Megatrends don’t simply happen on their own. They start with disruptive technologies that completely change the status quo, like gunpowder, the airplane, and the microprocessor. The trick lies in identifying potential disruptive technologies early on and then predicting where they might lead.
Back in 1976, I was part of a group at Tektronix tasked with retraining oscilloscope sales and field engineers to sell microprocessor development systems for the Intel 8008 microcontroller. Part of the challenge was to point to potential applications. After much head scratching, we told those engineers that perhaps these strange chips could be used to control elevators or washing machines.
Obviously, we hadn’t quite put our fingers on what effect they would have. So how many new technologies have the potential to similarly change the electronics we use every day?
One of tomorrow’s megatrends may be reflected in a recent Stanford University seminar. Leah Buechley, a post-doc at the Craft Technology Group of the University of Colorado’s Computer Science Department, offered her educational goals in “Computational Textiles and the Democratization of Ubiquitous Computing.” Specifically, she wants to involve non-technologists in creative applications for microcontrollers in clothing. Functions could be decorative, like flashing jewelry, or artistic, as part of a performance.
In her paper, Buechley describes “a reconfigurable costume \[that\] consists of a torso piece and an assortment of sensing appendages that can be snapped to the torso. Sensors in the appendages include muscle-flex sensors, accelerometers, bend sensors, and touch sensors. Sensor data is relayed to a computer, via a Bluetooth module embedded in the torso, where it can be used to control or generate music, video, and other multimedia content.”
Assembly seems relatively simple. “The costume, built using my version 2.0 e-textile construction kit, is form-fitting and stretchy,” the paper explains. “The electronic modules are kept as small as possible so they do not interfere with the dancer. The costume was used in an improvisational performance in May 2007 to control a player piano.”
To make that kind of design practical, Buechley investigated sewable conductors and developed the LilyPad, a fabric socket for an Arduino microcontroller (Fig. 1). Arduino is an open-source platform based on Atmel AVR microcontrollers and peripherals. It runs the Processing programming language and integrated development environment.
Like Arduino, Processing aims to get people in the electronic arts and visual design communities over the “math is hard” hump. The programming element is derived from Wiring, a C/C++-like language. The development environment is Java-based. Buechley chose Arduino because it already had an established user base and ready-to-use hardware (available from www.arduino.cc).
To reach out beyond what she has personally accomplished in small classes with middle and high schoolers, Buechley developed her LilyPad from a labor-intensive, cottage-industry fabrication of cloth, conductive thread, and circuitry to a mass-produced product that’s now sold online by SparkFun Electronics (www.sparkfun.com/commerce/advanced_search_result.php?keywords=lilypad).
A complete LilyPad kit, consisting of a mainboard, power supply, tri-color LED, light sensor, USB link and mini USB cable, and spool of conductive thread costs $86.65 in single quantities (Fig. 2). Since the products are intended for group projects, significant quantity discounts are available.
MAGLEV PEOPLE MOVERS
In April, NASA celebrated Yuri’s Night, an annual commemoration of the first manned space flight on April 12, 1961. Cosmonaut Yuri Gagarin met an untimely end in a training accident in 1968, but his adventure is remembered around the world. I attended my local event, a kind of science fair for grownups with the flavor of Burning Man, at NASA’s Ames Research Center.
NASA has a number of green-energy programs, and a company involved in one of them, Unimodal Inc., would be presenting that evening. Unimodal’s SkyTran project, a proposed mesh of overhead people movers that would cover entire cities, would be disruptive in its own way if realized (Fig. 3). Underlying SkyTran, however, is a potentially disruptive propulsion technology much closer to being realized in production hardware.
SkyTran will deliver “zero-emission public transit with the convenience of a car but without the need for government subsidies to build and operate the system,” says Unimodal. “On SkyTran you travel the city in a small, computercontrolled, magnetically levitated vehicle. The elevated network of solar-powered guideways provide you with fast, on-demand, point-to-point, non-stop, personal rapid transit.”
Continue on Page 2
Compared to other on-demand transit systems, such as Personal Rapid Transit (PRT) in Morgantown, W.V., SkyTran offers a low-cost and small physical footprint. These qualities encourage city-wide integration, like a network of bus routes, but with on-demand service and personal privacy and without traffic congestion and diesel fumes.
“The challenge of mass transit is getting people out of their cars. The physical size of rail and monorails limits their ability to reach people where they live, work, and play,” says Unimodal. “In contrast to a monorail’s expensive, massive, and visually intrusive support columns and trusses, the SkyTran design is lightweight and agile. SkyTran can be suspended over residential sidewalks, attached to building exteriors, and even routed directly to gates at airport terminals or through the interiors of shopping malls and office buildings.”
