Automotive GaN FETs enable higher operating frequencies and robustness

“In order to accelerate the adoption of electric vehicles (EVs), automakers around the world are eager to increase battery capacity and reduce charging time, while maintaining vehicle size, weight, and and device costs remain the same.

“

In order to accelerate the adoption of electric vehicles (EVs), automakers around the world are eager to increase battery capacity and reduce charging time, while maintaining vehicle size, weight, and and device costs remain the same.

Electric vehicle on-board chargers (OBCs) are experiencing rapid growth, allowing consumers to charge batteries directly from AC power at home, at public charging points or at commercial outlets. To increase charging speed, the OBC power level has been increased from 3.6kW to 22kW, but at the same time, the OBC must be installed within the existing mechanical enclosure and must always be carried in the vehicle so as not to affect the driving range. OBC power density will eventually increase from below 2kW/L today to above 4kW/L.

The effect of switching frequency

OBC is essentially a switch-mode power converter. It mainly consists of passive components such as transformers, inductors, filters and capacitors, as well as heat sinks, which make up the bulk of its weight and size. Increasing the switching frequency requires shrinking the size of passive components. However, higher switching frequencies result in higher power dissipation in switching elements such as power metal oxide semiconductors (MOSFETs) and insulated gate bipolar transistors.

Downsizing requires further reductions in power loss to keep the device temperature constant, as the area for heat dissipation is reduced with downsizing. Both switching frequency and efficiency need to be increased to achieve this higher power density. This presents a huge design challenge that is difficult to solve with silicon-based power devices.

Increasing the switching speed (the rate of change of voltage and current between device terminals) will radically reduce switching energy losses. This process is essential, otherwise the actual maximum frequency will be limited. Power devices with low parasitic capacitance between terminals (well-designed in low-inductance circuit routing) can do this successfully.

Better performance than silicon devices

Power devices built using wide-bandgap semiconductors such as gallium nitride (GaN) and silicon carbide (SiC) have unique physical properties that significantly reduce capacitance while ensuring equivalent on-resistance and breakdown voltage. Higher breakdown critical electric fields (10 times higher for GaN than silicon) and higher electron mobility (33 percent higher for GaN than silicon) effectively enable lower on-resistance and lower capacitance. This allows GaN and SiC FETs to inherently operate at higher switching speeds and with lower losses than silicon.

The advantages of GaN are especially obvious:

・GaN’s low gate capacitance enables faster turn-on and turn-off during hard switching, reducing crossover power losses. The gate charge quality factor of GaN is 1nC-Ω.

・GaN’s low output capacitance enables fast drain-source transitions during soft switching, especially at low load (magnetizing) currents. For example, a typical GaN FET has an output charge figure of merit of 5nC-Ω, compared to 25nC-Ω for a silicon device. With these devices, designers can use the small dead time and low magnetizing current, which are necessary to increase the frequency and reduce cycle power loss.

・Unlike silicon and SiC power MOSFETs, there is no body diode in the GaN transistor structure itself, so there is no reverse recovery loss. This enables new high-efficiency architectures such as totem-pole bridgeless power factor correction to become feasible at several kilowatt-hours, which was not possible before with silicon devices.

All these advantages enable designers to use GaN to achieve high efficiency at higher switching frequencies, as shown in Figure 1. GaN FETs rated at 650V can support applications up to 10kW, such as server AC/DC power supplies, EV HVDC/DC converters, and OBCs (up to 22kW when stacked in parallel). SiC devices can supply voltages up to 1.2kV and have high current-carrying capabilities, making them ideal for electric vehicle traction inverters and large three-phase grid inverters.

Figure 1: GaN outperforms all technologies in enabling ultra-high frequency applications

High Frequency Design Challenges

At switching frequencies in the hundreds of volts, the typical 10ns rise and fall times require careful design to avoid parasitic inductive effects. Common source inductance and gate loop inductance between the FET and driver have the following key effects:

・Common-source inductance limits drain-to-source transient voltage (dV/dt) and transient current (dI/dt), which reduces switching speed, increases overlap losses during hard switching, and increases transition time during soft switching.

・Gate loop inductance limits gate current dI/dt, which reduces switching speed and increases overlap losses during hard switching. Other negative effects include increased susceptibility to Miller turn-on effects, risk of additional power loss, increased design challenges to minimize gate insulator voltage overstress, and reduced reliability if overstress is not properly mitigated.

As a result, engineers may need to use ferrite beads and damping resistors, but these slow down switching and defeat the goal of increasing frequency. While GaN and SiC devices are inherently suitable for high-frequency operation, system-level design challenges need to be overcome to fully exploit their benefits. The adoption of this technology can be accelerated if a well-designed product is available that combines ease of use, stability, and design flexibility.

GaN FETs with integrated driver, protection and power management

Texas Instruments (TI)’s 650V fully integrated automotive GaN FETs deliver the high-efficiency, high-frequency switching benefits of GaN without the associated design and device selection pitfalls. The GaN FET and driver are tightly integrated in a low-inductance Quad Flat No-Lead (QFN) package, which greatly reduces parasitic gate loop inductance, freeing designers from gate overstress and parasitic Miller turn-on effects, and the common source inductance is very low. Low, can achieve fast switching and reduce losses.

The combination of the LMG3522R030-Q1 with advanced control features in C2000™ real-time microcontrollers such as the TMS320F2838x or TMS320F28004x enables switching frequencies above 1MHz in power converters, which are faster than existing silicon and SiC solutions. Magnet size has been reduced by 59%.

Compared to discrete FETs, the demonstrated drain-source slew rate greater than 100V/ns reduces switching losses by 67%, while its adjustability is between 30-150V/ns, a trade-off between efficiency and EMI, Thereby reducing the risk of downstream product design. Integrated current protection ensures robustness and adds many new features, including digital PWM temperature reporting for active power management, health monitoring, and ideal diode mode (as provided by the LMG3525R030-Q1), allowing designs Humans do not need to implement adaptive dead time control. The 12mm x 12mm top-cooled QFN package also enhances thermal management.

TI GaN devices have passed more than 40 million hours of device reliability testing and have a failure rate of less than 1 over a 10-year lifetime, meeting the durability expected by automakers. TI GaN is built on generally available silicon substrates and fabricated using existing process nodes at all in-house fabrication facilities, offering definite supply chain and cost advantages unlike other technologies based on SiC or sapphire substrates.