I had the pleasure of interviewing the CEO & co-Founder of NexGen Power Systems, Dr. Dinesh Ramanathan, about NexGen’s ground-breaking power system technology and their higher-performance LED drivers.
Shiller: First, I want to thank you for agreeing to speak with me, today. Maybe a good place to start is a brief description of NexGen Power Systems.
Ramanathan: Sure. NexGen Power Systems is a vertically integrated company that makes power systems which are very efficient, very small, very lightweight, and very environmentally friendly. The way we do this is we have three technology cornerstones on which we have built our technology. And these three technologies have been put together to allow us to make a power platform. That power platform is then applied to each one of the verticals that we end up serving. From a power systems perspective, the three key properties of what we do are our transistor, which is GaN on GaN, vertical gallium nitride power transistor. The second is our Merlin Power Engine, which basically has software and algorithms which allow the power supply and power system to work at one megahertz plus switching frequency. And the third is innovations that we have done on mechanical and thermal designs, which allow us to operate these power systems at the thermal boundary, which means we make them as small as we potentially can, and still make sure that they don’t get too hot.
Shiller: How was the technology received at LightFair? Did the lighting industry react to your value proposition, and do you think it understood your value proposition?
Ramanathan: They did, and this was our first time at LightFair, so we had a few things to learn, ourselves. But the overarching theme that we encountered was that everybody that is making LED lighting fixtures is looking for their drivers to become smaller, more efficient, and more lightweight. So that kind of fell directly in with our value proposition. That’s what we do. Now, they are not as savvy about the technology that goes into making the systems smaller, more efficient, and lighter weight. But what they care about is the outcome, which is making sure that the LED drivers that we put out into the marketplace are smaller, more efficient, and lighter weight.
Shiller: So a great black box is good enough for them.
Ramanathan: Yes. And that was the learning for us, because we were trying to explain how the technology works internally. What makes it smaller. And I don’t think that they care that much about the details of what goes in. What they care about is what they’re getting out from it, which is perfect because we can provide the solution that the customer wants without complicating their lives too much.
Shiller: The most striking thing I remember from our conversation at LightFair was that your vertical GaN transistors create drivers that are 6% more efficient, enabling a 250% increase in lumen output within the same size luminaire or smaller. And that’s a remarkable increase in light output. And I’d like to unpack that a little bit. Can you explain how your vertical GaN transistors create higher frequency switching and the 6% higher efficiency while also being smaller? I mean, it’s a lot going on.
Ramanathan: Let’s take them one step at a time. So fundamentally, from a 10,000 foot view, what tends to typically happen when you run power systems at higher switching frequencies is you shrink the size of all the passive elements that are there in a power supply. The passive elements are inductors, capacitors, transformers, all these basic elements become smaller. They become smaller because the total amount of energy that is now being transferred from the primary side to the secondary side in a circuit has become smaller with the higher switching frequency.
Shiller: The area under each curve, if you will?
Ramanathan: Yes, the area under each switching cycle is much smaller. You only need small amounts of inductors and capacitors to transfer energy from the primary to the secondary. So you don’t need as much local storage as you previously did, when there was more energy per cycle. Another very important thing that also happens is the filter that you need for conductive electromagnetic emissions becomes smaller. Why? Again, for the same reason, there’s only a small amount of energy that you’re transmitting per cycle. That’s the energy that you have to filter for any noise that gets injected into the AC line. For radiated emissions, we have a little heat spreader that we put around the system to make sure that the heat is uniformly distributed, that also acts as our EMI shield. So using that feature, we are able to make the entire system much smaller and also be able to bypass all of the constraints that are typically imposed on systems like this.
Shiller: That’s a great explanation with the frequency. But the next level up where the 6% improvement in driver efficiency enables a 250% improvement in light output. Is there a way to break that down? How much of that’s due to reduced heat versus reduced size?
Ramanathan: Right. So the way to think about how we’re able to produce more light is actually to do an apples-to-apples comparison. So it’s all about size. If you take a look at how small we can make the LED driver and then you look at how much power is actually generated by any other LED driver in that same form factor. Because we are able to reduce the size by roughly 60%, we can actually produce in that smaller form factor the same amount of energy by something that’s two times as big.
Let’s take an example. So let’s say our driver had a volume of two cubic inches. If we were able to put out, let’s say, ten watts of power. The competition is using four cubic inches to put out the same ten watts of power. So in four cubic inches, we could actually put out twice the amount of light or twice the amount of power.
