This year promises little growth in the electronics business. Companies hope that by carefully choosing the right evolving technologies, they will gain an edge as those technologies mature. Of course, this leads them to particular strategies in new product development to meet the needs of certain vertical market segments.
For example, Maxim Integrated offers its MAX36025 as a key element for securing elements of the Smart Grid, ranging from backhaul communications from smart-meter concentrators to access to substation controls. Using multiple layers of technology originally developed for secure communications between banks and point-of-sale terminals and ATMs, the MAX3265 employs a tamper-reactive dual AES cryptographic engine architecture for multiple cipher channels across compartmentalized system nodes and storage elements (Fig. 1).
Linear Stays Ahead Of The Pack
Linear Technology’s experience in LED-driver ICs reflects this trend in developing products that allow differentiation. There are high volumes in incandescent bulb replacement, but poor margins. Dealing with the automotive industry is tougher, but the volumes are just as high and sockets are stickier—that is, product life cycles are longer.
The latest LED segment with high barriers to entry and rich rewards is high-bay lighting, the kind of LED fixtures that illuminate warehouses and factories. There is a strong motivation to switch to LEDs, because of the downtime and service costs associated with sending qualified electricians out in cherry-pickers to replace or service burned-out conventional gas-discharge bulbs.
“We’ve had some looks at the guys like Phillips, who are making LED replacement bulbs, and decided that’s not where we can get good business. But things like industrial high-bay lighting, factory floor lighting, stadiums, stuff like that, where the cost of maintenance is really high and products need to last forever, were attractive,” said Linear Technology vice president Steve Pietkiewicz.
The product differentiator Linear chose was in the architecture it selected for the drivers. Pietkiewicz said that customers don’t like opto-isolators, and they do like a small external parts count. The Linear design uses a planar transformer for isolation and a single-stage flyback design that has built-in power factor correction.
The large LEDs used in office projectors and theaters represent an even more specialized market. “We’ve got products that drive those 20-, 40-, 50-A LEDs. They’re made by a company called Luminus, using a technology called photonic lattices,” Pietkiewicz said (see “Backlight LEDs Promise A Bright Future At SID Conference”).
For Linear’s automotive customers, one hot product area is buck regulators for the vehicle’s many engine control units (ECUs), with a special characteristic—they must turn in 30 ns. The car makers, then, can strip out a voltage-distribution stage. “You can go straight from car battery to engine-control-unit core processor, delivering 1 V at 3 A. That replaces the old two-step approach (12 V down to 5, then 5 V down to 1),” Pietkiewicz said.
Also, automotive dc-dc converters can get complicated. “The car companies like to have the switcher run above 1.6 MHz, so it doesn’t interfere with the AM radio band. But if you’re running at those kinds of frequencies, let’s say you’re running at 2 MHz and you want to step from 12 V to 1 V, that means you’ve got to have a really short ‘ON’ time in the switcher, which means you’ve got to have a really high-speed switcher,” said Linear CTO Robert Dobkin.
Linear offers switchers that will do that on 2 μA of quiescent current, with efficiencies in the 90s. That’s significant, because it means that none of the dc-dc converters ever has to be switched off, except by the ignition key.
How many ECUs does that represent? At Electronica in November, Dobkin met a German auto engineer who told him that there were 100 ECUs, 400 LEDs, and 122 electric motors in one of his vehicles, and they all need silicon to drive them. The fewer times they must be turned on and off, the better, as long as their quiescent requirements are small enough.
Another curious product driver in the automotive market is a response to exactly the opposite problem: the need for line-drop compensation in the dc-dc regulators used to support USB power points. Frequently, the active device is some distance away from the device being powered, and at the same time, that device is drawing significant current.
In 2013, Linear will be putting line-drop compensators in some of its regulators. The first part will be an external device, not a virtual remote sense. It will measure the current and jack the regulator output voltage up, assuming the car maker knows how much the IR drop is going to be at the far end of the internal wiring.
Not everything new in product drivers for small power ICs is automobile-related, though. Linear has been considering thermal energy harvesting as it relates to its recent acquisition of Dust Networks, the “mesh-to-edge wireless” sensor-network company whose technology is distinguished, among other things, by energy efficiency based on its use of deterministic time slots for transmissions.
An upcoming sensor device from Linear will draw less power than its previous generation, even though it includes an ARM core and a lot of memory. It will be built into a module that runs on a lithium-thionyl chloride battery. The energy harvesting becomes valuable when it’s necessary to extend battery life or to power the sensor.
There will be more details upon the product’s announcement. In the Dust Network scenario, the most common source of energy for harvesting is likely to be thermal, not based on a continuous delta-T but on daily cycling of ambient temperatures.
In electric cars, Linear has had battery charge-balancing chips through four generations. Dobkin is bullish on their potential for being a breakout technology in 2013, and not just in electric vehicles.
“I’m seeing increase in traction in terms of energy storage for averaging peak power demands. Linear will soon have several solutions for balancing very large stacks of batteries that will, depending on what your game is, increase the capacity of the battery, the longevity of the battery, or the cycle life of the battery.”
This means balancing currents of 2 A to 10 A or more, in big batteries, in solar farms. “Even today in Asia, you’ll find seas of solar farms that need to have large-scale storage systems attached. People want to use wind and solar and other power. They want to harvest that energy when it’s available, and they need to store that energy somehow and use it when it’s dark,” Dobkin said.
Even when it comes to vehicles, it’s wrong to think exclusively in terms of private automobiles. “We’ve now started to talk about the ‘transportation industry’ as opposed to ‘automotive,’ because then you end up with a market that consists of heavy vehicles, and farm equipment, and forklifts, and trains,” he said.
