12 December 2017: According to Yole Développement’s recent report Power GaN 2017: Epitaxy, Devices, Applications, and Technology Trends, in 2016 the power GaN market reached US$ 14 million. It is still a small market compared to the impressive US$ 30 billion silicon power semiconductor market. However, its expected growth in the short term is showing the enormous potential of the power GaN technology based on its suitability for high performance and high frequency solutions.
A future market for the charging of the high voltage drive battery, the DC-AC auxiliary power supply, as well as the DC-DC buck conversion to 12V and future 48V battery is possible for GaN due to its high-speed switching capability. Some main players, as Transphorm, have already obtained qualification for automotive, and this would help to finally ramp-up GaN production for EV/HEV in the coming years.
Few days after receiving a $15 million investment from Yaskawa Electric Corporation, we spoke with Philip Zuk, Vice President, Technical Marketing at Transphorm, a global semiconductor company that develops fully-qualified 650V GaN power devices for high-voltage power conversion applications. We discussed new developments and future prospects at Transphorm as well as global industry trends.
(Source: Power GaN 2017: Epitaxy, Devices, Applications, and Technology Trends report, Yole Développement, October 2017)
Yole Développement: Can you introduce Transphorm’s history and products?
Philip Zuk: Transphorm was founded by Primit Parish and Umesh Mishra in 2007—after they started, ran and ultimately sold to Cree a successful GaN LED business called Nitres. Over the past ten years, Transphorm has been focused on bringing high voltage (HV) GaN FET technology to market. This focus has encompassed designing, manufacturing and selling the industry’s highest quality, highest reliability GaN technology for the power electronics market (Merchant power supplies for data center servers, PV inverters, induction/servo motors, broad industrial and automotive). In 2013, Transphorm released the industry’s only JEDEC qualified device. In March 2017, we continued that first-and-only trend and released the market’s only 650V automotive (AEC-Q101) qualified device.
TPH3205WSBQA – AEC-Q101 qualified FET (Courtesy of Transphorm)
Transphorm is unique in that it:
• Offers the only JEDEC and AEC-Q101 qualified technology.
• Has one of the largest GaN patent portfolios.
• Offers the most diverse device/packaging solutions in the market.
• Is one of the only GaN suppliers with end customers in production today.
• Owns its EPI, design, , and process stages of the device value chain: this gives Transphorm multiple “knobs” that can be individually adjusted to maximize our GaN platform’s quality, reliability and performance.
YD: Would it be possible to share some of the new advances in Transphorm’s GaN technology development? Will Transphorm produce and qualify 1200V devices as well?
PZ: We are working on releasing our 3rd generation normally-off GaN FET technology. This technology is evolving with an increased threshold of 4.0 V, increasing its noise immunity, improving performance, eliminating the need for negative gate drive, and reducing cost over its previous generation. The new GaN technology will be available in traditional through-hole packages (TO-XXX), and surface mount discrete packages (SMD).
In parallel, Transphorm is developing an E-Mode technology based on a Metal Insulated High Electron Mobile Transistor (MISHEMT) D-Mode device, which is our normally-off cascode GaN FET (*). This GaN structure is different than what is currently offered by any other E-Mode supplier.
Our E-Mode will deliver definite advantages, such as an insulated gate, resulting in a widest margin between fully enhanced to maximum rating Vth available today (based on our D-Mode HEMT, AEC-Q101 test results shown below). It will also offer increased noise immunity with a ≥ 1.5 V threshold, reducing the need for a negative gate drive.
As of today, we have no plans to develop a 1200V platform.
* Transphorm’s E-Mode will not replace its current normally-off GaN FET device. Instead, it will offer flexibility to the end customer allowing them to select what best suits their designs.
YD: What are the strengths of Transphorm compared to other companies in the GaN industry?
PZ: We see several. As stated above, we have a vertically integrated business approach across the full product value chain—EPI, design, and fabrication, and application support. This allows for innovation at every development stage. Which, in turn, allows for higher quality and reliability.
We also have reached several first-and-only milestones within the industry. Transphorm designed and produces the only JEDEC-qualified technology, which is backed with publically-available data verifying lifetime reliability (HTOL, HVOS, HTDC). We also designed and produce the only AEC-Q101-qualified platform, meeting automotive industry preferred benchmarks.
Our GaN technology offers the highest threshold voltage (4.0V) and highest increased noise immunity in the market.
We offer standard easy-to-use TO-xxx packages that provide a 40% increase in power output versus the competition. We also offer specific drain and source SMD devices used for high side and low side locations respectively.
Lastly, Transphorm’s GaN is in production applications, some of which have been publically announced (Servers, Servo motors, Power supply power stages, GaN design modules).
YD: Today, SiC MOSFET is developing very fast and penetrating into a lot of markets. How would you picture the scenario around the 600V range?
PZ: The 600V/650V FET application market is huge (superjunction (SJ) being between $700M and $800M Total Accessible Market (TAM)). It includes everything from running the popular worldwide standard single phase line-to-neutraor line-to-line input from 85 Vac to 264 Vac. Naturally, this is a voltage that all want to participate in.
Indeed, SiC is penetrating this market due to the opportunity it poses. It has had a longer period to reduce its costs than GaN-on-Si. However, while SiC may seem like an attractive option today—we anticipate that as HV GaN-on-Si matures, you’ll see higher Return on Investment (ROI) and increased performance with current and future technological advances. There will always be customers that will use SiC within the 600V market(s), but ultimately this application space will be best served by high reliability GaN.
