Electric power in the developing world is usually spotty, where it is available at all. And the OLPC people have clearly taken that to heart. In addition to charging from the electric grid, their machine can also charge from a 12-volt car battery, which frequently doubles as a power system in developing nations.

And it can be manually charged, too. Some preliminary proto types had hand cranks that charged the unit's lithium-ion battery. But the cranks used small muscle groups in the wrist and forearm and proved difficult for young children to turn for more than a few minutes—far short of the half hour of cranking that might be required to power the laptop for a school day's use.

IMAGE: Potenco

Yo-yo a go-go: This portable pull-string power generator may prove to be a successful commercial product in its own right.

Members of the OLPC team, working with product design firm Squid Labs, in Alameda, Calif., found that a small child, using her arms to power a hand crank, could produce only 5 to 10 watts for a few minutes before her arms got tired. It would take about 10 minutes of cranking out 5 W to power the machine for 20 minutes in its color graphics mode.

The OLPC designers considered solar cells but rejected them as too expensive. They then turned to Squid Labs, which came up with an ingenious solution: a microgenerator powered by a pull string, similar to the assembly used to start a lawn mower. The first prototypes arrived in January and were quickly dubbed yo-yos [see photo, “Yo-Yo a Go-Go”]. Squid Labs has since spun off the project into Potenco, also in Alameda, which plans to commercialize the product.

The user holds the yo-yo—a separate accessory about the size of two hockey pucks stacked together—in her left hand, grips a handle in her right, and pulls a meter-long cord. The cord spins a fine shaft at roughly 2000 revolutions per minute. An embedded microcontroller adjusts the flow of power to the battery so that the generator operates at maximum efficiency, even while the generator speed slows as the child's arm tires out.

The designers expect adults and children 12 and older to be able to produce 20 W using this generator for short periods of time and 10 W for longer periods. Again, children younger than 12 will tire quickly. The device will cost about $10 to manufacture in quantity, Bletsas says. Potenco hopes to eventually sell the yo-yos commercially to charge cellphones and other devices.

People tend to love or hate the yo-yos. Supporters note that as separate units, they can be easily replaced. That's important because, like all moving parts, they're bound to wear out.

But Felsenstein isn't thrilled with them. Using a spring for power generation is not an advantage, he says. Each time you pull the string, some of the potential power is sidetracked into rewinding the cord. And, because the human body is bilaterally symmetric, a device using both sides of the body—both hands or both feet—to generate power is much more appropriate.

Whatever their disadvantages, the yo-yos will make it possible, at least, for kids without access to electricity to use the machine. Bletsas says it will likely take 2 and a half hours to fully charge the battery using the pull-cord microgenerator. And a fully charged preproduction model will run for 25 hours in its lowest power-consumption mode—that is, displaying pages of text with the backlight off. Web browsing with the backlight on will deplete the battery within 6 hours.

Making a device that can run on kid power means making design choices that favor power efficiency [see "Little Green Kid Machine"]. Typical laptops today can consume as much as 30 W of power, depending on what they are doing. The target for the first generation of the laptop was 2.5 W with the processor and color screen active—an astonishing 92 percent reduction in consumption; the prototypes run at about 3 W for normal use. That, Bletsas says, is still not as low as it needs to be to make sense as a human-powered device, although it does make human power a useful auxiliary source of electricity.

Much of the efficiency stems from the use of a low-power microprocessor, the Geode GX 500@1.0W made by Advanced Micro Devices, in Sunnyvale, Calif. It is a 32-bit microprocessor with an integrated graphics subsystem and a memory controller that operates at 366 megahertz and draws less than 2 W of power. It isn't cheap: it costs more than $20 in quantities of at least 10 000.

Aside from the processor, the most power-hungry subsystem of a laptop is the display. The OLPC design team found ways to cut power here, too. When the child uses the computer as an e-book, the display buffer stores a copy of the screen being displayed—this allows the central processor to shut down until a new image needs to be produced. Displaying in gray scale, with the processor off, the machine draws a mere half watt of power.

The ability to switch—from color to gray scale and back again—is perhaps the computer's most exciting innovation. Says Geekcorps's Vota, “When I first saw the screen, I was so in awe I forgot to take photographs—and I'm always taking photographs.” And its manufacturing cost, some $30 to $35 instead of $130 for a conventional laptop LCD screen, is also impressive.

