This article was originally published in Power Electronics News by Alix Paultre, Editor-in-Chief, on October 14, 2018.
The revolution in power electronics design and development is going forward full tilt, and the expansion of the Cloud, Industry 4.0, and the migration to the electric vehicle (EV) is changing how people consider how energy is generated, routed, and managed. We were able to sit down with Jim Witham, CEO of GaN systems, and talk to him about the state of the art in power semiconductors, and where he thinks things are going.
PEN: From smart phones and the Internet to our factories and vehicles, power electronics have become more and more a part of our lives. And we’ve gotten to the point where wide-bandgap devices are leading the charge. Everybody’s been screaming it from the rooftops for the last, say, 14 to 18 months, “Wide bandgap power transistors are driving improved solutions, cost-effective solutions, viable solutions”. Would you agree?
Jim: Absolutely. Power electronics are booming with GaN as a cornerstone technology in power-reliant industries from data centers, automotive, renewable energy, and industrial to consumer electronics. And there’s a reason for that. Wide-bandgap GaN transistors are enabling power electronics a 1/4 of the size, a 1/4 of the weight, 1/4 of the power loss as heat, all while making the system cost less expensive. And so, when you’ve got a basic building block like GaN, it can fundamentally change the economics of the power electronics industry so, if you’re a power electronics maker or a power electronics user, you’re going to be paying attention.
PEN: Well, you know, and, Jim, I’m really glad you brought that up, because one way you can look at it, though, is, yes, wide bandgap is better, but why has the industry come to this point? Because, if you think about it, 20 years ago if you were a tethered application, you didn’t care how efficient your system was. It’s just power coming out of the wall. Why this sudden, dramatic, and aggressive focus on efficiency? I mean, nobody cared about efficiency up until recently. Why now?
Jim: I think it’s because power electronics have become such an important part of our lives. Let me provide an example. I think it’s interesting to take a look at three industries that have for the most part operated independently in the past: the automobile, data center, and renewable energy industries. But now we’re seeing big changes in those individual industries and how they are becoming interrelated. The common thread is the impact these industries have in the power and energy footprint – which is why there is a heavy focus on efficiency. And they’re feeding off of each other, and that change creates kind of a multiplying effect, and that’s why we’re seeing this kind of rise in importance in power electronics across multiple industries.
PEN: Not only that, how GaN plugs into all of this. Because, I mean, I think we can all agree that these are growing and challenging industries, but I’m curious what your view is on how power became so prominent within each of these industries, and then how wide bandgap, especially GaN, can address that demand?
Jim: Sure. So first, let’s take a look at the automotive sector. Two big market drivers in that market are electric vehicles and autonomous vehicles. The drivers for the electric vehicle are pretty simple: Global CO2 emissions and global warming. We’ve got to do something about CO2 emissions and the electric vehicle is a major element in making that happen.
PEN: There are going to be the inevitable people who are arguing, and I’ll play devil’s advocate for them, some would make the argument that driver for electric vehicle market penetration is lower maintenance per mile driven – the innate advantages of electric vehicles is in lower maintenance, higher mileage potential. The top 10 automotive repairs don’t apply to electric vehicles. Do you think any of that has anything to do with it, or is it really driven by the pollution side?
Jim: The driving factor is CO2 emissions – and energy efficiency and sustainability. We will see simpler cars from the drive-train perspective, and that just allows more things (more applications) being put into the car. Things that have to do with data, accessibility, mobility, and convenience will come into the car. So, we would seem to, even if we simplify some parts of it, make other parts of the car more complex.
PEN: Got it, got it. Okay, then, where does GaN fit in? Because obviously, wide bandgap can fit into a lot of spaces in a car, but there are a lot of electrical systems, and there are sweet spots that GaN can address. I mean, Alex Lidow has often said about the LiDAR, but that’s just one system, and there are a lot of systems.
Jim: The big three applications in automotive power electronics are the charger, DC to DC converter, and the traction inverter. And, of those three uses, the traction inverter by far can benefit from GaN transistors, and so it has the biggest bang for the buck using wide bandgap devices. The real key here is if you can make your electronics not only less expensive, but make them a quarter of the size and/or a quarter of the weight, and increase energy efficiency, then the car goes farther. You also can use less batteries, your cooling system requirements go down, and the power electronics get smaller because you’re more efficient. And so, it has this multiplying effect inside of the car.
Then you start to go to the autonomous vehicle and there are sensors all over the car. LiDAR sensors are clearly a place where GaN transistors provide improved performance. You can get better resolution of the things that you’re looking at for your collision-avoidance systems. Another application is wireless charging inside the car for your phones and your computers. But also, we’re seeing people talk about getting power to all of those sensors in the car and eliminating some wiring boxes and connectors.