YD: Will different technology co-exist for a long time? Or, will one eventually become dominant?
PZ: Si, GaN, and SiC will continue to co-exist for a long time. Many stated that 74ACxxx logic would disappear and it is still alive and well in the market. There are many customers that will continue to use IRFxxx MOSFETs that have been around since the mid-80’s, because the technology is “good enough.” Plus customers will push back on having to do any re-qualification to implement newer technology (even if prices are raised).
During this co-existence, it’s important to keep in mind that one technology will be more suitable than another depending on an application’s power level, required performance, and price point. However, as systems drive to smaller form factors with increases in power levels, GaN will penetrate other power technologies footholds more and more.
YD: We have heard that new package solutions need to be implemented in order to fully benefit from GaN performance. What’s your view on that?
PZ: No. This is incorrect. GaN is different than SiC. With SiC, if one wants to use the high temperature capability, then new high temperature packaging is required. Even though GaN has faster edges (rise and fall times) than Si and SiC, traditional packages can and will continue to be used.
For example, we’ve removed all Kelvin source connections from our PQFN 88 packages as there was no performance/noise immunity advantage. (Si SJ uses Kelvin sources today.) We also increased the pads due to power suppliers using multi-layer boards (>10 layers) for increased board-level reliability. With our PQFN 88 packages not requiring Kelvin connections, this allows standard TO-xxx packages to be offered allowing increased thermal heat sinking capability for higher power levels of operation.
Additionally, with respect to packaging construction, bond wired solutions are not becoming extinct. Regardless of GaN packaging, development will continue to evolve such as clip versus bond wires and top-side cooling. In fact, these features will benefit all technologies.
YD: Most of the companies are focusing on entering the 600V market today. Do you think there would be an opportunity for newcomers in lower voltage (0-200V) GaN as well?
PZ: Applications that require 12V and 24V outputs use low voltage GaN devices today in order to increase the efficiency in DC to DC converters. This is where you see HV and LV/MV GaN complimenting each other in AC to DC converters such as server, storage, and network supplies. Wireless charging and LiDAR (automotive) are also large markets for low and medium voltage GaN devices. Most GaN companies have selected either HV or LV/MV technology development areas.
YD: Which market do you think would drive GaN technology to mainstream?
PZ: The fastest growing market today and moving forward is automotive. Beginning in the early 2020’s, you’ll see GaN in on board chargers (OBC), DC to DC converters, and DC to AC inverters. The internal combustion engine (ICE) is not going away, but the electric vehicle (EV) are growing and autonomous vehicles are being developed. You have car companies saying that they will no longer sell pure internal combustion engines after 2020 (Ford and Volvo). What’s more, it’s estimated that 30% of all cars sold in the world will be non-ICE by 2030.
However, the following applications markets will also continue to drive growth, though more slowly, in the low-voltage, mid-voltage and high-voltage GaN technology markets:
• Broad industrial (i.e., UPS, auxiliary power supplies, applications that cannot achieve the efficiency levels required)
• Automotive (LiDAR)
• Some Consumer/Computing applications for low power integrated power devices or monolithic driver + GaN FET devices
GaN in vehicles – Automotive High Voltage GaN applications (Courtesy of Transphorm)
YD: Do you feel that the market is ready to adopt GaN? Which market(s) would you consider as a driving force?
PZ: Yes and no. We have early adopters in all markets, but there are more “fast followers” that are sitting on the sidelines waiting for a time to jump in. This “time to jump in” is multi-dimensional and based on many factors such as:
• Quality and Reliability
• Multiple source opportunities
• Perception of the technology (readiness)
• Experience with the GaN technology (hardware (bridgeless totem-pole) and firmware development)
• Technical resources to investigate the GaN technology
• Understanding the technical advantages over incumbent technology
• Understanding that GaN technology is about “system cost” versus part-to-part cost
GaN is a lateral device that came from the RF and LED world, so it is new to power electronic engineers. Education and training is key to adoption, which is why we continue to build out our design resources library and work diligently with customers to support their design development. We have also developed and produced evaluations kits and will be releasing several reference designs to help engineers build more confidently with GaN.
Regarding a driving force, I would consider that to be any Merchant power supply being used in a market of 500W or more.
Totem Pole PFC (4kW) evaluation kit (Courtesy of Transphorm)
YD: Today, big foundries are entering the GaN market. How would you comment on the fabless model compared to standard IDM mode in the power electronic industry?
PZ: There are advantages to both with respect to burn-rate, size of resources required to develop and bring the technology to market. The drawback of the fabless model is that you have to use someone else’s standard process, which can be cumbersome. This may cause a GaN device supplier to compromise on the performance and/or quality and reliability of their offering.
We also know that the “secret sauce” in any FET technology is the EPI. So, having control and IP around the EPI is crucial for not only developing a single platform, but also evolving the technology over time.
If you look at today’s currently available E-Mode technology, there hasn’t been any generational increase or evolution in the products. Owning the process, EPI, and design allows a supplier to not only fine tune their technology over time, but also to increase or decrease voltage offerings. This is really the most effective way to develop disruptive technology against incumbent solutions.
We’ve opted for a foundry partnership in which the engineers working the line are Transphorm employees. This allows us to control quality while also learning first-hand more about how we can evolve the technology at the manufacturing stage as we develop new generations.
YD: How do you see the future of Transphorm in the next 5 years? What major evolution do you see in the future?
PZ: EXCITING!!! If I told you all we have planned for the future, then what fun would that be? Don’t worry. Transphorm has some exciting future happenings coming down the pipe.