How do high performance passives help make the most of silicon carbide efficiency gains?
Global consumption of electricity is steadily increasing. Electricity’s share of global energy consumption is also steadily rising, according to consultancy Enerdata. These macro trends are reflected in our own experience. Who knew, 30 years ago, that we needed motors to juice our lemons or wind up our car windows? But apparently, we do, and we choose electricity to drive them. Who would have predicted, 20 years ago, that we would be installing photovoltaic (PV) panels on our roofs to supplement grid energy? But we are.
There’s always a financial and environmental cost to generating, distributing, and applying electrical energy, and so it makes sense to use it as efficiently as possible. The power electronics business has been making incremental improvements to power devices, circuit designs, packaging techniques, and control algorithms to this end for decades. What has changed recently is that wide bandgap semiconductor devices, based on materials such as silicon carbide (SiC), have matured to the point that they are a viable alternative to Si devices. The shift to SiC devices will bring a step-change increase in the efficiency with which power is transmitted, stored, and used in a wide range of applications.
SiC’s wide bandgap gives SiC devices a breakdown voltage ten times that of Si. This makes it easier to use them in high voltage switching circuits. SiC can also be doped to have a lower resistance than Si, making the conductance losses in a SiC device’s channel less than those of a Si equivalent. The high breakdown voltage also means that SiC devices can be driven with a higher gate-source voltage than Si, which further reduces channel resistance.
The wide bandgap of SiC means that, although semiconductor performance usually derates with increasing temperature, it has enough headroom to operate effectively at higher temperatures than Si can sustain. SiC also has much greater thermal conductivity than Si. This means that SiC circuits can run at much higher temperatures than Si without cooling, where this is practical. If cooling is necessary, it takes less work to cool a SiC device than a Si device. This enables simpler, smaller, lighter, lower cost cooling systems.
SiC devices can also be switched more quickly than Si equivalents. This means that passive components, such as the inductive and capacitive parts used as energy reservoirs in DC power-conversion circuits or for electromagnetic interference (EMI) control, can be smaller. Higher frequency switching also creates less powerful ripples in a converter’s output, simplifying the conditioning circuitry necessary to clean up the DC output waveform.
The enhanced conductance of SiC means that a SiC device can be smaller for the same performance than a Si part, reducing its cost. A smaller device has less self-inductance than a larger part and will also experience less capacitive coupling to its package. Both these effects enhance its ability to switch at higher frequencies.
Although SiC parts may be more expensive than equivalent Si parts, at the moment, they can enable systemic savings that more than balance this out. SiC parts can be used to make denser power-conversion devices than possible with Si, which use simpler circuits with smaller passive components, less aggressive cooling strategies, and smaller enclosures, achieving higher reliability and a lower total cost of ownership. And crucially, they can be configured to use electricity more efficiently than Si alternatives.
Companies such as Infineon have made substantial investments in the development of SiC production capacity; device, package, and driver architectures; product ranges; and engineering support.
For example, it has launched a range of CoolSiC Schottky diodes that can operate at up to 1200V as standalone parts or up to 1700V in modules. The parts don’t experience any reverse recovery charge, reducing switching losses, and can operate at up to 175ºC. This makes them more efficient than Si-based diodes and enables higher switching frequencies and greater power densities. Applications are likely to include in server and telecom power supplies, solar energy capture and storage, motor drives, lighting, and welding, among others.
The graph below shows the switching characteristics of three types of 600V diodes, with the SiC Schottky diode having the cleanest waveform.
Infineon has also developed a range of 650V CoolSiC MOSFETs optimised for low losses and high reliability. The parts have low capacitances and optimised switching behaviour at higher currents. They also have an intrinsic fast body diode with low reverse recovery charge, which is robust to the challenges of hard commutation in motor drives. Infineon claims the parts also have superior gate-oxide reliability, excellent thermal behaviour, and are well protected against avalanche conditions. To make them easy to use, they work with standard driver circuitry.
A shift to SiC devices creates opportunities for more efficient power systems, at the cost of making greater demands on the supporting passive components, especially the capacitors and inductors. To capture SiC’s advantages, these ancillary parts must be uprated to sustain higher operating voltages and switching frequencies of more than 100KHz, and to operate for long periods at temperatures of up to 200ºC. Ideally, they should also be small, so that the power-density advantages of the SiC parts are not wasted by having to combine them with bulky passive components.
Infineon’s 7.5kW three-phase motor driver evaluation board (Source: EBV/Infineon)
The 7.5kW three-phase motor driver evaluation board from Infineon (pictured below) uses CoolSiC MOSFETs. It shows some of the critical passives used in such circuits, including the EMI filter (3), the large DC bus capacitors, balancing resistors, and DC bus connector (5), the high-side MOSFET drivers (7), and the position of the power module (6), which is mounted on a fan-cooled heatsink on the back of the board. It’s clear that the passive devices have a considerable impact on the size of the board.
