here, and on the other side, two more USB ports here. In addition though, those little antennas also server as port covers. We're trying to seal the machine, keep the dust and dirt out. So when you flip them down, they help protect those ports and prevent them from dust and dirt getting in the machine. And they do one other thing. They are also the system's latch. When you close the unit, and put the antennas down, they hook in and hook into little hooks in the side of the system, and that's what keeps the system shut. So they actually serve a triple roll.

It's just some of the neat little tricks that are in here, going after the mission of how do we deliver something that's going to last a long time for kids.

Input devices. You probably saw on that previous page, under the thumbs, there were a couple of game pads. There's a little game four direction thing, plus a game button, and yes, it is absolutely for games. We don't try to hide it as anything else. A big part of education when you do it right, is getting kids engaged and keeping them engaged, is writing content that's entertaining. And games are entertaining, and we know it is real, and we know it's a fact, and we know it's what kids want. So, absolutely, we built in a game controller.

Keyboard. It's a rubber keyboard with all kinds of funky stuff on it, unfortunately, you can't really tell, you can't see some of the neat features in the keyboard. But for example, just along here there are three long function keys, and what's actually going on is that under each of those sections of rubber, there are seven separate switches. Why? Because we can use them as an analog controller. If you move your finger across it or you're doing some software, it takes these discreet switches and turns them into an analog controller.

More opportunities for interesting things like guess games. Certainly controls, but science experiments. We've got some really cool things going here, and what we can do with science projects. The keyboard also has a light. You can turn it on and off right at the control, but right behind the keyboard up where the body comes up, there are two LEDs aimed right down at the keyboard. Why? Again, you realize in this experiment that Nicholas did in Cambodia, The backlight in the LCD was the only light in the house. Kids, this kind of challenge that they're facing, yeah, something as silly as a couple of LEDs built into the box that allows them to see what they're typing can actually have a big impact on their ability to learn.

Another thing that's really neat. No one else has this that we know of in the world, is a dual mode touchpad. What this means is, it's actually split into segments, and you can see down here in the center, there's a conventional touchpad area. Just like you have in a normal notebook. That's capacitive sensing. So that's what it's detecting is proximity via sense capacity via RF field. However, in addition, the entire width of that pad is a resistive sensor.

And traditional resistive sensor [xx] [xx] conductive as you push together it measures that resistance so it knows the point. Why did we do that? I'm going to have to simulate, just go you could do that. We're giving these systems to kids that may not be able to read or write, and we want it to help teach them how to write. Here's how you can do it. You can write on the touchpad using a stick or stylus. If you put down a piece of paper you can use a pencil or pen, but it's something that, if you think about it, how would you do it otherwise?

[question from audience]: Are you going to ship unicodes to foreigners?

Mark: Of course. Not just unicode but the machine is, the only place we're going to ship English machines, is pretty much to ourselves, and our developers. Absolutely a major task is internationalization and localization for everything we do. Continuing on a theme I guess, the microphone. We've done some neat little things with our microphone. We have an ability to bypass all the normal stuff that coupled the microphone to the system. And if you want to get techie real quick, that is a DC blocking capacitor, which normally you put a normal voltage into it, and the other thing that's always there, is a voltage, a bios voltage.

I don't know if you know how microphones work, but you need a voltage to make them do their thing, and your microphone jack has it, and unfortunately that gets in the way if I want to measure things. So we include a couple of internal switches in the machine that let us bypass that stuff, and now I can go find a 25 cent photo cell, and plug it into the machine and learn about light. Cheap inexpensive things, we've got our voltmeter here so that again, something approachable giving kids a chance and some tools to help learn. And then of course software to help exploit these capabilities.

Same thing with the camera. Obvious conventional still plus video, but also sensor modes. We're trying to make sure that we can provide better measurement capability of light rather than the convenience that you're normally looking for of auto-white balance, auto-color, all that. Instead you have the ability to have it neutral and actually use it as a legitimate measuring device. And that's the kind of thing that we're trying to pull off. I got a great request today, I don't know if I can do it or not, but it got me thinking.

Somebody was asking us, can we figure out a way to provide a convenient means of adding another lens in front of the camera so you would have a really cheap microscope. Isn't that a great idea? Now we've got to think about it. We'll see if we can pull it off, I don't know. Not committing to that one, but that's the kind of thing that happens at OLPC is that the world is helping us. They're looking at us and they're recognizing what could be done, and we're getting some really great ideas, and a lot of people helping. And again, we can't do them all of course we can't, but we try to think, is there some way that we can do this cheap? We'll try.

Power system. In these environments, you can't appreciate how difficult and nasty power is. Some of the worst stuff out there, actually the AC generators that people run, as the brushes start to go and these things were throwing out huge spikes, and it will trash all kinds of machines, so this is a common problem we actually protect against this, we need to do something about it. And so, very wide range in DC input with over range polarity and surge protection built into the system. It's the real world and that's the way it is, and we've got to deal with it.

And certainly safety first. I think everyone has seen the news lately about all the lithium Ion batteries blowing up and these recalls. We are not going lithium Ion, safety was a major consideration in that decision. We're going Nickel Metal Hydride. They're much safer and they burn at about one tenth the temperature. literally. Lithium Ion is a failure, this is just a much safer chemistry, but it's a special one. Here we go into chemistry again. Lots of disciplines involved in making a portable. We want to deliver 2000 battery cycles. Now I don't know if any of you upgrade your machines too often, but the reality is for most people, a battery lasts about a year and a half if you're a reasonably heavy user of it.

