Replacing Mechanical Circuit Breakers with SiC FETs
December 1, 2021
This blog post was first published by United Silicon Carbide (UnitedSiC) which joined the Qorvo family in November 2021. UnitedSiC is a leading manufacturer of silicon carbide (SiC) power semiconductors and expands Qorvo's reach into the fast-growing markets for electric vehicles (EVs), industrial power, circuit protection, renewables and data center power.
In the world of engineering, there is an old adage – "If it moves, it will break." We all know that anything mechanical like a fan or a relay is normally the first thing to fail and in critical systems, you need a program of pre-emptive maintenance and change-out of these items 'just in case.' It's worse still when the mechanical component runs with a high normal stress level, then must react reliably in an emergency, such as a contact breaker in series with an EV battery.
In this situation, the running current can be hundreds of amps or thousands under short-circuit conditions when the breaker has to open. The voltage is high, typically greater than 400VDC and it spikes higher still due to connection inductances when the fault current is interrupted. The voltage causes arcing, which can vaporize the breaker contacts and persist because it's DC and there is no zero crossing to extinguish the arc, as you would have with AC. Making and breaking is slow as well, in the order of tens of milliseconds, potentially allowing damaging let-through energy under short-circuit conditions. As the breaker ages, it also gets slower and more lossy. All in all, it's a tough life for a high-current mechanical circuit breaker, so they must be built robustly and sometimes with exotic methods to clear arcing, such as generating puffs of compressed gas or using magnetic 'blowout' coils.
Naturally, solid-state versions of circuit breakers (SSCBs) have been designed as alternatives and have been fabricated using just about all the available semiconductor technologies, from MOSFETs to IGBTs, SCRs and IGCTs, and they solve the arcing and mechanical wear problems nicely. Their big downside, though is voltage drop – an IGBT, for example, might lose 1.7V at 500A, producing an embarrassing 850W of dissipation. An IGCT is potentially lower but physically very large. MOSFETs don't show the 'knee' voltage of an IGBT, but instead, exhibit on-resistance. To improve on an IGBT, this RDS(on) would need to be less than 3.4 milliohms with a voltage rating better than 400V, which isn't currently achieved with a single MOSFET. Many in parallel would do it, but then costs spiral and double again if you need bi-directional capability. Electromechanical circuit breakers are not cheap but would still have the cost edge.
Does SiC make a difference?
So, do the new wonder technologies of wide band-gap semiconductors close the gap? Silicon carbide switches offer around 10x better on-resistance for the same die area as silicon and can cope with double the maximum temperature with a much better thermal conductivity to lead any heat away. This opens up the possibility of paralleling enough die in a small package to improve on an IGBT as an SSCB and the SiC FET is an ideal candidate. This cascode of a SiC JFET and Si-MOSFET is easy to drive and has the best RDS(on) x A figure of merit amongst current switch technologies. As an SSCB demonstrator, UnitedSiC put six of their 1200V dual-gate die in parallel and achieved 2.2 milliohms resistance for 1200V and 300A rating in a SOT-227 package. In tests, the prototype safely interrupted a fault current of nearly 2000A with the waveforms shown.
Figure 1. A SiC FET SSCB interrupting nearly 2000A safely
If the internal JFET gate is brought out to a separate pin, this allows more direct control of edge rates in fast switching applications and provides effectively selectable normally-off or normally-on operation, which can be desirable in some applications such as SSCBs. The ability to bias the JFET gate slightly positive also improves on-resistance a little. There is another feature that appears though – above about 2V positive, the channel is fully conducting, and the gate appears as a forward-biased diode. Now, if a fixed low current is injected, the actual knee voltage of the diode has an accurate relationship to die temperature. This can be measured and used for fast over-temperature detection and even long-term state-of-health monitoring if the trend of temperature is logged.
The trends are in the right direction for SiC FET SSCBs to displace electromechanical versions
SiC FETs have opened up the SSCB application for higher currents with losses that will only decrease as the technology advances. Paralleling devices to achieve ultimate loss-parity with mechanical circuit breakers is potentially possible, with costs not necessarily a deal 'breaker,' as die improve and fewer are needed for a given resistance. SiC wafer costs are also set to decrease by half in the coming years, along with economies of scale from the ballooning market for circuit breakers driven by EV sales. Factor-in maintenance and changeout costs of an electromechanical solution and the argument is even more compelling.
There is another engineering adage – "If it ain't broke, don't fix it." I'd say, don't wait for it to break — consider a SiC FET SSCB for a worry-free solution.