The key challenge for EMI suppression in SiC circuits is that the fast rise times of SiC switching waveforms create more high-frequency harmonics than the slower switching of Si parts, and also spread the harmonics across a wider frequency range. Designers therefore have to choose which material to use in the core of their inductive components carefully. Manganese-zinc (Mn-Zn) cores are effective at relatively low frequencies, while nickel-zinc (Ni-Zn) parts are more effective at higher frequencies. Nanocrystal choke cores work across the whole range of operating frequencies, but above a certain saturation current, the inductance of nanocrystal cores is lower than Mn-Zn or Ni-Zn parts.
Toroidal coils help keep the magnetic flux in check and reduce stay inductances, making efficient use of the magnetic field and minimising insertion losses. Using a divided, rather than undivided, winding strategy on a coil will minimise the capacitances between the start and end of the winding, improving high-frequency performance. A similar effect can be achieved on gear coils by dividing the bobbin on which the coil is wound.
Examples of toroidal and gear choke coils (Source: KEMET)
It’s also possible to make chokes that reject both common-mode and differential-mode noise. This creates options to reduce the size of the differential-mode choke, to omit it entirely and increase the size of its related capacitor, or to adjust the relative size of the two parts to achieve a different design tradeoff.
Capacitor choices are also affected by a shift to SiC switching. Film capacitors can sustain high operating voltages and ripple currents, have low leakage currents and are stable with frequency. Critically, they also have self-healing proprieties that enhance system safety. Drawbacks include their size, performance at high temperatures, and cost.
Materials choices in the construction of film capacitors have an impact on performance. Polypropylene film capacitors will work at high voltages, have low dissipation and leakage, strong self-healing properties, and a good frequency response up to 100KHz. Their main disadvantages are a low melting point and a lower capacitance per unit volume than some alternatives.
Pulse or resonant film capacitors are used in resonant stages and snubber circuits, applications that need to support high ripple currents and very fast waveform edges. This type of capacitor has thicker metallisation than typically used in DC link or EMI capacitors, giving it a low equivalent series resistance (ESR) and so a high current capacity.
KEMET has a number of parts optimised in different ways for these roles. The R75-H series has a thick metallisation layer and is rated to operate at 125ºC for 2000 hours, rather than the 105ºC for which such parts are usually rated. The R76-H series has a double layer of metallisation that enables it to operate at higher AC and DC voltages. It is meant for use in resonant and snubber circuits.
Film capacitors have disadvantages in DC link applications because of their relative bulk and their high-temperature characteristics. For DC link and DC filter applications, KEMET offers C4AQ-M capacitors that have good charge storage density and are available in a low-profile version. Although there is a slight tradeoff between the size of the parts and the ripple currents they can sustain, the relatively small size of the C4AQ-M capacitors reduces their ESR and hence reduces the number of filtering capacitors needed later in the circuit.
Multilayer ceramic capacitors (MLCC) can also be used as DC links in fast switching-based inverters, because of their high operating voltages, high ripple-current capability, good high-frequency behaviour, and ability to withstand high dV/dt rates. The key figure of merit here is the ratio of supported ripple current to device capacitance. KEMET offers KC-Link MLCC parts, which use Class 1 dielectric materials to achieve a high operating voltage, and a high ripple current to capacitance ratio that makes them useful for resonant tank circuits.
KEMET has also developed a new MLCC packaging technology. KONNEKT packaging doesn’t use a lead frame and enables multiple devices to be connected together using a form of sintering. The resultant packaged parts can be surface mounted to a PCB using the same assembly processes as for a standard MLCC, stacked on top of each other, or mounted vertically and then connected side to side to reduce ESR, improve thermal performance and enhance high-frequency filtering. KONNEKT also saves PCB space, making it a good match for dense SiC power-conversion circuits.
Features of KEMET’s KONNEKT™ technology
Another way to implement DC link capacitances is with aluminium electrolytic parts. These have high capacities, but offer poor performance at high frequencies, relatively short operating lifetimes, and limited ripple-current capacity. Moving from liquid to solid polymer dielectrics, as KEMET has for some of its ranges, stops the capacitor’s performance changing over lifetime as the electrolyte does not evaporate. Solid polymer electrolytics also have a lower ESR than liquid dielectric equivalents, improving their ability to withstand high ripple currents, and better performance at high frequencies.
The promise of SiC devices is that they can make our use of electrical energy more efficient. Capitalising on that promise takes careful attention to the choice of the SiC devices themselves, and to the supporting passive components that enable their effective operation.
KEMET has developed a range of capacitors and magnetic components specifically to meet the requirements of SiC topologies, offering smaller footprints for more board savings, lower application costs, best-in-class performance and long lifetimes. Find out more about the range, or if you're ready to take the next step with your project, get in touch with our team of technical specialists to discuss your requirements.
