Embedded Puck
by Lindsey Vereen
You can hardly see a hockey puck on
television. A puck is minuscule,
travels incredibly fast, and is usually obscured by a group of burly
hockey players. On Saturday evening, January 20, at the NHL All-Star
Game in Boston, Fox Television found a way to make hockey pucks
visible to home viewers. Although the pucks appeared perfectly
normal to spectators at the game, on TV screens they were
surrounded by a blue halo (think of the blue circles that hide the
faces of people testifying at high-profile, televised trials). At
speeds above
75MPH, the blue halo turned red, and a trail extended
behind it, like a comet's tail.
To accomplish this feat, the hockey pucks used in that game were
each cut in half, rigged with a battery and a circuit board containing
20 infrared emitters, and then epoxied together. The pucks were
activated by an electronic signal and had a battery life of about 10
minutes. An array of sensors surrounded the rink to track the puck
and relayed information back to the camera following the puck. The
information
was then routed to the "Puck Truck," where the signal
was massaged and then transmitted live to viewers. The delay time
between the event on the ice and the transmission of the signal was
less than 350ms.
Speaking of bizarre embedding practices, a number of semiconductor
companies are betting their farms on the future of the x86
architecture as an embedded system foundation. Several companies
have introduced products and established product road maps. The
most dramatic recent example is AMD's
announcement that the
company will cease further development on the 29000 family and
will instead devote its resources to the x86 architecture. SGS-
Thomson and National Semiconductor have made x86 product
announcements. Intel is focusing its 486 efforts in the embedded
market. As one wag put it, everyone hates the x86 architecture, but
everyone understands it. The products being introduced are based on
cores ranging from the 186 to the 486. For those of you who assumed
that 486 cores are enormous
and generate enough heat to toast your
croissant, fear not: design shrinks to 0.5 and 0.35 micron design
rules coupled with lower power operation are making it possible to
embed a 486 core on a die and still leave two-thirds of the die space
available for peripheral functions.
The driving force for embedded developers is always cost, which can
be evaluated in a number of ways. Time to market is one way. The
time at which you enter the market will affect the total number of
units you can sell.
If you lose a third of your sales by being late to
market, then it may not matter if you've saved some money on your
processor. If you build a product based on a widely understood chip,
with abudant development tools available, you can accelerate your
market entry at the expense of a higher parts cost. The x86
architecture's appeal is its plethora of software development tools
and the legions of engineers educated in the architecture.
Not only is embedded technology extending to niches you'd
never
expect, the likelihood that it will be based on x86 architecture is
growing. Now, how long do you think it will take to embed a Pentium
Pro-based system into a golf ball?