More Electric Aircraft Movement Looks to Innovative Technologies: Q&A with Microsemi
Aerospace applications are benefiting from higher processing power FPGAs and the steps taken to optimize power core modules.
You may be a jaded traveler by now, but the early pilots, and first-time pilots today no doubt, find becoming airborne an electrifying experience. Now, a push is on to tie flight and electricity together in a more literal sense. More Electric Aircraft (MEA) will rely on advances in technology to strengthen aviation performance and reliability, among other benefits.
Microsemi is one of the companies making the technology improvements the MEA movement demands, including with its announcement last month that it had introduced an aerospace power core module (PCM) designed to optimize power electronics. Our thanks to Jim Aralis and Shane O’Donnell of Microsemi, who responded to our questions about the new PCM and MEA goals following the announcement. Aralis is Microsemi’s CTO. O’Donnell is Aerospace Product Development & Technology Manager, Space & Power Management Group, Microsemi. He is also manager for Microsemi’s Aviation Center of Excellence.
EECatalog: What are three steps Microsemi took to make the new power module construction innovative
Shane O’Donnell, Microsemi: Three of the steps involved in making our new Aerospace Power Core Module innovative while optimizing for power density, performance, reliability and weight are:
- The product architecture has been designed for aerospace applications. The power substrate, containing the power SiC MOSFETs and diodes, is integrated seamlessly with the driver and controller printed circuit board assemblies, resulting in a low-weight solution with excellent electrical, mechanical and thermal properties. Furthermore, the architecture facilitates numerous customization options, such as adding power capability or functionality without increasing the module sizing.
- Microsemi has selected materials and components to achieve the highest performance and smallest form factor while designing for reliability, cost and manufacturability. For example, the PCM contains an AlSiC baseplate, SiC MOSFETs and an FPGA within a 5 kVA module, which weighs less than 300g, and with an MTBF in excess of 300,000 hours in a harsh environment.
- Using advanced electrical, thermal and system analysis, Microsemi can customize the PCM, not only for different aerospace applications, but for the mission of the aircraft. For example, Microsemi can differentiate the performance of the PCM used in an aileron system on narrow or wide body aircraft and adjust the unit construction accordingly to meet the customer’s requirements.
EECatalog: One topic of the MEA conference last month was the need for aerospace manufacturers to change their thinking for electric aviation. What are your thoughts on this and how can Microsemi’s knowledge and expertise guide this change in thinking?
O’Donnell, Microsemi: Power Electronics is an enabling technology in MEA. The move toward more electric aircraft (MEA) continues to drive demand for increased levels of performance, reliability and integration in the area of power electronics. When the reliability has been proven, traditional aerospace manufacturers need to consider the adoption of new technologies such as wide bandgap (WBG) semiconductors instead of their silicon alternatives, with strong heritage, due to the significant gains in performance, size, weight and reliability which can be obtained.
Microsemi has been supplying products into high reliability application areas such as commercial aviation, defense and space for over 55 years. The company’s high reliability philosophy, together with our technological expertise in semiconductors, has resulted in the development of wide bandgap devices such as silicon-carbide (SiC) MOSFETs and SiC diodes suitable for MEA.
While aerospace manufacturers may traditionally be focused on low-cost and low-weight solutions, a more comprehensive and integrated systems perspective deserves consideration. Although SiC MOSFETs are more expensive than silicon IGBTs, their improved power density, higher switching speed capability and lower switching losses can reduce component size and system weight and improve reliability.
Microsemi’s Power Core Module integrates the power switches, driver circuitry and high-speed communications interface in a single package. The replacement of Si IGBTs with SiC MOSFETs results in a power dissipation improvement of approximately 25%, with a resultant system gain in thermal performance (Figure 2).
EECatalog: What is it important to know today about FPGA development environments and the options designers have?
Jim Aralis, Microsemi: The density of course has gone dramatically up, so the things that we can do with FPGAs are much more impressive. We used to relegate them to doing things that were slow or were not too big or were specialized control functions, as opposed to being in data paths with data processing elements.
So the performance of FPGAs is getting to the point where a lot of things that could not be done by them before they can now do. FPGAs are coprocessors in data centers and servers, as well as taking on such tasks as voice recognition and server applications for doing baseband generation. Baseband virtual environments are being enabled because FPGAs cannot only process the data at high data processing rates and complexities, they can also be retargeted to new applications on an ongoing basis, which, again, is another new way of people using FPGAs, not just in embedded systems, but as well in systems where the complexity is a level above that, meaning servers, basestations, and things of that nature.
