New Power Transistors Look to Make History

Sep 1, 2008 12:00 PM
By David Morrison, Editor in Chief

Good things come in threes. We've all heard that expression. But in power electronics, you could make the argument that good things come in 30s, as in 30 years. Consider that the first commercially successful power MOSFETs were introduced roughly three decades ago.

Then, though this may be stretching the point, you could look back another 30 years to the late 1940s and note the invention of the first transistors. Though those first germanium devices didn't have much power-handling ability, they did pave the way for silicon bipolar junction transistors (BJTs), which would find a home in power supplies.

Now, jump forward three decades to the present and witness the emergence of a new class of power transistors based on compound semiconductors. Recently, International Rectifier announced a new technology platform based on gallium nitride (GaN).

The power semiconductor company plans to develop discrete devices such as MOSFETs and IGBTs and modular components using the new material. Technology roadmaps will be unveiled this month at both the Digital Power Forum and the Embedded Power Conference, and first prototypes will be shown publicly at Electronica in November.

News of the pending GaN devices comes in the wake of similar recent developments in silicon carbide (SiC) transistors. In the past year, TranSic, a Swedish developer of SiC devices, indicated it was sampling SiC BJTs. Meanwhile, Cree, the Durham, N.C.-based manufacturer, now has prototypes of SiC MOSFETs, which are the subject of this month's cover feature.

Although the vendors have yet to make firm commitments about their product launches, we could see the introduction of MOSFET or IGBT products in SiC or GaN as early as next year.

If the history of the power MOSFET is any indication, we could expect SiC and GaN transistors to have a similar revolutionary impact on the power electronics industry. Consider the experiences related by Power Electronics Technology's 2008 Lifetime Achievement Award winner, Rudy Severns. His story is told on pages 40-44 in this issue.

The arrival of the power MOSFET had a big impact on his work and his career path. In the late 1970s, when Severns suggested a possible move to higher switching frequencies, it was based on designs he had done using BJTs. Although power-supply designers could have raised their switching frequencies using those same transistors, it became much easier to do so with the power MOSFETs that soon would be on the market.

The arrival of power MOSFETs also made it possible for switch-mode power-supply (SMPS) technology to make the leap from military and aerospace applications to commercial power-supply applications in the 1980s.

For Severns, the MOSFET was also responsible (pardon the pun) for another switch. It enabled him to go from working as a power-supply designer for OEMs to working as a field applications engineer for semiconductor manufacturers. As an FAE responsible for teaching customers how to use MOSFETs, Severns would document a long list of power-supply topologies that could be implemented using MOSFETs, which led to his book on topologies.

A few years later, when Severns had enough of the corporate world, the MOSFET also made it easier for him to become a consultant since it had helped to proliferate the switcher, creating great demand for those with experience in SMPS design.

Now, will GaN and SiC transistors have such great and varied impacts? The truth is, we don't know. In the 1980s, the personal computer did a lot to create demand for switchers. In the coming decade, will new applications like renewable energy systems or hybrid vehicles spark a similar demand for better-performing SiC or GaN transistors?

For now, we can't know whether both SiC and GaN power devices will succeed, or when the new transistors might become mainstream. Severns notes that it took a decade before silicon MOSFETs came down enough in price to be widely used.

Although we'd like to believe technology advances more rapidly these days, the history of power electronics suggests otherwise. The answers should be perfectly clear — in 30 years.

Acceptable Use Policy blog comments powered by Disqus