The boundaries between IGBT, SiC, and GaN are blurring, leading to a complex choice that considers price, size, and reliability.
"SiC to eventually replace even megawatt-class IGBTs"… Demand for humanoids and robots drives WBG adoption

The formula for selecting components for industrial drives is breaking down. The days when components were largely determined by voltage and capacitance alone are over. As wide bandgap (WBG) components, represented by SiC (silicon carbide) and GaN (gallium nitride), flood the market all at once over the silicon IGBTs that have supported the market for decades, the choice of which component to use within the same power range has become a matter of weighing price, size, switching, and reliability.

Dabo Corporation’s technical webinar on industrial drives, scheduled for July, also addresses this transition head-on. We met with Kim Yi-kyu, a Research Fellow (FAE) at Dabo Corporation who will be giving the presentation, to hear about the dynamics of the device transition and changes in design requirements as experienced in the field.

IGBTs are still mainstream, and WBG is rising above them.

The main component for industrial drives is still silicon-based IGBTs. Most industrial drives operate at switching frequencies below 10 kHz, and there is a wide range in which the dU/dt (rate of change of voltage) is limited by design without an output filter. IGBTs are devices refined to withstand short-circuit tolerance and power cycling under these conditions. They also match well with the characteristics of drives where the switching frequency decreases as the output increases.

SiC and GaN are rapidly advancing into the upper tier. These two devices enable high switching frequencies and exhibit lower losses than IGBTs even at low switching frequencies. Infineon Technologies estimates that if SiC replaces IGBT modules, investment costs can be recouped within 1.5 to 3 years. Reduced heat generation leads to lighter cooling designs, and smaller components result in a smaller overall system.

Behind this transition lies a time lag between device material properties and manufacturing technology. Commissioner Kim stated that the demand for wide band gaps has existed in the market for a long time; however, the technology to manufacture products using such materials simply did not exist. "Now, as the technology of manufacturers like Infineon advances, products are flooding the market," he explained. This implies that latent market demand is shifting into actual adoption as supply catches up.

The positions of SiC and GaN have not yet been clearly divided.

The regions of the two devices are not clearly separated. GaN products are currently available up to the 700V level, so they are mainly used in the low-voltage range. Although the lack of high-voltage products was a significant weakness, voltage limits are rising as research progresses. Commissioner Kim noted that products in higher voltage ranges could emerge once cost and technical challenges are resolved. However, considering the user's need to weigh price and reliability, he predicted that it would take more time for these products to establish themselves in actual mass production.

Although SiC can be used at low voltages, its adoption in the low-voltage range is slow because there are already so many devices available for the same position and the unit cost is relatively high. Instead, its strengths become prominent when high switching is required at high voltages. However, this does not mean that SiC is exclusively for high output. Committee Member Kim stated that SiC can be the answer for applications where high switching is required to reduce product weight even at low outputs, or to extract high output in confined spaces. Ultimately, the same device is utilized in completely different ways depending on the customer's requirements.

Device manufacturers do not draw lines based on application either. Commissioner Kim stated that Infineon does not define device boundaries by setting specific targets. He explained, "In some positions, GaN and SiC are considered together, while in others, SiC and conventional silicon MOSFETs are considered together." He described the current situation as having "a jumbled mix of positioning." In the past, devices were selected based solely on voltage and capacitance, but now, as complex demands—such as "high switching performance while keeping prices low"—increase, the range of options has broadened. Consequently, selection guides have become more complex, and this is precisely the point where users find it most difficult. Commissioner Kim said that he would sort out this part as much as possible during the webinar.

Efficiency depends on the system, not the motor.

The driving force behind the transition to electric vehicles is ultimately efficiency. According to Infineon's analysis, power consumption by electric motors in industrial settings will increase to approximately 9,500 TWh by 2030, accounting for around 30% of global electricity demand. Calculations show that increasing motor drive system efficiency by just 10% could meet the power needs of all electric vehicles worldwide by 2030.

