Selecting the Right Gate Driver for SiC MOSFETs

Article By : Gina Roos

While SiC offers a range of benefits, it also introduces several design challenges that can be solved by choosing the right gate driver.

Silicon carbide (SiC) MOSFETs have made significant inroads in the power semiconductor industry thanks to a range of benefits over silicon-based switches. These include faster switching, higher efficiency, higher voltage operation and higher temperatures that yield smaller and lighter designs.

Those attributes have led to a range of automotive and industrial applications. But wide-bandgap devices like SiC also introduce design challenges, including electromagnetic interference (EMI), overheating, and overvoltage conditions, which can be solved by choosing the right gate driver.

Since gate drivers are used to propel power devices, they are a critical piece of the power puzzle. One way to ensure an optimized design using SiCs is to carefully consider the choice of the gate driver. Simultaneously, it requires a close look at the key requirements of a design – efficiency, density, and, of course, cost – because there are always trade-offs, depending on application requirements.

Despite the inherent benefits of SiC, cost remains a roadblock to adoption. When looking at SiCs versus silicon on a part-comparison basis, they will be more expensive and difficult to justify unless designers look at the total solution cost, according to power IC manufacturers.

Let’s first address the applications, benefits and trade-offs of SiC versus silicon MOSFETs or IGBTs. SiC FETs offer lower on-resistance (thanks to a higher breakdown voltage), high saturation velocity for faster switching and a three-fold increase in bandgap energy. The result is higher junction temperature for improved cooling. Meanwhile, a three-fold improvement in thermal conductivity translates into higher power density.

There is industry agreement that low-voltage Si MOSFETs and gallium nitride operate in the <700-V range. Above that is where SiC comes into play, with a bit of overlap in the lower power range.

SiC mostly replaces silicon IGBT-type applications over 600 V and above 3.3 kW, and even more so at about 11 kW. That’s the sweet spot for SiC, delivering high-voltage operation, low switching losses and a higher switching frequency power stage, said Rob Weber, product line director at Microchip Technology.

Those attributes allow the use of smaller filters and passives while reducing cooling needs, Weber said. “We’re talking system-level benefits versus IGBTs, which is ultimately a reduction in size, cost and weight.

“You can reduce the losses up to 70 percent, for example, at a 30-kHz switching frequency, and that is a result of some of the different characteristics of silicon carbide in terms of the breakdown field, electron saturation velocity, bandgap energy and thermal conductivity,” he added.

Source: Microchip Technology (Click on image to enlarge.)

A key engineering benchmark is efficiency, which results in levels of improvement. Weber said another SiC cost-benefit metric is emerging: system-level advantages over IGBTs.

“With silicon carbide you can operate at a higher switching frequency that enables you to have smaller external components that surround the immediate power stage like filters, for example, which are big, heavy magnetic devices.” They also “operate at higher temperatures or operate cooler due to the lower switching losses, replace a liquid-cooled system with an air-cooled system and shrink the size of the heat sink,” he explained.

Reduction in component size and weight also translates into lower cost, meaning SiC provides more than just greater efficiency, Weber asserted.

In a part-to-part price comparison, however, SiC remains more expensive than traditional silicon-based IGBTs. “The SiC module will cost more from every manufacturer, but when you look at the total system, the SiC system costs are lower,” said Weber.

In another example, a Microchip customer was able to achieve a 6 percent reduction in system costs when using a SiC MOSFET.
Once designers make the switch to SiC, they also need to consider trade-offs. Power semiconductor manufacturers agree there are “secondary effects” like noise, EMI and overvoltage that must be overcome.

“When you’re switching these devices faster, you potentially create more noise which will translate into EMI,” said Weber. “In addition, while SiC is great at higher voltage it is also much less robust than IGBTs for short-circuit conditions and [when] you’re getting variability in your voltage. You get overvoltage conditions, which is causing some designers to use higher voltage-rated SiC devices so they can control the overvoltage [and overheating] better.”
This is where the selection of the gate driver plays a key role. SiC brings unique requirements for characteristics such as supply voltage, fast short-circuit protection and high dv/dt immunity.

Selecting gate drivers

Selecting the right gate driver for SiC switches requires a new mindset when comparing them to silicon-based devices. The key areas to look at include topology, voltage, bias and monitoring and protection features.

Previously, it was acceptable to use a sequential approach in selecting the gate driver, said Weber. “Prior to SiC, you’d pick the IGBT first, then the gate driver second, then the busbars and the capacitor” he said. “It’s totally changed. You have to look at the whole holistic solution that you’re building and those tradeoffs at every step instead of this sequential approach that you have with IGBTs. It’s been an education for a lot of customers.”

In addition, there are a variety of SiC gate drivers that differ in terms of features, integration and price, targeting simple to more complex designs.

Topology, power level, protection and functional safety requirements along with the SiC device generation in use dictates the kind of driver needed for an application, said Lazlo Balogh, system engineering lead for high voltage power at Texas Instruments

For example, a non-isolated driver, which could require much extra circuitry, Balogh said.

There are also isolated drivers that can handle negative bias and isolation issues but will still need some type of system monitoring. That’s true up to devices that offer further integration such as monitoring and protection circuitry and functional safety for automotive applications, he added.

“The checklist for deploying SiC the right way is to look at the topology and what kind of devices you have to drive, then pick the gate driver, optimize the bias, figure out what kind of protection is needed and then optimize the layout,” Balogh said.
Among the roadblocks hindering SiC adoption is the need for special packaging at higher switching speeds to eliminate source inductance.

That’s typically done with the Kelvin source connection, said Balogh. “Source inductance can be nasty and cause a lot of ringing and additional power losses because it slows down the switching action.”

Source: Texas Instruments (Click on image to enlarge.)

“This is where the layout engineer becomes your best friend because you really have to look at the layout to mitigate the ringing and optimize it for high speed switching,” said Balogh. This includes minimizing the trace inductance and separating the gate and power loops along with proper bypassing of switched current paths and broad frequency band by selecting the right components, he added.

What’s really critical is connecting the driver to the switch, Balogh said. Designers must connect the ground of the driver directly to the source of the power switch because of stray inductances that can increase switching losses, he said.

Suppliers agree it is possible to use standard drivers to control SiC devices. Still, designers must determine the magnitude of the tradeoff, which usually entails extra circuitry or larger external devices. For example, one way to reduce ringing and overvoltage when using a standard driver is by increasing the size of the gate resistor.

TI’s Balogh said other considerations include protection functions, under-voltage lockout, higher-frequency operation, faster switching and die hot spots, all of which can determine power losses, EMI and size.

In addition, the extra circuitry usually takes much more space than a dedicated SiC. Hence, higher-end designs often opt for a dedicated SiC core driver, which take account for faster switching, overvoltage conditions and issues around noise and EMI, he said.

“You always can use a standard gate driver but you have to complement it with additional circuitry and usually that is the trade-off,” Balogh said.

Read the full story here at Electronic Products.

Gina Roos is the editor-in-chief of Electronic Products.


Leave a comment