To sum it up, auto makers are taking elements out of the system that are both heavy and bulky and inefficient and replacing it with small, efficient power electronics. There are some real interesting things happening in the power electronics drive-train, sensor, and in the personal electronics areas.
PEN: Right. Now Jim, I had said in the pre-brief, I had said how I think that it’s more the wide bandgaps versus the silicon industry. I don’t think there’s much fracture site within the wide bandgap industry. However, there are two camps, and I’m curious how you would position GaN against silicon carbide within the automobile?
Jim: At lower voltages, 100 volts and 650 volts, GaN always wins. It has better performance and is lower cost, period. Above 1,000 volts, there aren’t GaN devices, so that’s silicon carbide’s territory.
PEN: Got it. So it’s almost going to literally break down to a voltage point somewhere around above or below 1,000 volts?
Jim: Yes. Exactly. And the reason is the fundamental physics are there. GaN has better performance than silicon carbide in switching power systems and is less expensive than silicon carbide. Silicon carbide is complementary and not competitive to GaN.
PEN: Got it. Got it. Well then, let’s move on to the data center, then. Because, I mean, automotive is, a lot of the automotive argument is driven by emotion, but there aren’t people sitting around saying, “I like to hear the data center hum”.
Jim: Many years ago, a data centers was just a “server in a closet” business, and servers were not a big user of energy or power in the world. However, data centers now already consume two percent of global power usage, and could reach five percent or more per some forecasts. The amount of data that are being processed will only grow larger not only from mega-data centers for companies like Amazon, Microsoft, Google, and Alibaba but data from autonomous vehicles. GaN will play a big role. By using more efficient and smaller power supplies in the server racks using GaN increases energy savings and allows for more servers in the same rack space, which means more revenue through that data center. By using the existing resources more effectively, it also means delaying the CapEx needed for new data center build outs.
PEN: Do data center operators care about light-weight and small size, or do they care about energy efficiency?
Jim: Efficiency. So, where in automobiles they really care more about making things lightweight and small. In data centers they care about the monthly utility bill. ~40% of the cost of running a data center is the cost of power. When you make the server racks more efficient, you also make the cooling systems smaller.
PEN: The reliability, I would imagine, is more important than efficiency of them.
Jim: Reliability and safety are number one in most every industry we service. It’s required just to play in the game. Then you’ve got to bring something to the party in terms of performance and/or cost. Energy reduction is an important issue in the data center. Microsoft has grown 5X in the last five years in their data cente
r. Other companies have done similar things, and they’re predicting another 5-10 times over the next five years.
Jim: Now, let’s tie this back to the automotive industry. Car companies are also predicting 10 times growth in their data centers. The autonomous vehicle will collect a ton of data and send some of that data to the cloud, which will then be needed to be analyzed and, with AI and machine learning, drive specific actions automatically.
PEN: Agree 100%, Jim, because, the car is literally the convergence of every developing technology we have. The cloud, AI, power management, smart grid. It all comes back to the car.
Jim: Yeah, yeah. You’re right. And that’s why in this multiplying effect of many industries kind of feeding off of each other, and really, pointing, like you said, pointing towards the automotive industry, is why power electronics is booming so much. The whole idea about reducing emissions in the car and the whole idea about electrification falls apart unless you have renewable energy and infrastructure for charging stations, right? Think about it. The demand for energy is rising. The world is trying to reduce CO2, and if I go burn coal in a plant to produce electricity to go charge my electric vehicle, I’ve just shifted the ecological problem somewhere else in the chain. I haven’t solved the problem in terms of the environment. I must produce the electricity for automobiles (and for data centers) from renewables or the goal of reducing CO2 levels is not met. So, there’s a huge focus on renewables to meet these energy demands.
PEN: You have mitigated it. Because the thing is, I do hear it all the time from people. But the fact is, a coal plant is about 30% efficient and a car is only about 10%. So even if all you’re doing is shifting the burden to the coal plant, at least you’re getting a marginal efficiency increase on an electric over gas.
Jim: You make a valid point. But why not go all the way!
PEN: True, but at least you’re not shooting the horse if you’re putting electric cars on the street powered by coal, that’s all I’m saying, to address the people who are going to be reading this and who are thinking to themselves, oh, no. But the bottom line is, it’s okay.
Jim: You’re right, it’s better to say it’s a two-step process. Number 1 is, you have the benefits of the electric vehicle and you get more efficient use of energy. Number 2, you switch it over to renewable energy, then you’re eliminating the CO2 from the whole system. Eliminate gasoline in the car, and then if you look from the data center perspective, they can also be clean energy and power in the data centers, and we’re seeing that happening. Google announced last year, I think it was in 2017, where they announced that they now generate 100% of their energy renewably for their data centers. So, they’ve already taken that step, and we’re seeing others take that same step, so that they’re putting a lower ecological footprint on the earth while still getting us all our data crunching that we need.
