More SiC FET options means more flexibility for cost-effective solutions

  • Dec 02, 2021
  • UnitedSiC

Everything in life is a balancing act, and that can make decisions difficult. If money is involved, it brings in another dimension – if I buy an EV, will the extra cost payback? Over what time? What is the value of saving CO2 emissions? Which model has the best residual value? Deciding factors might be subjective and dynamically change, but when you are selecting a semiconductor in the design of an EV power converter, you would hope it’s a bit more scientific.

If you have made a wise choice and decided on using wide band-gap SiC FETs from UnitedSiC, the trade-offs are there to consider – number and rating of parallel devices, switching frequency, operating mode, efficiency target, acceptable junction temperature rise, apportioning of conduction and switching loss, and cost, to name a few. The choices are often narrowed down by external factors: for example, a totem-pole power factor correction stage at say 3.6kW, used in an EV charger, will typically generate 400V and operate around at 60kHz with around 20% inductor ripple current in continuous conduction mode. Given these conditions, 750V SiC FETs in the ‘fast’ leg are an excellent choice for their low losses and are available down to 6 milliohms on-resistance.. In real life, cost is always an issue, so can we trade off some losses for a lower cost part? UnitedSiC makes this easy to evaluate with the online FET-JET Calculator which allows selection from a wide range of power conversion topologies and operating conditions, calculating switching and conductor losses, and temperature rises for different devices. Numbers of paralleled parts can be selected and the heatsink performance specified.

To give the widest choice of options, UnitedSiC has made the Gen 4 750V devices available at a variety of on-resistance performance points, eight in total, from 6 to 60 milliohms. Plugging the most relevant of these into the FET-JET Calculator with our chosen conditions gives the plot below, with some interesting results.

Moving from an 18-milliohm (UJ4C075018K4S) to a 23-milliohm device (UJ4C075023K4S) shows no efficiency decrease, because although conduction loss has increased, switching loss has decreased more. There is however a device cost saving of 20% over the lower resistance part. Perhaps the 33-milliohm part (UJ4C075033K4S) is a good choice, dropping efficiency by 0.1%, dissipating 36W more, but with better than 40% reduction in the switch cost. Junction temperature increases by around 6°C for the same heatsinking but is still only about 102°C. The 60-milliohm part (UJ4C075060K4S) is less than half the cost of the best SiC FET considered, for the penalty of 22W extra dissipation and a junction temperature of 122°C. Better heatsinking can be considered for a cost trade-off with device type and temperature rise, but the extra size and weight is a down-side in EV applications.

Other options can be considered, if two of the 60 milliohm types are paralleled, the total resistance can be cut in half, while adding switching loss but cutting the overall thermal resistance, resulting in a lower junction temperature rise and a lower increase in temperature-dependent on-resistance. The FET-JET Calculator is your friend here – try multiple devices at the different on-resistance levels and heatsink options. You may even find a tipping point where the lowest loss combination with the higher cost devices actually reduces heatsinking to the level where liquid cooling could be eliminated, for example, outweighing the extra switch cost by far.