Silicon Carbide in Grid-Tied Applications
Article
Silicon Carbide (SiC) components have enabled higher efficiency and reliability for power delivery systems, especially inverters and active rectifiers that are grid tied and utilized nearly 24 hours a day, 365 days a year. As the industry pushes for more eco-conscious solutions, such as solar and electric-vehicle (EV) charging, the need for more efficient and more reliable components increases.
Thanks to advancements in SiC technology, grid-tied systems can benefit from higher power density, faster switching, lower operating temperatures, and lower overall costs. In addition to solar/wind/hydro and EV charging, other high-power applications such as industrial equipment, uninterruptible power supplies, and other power supply applications have driven the need for better power-factor correction (PFC) and additional capability such as bidirectional power flow.
Wolfspeed enables grid-tied applications through use of an innovative product offering and a toolset that includes reference designs and simulation software. This article will explore how devices and topologies can be best selected to match the application as well as what SiC components (discrete or module-based) and resources are available to assist the designer through development.
Device and topology selection of grid-tied applications
The increase in PFC for high-power applications has spanned many decades, starting with bridge (diode) rectifiers containing an LC component. Although this traditional configuration was very simple to implement, performance and size were key limitations. Currently, active boost PFC topologies dominate most applications and offer adequate performance at a reasonable cost but can provide only unidirectional flow (limiting its application base) and don’t quite meet today’s latest high efficiency standards.
Continuous-conduction–mode bridgeless totem-pole PFC configurations (Figure 1) have enabled lower losses, bidirectional power flow, and high-efficiency performance, which has allowed SiC technology to transform the marketplace for power conversion and delivery.
Si MOSFETs typically have a high reverse-recovery charge and time for the body diode (Qrr of about 10,000 nC and Trr of 500 ns for the latest 15mΩ/650V super junction Si MOSFET with fast body diode at high temperature), while the SiC MOSFET’s body diode can have much lower values such as 293 nC and 22 ns for latest 25mΩ/650V SiC MOSFET, respectively. Because totem-pole configurations require good body diode performance, the Si MOSFET isn’t a practical solution due to its Qrr and Trr during recovery after switching. And while the insulated-gate bipolar transistor (IGBT) is usable in a single-phase and three-phase totem-pole configuration, it usually comes with high switching losses and conduction losses which limits the maximum switching frequency. Multi-level IGBT converters alleviate some of these switching problems but have complex control requirements and often consist of many components and gate drivers, which drive system cost.
SiC enables simple and high-efficiency topologies such as single-phase and three-phase totem-pole configurations with added functionalities and at a higher power density. This lowers switching loss and allows for high-frequency applications, as well as increases power density and lowers the weight, size, and cost of the magnetics. With a lower RDS(on), efficiency can be higher than 99%. Also, because there are fewer components and a simpler form of control, it tends to lower overall system cost.
Wolfspeed’s SiC product portfolio supports a wide range of power (1 kW to 600 kW and higher) through use of both discrete components (for lower-power applications) and power modules (meant for higher-power applications). Voltage ranges of 650 V up to 1,700 V across 50+ products provide a high flexibility for topologies and meet specifications for power density, efficiency, and scalable solutions.
Depending on the application, discrete or module-based SiC packages can benefit the design based on power level and functionality. Discrete MOSFETs including a Kelvin source pin provide options for control of the source emitter that bypasses any package parasitics that might add inductive voltage during switching (see Figure 2 for an example of how the “KS” pin can be used for optimized control).
For higher-power applications, modules such as the Wolfpack family of devices can enable drop-in replacements of existing solutions while making use of optimized SiC performance with maximum current capabilities (due to minimized stray inductance). The BM2, BM3, FM3, and GM3 modules are contained in well-established footprints for easy sourceability, while the XM3 and HM3 modules are designed in optimized, custom modules for maximum performance and current capabilities.
Using SpeedFit with PLECS models allows the designer to find their fit through modeling and simulation of complete power electronics systems, including thermal/loss modeling and emulating datasheet characteristics. The models are based on real results and provide visual demonstrations of expected performance (see Figure 3).
Additionally, SpeedFit can be used for web-based simulations, including a full library of models that enable design flexibility with extra software. Common topologies are pre-loaded to enable quick simulations with accurate loss models that validate design choices.
When designing a gate driver for a SiC MOSFET, several key considerations must be made. Here are some typical specifications that should be met for Wolfspeed SiC MOSFETs:
- Isolation: Common-mode transient immunity greater than 100 kV/µs
- Insulation: Maximum working insulation voltage (VIORM)
- Driving capability: Up to 5 A
- Propagation delay time: 50 ns or better
- Channel mismatch time: 10 ns or better
- Active Miller clamp
- Gate supply voltage: 15 V/–4 V
To meet some of these standards, evaluation tools and resources are offered to assist in guiding the designer through topology and component selection. For instance, ADI AduM4146 by Analog Devices is a reference design that consists of an isolated, single-channel driver capable of driving Wolfspeed’s SiC MOSFETs, as well as the Si823Hx isolated dual driver from Silicon Labs (which can be used for half-bridge solutions). The UCC21710 by Texas Instruments can provide up to 10 A of driving capability. All of these come in off-the-shelf, plug-and-play platforms for evaluating Wolfspeed SiC MOSFETs.
