George continues his article series looking at all aspects of the basic structures that make semiconductors work. In Part 6—his final article of the series—he builds on last month’s exploration of MOSFETs by this time examining MOSFETs designed specifically to handle high power.
Last month’s article was dedicated to metal oxide semiconductor field-effect transistor or MOSFET designed to handle signals and power of relatively low level. Now let’s look at MOSFETs designed specifically to handle high power. Among the advantages of power MOSFETs are their high switching speed and good efficiency. At operating voltages less than 200V, the power MOSFET is the king. It is found in various applications from switching power supplies and motor controllers to analog amplifiers.
Power MOSFETs come in three major categories: N-channel enhancement, P-channel enhancement and N-channel depletion—with the N-channel enhancement being the most popular and achieving the lowest RDSon. P-channel depletion type power MOSFETs theoretically exist but I haven’t found one commercially available. MOSFETs offer high input impedance and a forward voltage drop decreasing with temperature. That ensures even current distribution. As a result, power MOSFETs can be paralleled to achieve the desired current handling capability. Keep in mind that—like all other MOSFETs—power MOSFETs also contain the intrinsic bulk diode.
Vertical Diffused MOS (VDMOS)—also known as Double-Diffused MOS (DMOS)—is the most prevalent power MOSFET structure today. Its cross-section is depicted in Figure 1. With the source electrode on the top and the drain electrode at the bottom of the device, the current flow through the device is vertical. The P wells and N+ regions are manufactured by a process called double diffusion—hence the “double-diffused” in its name.
Unlike most signal devices that are planar, the structure of most power devices is vertical. That’s because the current handling capability depends on the channel width and the vertical structure allows the die area to be minimized, which also keeps the cost low. Furthermore, the transistor has a cellular structure where essentially a number of transistor cells work in parallel. The most common cell shape is hexagonal, introduced by International Rectifier (now Infineon Technologies), With that in mind, they gave the name Hexfet to their line of power MOSFETs. For use in some analog applications, such as high-end audio amplifiers or microwave/RF, power MOSFETs with lateral structure (LDMOS) are manufactured.
SELECTING POWER MOSFETs
To select a power MOSFET, you need to decide on several characteristics. For starters—especially for switching applications—RDSon, the drain-to-source resistance when the device is fully turned on—is very important. It determines the voltage loss across the MOSFET and, therefore, its power dissipation. Devices with RDSon in the range of milliohms are commonly available.
You also need to make sure the maximum gate-to-source and drain-to-source voltages are not exceeded. Typically, the maximum gate-to-source voltage VGS is about 20V, but many power MOSFETS are fully on at a VGS of 5V or less. Increasing the drive voltage beyond that level makes no sense. It does not decrease the RDSon, but it does age the MOSFET reducing its life. For the maximum drain-to-source voltage VDS, I always try to select a device with at least double the operating voltage. Figure 2 shows the typical power MOSFET I-V characteristic. It is not dissimilar to the enhancement type signal MOSFET as discussed in Part 5 of this article series (Circuit Cellar 354, January 2019).
Obviously, not exceeding the maximum drain-to-source current IDS is just as important for the reliable operation of the transistor. Specification sheets contain graphs defining the safe operating area within which the MOSFET can perform without being damaged.
Dynamic behavior—in other words, switching behavior—of the MOSFETs is guided not only by their RDSon but also by their intrinsic capacitances and reverse recovery characteristics. The input capacitance CISS is the sum of the gate-to-source capacitance CGS + gate-to drain capacitance CGD. The output capacitance COSS is the sum of the drain-to-source capacitance CDS + the gate-to-drain capacitance CGD. And, finally, the reverse transfer capacitance CRSS equals to the gate-to-drain capacitance CGD.
To turn the MOSFET on and off, the driver must be able to overcome the gate charge, which can be expressed as an integral as follows:
To turn the saturated device off, the driver needs to absorb in a given time—also known as the storage time—the excess reverse recovery charge. Numerous gate drivers are commercially available to suit just about any MOSFET driving requirement. Figure 3 shows a simple power MOSFET switch.
