“Hope springs eternal,” Alexander Pope said. That’s true in power electronics, where a golden age of high-efficiency switching supplies and plentiful power delivered by a robust grid awaits. That’s where companies big and small are placing their bets. But these are still early days.

600-V GaN GETS STUCK

My surefire prediction for power products in 2012 is that somebody will start shipping 600-V galllium-nitride (GaN) FETs. But who will it be?

Designers who were hoping to see higher-voltage GaN devices were disappointed in 2011. Prototypes boldly displayed at the Advanced Power Electronics Technology (APEC) conference in March in Fort Worth, Texas, that were expected to be in production by year’s end have apparently slipped.

International Rectifier, which has been selling driver/bridge modules of its lower-voltage depletion-mode devices, exhibited 600-V devices at APEC and said they’d be sampling by year’s end. If that happened, it didn’t make the news.

Supported by the Office of Naval Research and the Defense Advanced Research Projects Agency (DARPA), GaN newcomer Transphorm boldly burst out of stealth mode to show a GaN-on-SiC (silicon carbide) fast diode in Fort Worth. In May at PCIM Europe, the company demonstrated its 600-V EZ-GaN transistor. Commercial products aren’t yet available, though, and no pricing or availability information has been released.

Efficient Power Conversion (EPC), which has had a commercial presence with a range of enhancement-mode FETs with voltage ratings of 200 V and below, has moved into its second generation with improved current handling and gate charge, but has not yet broken the 200-V barrier.

GaN power devices promise significant improvements in switching-supply and Class-D audio efficiency, due to lower conduction and switching losses. They’re still novel enough, and the differences between the processes that allow them to be made rate a company by company explanation of the technology.

International Rectifier

International Rectifier announced its GaNpowIR GaN-on-silicon epitaxial process and device fabrication technology at the end of 2008. It was developed to overcome the requirement for SiC or sapphire substrates, which had been holding GaN back until that point.

Silicon substrates were attractive because the use of silicon wafers would lower substrate costs by a factor of 100. Unfortunately, they had been found unusable because epitaxial depositions inevitably contained lattice defects and deformations because of the mismatch in lattice constants and thermal expansion characteristics between the substrate and epi.

The company’s GaNpowIR heteroepitaxial process technology evolved from industry-wide research on the manufacture of GaN RF devices and LEDs. It led to a breakthrough that enables the use of silicon substrates.

IR’s power FETs are high-electron-mobility transistor (HEMT) JFET devices in which a bulk layer of undoped GaN is deposited on the substrate and the drain, source, and gate “float” on a two-dimensional electron gas (2DEG) that forms spontaneously when a thin layer of aluminum gallium nitride (AlGaN) is deposited on top of the pure GaN.

The FET gate controls conduction through the channel by applying voltages to titanium-aluminide (TiAl) drain, source, and gate ohmic contacts above the electron gas. When the voltages are applied, an insulated or rectifying metal gate structure forms between the ohmic contacts. The resulting fields modulate the electron gas.

Normally, this would result in an enhancement-mode JFET, in which the channel would conduct in the absence of gate drive. But IR has some secret sauce that makes its devices only conduct in the presence of a gate voltage.

IR’s first GaN products, the iP2010 and iP2011 modules for point-of-load (POL) applications, contained both a driver IC and the monolithic GaN-based power device. The iP2010’s maximum switching frequency is 3 MHz, its input voltage range is 7 to 13.2 V, and its output voltages could be selected at standard values from 0.6 to 5.5 V with output currents to 30 A. It operates up to 3 MHz. The iP2011 could switch at frequencies up to 5 MHz, but its maximum output current is 20 A. At the time of the announcement, pricing in 2500-unit quantities was $9.00 and $6.00.

EPC

Efficient Power Conversion announced its first products in 2008 and offers 16 N-channel enhancement-mode devices with V dc ratings from 40 to 200 V and drain currents from 33 A (40 V) to 3 A (at 200 V). Another 200-V device has a current rating of 12 A.

