An introduction to EV charger design: the power stages 

At heart, DC electric vehicle chargers are just AC-DC power supplies, but the application is quite novel – users only pay for energy supplied and any lost in conversion inefficiency is down to the charger operator. The initial capital expense has to be paid off and maintenance minimized over a long expected service lifetime, so low cost, reliable and efficient equipment is necessary.

There are other practical differences from a typical AC-DC power supply – the output of an DC EV charger is high voltage and needs to be variable to suit different batteries, with a charging regime of timed sequential constant current and constant voltage. Power levels are also high, with the ultra-fast chargers rated in fractions of a megawatt. The option for bi-directional energy flow is increasingly another consideration.

AC-DC and DC-AC power conversion in conservative industries such as Utility AC supplies and CAV/rail traction have dealt with these technology issues over decades, but only as small parts of much larger systems. These industries have therefore found it practical to use older, established technologies such as IGBTs for semiconductor switches with consequent low operating frequency and large, costly, magnetic elements.

DC EV chargers have design challenges

For a DC EV charger, design engineers have been challenged to reduce losses, cost and size and have therefore explored more recent technologies that give a performance edge in these respects. In many power applications, silicon MOSFETs have replaced IGBTs, enabling much higher switching frequencies and they can operate as switches or synchronous rectifiers in the same circuit, allowing bi-directional energy flow. Si-MOSFETs are commonly available with voltage ratings to 1200V, so suit the typical DC-link and output voltage of the charger. However, the devices are resistive when turned on, so at higher power and higher consequent current levels, the dissipation in each device exponentially increases. They can be paralleled for lower loss, but the cost and complexity spirals. There is always a power level therefore where Si-MOSFET viability starts to diminish. Another issue is their inherent body diode – some conversion topologies force the diode to conduct and store charge in the dead time between hard ‘on’ and ‘off’ states and this causes extra dissipation when the diode is subsequently reverse biased, with a high spike in ‘recovery’ current and losses as the charge is cleared. The effect scales directly with frequency so limits the other benefits of switching faster.

Wide band-gap devices (WBG) in Silicon Carbide (SiC) and Gallium Nitride (GaN) are hailed as the future of power switches, as they have relatively lower dynamic and static losses than silicon and can operate theoretically at much higher temperatures, although this is practically limited by packaging. They are not a panacea though, as both SiC and GaN have more particular gate drive requirements than silicon, the SiC body diode is fast but has a high forward drop and GaN has no body diode but exhibits a reverse voltage drop during dead time that depends on the gate drive voltage. GaN also has no protective avalanche effect, so device manufacturers have to rate their parts at relatively low voltages to give an adequate operating margin before catastrophic breakdown.

SiC devices are now commonly seen however and an optimum design will often be a combination of Si- and SiC-MOSFETs and SiC diodes. A typical three-phase charger powertrain outline might look like Figure 1.

Here, a ‘Vienna rectifier’ stage provides mains rectification and power factor correction, followed by a full-bridge ‘LLC’ converter. The Vienna rectifier allows switches rated at half the DC-link voltage, so Si-MOSFETs could have an advantage here on cost and on-resistance, and 650V rated devices could be used up to three-phase 480VAC input. The circuit is only uni-directional however without significant further complexity. The LLC stage could safely use 1200V SiC MOSFETs up to a DC link of about 1000V, as the full-bridge arrangement clamps the maximum voltage seen by any device to the DC rail. For Lower voltage DC-links, as seen for example in single phase AC inputs, GaN devices are an option. For uni-directional operation, diodes could be used for the output with some extra conduction loss and SiC types would be suitable with their absence of reverse recovery effect.

Chargers are usually modular

The trend in DC EV charger design is for modularity, where higher power is achieved by stacking or paralleling conversion stages. This can have the advantage of reducing semiconductor stresses and when implemented with clock interleaving, EMI and stress on associated passive devices can be lower as well. The concept is taken further with complete charger sub-units generating around 25kW assembled in multiples for higher power, up to the current maximum seen of around 350kW. As each sub-unit is autonomous, they can be switched in as required for best overall efficiency and a single failure will only reduce potential maximum power, rather than disabling a complete charger.

Diving deeper

There are a lot of options to consider when it comes to the power stages of EV charging stations. Here we’ve only scratched the surface. However, if you’re looking for detailed design advice and component recommendations on the individual stages, download the first volume of the EV Charging Infrastructure Designbook. In it we’ll cover:

• Charger types
• AC and DC station architecture
• Power stage functions
• Technologies and conversion topologies 

Alternatively, if you’re ready to take the next step with your design, reach out to our EV charging experts to discuss your requirements in detail.