So how does it work? It’s a system of overhead maglev rails arranged in a mesh across the city. Suspended from the rails are inexpensive, two-person pods. You buy a ticket to a node on the system, hop aboard an already waiting pod, and you’re shunted to your destination node. It’s somewhat like packet switching, and you’re the packet.
You can even think of it in terms of data being broken up into packets. Suppose you and your spouse are headed for the airport with the usual ton of luggage. You buy a ticket for the luggage and another for you and your spouse. You put the luggage in the first pod and yourselves in a subsequent pod. The system then routes you to the airline terminal, with an extra stop for the luggage at airport security.
Presumably, then, with a little work, packetswitching protocols will scale upward to human size. Okay, a “little” bit of coding is going to be involved, but it really is primarily a scaling problem. Also, the RFID tags and readers that replace the packet headers already are mature technologies. But what about propulsion? If SkyTran is going to work, it’s going to need a disruptive propulsion technology. If it’s just something with moving cables and clamps, it’s a glorified ski lift.
SkyTran’s proposed levitation technology is something new, based on programs from Lawrence Livermore National Laboratory (LLNL). The researcher most involved with those programs, Richard F. Post, has been working on magnetic levitation at the lab for decades. He’s only indirectly involved with SkyTran, but his recent work focuses on the same kinds of problems, albeit directed at more conventional tracked intra-urban vehicles.
“Urban and inter-city maglev systems could represent a practical and energy-efficient solution to pressing transit needs,” says Post. “To maximize the energy efficiency of maglev systems, both the levitation means and the propulsion system must be optimized. A candidate system is the Inductrack, employing permanent magnets on the moving vehicle to achieve levitation.”
The Inductrack is a maglev technology initially developed at LLNL for high-speed trains. One version under way at LLNL has the potential to be a disruptive technology. Another generic urban version is already being developed for a public transit system in Pennsylvania, with a preliminary design currently running (Fig. 4).
“There are many reasons why magnetically levitated trains could be preferred over conventional transit systems. \[For\] Inter-city transportation, much higher speeds are possible than with steel-wheeled trains,” says Post. Other advantages include “lower noise, greater passenger comfort, increased safety against mechanical failures, reduced maintenance, greater rider comfort, \[the ability to\] climb steeper grades, and higher energy efficiency than conventional urban transit systems.”
For long runs, two different types of maglev trains have already been built and demonstrated at full scale at speeds up to 500 km/h. Electromagnetic-suspension (EMS), magneticattraction systems employ servo-controlled electromagnets on the train car, which is attracted upward to an iron-plate rail. In electrodynamicsuspension (EDS), magnetic-repulsion systems, cryogenically cooled superconducting magnets are installed on the moving car. Currents induced in coils embedded in tracks on each side of the car repel these supermagnets.
The best known EMS system, the German Trans-Rapid TR08 demonstration train, has been run on 30 km of test track with operating speeds up to 450 km/h. On the other side of the globe, the Japanese Yamanashi demonstration train is an EDS system that has achieved 500 km/h on an 18-km test track.
There’s one key drawback to these systems, which employ linear synchronous motors (LSMs) for propulsion, in small-scale urban implementations. The LSMs use inverters to drive high ac currents through three-phase windings embedded in the track. To keep I2R energy losses (resistance to current flow in the conductor) low, the “block length” of these windings must be limited, using “block switches,” adding to the system’s cost and complexity.
Continue on Page 3
Another LLNL program is tackling these hurdles. In connection with a NASA-sponsored study of the rocket-launching capabilities of maglev, researcher Ed Cook developed a modular, pulsed LSM drive that promises much higher efficiency than that possible with conventional LSM drive systems. Cook’s modular, pulsed LSM-drive system uses cost-effective condensers and solid-state components. Controlling the phasing of the pulses allows either acceleration or deceleration with regenerative energy recovery.
Yet as Post observes, if you’re not going 500 km/h, design gets simpler. Like the Japanese system, the LLNL technology uses EDS, but it does so without the cryogenically cooled superconducting coils and the control circuits that maintain stable levitation. In fact, Inductrack simply uses arrays of permanent magnets, rather than coils.
Just as a linear induction motor “unrolls” a more conventional rotary induction motor, Inductrack unrolls a permanent magnet (PM) motor. Rotary PM motors have been a hot topic lately, because they’re no longer substantially more costly than induction motors, due to huge increases in the cost of copper wire.
Unimodal proposes to use Inductrack in its SkyTran systems. There would be two components: an array of permanent magnets mounted on the vehicle (the unrolled rotor) and the coil-wound track (the unrolled stator). The permanent magnets are arranged in configurations called Halbach arrays, named after Klaus Halbach, a retired Lawrence Berkeley National Laboratory physicist who came up with the idea in connection with problems concerning focusing particle beams. (The basic effect was discovered by another researcher in 1973.)
