Many power conversion tasks involve taking an input voltage and turning it into a different output voltage, which has often been stabilized, converted from AC to DC (or vice versa), and then given galvanic isolation. This is what happens when you use an AC mains adaptor to charge your phone with DC, or an inverter to turn DC from a car battery into an AC mains supply.
These are one-way conversions, but rising interest in alternative energy schemes and electric vehicles (EVs) means there is growing interest in making it possible for power to flow efficiently in both directions. This might be useful, for example, in a photovoltaic cell installation in which excess DC solar power is fed into the AC grid during the day, and then when the local storage batteries are depleted they can be charged from the grid through a bidirectional converter/inverter. Another example is in EVs, in which a bidirectional DC-DC converter drops the 400V traction-battery voltage down to 12V to drive auxiliary equipment, but converts 12V back to 400V if the traction battery’s charge falls too low.
Such bidirectional energy flow to and from batteries needs careful management. The 12V lead-acid cells widely used in cars need a controlled current flow until they are fully charged, and then a trickle current. In contrast, the 400V arrays of lithium-Ion cells used for electric vehicle traction need a carefully controlled constant voltage.
Building a bidirectional converter
There’s little point in enabling energy to flow to and from batteries in this way if large amounts of it is lost in the conversions made in each direction. This means using highly efficient power converters, and that usually involves increased circuit complexity. It is possible to go halfway, by linking two unidirectional converters in ‘anti-parallel’ with sensing circuits that energize one or the other. This may be easy, but it means double the component count, cost, and, significantly in vehicle applications, weight.
A more elegant and cost-effective approach is to configure power components to operate in both directions.
Consider isolated, bidirectional energy exchange between a 400V traction battery and a 12V auxiliary battery. The preferred power-stage topology for conversion from 400V to 12V is a full bridge, which limits switch stress and uses the isolation transformer efficiently. The output stage is a bi-phase rectifier, which minimizes circuit stress and component count, and is shown in Figure 2a.
It may not be obvious how 12V power can then be converted back into 400V, but Figure 2b shows that the 12V output diodes can be replaced by synchronous rectifiers and switches Q1-4 can be turned Off, effectively leaving just their body diodes D1 - D4 in circuit. Read the circuit from right to left, though, and it looks familiar: a current-fed push-pull power stage with a full-bridge output rectifier. The power components and magnetics are the same, but are used differently to set the direction of energy flow. Q1 - Q4 can also be actively switched as synchronous rectifiers to improve efficiency, although at 400V the gains of doing so may be limited.
Making efficient bidirectional power conversion a reality demands sophisticated control chips, which are usually located on the low-voltage side, so they can then conveniently take start-up power from the 12V battery. If the high-voltage side of the converter uses a phase-shifted full-bridge topology, the control IC can easily pass gate-drive signals across an isolation barrier using simple transformers. As the signals are of a fixed width, just phase-shifted with respect to each other to provide regulation, the transformers do not face problems with variable pulse-widths causing different peak positive and negative gate voltages.
It’s possible to do a similar exercise with an AC-DC converter, with an active bridge rectifier configured as the legs of an inverter for reverse energy flow. One modern approach to this is to use a totem-pole rectifier and power factor correction stage, which can be easily reconfigured as an inverter, as shown in Figure 3.
Wide-bandgap devices in power conversion
Wide-bandgap (WBG) silicon carbide (SiC) and gallium nitride (GaN) semiconductors are now available to replace silicon devices. Acting as switches, they offer lower On resistances, faster switching rates, and higher temperature operation than silicon. Discrete SiC diodes don’t suffer from reverse recovery charge, and can work at high voltages. SiC switches have fast body diodes, and are robust, with high avalanche energy and excellent short-circuit current ratings. There are SiC version of JFETs, MOSFETs and cascodes – a normally-Off combination of a Si-MOSFET and SiC JFET with close to ideal switching characteristics (Figure 4).
The cascode configuration of a Si-MOSFET and SiC JFET
WBG devices are particularly suitable for bidirectional converters where efficiency and size are concerns. Fast switching edges result in low losses when operating at high frequencies, which in turn allows the use of much smaller passive components.
If switches Q1 - Q4 of the circuit in Figure 2b are configured as synchronous rectifiers, rather than turning them Off and allowing their body diodes to act as rectifiers, they could be implemented using high-voltage Si-MOSFETs. However, these devices would have a greater conduction loss than SiC, and poor body-diode reverse-recovery characteristics that can lead to device failures. On the other hand, SiC cascodes that are rated for high voltages nonetheless have the body-diode characteristics of a low-voltage Si switch, with very low forward voltage drop and fast recovery, enabling low-loss operation.
If the full-bridge Q1 - Q4 is being used as a power stage it will often be run in a resonant mode with phase-shift control. This approach offers the best efficiency above a few hundred watts and achieves zero-voltage switching when the switch is turned on, with an external inductance resonating with the transformer’s capacitance and the switch’s output capacitance COSS. SiC devices, particularly cascodes, have a very low value for COSS, so designers can use a relatively small external inductance to achieve resonance, which helps to increase the duty-cycle range and/or the maximum switching frequency.
SiC and bidirectional power conversion complement each other
SiC devices work well in bidirectional power-conversion strategies and have the right characteristics to enable low losses. A wide range of SiC diodes, SIC JFETs and SiC FET cascodes are available from UnitedSiC, backed up by a wealth of helpful application data.