The upcoming generation will feature some interesting innovations, such as wireless coupling (through planar coils top and bottom) for communications up and down the battery stack, eliminating noise in the communications link.
There also will be greater precision in measuring battery voltage than ever—a necessity because the extreme flatness of the discharge curves for the latest battery chemistries requires that to determine state-of-charge. So Linear will use its subsurface Zener voltage reference technology, and every time the battery gets a full charge, the system will recalibrate.
Intersil Seeks New Sockets
Intersil CEO Dave Bell is looking for ways to beat what looks like a period of flat growth, which means picking the right market segments to support. For example, total semiconductor content in vehicles continues to grow, particularly in electric cars and hybrids. Semiconductors are now around $250 per car and should increase, in Bell’s opinion, to around $400 per car on average in several years.
The future doesn’t look as good for PCs, though, especially desktops and traditional notebooks. But smart phones, tablets, and ultrabooks (a “tablet with a keyboard,” in Bell’s words) are a different story. Unlike traditional laptops, tablets and ultrabooks require more than 10 hours of batery life, instant-on performance, and a touchscreen, though Bell said that the kinds of ultrabooks and portables he’s considering haven’t hit the market yet.
“Today, there really aren’t any true ultrabook-like products out there. If you tear apart a MacBook Air and compare it to maybe a clunkier, full-sized notebook computer, there is really not a whole lot of difference from an architectural standpoint,” he said. True ultrabooks, he added, will be shipping early in 2013. Companies that concentrrate on analog and power ICs can focus on integration to crack this market.
“Today, if you look at a conventional notebook computer, the power system is quite fragmented, with dc-to-dc controllers for core power, additional power controllers for system power, and an external battery charger. As the market starts going to true ultrabooks, that functionality will become more integrated. In particular, as new architectures allow power levels to shrink, it will be possible to integrate on-chip power MOSFETs,” he said.
Apple’s latest iPad represents the current level of integration in power systems. A single big power-management IC (PMIC) handles almost all of the system-power functions. If there is a cellular interface, another PMIC drives the base and cellular functions.
“Some pretty high levels of integration are already in smart phones and tablets, and the bottom line is, as true ultrabooks emerge, you’re probably going to see similar levels of integration in those products as well,” Bell said.
Power levels are coming down in ultrabook processors, which may reduce current-demand rise and fall times. However, some quad-core ARM processors can sink currents of 20 A or more.
“If you try to handle that with a very small handheld device, you’re going to have to start using architectures that involve maybe two- and three-phase power converters, but with much smaller inductors and capacitors than you find with a full-sized notebook,” Bell said.
Product design for such radically different market segments as automobiles and personal communications and computing can be challenging. To succeed, though, companies need to respond to “the notorious impatience of Wall Street,” Bell said.
“When you’re talking about things like smart phones, they get all excited because they know that you can get very rapid revenue results from an investment. They like to see revenues growing very rapidly. But if you start talking about automotive ICs, where you can spend maybe four or five years developing a product before you’re going to start seeing early production quantities, Wall Street just doesn’t have the patience for that,” he said.
“At the same time, you do need to plan on product longevity. In many cases, the automotive companies want assurance that you will be able to provide this same product for 15 years,” he said.
Power IP For The Fabless At S3
Engineering service companies such as S3 Group of Dublin, Ireland, provide custom core intellectual property (IP) for ASICs that approach the performance of discrete ICs from the independent device manufacturers (IDMs). For some time, S3 has offered core IP for both low-dropout regulators (LDOs) and digital dc-dc converters in simpler chips. In certain areas, though, there’s a trend toward integrating more power-management functionality directly on systems-on-chip (SoCs).
“There are so many power domains now on complex SoCs. Having one or more on-chip dc-dc converters with a fixed set of LDOs would not be a huge challenge,” said Dermot Barry, vice president of consumer silicon at S3.
Companies are now taking functionality that used to exist in an external power-management chip and integrating it.
“From the system customer’s point of view, if they can remove those power-management chips, it helps, not only from a cost perspective, but from a size and volume perspective as well,” Barry said.
Specifically, the advanced work that S3 is involved in focuses on integrated dc-dc converters, using TSMC’s 28-nm process. That’s a follow-on to a dc-dc converter on an external power-management chip that uses the more mature 0.18-μm CMOS technology.
Still Room For Innovation
From all the talk from senior executives about high levels of integration targeting the needs of vertical markets, there is still room for innovation in traditional semiconductor devices, particularly in reducing physical size and external parts count.
In November, Bell Labs power spinoff Enperion announced its EL700 PowerSoC family of low-power point-of-load dc-dc converters based on electroplated wafer-level magnetics. The chips in the 42-mm2 EL700 family integrate MOSFET switches, controller circuits, compensation, and a tiny silicon inductor (Fig. 2). The inductor enables the devices to switch at up to 18 MHz for low ripple and fast transient response. The 1- and 1.5-A PoLs have an input voltage range of 2.65 to 5.5 V.
The company’s wafer level magnetics (WLM) have the potential to transform magnetic components from three dimensions to two in a thin-film form that can be deposited with standard wafer processes on top of CMOS wafers. The WLM technology is fully qualified for full-scale mass production in a high-volume foundry.
Although the EL700 buck converters aren’t as highly integrated as some other products, the underlying IP begs for application in a broad range of more narrowly focused applications. For example, the WLM technology can be transferred to other micro-magnetic applications such as micro-transformers for signal isolation, micro-electromagnets for life sciences, integrated magnetic sensors for navigation, and PMICs for portable consumer products.