Ratcheting down the power is achievable, but making sure the device still works remains a challenge.

Energy is everywhere for the taking, but being able to harness it has been only part of the problem. Transforming that energy into usable form so that it can be stored in a battery or funneled into a large distribution grid has proved to be an even bigger problem, and until now that has been the weakest link in the chain production chain.

But there’s a new challenge on the horizon, and it’s one that falls squarely in the hands of the engineers who design and build electronic components. It’s making sure the device works as planned each and every single time, without interruption.

What’s driving this shift is a fundamental change in how companies and universities are approaching the low-power problem. Rather than just making things more efficient, they are now experimenting with the idea that some devices don’t need much power at all—ever. The catch is that they are bound by duty cycles, and those devices had better fire up the first time.

Consider a lawnmower, for example. It’s okay if you pull the starting cord once and it doesn’t catch. That’s one duty cycle of the starter device. But what happens when a pacemaker misses the duty cycle and it only has enough power for one charge? And if it a device has been left in a harsh environment for years without testing, such as in the woods or inside the concrete on a bridge or road, it very well might not start up.

The trouble is that shaving energy consumption down to the absolute minimum can cause its own problems. Redundancy is one way to solve that problem, but that also increases the need for more stored energy, which in turns means either more size or thickness in the storage device. In the case of thin-film storage, that can mean significantly more area for storing energy.

–Ed Sperling

Energy is everywhere for the taking, but being able to harness it has been only part of the problem. Transforming that energy into usable form so that it can be stored in a battery or funneled into a large distribution grid has proved to be an even bigger problem, and until now that has been the weakest link in the chain production chain.

But there’s a new challenge on the horizon, and it’s one that falls squarely in the hands of the engineers who design and build electronic components. It’s making sure the device works as planned each and every single time, without interruption.

What’s driving this shift is a fundamental change in how companies and universities are approaching the low-power problem. Rather than just making things more efficient, they are now experimenting with the idea that some devices don’t need much power at all—ever. The catch is that they are bound by duty cycles, and those devices had better fire up the first time.

Consider a lawnmower, for example. It’s okay if you pull the starting cord once and it doesn’t catch. That’s one duty cycle of the starter device. But what happens when a pacemaker misses the duty cycle and it only has enough power for one charge? And if it a device has been left in a harsh environment for years without testing, such as in the woods or inside the concrete on a bridge or road, it very well might not start up.

The trouble is that shaving energy consumption down to the absolute minimum can cause its own problems. Redundancy is one way to solve that problem, but that also increases the need for more stored energy, which in turns means either more size or thickness in the storage device. In the case of thin-film storage, that can mean significantly more area for storing energy.

–Ed Sperling

For years, vendors have tried to push low-power design to consumers. They have had limited success. Even the U.S. government has gotten involved with its EnergyStar ratings, rebates on solar panels and, at least in California, special dispensation to use commuter lanes on roads or electric vehicle parking spaces at airports.

None of that was particularly effective over a period of two decades. The first round of electric cars was a dismal failure, and demand for low-power versions of chips often trailed faster versions of the chips by one to two generations.

All of that is changed, in large part because consumers are now demanding low-power designs. This is reflected in sales across a wide swath of markets. In netbooks, for example, devices with a battery life of nine hours are flying off the shelf while those with a battery life of three hours don’t move at all.

In automobiles, most of the cars still selling are highly efficient hybrids, although in Silicon Valley the car for top executives who can afford it is the Tesla. Sales are now so robust, in fact, that California no longer allows single-driver vehicles that reach a minimum efficiency threshold to use the carpool lanes.

And in enterprise IT, the selling point is low power consumption rather than performance—IT managers still demand some performance increase, but the bigger concern is the cost of power both in running the machines and in cooling them.

This is being reflected across the entire supply chain. At the submicron level, one of the primary drivers of all chips is the power budget. It is now part of the architecture rather than an afterthought. And in the IP world, low power is a prerequisite to a sale.

What’s interesting is that we’ve just scratched the surface on reducing power. Energy scavenging, better software, more efficient use of cores in a multi- or many-core chip and better chip design with a sharp reduction in margins are all the subject of massive research projects by chip makers, universities and even governments.

For engineers and chip architects, this may require a shift in thinking from an emphasis on performance to an emphasis on power. But there are clear signals from larger markets such as automobiles (witness what’s happening at General Motors vs. Toyota) that this is the way of the future—and that there’s no going back.

–Ed Sperling

There are two major concerns in chip design these days. First, is getting a product out the door. The second, is making sure it fits into the power envelope so that a battery lasts long enough to be a competitive marketing advantage.

For the most part, consumer technology is roughly equivalent these days, regardless of brand. In fact, the real differentiator in many cases may be the brand name itself. You might buy an Apple iPhone because of the brand or the number of available applications, but when it comes to actually using the features in the phone there probably isn’t much difference between the iPhone and the Palm Pre or the Blackberry, for that matter. The buyer gets an extra feature here or there, and it may or may not work as well as they had hoped.

But given the demands those applications put on battery life, consumers increasingly are willing to sacrifice brand and full function for time between recharges. A netbook with 9 hours of battery life may not be able to play games as well as a supercharged quad-core notebook, but that’s not the big consideration for mobile users.

Even in the enterprise, where racks upon racks of servers consume millions of dollars worth of energy each year, power is becoming the biggest determining factor in purchases. It’s not just the amount of power used to run the machines. It’s also the cost of cooling them to avoid data corruption or a complete server room meltdown.

Left to their own wits, it’s unlikely any of these design considerations would have crossed the minds of systems builders, because power and performance have been the only drivers of computing of all types for the past six decades. That is changing rapidly, and it is becoming a design consideration at every level, from the initial architecture to what IP is used and how many sleep states are present. But what’s really changed is that these requirements are now being pulled by the buyers rather than pushed by the sellers.

That’s a fundamental shift in electronics, and it’s likely to demand changes in areas we haven’t even thought about. Arguably for the first time, and certainly for the first time at this scale, the power is really in the hands of the people who buy the devices.

–Ed Sperling