However, efficiency is not determined by a single motor. Actual efficiency is the product of the efficiencies of the drive, motor, and load unit. If the drive is 97%, the motor 90%, and the load unit 60%, the system efficiency drops to approximately 52%. Even if the motor is upgraded to the IE5 grade, the savings are diminished if the load or drive is a bottleneck. Under the International Efficiency Rating (IEC 60034-30), IE3 allows for motor losses of up to 12%, while IE5 requires less than 7%. Nevertheless, as of 2025, only 26% of the approximately 820 million motors installed worldwide will be driven in conjunction with a drive. This means there is still significant room to improve efficiency.

The reality on the ground is another matter. Commissioner Kim stated that while there is a common consensus in the industry to increase efficiency, neither companies nor society are taking this seriously. Therefore, it is necessary for experts to separately persuade and explain its value to users. Energy efficiency regulations were also viewed as having a strong indirect nature, naturally following the process of reducing consumption, rather than directly affecting drive design.

Servo, general purpose, and medium voltage: different requirements require different designs.

Industrial drives are not a single unit. They are divided into servo drives requiring precision control, general-purpose drives (GPDs) that handle a wide range of applications such as pumps, fans, and process automation, and medium-voltage drives responsible for auxiliary power generation equipment and large-scale facilities. Even looking at the output range alone, there is a significant gap: servos range from 50W to 315kW, general-purpose drives from 100W to 1,250kW, and medium-voltage drives from 250kW to 36MW.

The difference is immediately apparent in the component specifications. Overload characteristics are a prime example. Applications that do not require high torque at low speeds, such as pumps and fans, are sufficient with a 110% overload (light load), whereas applications requiring low speed and high torque, such as industrial automation, demand a 150% overload (medium load). Servo drives may also need to instantaneously withstand up to 300% of their rated current during acceleration. The same applies to switching frequency. For general-purpose drives, the frequency drops from 4–8 kHz at less than 100 kW to 2–4 kHz at 100 kW or more. The servo is based on 4~8kHz, but is raised up to 16kHz with derating.

Commissioner Kim’s assessment is that the overall framework of the requirements has not changed significantly. Higher efficiency, faster responsiveness, and higher reliability are required just as they were in the past. What has changed is the level of sophistication. Increasing power density means increasing efficiency, which leads to a demand to maintain output while making products lighter and smaller. With the addition of IoT and AI capabilities, these requirements have become even more demanding.

"SiC has permeated even railways"... Replacement is just a matter of time

Then, where is the turning point where SiC begins to seriously displace IGBTs? Commissioner Kim stated that it is difficult to draw a line based on power output ranges. This is because SiC products are already being released that cover the entire spectrum, from low-wattage to high-wattage levels. He explained that even for the same product, there are specific areas where demanding requirements arise due to confined spaces or environmental and mechanical constraints, making it difficult to differentiate them based solely on power output.

He expressed a clear stance on the direction. "I dare say that an era will come when SiC replaces silicon IGBTs in the medium-to-high capacity range, even up to the megawatt level," he said. He also added advice that it is better to secure opportunities for early development and learning with currently available products before that era arrives.

The basis for this is that the scope of application is expanding into unexpected areas. SiC is already being incorporated into the railway industry, a conservative sector that is most strict about safety. Commissioner Kim drew a line, stating that the reluctance to adopt the technology is not due to concerns that safety may decrease with higher switching rates. He explained that the greater factor is the anxiety over what might happen during long-term operation, as the device has only recently been used in applications. He noted that overseas countries began applying the technology before Korea and already have references from years of operation.

This anxiety is also evident in the questions customers ask when reviewing SiC. Commissioner Kim stated that engineers mostly ask about reliability, while the purchasing department asks about price. Since FAEs primarily deal with engineers, emphasis is placed on reliability verification. When an unfamiliar requirement arises, they collaborate with Infineon to verify and test it. "It is burdensome for engineers to take responsibility for things they are not familiar with, but in such cases, you can think of it as the vendor taking on the responsibility," he explained.