At Microsemi, we represent FPGAs that are for whole systems. We have nonvolatile memory for use inside, but also for programmability, meaning you don’t have this high rate of interface to the outside world to get your programming materials—it can be a standalone part and look just like a part you buy that has already been programmed, as opposed to the system-level stuff that has to be done with other people’s FPGAs.
We are seeing a high level of security. The security that is being built into the FPGAs now— both ours and our competitors—to enable network security as well as data security is unprecedented even in dedicated ASICs because these algorithms and these capabilities are evolving at a very high rate.
FPGAs have core capabilities like [physically unclonable functions] PUF, which are unique functions that identify each individual FPGA. We have very high level elliptic curve cryptography. We have a lot of very sophisticated blocks that build security capability, but then you put them together with an FPGA, where you structure an architecture security for your applications separately, and they become very difficult for somebody to go back and decode or break. Also, the economic benefit of breaking it lessens because the application is not all-encompassing.
Also pretty far along and getting even more important is the system on a chip inside an FPGA. Processors, memories, multiprocessors, and digital signal processing is now inside chips. We’re also putting in specialized accelerators and data accelerators that can be used with the fabric as well. There is a fundamental change coming with our next generation of FPGAs, and probably with the next generation of our competitors as well: you almost can see it as an embedded FPGA because so many capabilities are hard wired in because of the evolution of SoCs and systems in general.
To take what I’ve just mentioned a little further, we build a lot of SoCs for storage, for networking that are not FPGAs, and 80 to 85 percent of those circuits are actually IP—meaning they are almost common between something even as disparate as a PCIe controller, memory controller and switch—they have very similar elements in them—the only difference is how you hook them up and some of the dedicated logics. So you end up with an FPGA in which 80 percent of its functionality is in the number of applications processors, the number of DSPs, the number of controllers, memory, and then some dedicated logic, which can be implemented in an FPGA. You will find an FPGA that will compete directly with an ASIC in those areas because not too much of it has to be programmable in a logic sense. It’s programmed by micro-code and even Linux code directly.
We are going to run out of steam in process technology somewhere at four or five nanometers, which is one generation from where we are doing our designs right now at seven [nanometers]. When that happens the complexity of these chips is going to be so great that the implementation methodology of preference will be processors and FPGAs.
It costs too much to build a custom chip for every application, and nobody will be able to do it. So the people who know how to mix processors and FPGAs will be the people who will win the application space for doing economical application-specific SoCs.
EECatalog: What does greater processing power mean for the More Electric Aircraft movement?
Aralis, Microsemi: Much more analysis of the data can occur in real time, and we can make adjustments. In the past we had to take data from, let’s say, a jet engine, and use that data (very good data, “big data,” as everyone calls it) to optimize the jet engine for the general case of when you took the data with maybe a couple of points that say: this is takeoff configuration; this is high cruise; this is low cruise; this is landing—and so you could configure the jet engines to have performance at certain different areas. Say it was cars rather than jet engines, you would have: driving offroad; driving onroad; driving slippery.
But now we’re seeing the advent of this capability of very, very high processing and that of deep learning, the idea of using learned data in order to do things. We combine those things now to monitor jet engines, for example, and learn how to dynamically change the engine’s performance. We can do the equivalent of all that data analysis that in the past took place around one week after data collection. But now rather than doing the analysis a week later, it’s done real time—so it is no longer low cruise/high cruise—it is wherever you are. It’s optimizing the engine on a very quick cycle so that you can have it optimized no matter where you are in the cycle.
We couple that with the ability to do things, even acoustic vibration, [both] sub-acoustic and super-acoustic where we look at what is going on in the engine in order to predict its reliability. The work we are doing with some of our partners is to make aircraft more reliable, and cheaper.
An aircraft is one of the few places where if you make something last longer, you actually reduce the costs—If you can make an engine last twice as long, you cut its cost in less than half because you don’t have to replace it and you can use it twice as long. It is not like a pair of shoes or even an iPhone that goes out of style. If you double in time the length that it works, you get less than half of the cost.
Our heuristics can predict closer to where it is going to fail. If you can do that you can keep the engine in the airplane much longer and even at a safer level to cut the costs. You can cut it in half. Our goal is to build these systems with twice the MTBF and with that you have actually cut the thing in less than half cost of operation.
So those are the kinds of things we apply directly to aircraft in order to increase reliability, and lower their cost as well as their weight.
As well as seeing to it that when you do have to change something, you do it at a time and place that is much cheaper. You don’t want your planes to show failure in a remote location—it takes a lot of money to get the parts and the people there.