Shiller: So because you’ve halved the volume, you’ve essentially doubled the power density, and halved the thermal density.
Ramanathan: Exactly right. So the 6% improvement in efficiency has to be there, because otherwise we can’t shrink the size by 50%. And the way to think about a 6% increase in efficiency is to look at the losses and look at the fact that we’re actually cutting the losses down by roughly 50%.
Shiller: That occurred to me. So you’re going from roughly 15% losses to say 8 or 9%, maybe?
Ramanathan: Yes. We’re cutting 15 point worth of loss to roughly 8 or 9 points of loss.
Shiller: Which is roughly a 40% reduction in losses?
Ramanathan: It’s more than 40% of the overall loss that’s actually eliminated.
So these two mechanisms, reduced heat and reduced size, put together is what allows us to generate a lot more light in the same form factor. And I think the key issue is to look at the form factor.
I’ll give you another example from the data center applications that we’re working in, where our power supply size is fixed. Because we can provide more power density, we can actually give twice the amount of power in the same form factor. And if you’re able to give twice the amount of power in the same form factor that our customers are looking for, then they can deploy more server racks. The same thing happens in the LED lighting space.
Shiller: So, it’s hard to separate the reduced heat and size because one enables the other and then they compound. Can you elaborate a little bit on the high frequency transistors reducing EMI. Can you talk a little bit more about the consequences, how it eliminates a lot of components?
Ramanathan: So this is a two-step process. First, because of the technology we’re bringing to the table, we’re actually able to shrink the size of all the components. The actual number of components isn’t shrinking. They’re actually becoming slightly larger. And the reason for them becoming slightly larger is because we have an external gate driver outside the Merlin Power Engine, which is what we use to drive our gallium nitride transistor. Most of the controllers that you will see in the marketplace, especially in the LED space, they have integrated all these drivers into one chip because they have made these ASICs. We haven’t because it’s new technology. So, the next step in our program is to take some of these circuits that we have had to put outside and actually build it into ASICs. And when we do that, that reduces the number of components that we have outside. It will also allow us to make sure that the design is much more cost efficient than it is today. It’s already cost efficient primarily because it shrank the size of all these components. But now we’re actually able to integrate some of those external components into our controllers. And then that causes the next wave of cost reductions that we can bring to the table.
Shiller: At LightFair, you shared that the company recently emerged from stealth mode and that you were pursuing lighting, automotive and computer hardware industries. Is that an accurate statement?
Ramanathan: That’s correct. So we’re pursuing LED drivers. We’re looking at power supplies for high end laptops, for data centers, and for electric vehicles. So you’re exactly right.
Shiller: Circling back to the ASIC conversation. When you develop ASIC drivers, do you see them being small enough for use within lamps, or what most people call bulbs, because that’s how they can get drivers into a screw shell, for example?
Ramanathan: We fundamentally believe that they will become much, much smaller than they are right now. And we do that by integrating everything that’s outside into an ASIC. There are a lot of components that we have outside, primarily voltage adjusters and things along these lines which don’t need to be outside a chip at all. They should actually be inside a chip. The semiconductor industry makes ASICs that are specific for lighting applications. That’s what we intend to do with our control engine. Now, all those engines typically run between 60 and 100 kilohertz. Our ASICs will basically be operating these switches at a megahertz-plus switching frequency. That’s the next way for us to bring costs down and also shrink the size. The other thing to also point out is that one megahertz is where we are starting today. From one megahertz, we can go to 1.25, 1.5, 1.75. 2. And at each one of these levels, the components that we talk about shrink in size. So that’s an advantage that we bring to the table.
Shiller: So that’s a peek at your roadmap, right?
Ramanathan: Yes, we just keep going. To some extent, we have to wait for some other components to catch up to our increasing switching frequency. So, for instance, the magnetics have to catch up with where we are. So if you’d asked the same question about four years ago, most magnetics would have worked at about one megahertz switching frequency or they would have been characterized up to one megahertz. Today, when you go look at magnetic material, the magnetic suppliers are characterizing them at two and even three megahertz switching. So, we know that as we push the switching frequency further and further up, the magnetics guys are doing their part in making sure that the ecosystem is actually a place where these technologies can come to bear.
Shiller: Can you share a little about which aspects of your manufacturing are occurring in India versus the US?
