Advances in wide-bandgap semiconductors for vehicles

Wide-bandgap (WBG) semiconductor switches are now commonly designed into new power conversion products in many application areas. Cost reductions make them increasingly viable for commercial products and in aerospace, for example, their natural radiation hardness gives them a valuable edge.
The electric vehicle market however, has been slow to pick up the technology for traction drives, on-board chargers and DC-DC converters. There are several reasons: the industry is conservative, with technology changes introduced only slowly; costs must be very low but performance and reliability very high, suiting established technology; and, perhaps surprisingly, WBG devices lose their major advantage of small magnetics in the important application of motor drives.
WBG switches can operate at MHz rates, reducing DC-DC or AC-DC converter transformers and chokes down to miniature dimensions, but the magnetic in an electric vehicle (EV) drive is the motor itself. This means that drives might switch at typically less than 10kHz, so dynamic losses are very low, even with IGBT technology. High frequencies also risk high electromagnetic interference (EMI), especially when the connection from drive to motors may be a few feet of cable.
‘MOSFETs are better than IGBTs.’ Sometimes, that is.
An obvious advance 10 years ago might have been to use silicon MOSFETs, which had become ubiquitous in other power conversion markets. Without the benefit of their high frequency ability, they have practically been excluded from high-power, high-voltage EV drives switched at low frequency. Even more striking, under these conditions, MOSFET resistance-based conduction losses can be worse than that of IGBTs of the same voltage and current class, with their fixed saturation voltage.
IGBT vs. MOSFET conduction losses

WBG devices must also prove their robustness in an EV drive application. Short circuits from stalled motors are possible along with high voltage transients, shock, vibration and high temperatures.
SiC MOSFETs can make a difference
Things are changing, however, with the latest generations of WBG devices, particularly silicon carbide (SiC). The crossover point in Figure 1 up to which conduction losses of SiC MOSFETS are lower than IGBTs, has moved higher. As unit costs decrease, it is feasible to parallel MOSFETs for a proportional reduction in conduction loss. Two parallel MOSFETs actually reduce total conduction losses by more than a factor of two. As each die runs cooler and its on-resistance reduces, the paralleling IGBTs only share the dissipation without substantially reducing loss. Given the use of SiC, switching frequencies can be pushed up to an extent without significant increase in dynamic losses and this does give smoother motor control and less mechanical wear.
Through the use of WBG devices, further consequential benefits arise:
- Drive circuitry takes significantly less power.
- The inherent body diode of a SiC MOSFET can be used in some cases for current “commutation” where an IGBT would need an extra added parallel diode.
- Bi-directional energy flow can also be far more efficient with MOSFETs than with IGBTs.
- MOSFETs generally have the advantage in motor control that their voltage drop at light loads can be very small. IGBTs have a “knee” voltage at turn-on, which means there is always volt drop at any load while consequent dissipation and smooth control at light load is compromised.
- IGBTs typically use lossy “snubbers” to absorb switching energy that would otherwise result in damaging voltage overshoots, whereas SiC devices have far less switching energy and therefore need to dissipate much less in snubbers for a given overshoot, making them smaller and lower cost.
WBG edge rates can be embarrassingly fast
Another hurdle to the introduction of WBG devices is actually one of their advantages in other applications: Even if switched at low frequency, their edge-rates can be phenomenally fast, measured in nano seconds (see Figure 2). In motor drives, the waveforms therefore must be deliberately slowed to avoid excessive EMI and interactions causing phantom turn-on. This phenomenon occurs when a device spuriously conducts in the presence of high dV/dt with consequent shoot through and potentially catastrophic damage. For a practical design using WBG devices, the old layout practices typical for IGBTs are inappropriate and designs must be re-thought from the ground up.
SiC MOSFET typical switching waveform

The advantages of WBG switches in the on-board charger (OBC) and auxiliary DC-DC converters in EVs are more tangible, with the prospect of higher efficiency with smaller magnetics size, weight and, consequently, cost. In some cases, the performance of SiC devices enables the use of topologies that would have been impractical with IGBTs and silicon MOSFETs, such as the ultra-high efficiency bridgeless totem pole power factor correction stage for the OBC. Designs using SiC for the OBC and DC-DC auxiliary converter are now commonplace and increasing familiarity with the technology is opening the door for use in the traction inverters themselves. As WBG devices begin to show measurable increases in overall system efficiency, a virtuous circle emerges of lower losses yielding more range from a battery charge along with smaller and lighter heatsinking, which in turn is less load on the vehicle, allowing yet further range.
The gap is closing, but SiC MOSFETs remain more costly than IGBTs
WBG devices can yield enough performance advantage in EV power conversion to measurably increase range, help promote EV sales and more widely to make a difference to the environment. As with any new technology, costs of switches using SiC have been, and are likely to remain, higher than an IGBT with the same headline voltage and current ratings. For example, if we compare an IGBT, a silicon MOSFET and a SiC MOSFET rated 600/650V and 60A, currently, through distribution, the IGBT is $1.42, the Si MOSFET $5.30 and the SiC MOSFET $20, all at the 1,000 pcs rate. However, as discussed, significant savings are made elsewhere with the SiC device: Less heatsinking, smaller snubbers and fewer additional components such as parallel diodes.
In the on-board charger and DC-DC converter applications, the savings with SiC are even more, with much smaller and lower cost magnetics and, as time passes, the SiC device will continue to drop in price, closing the gap.
What price do you put on reduced emissions and increasing the attractiveness and viability of an EV with a longer range?

