Let’s begin by pointing out that there are some myths associated with silicon carbide (SiC) MOSFET technology today. One is that it’s taken as an expensive technology, and it may be true that in some voltage classes, SiC MOSFETs are more expensive than similar silicon devices. However, in all cases we have to look at the overall system cost, and there are already many applications today where the system cost can be at least equal or even lower than a solution based on silicon components.
By Alix Paultre, Editor Power Electronics News (based on a technical paper presented by Christian Felgemacher, ROHM Semiconductor)
There are three main advantages for silicon devices. One is the lower specific on-resistance of these devices, operating at much higher frequencies, and at higher temperatures, provided that certain issues at packaging level can be addressed.
The figure above is an overview of some the reliability tests done on SiC devices typically being done for ROHM trench MOSFETs. All the typical tests: high temperature reverse bias, high temperature gate bias, storage tests at high and low temperature, and tests involving humidity or thermo cycling. We can see that silicon carbide MOSFETs undergo reliability tests that are very similar to those for silicon MOSFETs and IGBTs.
A simple SiC solution for an auxiliary power supply
If we look at any typical power electronic conversion system, we usually have some kind of input voltage and input current, also an output, and there is a main converter. In addition, you will also need some form of auxiliary power supply to operate all your control ICs, to provide voltage levels for your gate drivers, for sensors, for human interfaces, etc. Typically, that’s a small power converter that could reach power ratings, depending on application, in the range of one-tenth or one-hundredth of a watt.
This auxiliary voltage supply is usually separated from the main power path and needs to provide an isolated low voltage output, usually from a high-voltage input. Typically, some type of flyback converter is used to address this issue. Usually, you have some kind of retractive file, for example, rectifying 3-phase AC to DC voltage, and you have a flyback converter to convert that to 12 or 24 volt.
If we look at typical industrial power supplies, the AC input range could range anywhere from 210 to 690 volts AC, and this gives you a rectified voltage of up to 1,000 volts DC. Then you also have some reflected voltage from the flyback circuit, and you have some overshoot on the turn of the MOSFET. If you look at these voltages and think about what the maximum voltage of the MOSFET in this situation has to be, you realize that you need about 1,300 volts. Obviously, this depends on the input voltage rating that you’re working with and then you want some design margin, and then you end up with the requirement that you need to use a device rated for 1,500 volts or above.
Silicon vs. SiC
Typically, if you use a six-Ohm silicon device, you end up with a very high gate charge Qg and that results in high driving losses in this device, or you have leakage currents for some of these MOSFETs; especially if you operate them on high junction temperatures. Due to the on-resistance, you also have significant losses.
Another approach would be to use a series connection in some form of 800V silicon MOSFETs. This, of course, makes the whole control of the circuit much more complex; you either need static voltage balancing that works, or you need to think about how you control the turn-on and turn-off of those devices. One further option would be to use a two-switch flyback topology, but that introduces additional complexity. You’ll need an isolated gate driver for the high side device and a relatively large heat sink.
Why should we use a silicon carbide MOSFET in this application? If you look at the device characteristics in the figure abve, the silicon carbide device is actually rated for 1,700 volts, whereas the silicon devices are all rated for 1,500 volts. The most significant advantage here is that the silicon carbide MOSFET gives you a much lower Rds(on) in the range of about one Ohm. Also, if you compare this with the other devices, comparing Rds(on) and gate charge, you obtain a much reduced capacitance compared to the silicon devices.
There is one slight drawback in this case, which is the high gate-resistance of the silicon carbide device, and this will put some demands on your gate drive, such as low-impedance to quickly turn on and off the device. The underlying advantage is the trade-off that you can have a significantly reduced Rds(on) times area value for silicon carbide MOSFET in relation to silicon MOSFET. This helps you giving you a low Rds(on) device, but at the same time at very low capacitance.
The advantage is that you need only a single switch, and the solution is an isolated package, and we also often adjust the power MOSFET but also suitable control IC, which targets specifically this application using silicon carbide MOSFETs. Though, the control IC, in this case, is well matched to the power device. Also, in this application, the heat sink is not mandatory if you operate at less than 40 watts.
The figure below shows both the control IC and the power device on an eval board, targeting an input voltage range between 300 and 900 volts DC. There is also a rectifier on this board, so you can also operate this at AC. This is for 12 volt DC output and can operate at up to 40 watts output power without using a heat sink. In some cases, it might even be possible to exceed this power if you attach a small heat sink to certain components.
The switching frequency is variable because we use quasi resonant switching, so it’s between 90 and 120 kilohertz. On the board, you can see the discrete THT device and SiC MOSFET, highlighted in red in the middle on the front side of the component, just above the flyback transformer. Also, on the other side of the PCB, the control IC BD768xFJ, which is used to control the whole circuit, but also to drive the silicon carbide MOSFETs. This board, for example, is available as an evaluation board.
The key features of the control IC, that it is optimized for this application using and driving a silicon carbide MOSFET, it implements quasi-resonant DC/DC control for this application, and the maximum switching frequency is 120 kilohertz. There are also a load of other features, such as brown out detection, over voltage protection for the VCC supply of this device, soft start features, etc. The operating VCC range is between 15 and 27.5 volts, and the operating current that this IC draws is in the region of 800 micro amps.
The control IC quasi-resonant switching results in reduced effective switching voltage and turn off of the silicon carbide MOSFET, which further allows you to optimize the efficiency of the circuit. An auxiliary winding supplies the IC with a start-up circuit to allow initial supply. Aside from reducing the switching losses, this quasi resonant operation is also helping you to achieve lower EMI noise.
The next step
One approach would be to integrate this solution. For example, we could think about integrating the silicon carbide MOSFET and the controller in one package to further reduce the amount of components needed, and simplify the design. This is just an idea of what the next logical step could be.
An auxiliary supply solution using an advanced control IC like the BD768, together with a 1,700 volt silicon carbide MOSFET, is a very good alternative to other forms of auxiliary supply circuits, which today use series connections of silicon MOSFETs, or other complex topology to achieve the desired voltage blocking capability. If you take advantage of the SiC device benefits, you can realize system cost advantages on the system level.