Ramanathan: Sure. So the way we run our organization is that our transistors, which is the key portion of our technology happens in the United States. Our fab is located in upstate New York. It’s in Syracuse. And it’s a big fab that does both manufacturing, as well as R&D. For us, that key technology that we develop is based in the United States. The rest of the system development front, we do between India and the US. We have a group here in Santa Clara and another group in Southern California. They do most of the prototyping work of some of these really high frequency, advanced designs. And then those designs are taken over by our team in India, that takes and replicates them on various levels. The team in India actually does mechanical design. They also do some of the software work. They do all the EMI work and the full system validation gets done in India. The contract manufacturer that actually builds these systems for us is also based out of India. We took into account some of the geopolitical issues that we’re seeing in the world today.
Shiller: That’s where I was headed next. At LightFair, you referenced an enormous increase in Indian manufacturing that’s occurring with global exports. Would you like to speak to that at all? And whether you see India challenging Chinese dominance in global manufacturing.
Ramanathan: Yes, I think India will, at least based on what we are hearing from our contract manufacturers and others, in India. Their output has gone up significantly. And I’m not telling you anything that is hidden or not known. There have been articles that say that iPhone 13 manufacturing is actually happening in India and it’s being done by Foxconn and other players that do work for Apple. I would consider iPhone 13 one of the most complicated pieces of electronic hardware that gets put together. And if that’s getting done in India at volume, you can understand the sophistication that’s actually at play at this point in time. Our contract manufacturer is increasing their capacity by about 5X, and the reason they’re able to increase it and they’re going 5X is just to make sure that they have the capacity to look at a lot of contract manufacturing that is moving away from China and into India. And an interesting part is, our contract manufacturing, for instance, is actually largely owned by a Chinese company. And the conversations with those contract manufacturers tell us that as long as we’re able to get contract manufacturing done in whichever country that makes sense for us, they’ll continue making it in those countries. I think some of the business folks inside China have also realized that having shutdowns and stoppages of manufacturing doesn’t help them. It doesn’t make their customers any happier. So they’re thinking “how do we make sure that our supply chain is as robust as we can?” And that means setting something up outside of China makes perfect sense.
Shiller: The pandemic issues are just one challenge with the Chinese supply chain. There are other geopolitical obstacles growing.
Shiller: What can you share about your commercialization timeline?
Ramanathan: Our full commercialization is going to happen in the fourth quarter of this year, towards the end of this year. And that could possibly slip by a month or so. So I’d say towards the end of this year or very early part of next year is when all our products will actually go into the market. The LED products are a little further advanced compared to the rest of the products, but it’s no more than a month or two. So we expect to start shipping in volume towards the end of this year.
Shiller: Very good. Do you plan to sell your drivers through electronic distributors like Arrow, Mouser or Future? Or through an internal OEM sales organization that you’ll have to build?
Ramanathan: Initially it’ll be an OEM sales organization and most of what we’ve been doing from our marketing and sales approach is a very targeted digital marketing campaign, that we’ve been putting together and that targets very specific customers and very specific engineering organizations, inside those customers. So we expect to end up working with the top 10 to 15 customers in this particular space because we are trying to work our way from top down. Once we get those customers lined up and show them the technology that we’re bringing to the table, then it’s a matter of how to scale that? This becomes when Mouser and all the other distributors kick in.
Shiller: I see all of the benefits for a luminaire maker, but I see even more profound benefits to lamp makers because they are just so much more space constrained and thermally constrained. Can you speak at all to a roadmap or a timeline to benefiting lamp makers as opposed to luminaire makers?
Ramanathan: Yes, so lamp makers are a little trickier, and the reason they are trickier is primarily because cost is one of the primary requirements that they have. So we get to that lamp market once our technology starts to ship in some volumes.
We could even think of taking our our gallium nitride chip and actually making and integrating it into the package that we put together so we can have an ASIC with our gallium nitride all on the same chip. That then gives them the smallest form factor, but that’s at least two years down the road for us. It’ll take roughly 18 months to put our basic product out into the marketplace. Pursuing the lamp market is going to be a function of how small we can make it, and how effective it is for them.
Shiller: Thank you, Dr. Ramanathan, for sharing your exciting technology with our readers.
Download the NexGen whitepaper Miniaturization of LED Drivers with NexGen Vertical GaN Technology here.