Maximum RDS(on) with 5-V gate drive ranges from 4 to 100 mΩ for the 3-A, 200-V device. It’s 25 mΩ for the 12-A 200-V device—a generational and cost difference. Gate charge also varies by VDS and generation. Of course, there are no reverse-recovery issues. EPC also offers development boards and a demo circuit, a 19-V to 1.2-V buck converter.

EPC says that its manufacturing process begins by growing a thin layer of aluminum nitride (AlN) on the silicon base wafer to isolate the device structure from the substrate (see the figure). This is followed by a thick layer of highly resistive gallium nitride, which provides the foundation for the actual GaN transistor.

An electron-generating material that creates a GaN layer with an abundance of electrons directly below it is then applied. Further processing forms a depletion region under the gate, so turning the device on is essentially like turning on a conventional n-channel, enhancement-mode power MOSFET.

Current-handling ability is related to the number of devices and thus to die size. VDS is determined by how long you make the channel, which works out better with GaN than with silicon because the resistivity of the channel is so much lower.

An application note on EPC’s Web site, “Fundamentals of Gallium Nitride Power Transistors” by Stephen L. Colino and Robert A. Beach, clarifies the issue of QRR by noting that the last part of the performance picture is that of the so-called “body diode.”

The company’s GaN transistor structure is a purely lateral device, absent of the parasitic bipolar junction common to silicon-based MOSFETs, the note explains. Reverse bias or “diode” operation, then, has a different mechanism but similar function.

“With zero bias gate to source, there is an absence of electrons under the gate region. As the drain voltage is decreased, a positive bias on the gate is created relative to the drift region, injecting electrons under the gate. Once the gate threshold is reached, there will be sufficient electrons under the gate to form a conductive channel,” the note says.

According to the company, no minority carriers are involved in conduction, so there are no reverse recovery losses. And while QRR is zero, output capacitance (COSS) has to be charged and discharged with every switching cycle.

“For devices of similar RDS(on), GaN transistors have significantly lower COSS than silicon MOSFETs. As it takes threshold voltage to turn on the GaN transistor in the reverse direction, the forward voltage of the ‘diode’ is higher than silicon transistors. As with silicon MOSFETs, care should be taken to minimize diode conduction,” the note says.

So which company will be the first to start selling 600-V devices? Infineon and MicroGaN in Europe are leapfrog possibilities. MicroGaN? Here’s the preliminary datasheet: www.microgan.com/includes/products/MGG1T0617T.pdf.

BEYOND THE SMART METER

One cannot discuss the power picture in 2012 without considering the Smart Grid. I am, however, suffering from “Smart-Meter Fatigue,” an ennui brought on by any mention of what some consumers think about the accuracy of meter readings or whether RF from the meters will make their brains fall out.

MAXIM

It was interesting to take part in a morning-long media event at Maxim Integrated Products in early December to see what the company thought the opportunities were in the wider Smart Grid. One telling point was that the six presenters were all VP-level executives—leaders of multiple product lines—from the U.S. and overseas.

Jay Cormier of Teridian, which Maxim acquired last year, discussed issues like communications and security for the most part. On the communications front, Maxim has been in the forefront of the evolution of power-line communications (PLC) from simple mark-space frequency shift keying (FSK) in the sub-500-kHz band to G3 PLC. (For G3 PLC specs, see “G3-PLC Open Standard for Smart Grid Implementation” at www.maxim-ic.com/products/powerline/g3-plc/.)

Security was a hot topic at the meeting, though necessarily lacking in detail. Maxim has long been silently active in security for point-of-sales (PoS) terminals. The technology transfers well from PoS to a range of Smart Grid application areas from metering to substation supervisory control and data acquisition (SCADA) applications.