PEN: Well, you know, Jim, I’ve always been fond of saying that I trust greed. The moment the technology becomes cost-reducing, even the most stubborn Luddite is going to say, hold it. Wait a second, it’s cheaper? I’m in.
Jim: Yeah. In the overall system, the cost for electronics goes down when you use GaN transistors, and we’re doing this with automotive companies, data center companies, and renewable energy companies. So, you’re right. When you can show bottom line improvements, it’s not just the engineers that are enamored with wide bandgap, but it’s also the CFOs. Also, in renewable energy systems, GaN power semiconductors are able to reduce energy lost in conversion by up to 75 percent, provide efficiency gains in bi-directional (charging and discharging) systems, and enable the design of smaller and more power-dense energy storage systems with lower system and lower installation costs.
PEN: Exactly. And that’s always been a risk when you’re having an engineering discussion, is, because the facts are so clear. I can’t do the consumer journalist thing and just play devil’s advocate, because there is no alternate position to cleaner, cheaper, better, more reliable energy.
Jim: Exactly. Maybe four or five years ago, I’d be talking to engineers but I spend all my time now talking to business people about the economics of their business and how using wide bandgap with their engineers’ designs can change both the top line and the bottom line for the company. And that’s where you really get momentum – when the business people – CEOs and the CFOs – get behind this change.
PEN: The bottom line is, is that we’re literally at that point. Luckily, we’re literally witnessing, somewhere historians are going to say, in like, in the next couple of years, could be two, could be five. But somewhere in the near future someone’s going to draw an imaginary line and say, this is where the tipping point came for electric vehicles, alternate energy and wide-bandgap semiconductors.
Jim: Yeah, I think you’re right. I think we’ve probably already passed that line, because we’re already seeing Solar ESS and Data Center power supply products in the market. And for automotive, it takes a good six years for a design to work through the cycle and we’re half way through that cycle with many, many automotive customers. Maybe half of that six years with the automotive engineers, and half that time with the product guys. We’ve already passed that line with the engineers and we’ll see that coming out as the product guys get them into production models here in the next couple years.
PEN: Okay, I’m going to ask you to stick your neck out a little bit. When would you say we will see viable, cost-effective, high-performance, by high-performance I mean reliable distance-wise, no performance anxiety. But when will we see at reasonable price points, like $30,000 or less? Are you willing to make an assessment of when that would be relatively mainstream? Not even mainstream, mainstream, but to the point where it’s approaching, the early adopters can be comfortable?
Jim: I’m not really the expert on that, but certainly the models that we see happening in Asia are totally geared towards the medium and low-end consumer, so the price points that you’re talking about are even lower. I don’t know exactly when those will come out, but things are happening much faster in Asia. So, don’t be surprised to see that price point hit in the near future, and then crashing past it as more and more models come out from some of these manufacturers and economies of scale take hold.
PEN: One of the things that a lot of people overlook, is that two- and three-wheeled, especially three-wheeled vehicles in the developing world, are going to be the real change, you know what I mean? Places like Shanghai are already cleaner because of three-wheeled vehicles are electric instead of gas.
Jim: Yeah. Absolutely. I was recently visiting with a leader of a transportation company in Asia, and they’re focused on Southeast Asian market. She said she doesn’t take taxis. She doesn’t take Uber or her own car around the cities in Asia. Her mode of transportation is the motorcycle fleet because it’s the only way to get around places like Jakarta, Saigon, or Bangkok and it saves hours and hours of time not stuck in gridlock. And so you’re absolutely right, electric three-wheels, electric two-wheels, are a really important component of that. And it’s not something we think about here in North America, nor in Europe. But in Asia, it’s an area they are acting on. These are smaller vehicles with smaller battery systems in terms of capacity and with battery voltage in the sweet spot of GaN.
And it’s likely not just a ‘$30,000 or less car’ because we have entered the Car Share era of Uber, Lyft, DiDi, Grab, and ZipCar. And today’s Car Share requires a Share Car – where the car is not owned by the user but by the operator. And the metrics for success are new. For the user, it’s convenience and low cost per use. And for the operator, the Car Share entrepreneur, the metrics are different too. Instead of 4% usage of the car, it’s 7-day/20-hour usage. So critical is not only low initial cost, but also low operating cost per mile, higher energy efficiency, and much longer lifetime.
There are lots of new concepts on the design table – a particularly provocative one is Toyota’s i-Road 3-wheel concept which has been trialed in both Tokyo and Grenoble. Whatever designs win, this is a huge marketplace for vehicle makers, and for GaN powered electronics and a big trend in the marketplace.
PEN: Very cool. So, Jim, any closing thoughts for our audience?
Jim: Power electronics is booming. And the reasons are clear. Electronics have become a bigger and bigger part of our lives from our automobiles and our data centers, to our renewable energy systems. And wide bandgap GaN is playing a huge part in making this all happen and making the world a better place.