Wolfspeed reference designs/kits and comparisons of SiC vs. IGBTs
The CRD-06600FF065N-K is a reference design configured as a single-phase 6.6-kW bidirectional on-board charger (OBC) for EV-charging applications, as well as general energy storage that may be “griddable.” It includes a totem-pole PFC (AC/DC) stage and an isolated bidirectional CLLC DC/DC stage, operating at a total of 96.5% peak efficiency and power density of 3.3 kW/L (see Figure 4). The design is a great starting point for EV and energy storage applications in development and allows for full evaluation and testing of Wolfspeed discrete components, as well as a set of design files including schematic/layout and related firmware.
A higher-power, three-phase bidirectional charging design with a high efficiency (98.5% at 22 kW) can also be referenced during development for guidance and discrete component evaluation. This design features Wolfspeed’s 1,200-V 32-mΩ SiC MOSFETs arranged in a two-level, six-switching PFC/inverter.
For higher-power applications that might warrant an array of MOSFETs, consider using Wolfpack’s power modules. For instance, a 25-kW active front end (AFE) might utilize an FM3-series module (containing six SiC MOSFETs) and configure them with the appropriate heatsinks, magnetics, gate drivers, voltage/current sensors, and a controller. Also, it’s always a good idea to add some safety-related functions such as soft-start, fuses, and EMI/EMC filters. Figure 5 shows a setup of this (MOSFETs are located in the “power stage”). This same setup was tested comparing the Wolfpack FM3 module with a Si IGBT, and it was found that SiC enables 5× faster switching (up to 100 kHz) and lowers losses by over 400 W (running at 25 kW with 480-V input and 800-V output). The performance boost seen with SiC also provides some additional benefits, such as smaller filter sizes, higher overall efficiency (up to 2% higher than IGBTs), and a reduced operating cost.
A similar comparison was conducted for a much larger wattage (200 kW) AFE/inverter. The CAB400M12XM3 SiC module enables dramatic reduction in losses, size, and cost compared with existing Si IGBT solutions due to the increased switching frequency and smaller components (magnetics, capacitors, and thermal cooling components). When comparing performance, it was observed that the total losses for the XM3 module were consistently about 1,000 W less when operating from 100 kW to 300 kW. Also, efficiency was about 0.75% higher at 200-kW output. XM3 kits are available for purchase as a core building block, including cooling, drivers, controllers, and sensing. Figure 6 shows a size comparison between the 200kW AFE/inverter utilizing SiC technology and an existing solution with Si IGBT components.
Magnetics, PCB/system layout, and how to optimize for maximum efficiency
One thing to consider during the development process of a high-wattage power supply is how selection of magnetics plays into density and efficiency. With higher-frequency operation (enabled by SiC), power density increases while power loss decreases on both the core and winding, which translates into higher efficiency. Generally, there are tradeoffs in terms of performance and system cost, but employing SiC MOSFETs help improve the performance and lower losses. When selecting an inductor for the application, consider the core material and its construction. Some poor constructed inductors have stronger “fringing effects” than others, which result in extra power loss due to eddy currents.
Figure 7 shows a table of several power inductors and a comparison of their parameters and performance. There are tradeoffs between core loss and DC bias, along with frequency range and capabilities.
Switching high energy can result in major dV/dt and di/dt changes on traces and nodes for a PCB. When conducting PCB layout, it is important to minimize the pad size of drain nodes to reduce coupling and parasitic capacitance, as well as keeping sensitive signals away from high dV/dt trace/nodes and magnetic fields, such as PFC chokes.
When high-energy drain planes are nearby gate traces/pads, parasitic capacitances can be formed that result in significant power loss. For example, a 1-cm2 PCB trace overlap in an 800-V bus switching at 100 kHz can result in up to 1.2 W of loss from 38 pF of parasitic capacitance (see Figure 8). The switching loss can be increased due to the larger gate charge, which reduces efficiency. Also, when stray capacitance is introduced around the gate (between source gate and drain gate), crosstalk can occur when there is a large ratio between them, resulting in more shoot-through and higher voltage spikes. The added capacitance on the gate signal (on the other side of the external gate resistor) can even enhance gate oscillation, which reduces overall reliability. All of these stray capacitances can be minimized by reducing trace lengths, minding the sensitive traces and potential for high dV/dt and di/dt, and carefully positioning power components as appropriate to avoid overlap between gate, gate-drive circuit, bias power supply for the gate drive, and the drain of the MOSFET. Those referenced designs mentioned above serve good examples of optimized layout practice for SiC applications.
For systems containing power modules, some of the same optimizations apply. When using a power module and bus bar operating at high switching speeds, it’s important to minimize total stray inductance, which maximizes efficiency. With IGBT based inverters, stray inductance doesn’t play such a critical role due to the limited switching speeds, but bus bar and capacitor design and selection should be optimized to take full advantage of SiC; see Figure 9 for a plot comparing the two and how stray inductance can be significantly lowered with smart design choices.
Additionally, when laying out the power modules with the rest of the system components, it’s better to keep the inductance between each module and capacitors equal while also allowing large surface areas to help dissipate heat. And it’s best to incorporate planes of laminated copper for the DC busbar (as opposed to strips).
Conclusion
SiC discretes and power modules from Wolfspeed can bring huge system-level benefits to grid-tied converter applications, including higher voltage, faster switching, increased power density and current capabilities, and an overall boost in system efficiency while reducing BOM costs for passive components. Wolfspeed has a wide portfolio of parts suited for different applications and web-based tools to quickly help customers evaluate parts for their systems, as well as several reference designs that accelerate time to market and provide more confidence to the designer.