Anup Bhalla, VP Engineering and Zhongda Li, NPI Manager at UnitedSiC
MOSFETs & Power MOSFETs, Power Conversion
@UnitedSiC #siliconcarbide #powerconversion #evecosystem #ev #electricvehicle #psd
Today’s hybrid and fully electric vehicles (EVs) are still in the early stages of the iterative process of innovation. Several generations of hybrid vehicles have been introduced to market and improved through a combination of technology advances and user feedback. A number of vendors now market competent EVs and although their ranges are limited, the reviews now tend to focus on when they’ll catch up with fossil-fuel cars, not whether or not they can. Analysts are now forecasting that perhaps 25 percent of all vehicles will be electrified by 2025, and that there could be up to a quarter of a billion EVs in use by 2040 (see Figure 1).
What’s interesting about the current EV market is that the development focus is largely upon the vehicles themselves, and to some extent their charging networks. But a successful EV ecosystem, with the credibility and robustness to supplant our century-long fossil-fuel habit, will demand systemic thinking and constant optimizations at, and between, every level of that ecosystem. Many of these optimizations will be enabled by a shift to more efficient power-conversion electronics, which will offer wide-ranging benefits.
Click image to enlarge
Figure 1: Electric vehicle predictions (source: International Energy Agency)
More efficient vehicle electronics
The most obvious application for this technology is in vehicle electronics. Many EVs are being designed with well-established silicon IGBTs and switching devices, sometimes even in discrete packages rather than the module format that has become popular for standard sub-circuits. There are lots of these devices in today’s EVs – perhaps up to 66 of them in a single Tesla vehicle – providing the powertrain inverter, and power switches for on-board battery chargers, inductive charging converters and auxiliary converters for driving legacy 12 V vehicle systems.
Using discrete IGBTs gives certainty, at the cost of excess circuit complexity and the loss of potential efficiency gains that could be made from moving to a more modern switch such as wide band-gap silicon carbide (SiC) devices.
Finding the sweet spot
Taking on a new technology such as SiC switches means making a systemic tradeoff between their performance advantages, the inevitable price differential with established technologies, and the cost savings made possible by the new device’s characteristics.
For example, as EV battery voltages rise to 800 V, designers may introduce SiC diodes and MOSFETs. If they push the switching frequencies of these devices to their theoretical maximums, the resultant circuits may be up to five times smaller and lighter than silicon alternatives, at the cost of higher switching losses at these high operating frequencies. Backing off the switching frequencies to around 50 kHz, and using the SiC devices in cascode arrangements, could deliver both a switching efficiency gain and enable the use of fewer, smaller components than an equivalent conventional silicon design.
The systemic optimization opportunity here is to consider all the gains possible from shifting to a device technology that is initially slightly more expensive, but which is expected to rapidly descend the price learning curve.
The greater efficiency of SiC devices at reasonable switching speeds will reduce the size of the supporting inductors, transformers, capacitors and other passive components, as well as their related heatsinks, relative to those used in a high-power IGBT automotive drive. This could mean, for example, a shift from using bobbin-wound inductors with fly leads that have to be hand-soldered onto circuit boards for the IGBT-based circuits, to small planar inductors, with windings formed from stamped metal or PCB traces, that can be surface-mounted for the SIC-based approach.
The extra cost of one type of component will soon absorbed by lower-cost manufacturing, smaller circuit sizes, greater reliability and repeatability of the resultant circuits, and so on. Moving the SiC parts from discrete packages into modules will create opportunities for further economies, in terms of lower mechanical and electrical connection costs and simplified cooling schemes.
More efficient chargers
Another place where more efficient devices can create opportunities for systemic optimizations is in charging circuits. Charging EV batteries from the mains means having to make an efficient conversion between AC and DC. This demands the use of complex conversion-circuit topologies and power-factor correction (PFC) techniques to enable the most effective transfer of energy to the battery.
One way to achieve PFC is to put 50 Hz or 60 Hz inductors in series with the power converters, at the cost of increased losses and larger circuit boards. An alternative is to use an active switching circuit that takes a rectified mains voltage and boosts it to a fixed DC level, while controlling the line current to be sinusoidal. This basic approach can lead to losses at high power because the circuit topology puts three diodes into the current path, each of which dissipates energy.
Replacing these diodes with silicon MOSFETs can reduce conduction losses, and further efficiency gains can be made by swapping out other diodes in the circuit topology for synchronous switches.
As ever in circuit design, though, there are further trade-offs to be managed. The MOSFETs have inherent body diodes, which can act as a rectifier during certain phases of the power conversion cycle and end up conducting the full output current. When the device is reverse-biased by the next phase of the switching cycle, a large reverse-recovery current flows, causing power dissipation and electromagnetic interference that cancels the efficiency gains made by replacing the original diodes.
The next optimization is to shift to SiC MOSFETs, which have low channel-conduction losses, operate at high speeds, and have a fast body diode. However, the forward voltage of the body diode can be 2.5 – 3 V, leading to high conduction losses.
One way around many of these issues is to use the SIC devices in a cascode arrangement, in which a high-voltage SiC JFET is coupled with a high-performance, low-voltage Si MOSFET in one package. The resultant device has low switching losses, due to its extremely low input, output and Miller capacitances and stored energy ratings. In addition, the body diode voltage drop is 1.5-2V,leading to lower conduction losses.
All these tradeoffs and more go into applying SiC cascodes in power-conversion circuitry. The systemic optimization question, though, is how good the SiC cascodes are in an onboard electric vehicle charger?
One way of assessing this is to see whether a SiC converter can meet the 80PLUS Titanium power-supply efficiency standard, which says it must exceed 96 percent efficiency at high line and half load. Meeting the standard demands that, if the mains-conversion stage is to achieve 97.5 percent efficiency, the PFC stage must exceed an efficiency of 98.5 percent. UnitedSiC makes a 1.5 kW demo board, using UJC06505K SiC cascodes switching at 100 kHz, which achieves this requirement with some margin (see Figure 2).
Click image to enlarge
Figure 2: Using SiC cascodes to achieve high power conversion efficiencies
The EV ecosystem optimisation opportunity
SiC devices, on their own or in hybrid cascode modules, can increase the efficiency of individual aspects of today EV ecosystem, such as in-vehicle electronics or charging circuitry. Once all these efficiency gains have been made, there will be other opportunities for systemic optimizations within the EV ecosystem.
For example, the uptake of EVs will create a large estate of highly distributed energy-storage units (i.e. the vehicle batteries) which could be used to sink, as well as source, energy. If EV owners can be persuaded to allow their vehicle’s batteries to be used in this way, perhaps nuclear power stations will be able to use them to absorb the energy they have to generate overnight at their base loading, which might otherwise go to waste. And perhaps grid operators will find more efficient ways to distribute energy by drawing on these localized energy stores, rather than having to install extra capacity just to service peak loads.