Mary Lou Jepsen, the CTO of the OLPC project and the former CTO of Intel's now-defunct display division, worked with Taiwanese display manufacturer Chi Mei Optoelectronics to develop this display [see “Dream Jobs 2007,” IEEE Spectrum, February]. At its heart, the display is a reflective black-and-gray LCD. The basic technology, ubiquitous in cheap watches, calculators, and other consumer electronics products, uses a polarizing film to control the reflection of light.

When a pixel is on, the chain of liquid-crystal molecules untwists, and because light isn't reflected back through the film, the pixel appears dark. Such displays are easily readable in bright light. In fact, the brighter, the better: the more light available to be reflected, the greater the contrast between darks and lights. The screen resolution of the OLPC's reflective LCD is 1200 by 900 pixels.

Pushing a button turns on a backlight—in this case, a panel of LEDs—and adds color to the picture. Conventional color LCD screens use a fluorescent white backlight, not LEDs. Filters at each tiny picture element absorb colors from the white light to define that pixel's red, green, and blue components. The problem is that these filters absorb 80 percent of the light emitted by the fluorescent light, wasting power, and the filters are a third of the manufacturing cost.

So the OLPC display uses white LEDs, which provide purer light. That purity means that the filters don't have to be as dense to block out unwanted wavelengths, so a lot more light gets through. The OLPC team also designed the colored elements of the display to operate at a much lower resolution than the basic gray display—800 by 600 pixels. Because the higher resolution black-and-gray image still shows, the perceived resolution is closer to that 1200-by-900 resolution. As a result, even in color mode, the screen uses less than 14 percent of the power of a conventional LCD.

The tradeoff for the increased daylight readability, lower cost, and lower power consumption is color saturation. That is, the colors look washed out compared with a more traditional display. That's okay, as far as Bletsas is concerned: “You don't need HDTV—you need to be able to read with a little color.”

Such a dual-mode display, to date, has never been available in a commercial laptop and will likely migrate into commercial products. “I would love to be able to turn off the backlight of my computer so I could read it outside,” says Vota.

Jamais Cascio, cofounder of Worldchanging.com, in Seattle, is excited about the display for another reason. “These will become the de facto evening lights in many homes,” he says. “Light at night is a deficiency in many developing nations that is underappreciated. Without artificial light, you can't read or do homework—the day is shorter. People will use these to read conventional books.” (Worldchanging is a Web site covering advances in science and technology that have the potential to do social good.)

Interesting as the power and display innovations are, the technical area most crucial to the laptops' success will probably be networking. The laptops are equipped with Wi-Fi radios costing about $10 each, allowing a group of laptops to establish a mesh network among themselves. These radios are built to IEEE 802.11s, a standard for mesh networks due to be finalized this year; the OLPC developers jumped the gun and designed their product according to a draft standard.

The laptop's “bunny ears” are external Wi-Fi antennas that typically provide 5 decibels of gain, better than the internal ÔªøWi-Fi antenna in a garden-variety laptop. Most of that improvement comes from the fact that the OLPC's antennas are mounted vertically above the display lid, rather than inside it, as in most laptops.

Bletsas says his design will provide node-to-node connectivity over 600 meters. Over a flat area without buildings and with low radio noise, that connection can stretch to 1.2 km. Students can put their computers on the mesh network simply by flipping the antennas up. This turns on the Wi-Fi subsystem of the machine without waking the CPU, allowing the laptop to route packets while consuming just 350 milliwatts of power.

Bletsas says that in this router mode, a fully charged computer will run for 24 hours. As a side bonus, students are likely to learn something about mesh networks—if you fail to turn up your antennas or keep your laptop charged, two of your friends might not be able to chat through your node.

The mesh network feature lets students in the same classroom share a virtual whiteboard with a teacher, chat (okay, gossip) during class, or collaborate on assignments. If the school has a connection to the Internet via phone or satellite, a computer with essentially the same hardware as the laptop but with the addition of an Ethernet interface and a hard drive will act as a server for the school network and a router for Internet connections. The OLPC program expects to be able to produce the servers for €100, or about US $130.

Internet connectivity in the developing world is rare today, but the presence of so many Internet-capable computers in a school may spark administrators to invest in Internet connections. That's the hope, anyway. And Bletsas says his team is doing everything it can to make that a reality, including negotiating for low-cost Internet connections for the schools in an entire country, and developing school servers with solar-powered repeaters.

The final key piece of computing hardware is the 512 MB of flash memory, costing as much as $20. “This is one place the designers clearly made a compromise to get the total component cost down,” says Ethan Zuckerman, a research fellow at the Berkman Center for Internet and Society at Harvard Law School. “The device would be more useful with 1 gigabyte of memory, but that would have cost an extra $20.”