Even such basic components as analog-to-digital converters (ADCs) have unique requirements when they are used for power-line phasor measurements in the transmission and distribution subsystems of the grid. (See “Advanced Power-Line Monitoring Requires a High-Performance, Simultaneous-Sampling ADC” at www.maxim-ic.com/app-notes/index.mvp/id/4281.)

CISCO

Smart Grid opportunities are even broader at Cisco, where top execs perceive the possibilities of something even bigger than the Internet. Jennifer Lin, who leads the product management and technical marketing team in Cisco’s Connected Energy Network Business Unit, says Cisco started developing technologies and solutions for this space a little over two years ago.

“We recognized that not only is there a huge opportunity in IT enabling the evolving grid evolution, but also because the company was asked to take a leadership role in the non-technical aspect as well,” Lin says.

“We have been very involved with NIST (National Institute of Standards and Technology) on technology standards. Our boss, Laura Ibsen, drove regulatory and government affairs for many years here at Cisco and we felt we could play a leadership role the evolution of some of the regulatory policy evolution, as this whole industry transitions,” she says.

“We’ve been referring to the new emerging business models, what we often refer to as ‘The Grid Exchange,’ with its emphasis on energy traders and their need for real-time information, plus the evolution of higher degrees of automation in the grid, for which there are some interesting business models evolving,” says Lin.

Cisco sees those opportunities coming in the less highly regulated territories where energy is unbundled, as it is in Texas, and in the U.K., where the company is starting to see competitive forces create some interesting new business models.

The company has taken an architectural approach, recognizing that “the business architecture drives the technology architecture.” Cisco has brought in some key players in the industry to help its engineers think about how the grid is evolving and how they could create a highly embedded communications interlay to support some of that evolution.

Cisco has been really looking at some of the more disruptive aspects of the grid: the integration of renewable energy and transitioning legacy data systems to a much more integrated IT-based network. Lin says Cisco has have gone far past basic metering issues. What the company essentially sees is one large convergence opportunity. Working with many other industries, Cisco wants to create a convergence of the proprietary network into a common pervasive and secure end-to-end IT platform.

This is an evolution from the collection of sub-networks today that the Electric Power Research Institute (EPRI) and NIST laid out. What exists now is a field area network, essentially from the premises to either secondary substations or the pole top, two tiers of communication there, and then back into the primary substation, and upstream of that, to your transmission network.

On that upstream side, Cisco has been developing ideas around networking phasor measurement units and how IT technology can help with very low-latency real-time distribution of very granular power quality measurements—with its timing and its management, trying to get real-time visibility into the sub-network.

Cisco started several years ago really playing to its strengths in traditional routing and switching capabilities trying to make those an appropriate for substation automation. The inflection point there was the efforts around IEC61850s, which is the common data model for substation automation, and how IT networks could support that essentially to create a network architecture that would support various substation-automation scenarios.

The company is already working with grid-technology partners, such as GE, ABB, and Siemens, who make the relays and substation gear to make sure Cisco could validate its communications infrastructure against their substation solution.

Some recent work has been with Firstwind, which has roughly 500 MW of wind capacity. In this case, Firstwind has been using connected grid routers and switches in its wind farms to do a variety of things, not just in the core network of energy generation, but in other situations where IP networks can help, like IT video surveillance across vast wind farms.

That work involved physical security and other things like wireless mesh technologies to enable remote field workers to communicate over long distances far from cellular towers. In that instance, Cisco took advantage of a company it acquired called Arch Rock.

Technology-wise, Arch Rock was focused on enabling embedded wireless sensor networks over IPv6. There are many design issues around how to secure and manage all the end points, update them with firmware upgrades, and accomplish real-time monitoring and diagnostics. That, along with correlating the information that comes from all of those end points, presents an interesting challenge, and the bar is very high in terms of network resiliency.

Cisco has also been public about similar work with GE and Siemens Energy on wind and solar generation, but Lin says that control-network security at a deeper level than surveillance cameras is the element that completes the picture. An important element of that, she says, is inserting a security layer without impacting latency.