WBG demand fueled by humanoids and robots

As investment in humanoid and collaborative robots surges, the demand for drives is growing accordingly. Commissioner Kim viewed robot drives as essentially no different from existing industrial designs. He noted that the core requirements of high efficiency, high responsiveness, and high reliability are applied as is, with only the standards continuously rising.

The variable that determines the design direction is the drive method. Most robots are powered by batteries. To extend battery life, efficiency must be high, and to extend life, the device must be lightweight. If it is heavy, it requires that much more power. This is why the conditions of being lightweight and having a long lifespan lead to a demand for wide-bandgap devices. Commissioner Kim stated that Dabo is actively recommending GaN-based materials to the robotics sector, and that several robotics companies are currently reviewing and testing them. He added that specific examples of collaboration are difficult to disclose due to concerns regarding client confidentiality.

Component as Solution… 22kW Reference Design

Infineon emphasizes that it can fill most of the active components within the drive with its own product family. The power stage includes discrete components such as IGBTs, MOSFETs, SiC MOSFETs, and GaN HEMTs, as well as power modules and CIPOS-based Intelligent Power Modules (IPMs), while EiceDRIVER supports the gate drivers. The sensors consist of the XENSIV series, which measures current, magnetic position, speed, and pressure, while control is handled by XMC, AURIX, and PSoC microcontrollers (MCUs) and memory. OPTIGA embedded security and AIROC wireless solutions are added to this. In its own materials, Infineon claims to be the market leader in power switches for industrial drives.

The embodiment of this strategy is the 22kW general-purpose drive reference design 'REF-22K-GPD-INV-EASY3B'. It is designed for 3-phase 400V AC systems and has a rated output of 22kW. The design entry barrier was lowered by integrating the IGBT power module (FP100R12W3T7_B11), XENSIV current sensor (TLI4971-A120T5), gate driver IC (1ED3131MC12H), 1.7kV SiC MOSFET (IMBF170R1K0M1), and MCU (XMC4800·XMC4300) onto a single board.

Dabo plays the role of bundling these components into solutions for Korean customers. Committee Member Kim stated that just as Infineon handles the entire spectrum from sensors to power devices and MCUs, Dabo is capable of supporting everything from controllers to power stacks. He cited expertise in motor drive applications as a key differentiator. He explained that since a significant number of engineers come from a motor drive background, including himself, the company is strong in the control and power stack sectors. There are also quite a few customers who do not use the recommended packages exactly as they are. If a customer who has experienced field issues with a specific package requests a different one, Dabo proposes an alternative product that delivers similar performance.

Technology is changing, but there is no one to keep up.

Commissioner Kim’s diagnosis is that it is people who ultimately determine the speed of the transition. The Korean industrial drive market remains highly dependent on Japanese and European finished products from companies such as Mitsubishi, Yaskawa, and Siemens. Domestic OEMs face many hurdles to overcome if they wish to transition to in-house design, and the first one he cited was a shortage of manpower.

He stated that it is not the difficulty of the technology that is being avoided. He explained that although the technology has been practiced for a long time, it must be relearned whenever trends change, and while new personnel need to adopt and carry on those trends, the influx has ceased. As he put it, "Only those who have been doing it keep doing it," and the average age of current employees is rising. A vicious cycle is in place where a shortage of manpower leads to reduced investment, and the inability to keep up with trends results in declining competitiveness compared to rival nations. He expressed concern that while some leading countries possess a workforce in power electronics equivalent to Korea's population, Korea has few applicants and lacks investment.

He also shared the reality that while the Davo Research Center receives investment in development and research through corporate support, the workload for a single researcher is excessive due to a lack of recruitment. Although the inflection point where devices transition from IGBTs to SiC and GaN has become clear, the heavier challenge remaining for the Korean market is how to cultivate the talent to follow that inflection point to the end.