PCIM Europe 2024: A Conversation with Balu Balakrishnan, CEO of Power Integrations

I'm doing very well, thank you. And thank you for inviting me.

So there are two products we are introducing. One was actually introduced in APAC, but it's still a brand new product, which is called InnoSwitch 2. This is the first product that can generate multiple outputs with a single magnetic element, which is a transformer. So you can eliminate DC-DC converters by using InnoSwitch 2. Most embedded power supplies have multiple outputs: they may have a 12V output, 5V output, 3.3V, sometimes they have a 48V output for backlighting and so on. The way it's been done in the past is you have one power supply followed by DC-DC converters. That means your efficiency goes down dramatically because instead of having 90% efficiency in the power supply, you have 90% times another 90%, you get 80% efficiency. By eliminating the DC-DC converters, you not only eliminate multiple inductors and the DC-DC converter stage, but you increase the efficiency by 10% because you don't have a second stage. So that's one product. The second one is we just announced today, called BridgeSwitch 2. This is a driver for motors that has not only the drive transistors but also a lot of features to protect the motor, to protect the driver, and also to improve the efficiency of BLDC motors. The BridgeSwitch 2 can go up to one horsepower without a heat sink. There's a demo you can go and see. We are able to get such high efficiencies that you can eliminate the heat sink. In addition to that, it has many other features: it can monitor the input voltage, it can monitor the temperature, it can monitor the vibration in a motor, so you can even determine whether the motor is about to fail. It has extremely good reverse diodes which are very soft but very fast. Best of all, it has extremely low no-load consumption, 10 milliwatts of no-load consumption. I don't think there is any other solution that even comes close to that. It's also self-powered from the high voltage; you don't need a separate power supply. So there are many more features, but those are the two major products that we recently introduced.

First of all, GaN brings a lot of advantages, but it's also very difficult to use because it's so fast. If you just have a discrete device, somebody has to figure out how to drive it, how to protect it, and you can't have too much parasitics between the driver and the transistor. Even the transistor itself can't have too much parasitics if you're in a half-bridge solution. So what I tell everybody, and which is our motto in the company, is we don't provide just a device; we provide a complete solution. In fact, we do such a good job of it, we were shipping GaN for a long time before people found out that we were shipping GaN. In fact, we were the first ones to ship GaN in high volume, and it was accidentally found out by a research firm, and they were somewhat embarrassed because they are following every GaN company in the world, but we are the only ones who are shipping and they didn't even know we had GaN. So that's exactly what we want to do: we want the customers to be able to use our product without having to pay any special attention, because we have solved all the problems for them, whether it's protecting the device, controlling the dV/dt of the device, driving it appropriately, or optimizing it for efficiency, and so on. We take care of it, so they can use the device as if it's a silicon device or any other device, but all of the system issues are resolved. That's something that we really push for. In fact, when I talk to my R&D guys, I tell them it's not sufficient to integrate; you have to innovate on top of it. You have to bring something to the market that is so revolutionary that it makes a huge difference to our customers. I think most people, including our competitors, would agree that we bring products that are very, very innovative. For example, we have FluxLink, which is an isolated communication link that gives a huge advantage in terms of efficiency because we can precisely control both primary side and secondary side because of the speed of FluxLink. FluxLink is also very rugged. For example, in automotive, they don't use optocouplers because of reliability issues, but FluxLink is so far superior to optocouplers, it's a perfect solution for isolation.

Great question. So, you know, we chose our approach for a very good reason, two reasons. One is reliability. I truly believe the approach we are taking is far more reliable than enhancement mode, especially when you go to data centers and automotive. Reliability becomes very important. As a result, most customers I've talked to are very concerned about using GaN; they feel that it's not as reliable as silicon or silicon carbide, and it's especially true in automotive. What we would bring is a device that's as reliable as silicon or silicon carbide but provides the benefits of GaN. The benefits of GaN are, of course, better efficiency, but also that it is cost effective, more cost effective than silicon carbide. In fact, within the next few years, it will be more cost effective than silicon. And we chose for those reasons our approach. To answer your second question, there are two challenges for GaN to go to very high power levels. Our current technology can go up to about 10-15 kilowatts, but if you really want to go into a drivetrain, then you're talking about 100 to 500 kilowatts. To do that, you need both voltage and current. The voltage we have already solved; our structure will allow us to go to very high voltages. For example, we are the only ones offering 1250 volt GaN commercially in the marketplace. We announced that last year. We also have 900 volt GaN, 750 volt GaN. Most other companies struggle to meet 650 volts, whereas we have 1250 already in the marketplace. So that's the voltage issue; I think we have solved that one. The next one is current. You need to get high enough current to be able to operate at 100 to 500 kilowatts. That's a solvable problem, but it still takes a lot of R&D work. We've been working at it for some time. We believe within the next 3 to 5 years, we'll be able to have a technology that can go up to very high power levels. Then it'll compete very favorably with silicon carbide because GaN is inherently less expensive than silicon carbide.

One way to look at it at a high level is that in the next 20 years, there'll be a lot of investment going into carbon net zero initiatives. 70% of carbon reduction is going to come from two things: energy efficiency, which is about 37%, and the remaining is renewables. We play in all of those things. For example, we play in renewables in terms of offering IGBT drivers and silicon carbide drivers for solar and wind applications. We also provide products for high voltage DC transmission systems, these are plus or minus 800,000 to plus or minus a million volt DC transmission systems that transmit power from either a solar farm or a wind farm to the grid. That has to be done efficiently; DC is 40% more efficient than AC transmission, and we facilitate that. The last one, which is the largest one, is energy efficiency. Efficiency of all the products that consume electricity, that's our major market. There we have been pioneers; since the '90s we have offered the most efficient products and lowest standby consumption products. So in the next 20 years, there'll be something like $1 trillion spent on carbon reduction. All I care is get a little bit of that; that will make our company a very large company. So our entire focus is on how do we make our products so attractive that we get our unfair share of this market, this carbon reduction market. And it'll affect all of us; we have to bring the carbon emissions down, otherwise this world will be very difficult to live in.

Thank you, thank you.