Specifically, a Halbach array is a stack of permanent magnets with alternating poles (Fig. 5). The magnetic field is enhanced on the bottom and cancelled on the top. In theory, each square meter of magnet array should provide up to 50 metric tons of lift. That’s more than the weight of the magnets—by about 50 times.
Most of us encounter one or more Halbach arrays every day in the form of the familiar flexible, flat refrigerator magnet. For fun, take a couple of refrigerator magnets and move their “working” sides against each other. You will feel them alternately attract and repel. Meanwhile, it’s possible to get an impression of how Inductrack would work in the SkyTran application by looking at what LLNL achieved with its tracked-vehicle research.
There are two versions of Inductrack. The first focuses on high-speed inter-urban transport. For more sedate intra-urban speeds, an Inductrack II configuration reduces electromagnetic drag forces at those speeds. Inductrack II employs dual Halbach arrays and a cantilevered track, with one array above and one below the track. The horizontal component of the magnetic fields from the two arrays adds, while the vertical fields cancel.
By adjusting the thickness or the width of the magnets of the lower array, relative to the upper array, an optimum level of induced levitating current can be achieved for a given levitated weight and magnet weight. The magnet arrays can work with a litz-cable “ladder track” or slotted, laminated sheet conductors with fiber composite reinforcement.
The LLNL experimental setup was more of a levitation demonstration than a propulsion demonstration. Since it involved railcars above a track, it was a sort of vertical mirror image of the SkyTran proposal.
The team mounted single neodymium-ironboron magnet arrays on the bottom of the car so their magnetic fields would induce currents in the track coils below the car, lifting it by several centimeters and centering it. The track contained a close-packed array of shorted coils. Auxiliary wheels support the train car at rest. Once a low-energy auxiliary power source pushes the car beyond a minimum speed, the arrays induce sufficient currents in the track’s inductive coils to levitate the train (Fig. 6).
During this year’s Sun Labs open house at Sun Microsystems, Sun Distinguished Engineer Hans Eberle presented a mechanical sample of a small-scale, four-port switch—a potential forerunner of a new switch fabric based on Sun’s Proximity Communication technology. This packaging concept provides wireless chip-to-chip communication.
On each die, layer-one metallization includes metal pads that form the plates of capacitors, with one plate on one die and the other plate on a second die, allowing the construction of stacked-chip modules without bond wires. Many parallel signals are coupled capacitively across a very short thickness of dielectric (Fig. 7). Simple in concept, the tricky part is the driver circuits.
According to Sun, “the chip must contain logic for driving and amplifying the signals, and the receiver circuit must tolerate far more variation than a wired connection. The voltages can vary widely, so Proximity Communication technology is engineered to work over about a factor of 10 voltage variation.” Because mechanical misalignments are inevitable, Sun has methods for compensating, as well as for dealing dynamically with effects such as vibration and unequal thermal expansion.
The disruptive factor lies in the combination of potential breakthroughs in density, cost, speed, latency, and power demand. Sun estimates that by shrinking the interconnect for the communication path, the power and the cost per bit transmitted all could decrease. Also, it will be possible to get tens of terabytes per second between packaged very large-scale integration (VLSI) chips.
The company has been pursuing Proximity Communication theory for several years, but Eberle’s model is the first concrete realization of the technology. While the dream is 10 terabits/s and thousands of ports, the results at Sun Labs’ open house were a little more modest, due partly to the use of a mature process technology for the proof of concept.
Eberle thinks the technology might someday radically change datacenter operation. He says people are just beginning to appreciate that data movement is probably more critical than how the data is computed in the first place.
“Virtualization is basically a layer of abstraction that lets \[datacenter operators\] think in terms of pooling resources—compute, memory, storage—and allocating them wherever they’re needed most at the moment. It doesn’t matter where the resources are physically located. They’re all part of the pool,” says Eberle. “But if you add a layer of abstraction, you need some additional bandwidth to make that possible. If your memory suddenly resides on another node, you have to transfer the data,”
Bigger, faster switches can play a role. “Today, the biggest \[single-stage\] switch has about 24 ports. We can easily do a 256-port switch and go up to 1000 ports,” he says. “We basically can look at this as a flat switch.”
Today’s large-scale switching systems are limited by the I/O bandwidth of the individual chips, forcing the hierarchical designs that are so common. Yet the high inter-chip bandwidth offered by Proximity Communication could radically simplify the creation of these large-scale switches. It will be possible to realize a large crossbar architecture in a single package that contains multiple Proximity Communication chips, all communicating at high speed.
Moving data faster also enables more qualityof- service guarantees. Streaming video imposes real-time constraints. Eberle says it’s much easier to provide that kind of service level in a switch such as he envisions, rather than in a multi-stage switch. And, there’s a manufacturing benefit to the contactless interconnect. Unlike wire-bonded multichip modules, it’s easy to rework a Proximity Communication module.