For the EV on-board charger (OBC), higher charging rates have led to an increase in power to 22 kW. But at the same time, the OBC must fit into the available installation space and must not overheat. This is already feasible today with gan-fets.
Gan components bring high frequencies to applications like on-board chargers in EVs. This means that passive components are smaller and losses are significantly reduced. (image: texas instruments)
In order to increase the penetration of electric vehicles (evs), it is important to address the main concerns of users, which are range, charging time and vehicle affordability. Automakers around the world are therefore striving for more battery capacity and shorter charging times, but without allowing the dimensions, weight or cost of the components to go through the roof.
evs’ on-board chargers allow users to charge the battery directly from the AC mains – whether at home or at a public or commercial charging station. Technically, there is a lot going on here. The need to increase charging rates has led to an increase in power from 3.6 kw to 22 kw, but at the same time the OBC unit must fit into the available installation space and be carried by the vehicle at all times, which must not reduce the operating radius. At the same time, efforts are underway to increase the power density of the on-board chargers from the current level of less than 2 kw/l to more than 4 kw/l.
The importance of switching frequency
The OBC is basically nothing more than a switched power converter, whose weight and volume is largely accounted for by passive components such as transformers, inductors, filters and capacitors with the corresponding heat sinks. The use of smaller passive components is made possible by increasing the switching frequency, but this also increases the losses in the switching components, which include power mosfets and igbts, for example.
Since there is less surface area to dissipate heat as the dimensions shrink, the losses must decrease to keep the devices at the same temperature. Increased power density requires a simultaneous increase in switching frequency and efficiency. This is precisely the challenge with which silicon-based power devices have their problems.
Increasing the switching speed, i.e. the speed at which the voltages and currents at the terminals of the relevant device change, generally reduces the energy losses that occur during switching. This speed increase is necessary because otherwise the frequency cannot be raised above a certain upper limit. Improvements are possible with power devices that have a lower parasitic capacitance between their terminals and are used in carefully designed circuits with low inductance.
Moving away from silicon
Power devices manufactured with wide band gap semiconductor materials such as gallium nitride (gan) or silicon carbide (sic) are characterized by significantly lower capacitance due to their physical properties at comparable on-state resistance and similar breakdown voltage. Higher critical breakdown electric field strengths (ten times higher for gan than for silicon) and the higher electron mobility (over 33 percent higher for gan than for si) effectively enable both lower on-resistance and lower capacitance. As a result, gan and sic fets are principally capable of operating at higher switching speeds with lower losses than occur with silicon.
Special advantages of gan
The low gate capacitance of gan allows for faster on and off switching during hard switching operations, which reduces transient losses. The gate charge figure of merit of gan is 1 nc∙Ω. Gan enables a low output capacitance and thus allows fast drain-source transitions during soft switching, especially at low load and switching speeds. magnetizing currents. A typical gan FET, for example, has an output charge figure of merit of 5 nc∙Ω, compared to 25 nc∙Ω for silicon. This gives designers the ability to use short dead times and low magnetizing currents, which in turn is necessary to increase switching frequency and reduce cyclic losses.
Unlike si- and sic-based power mosfets, a gan transistor does not have a body diode in its structure, so there are no blocking delay losses. This enables new, highly efficient architectures such as bridgeless totem pole power factor correction stages even at several kilowatts, which was not possible with silicon devices.
All these advantages give designers the opportunity to use gan to achieve high efficiency at significantly higher switching frequencies (see Fig. 1). gan-fets with a nominal voltage of 650 V can be used for applications up to 10 kw, e.g. B. AC/DC power supplies for servers as well as high-voltage dc/dc converters and obcs for electric vehicles (for up to 22 kw when connected in parallel). Sic devices, offered with nominal voltages up to 1.2 kv and high current capability, are well suited for EV traction inverters and large grid-tied three-phase converters.
Design challenges at high frequencies
Typical rise and fall times of 10 ns require careful design when switching several hundred volts to eliminate the undesirable effects of parasitic inductance. Common-source and gate-loop inductance between FET and driver are critical in several respects. The common-source inductance limits the steepness of the drain-source voltage edges (dv/dt) and the current edges (di/dt), which reduces the switching speed and increases the overlap losses during hard switching as well as the transition times during soft switching.
The gate loop inductance limits the steepness of the gate current edges (di/dt), which also reduces switching speed and results in higher overlap losses during hard switching. Other negative effects include increased sensitivity to unwanted turn-on due to the Miller effect, which poses a risk for higher losses and a design challenge to minimize excessive voltage stress on the gate insulator. the latter affects reliability unless appropriate countermeasures are taken.
Developers can use ferrite beads and damping resistors, but these reduce the switching speeds and thus run counter to the desired frequency increase. Gan and sic devices are fundamentally adaptable to high frequencies, but there are additional technical challenges to overcome at the system level to fully realize their benefits. A carefully designed product that takes into account the aspects of ease of use, robustness and design flexibility will accelerate the acceptance of the technology.
Gan FET with driver, protection functions and PWM
The fully integrated 650 V automotive grade gan fetches from texas instruments are designed to take advantage of the efficiency and switching frequency benefits of gan without the disadvantages of design and component selection. The integration of the gan FET and associated driver in a tight QFN package with low inductance values results in a significant reduction in parasitic gate loop inductances, eliminating concerns of possible gate overload and parasitic turn-on due to the Miller effect. At the same time, the low common-source inductances ensure high switching speeds, which in turn reduces losses.
In conjunction with the C2000 series of real-time microcontrollers (z. B. Tms320f2838x or tms320f28004x), the LMG3522R030-Q1 enables switching frequencies of more than 1 mhz in power converters, reducing the size of inductive components by 59 percent compared to si- and sic solutions.
Proven drain-source slew rates of over 100 V/ns enable a seven percent reduction in switching losses compared to discrete fets, while adjustability from 30 V/ns to 150 V/ns allows trade-offs between efficiency and EMI to mitigate risks to subsequent product design. Integrated overcurrent protection provides robustness, and new features include PWM temperature sensing for active power management, condition monitoring and ideal diode mode. These features of the LMG3522R030-Q1 make adaptive dead time control unnecessary. The 12 mm × 12 mm top-cooled QFN package also allows for improved thermal management.
With more than 40 million reliability hours and a FIT rate of less than 1 for a ten-year lifetime, tI’s gan devices deliver the ruggedness demanded by automotive manufacturers. Produced on commonly available silicon substrates using existing process nodes and 100 percent in-house manufacturing, they offer definitive supply chain and cost advantages that differentiate them from other technologies based on sic or sapphire substrates.