While there are many possible configurations, Figure 3a shows major concerns the circuit designer has to address. To begin with, resistors R1 and R2 are there to protect the sensitive gate. As a voltage divider, R1 and R2 ensure the gate voltage VGS is adequate to turn the MOSFET fully on, while at the same time it makes sure the VGS does not exceed the safe limit. R2 also ensures that, should the driving circuit become open, the gate will not float or become subject to static charges. The Zener diode D1 is not always needed, but if an analysis reveals a failure mode that could result in VGS exceeding the allowed maximum, D1 will safely clamp it below that level.
The requirement for including the freewheeling diode D2 in the circuit depends on the characteristic of the load. Inductive loads—such as relays or motors—do require it to dissipate the back electromagnetic force (EMF) generated when the load is disconnected. The back EMF, depending on the load characteristics and how fast the MOSFET disconnects, can reach hundreds of volts, exceeding the device’s breakdown voltage. But don’t be fooled by “purely resistive” loads. Just their wiring can introduce a considerable inductance into the circuit.
Complementary combinations of N-channel and P-channel MOSFETs, as shown in Figure 3b, are popular in half-bridge and full-bridge configurations to drive loads such as motors, when polarity reversal is a requirement. Here, once again freewheeling diodes need to be employed, but are not shown for simplicity. The bridge driver has an additional feature: It can short-circuit the motor armature and cause a reverse current to flow through the windings. That acts as an efficient electrical brake. In some applications, such as trains for example, the EMF is used to recharge storage batteries.
Another issue I want to bring to your attention is that, in a half or a full bridge configuration, you must make sure that in the complementary pairs, the two transistors are never turned on at the same time, essentially acting as a crowbar across the power supply. While Figure 3b is representative of CMOS logic and can be used in some low power applications, in high power circuits all four MOSFET gates have individual inputs to ensure they are switched at a correct sequence. For that you can design or purchase an appropriate driver circuit or use four separate output pins of a microcontroller.
CONTROL VIA PWM
MOSFET switches provide efficient power control through pulse-width modulation (PWM). Lights, heaters, motors and other devices are common candidates for this type of control. Unfortunately, not all electrical loads respond well to PWM. This is rarely mentioned in specification sheets, so it’s up to you to test it. Once I needed to control the speed of a couple of 12VDC fans. The PWM controlled the speed all right, but the fans emitted a rather unsettling squeal.
Lowering the repetition rate caused jerky (and audible) rotation of the fans, while increasing it beyond hearing range resulted in the fan motors to cease to respond. Not willing to buy and test another set of fairly expensive fans, I ended up integrating the pulses by a capacitor with a source follower into a DC voltage. It worked, but a large heatsink on the power transistor was the price to pay.
Because the silicon-based power MOSFETs have reached their theoretical limits, gallium-nitride (GaN) based FETs were introduced in 2010. GaN is a high electron mobility semiconductor (HEMT), providing the new transistors with many advantages.
GaN has an electric field strength higher than silicon, and in combination with its higher electron mobility GaN devices are able to be smaller with smaller RDSon and a higher breakdown voltage than a comparable silicon MOSFET. GaN FETs come as enhancement and depletion mode devices, which can replace the silicon MOSFETs in many circuits.
GaN devices—due to their lateral structure—feature a very low intrinsic capacitance, and are able to switch hundreds of volts within a matter of nanoseconds. There is no parasitic PN junction found in silicon devices, which is the reason for the intrinsic bulk diode. However, different mechanism creates a similar diode inherent in the GaN FETs. The low GaN FET capacitances responsible for its extremely high switching speed require precise PCB layout to avoid instabilities.
There are other semiconductor devices—such as IGBTs, thyristors, PUTs and others—that I didn’t cover in this article series. But I touched on those in my article “Electrical Power Converters (Part 3) in Circuit Cellar 320 (March 2017). Thanks to the present-day availability of many inexpensive ICs and system building blocks, circuit design with discrete components is becoming almost an arcane art. But then again, transistors are building blocks of ICs and many engineers make a living designing those. Even if IC design is not in your future, having an insight into the underlying principles of the solid-state technology can only help you with future projects.
Electronics Tutorials https://www.electronics-tutorials.ws/transistor/tran_7.html
Power MOSFET Basics
Electrical Power Converters Part 3, Circuit Cellar Issue 320, March 2017 by George Novacek
Infineon Technologies | www.infineon.com
PUBLISHED IN CIRCUIT CELLAR MAGAZINE • FEBRURARY 2020 #355 – Get a PDF of the issueSponsor this Article