About 500 cycles, that's about it. Battery capacity starts to go way down, forget it, go get another battery. Again, in the middle of the jungle in the Amazon, you can't run down to the local Radio Shack and get another battery. What do we do? And the cost alone, of having to replace the battery if it's 500 cycles, four or five times over five years would be a huge adder. So what we're doing is we're working with a battery manufacturer to make a new version of Nickel Metal Hydride. Modified chemistry that our target is to actually deliver 2000 usable real battery cycles. Not easy. Don't have the test results back yet.

But very serious work for making this happen including unique charging and discharging algorithms, unique sensors, in fact record higher precision on both the voltage current and temperature sensors. To pull this off requires very precise monitoring. Particularly in the face of high frequently changing ambient temperature which is exactly what we're living with. But that's one of the many things that we're trying to do in the box.

Gang Charger. Common situation that I mentioned is that you've got power at the school, but you may not have power at home. So what do you do? Well, we're trying to make sure we've got a gang charger so that a common mode for those environments are like this. Just going to swap your battery for another battery and run all day long. Battery life is a very big deal to us, we can deliver all day long battery life in any reasonable environment.

So where it's possible, the gang charger. And then there's this one. The source of both skepticism and enormous engineering challenges. But very real. Human Power. What do you do if you don't have AC power at home, you don't have AC power at school, what do you do? Human power. We're on the fourth iteration of this concept. We've gone through probably fifteen different versions, which you've probably all seen the picture of the crank in the green machine.

That was a wonderful picture. When I first joined OLPC I was trying to explain to people quickly what I do. I show them that picture and most people just go, Oh I get it. Something different in a PC and you're facing some environmental challenges. So what this is, this evolution of it, as you can see, we call it a yo-yo. That's our nickname for the ID concept.

It's a spring loaded pull device, see you pull and generate power and then it pulls the handle back in. You can see the strap located on the device. That's designed so you can easily attach it to something.

Because as it turns out, we're doing a lot of biometrics and physics here. When you're going like this, it takes you so much energy just to hold the device constant that you're not as efficient, you get tired faster. So if you instead tie this to a tree or branch or a doorknob or what have you, you can keep going until that arm goes tired and then use the other arm.

If you want to, you can use it with your foot as well. If you continue to use different muscle groups, you can generate power for much longer. I'm not kidding, it's really true, and we're actually doing some research, in fact, it may be Berkley, but a professor is going to be doing some analytical monitoring including CO2 model and actually determining human efficiency as we come up with ways to use this device, recommend ways to use it, etc.

But it's real. This is actually something we will be supplying. Some of the countries that we've talked to don't care about the human power, because they've got sufficient AC. Other's, this is one of their hottest points, they're really really excited about this device. Because they've dealt with the great challenges of trying to get improved technology into these places that have no infrastructure, and some of these are really tough, talking about places in Thailand, for example, where their village is way in the mountains, there's no way you're going to run AC there no matter what you do. Your best possible thing would be to try to put in a generator but then there's no way to get fuel there. These are tough locations, some of them.

[question from audience]: solar cells?

Mark: Solar cells are not a bad choice, again it depends where. In the middle of a rain forest, you get too much rain and not enough sun. In other locations, they may well be possible, but they're very expensive. And this is an incredibly inexpensive device. Solar will be part of our solution, absolutely, in fact, earlier, you saw we have the solar powered access points. When device power is low enough, we can probably get to the stage, and the frequency is low enough, that is the frequency of the number of devices required to support end notebooks, but if it's in the notebook itself we have to be extremely cost sensitive. It's the only way we can ensure we have a solution. And so, this is going to be about ten dollars.

[question from audience]: I was just wondering why, what are the reason for you not to use pedals?

Mark: Oh, they're great, and it would cost us more to ship the pedals than it would the device. They're big, they're bulky, it depends on what you're like, what kind of room you have to store them. The other problem is... Backup... When you're making a normal notebook, first thing you do is you define your target market. Certainly if you're a marketing person you do.

You don't do that here. We have a whole variety of situations, and we're encouraging folks to invent more, and we're providing tools for them, so I absolutely guarantee you that someone will have a pedal powered option. What we're focusing on at OLPC ourselves in conjunction with a partner of course, is a very inexpensive device that has good power output, that's portable, kids can carry it with them, and use it in lots of different situations. No question, pedals are great for a big device. Different order magnitude cost, for example, a shared use device in a school.

[question from audience]: How often do you have to pull that to keep the laptop going?

Mark: It obviously depends on what you're doing. To give you a rough idea, our general metric is we're targeting ten to one. I pull for ten minutes, and it runs for a hundred minutes. Now to do that, you need to generate about 20 watts, and that's right about what this device creates. But that means I need to get the system power consumption down to 2 watts. Not easy. That's exactly what our target is, and in fact as we move on, I'll talk about power management.

Do note that due to its length, the transcript of Mark J. Foster's speech was divided into three parts. Above is Part 2 of Mr. Foster's speech. Please continue to Part 1 and Part 3 for the full transcript.