MOSFET selection techniques

Source: Analog and mixed signal  Given the maturity of MOSFET technology, choose a MOSFET for the design of the surface is a very simple matter. Although engineers are well versed in the MOSFET data sheet on the quality factor, but in order to choose the appropriate MOSFET, engineers must use their professional knowledge of all the different specifications for specific applications overall and careful consideration. For example, the server power load switches in such applications, the MOSFET is basically always been in the conduction state, it does not matter the MOSFET switching characteristics, and on-resistance (RDS (ON)) it is likely that this application key quality factor. However, there are still some applications, such as switching power supplies, the MOSFET as an active switch, so engineers must evaluate other performance parameters of the MOSFET. Let us consider some applications of MOSFET specifications priorities.
 MOSFET may be the most common applications in the power switching elements, in addition, they also benefit from the power output. Servers and communications equipment and other applications are generally configured with multiple concurrent power to support N +1 redundancy and continuity of work (Figure 1). The average load-sharing power supplies in parallel to ensure that the system even in a case of power failure still be able to continue working. However, this architecture also needed a way to connect the output of the parallel power supply, and to ensure a power supply failure will not affect any other power. The output of each power supply, a power MOSFET allows the public to share the power load, while the power supply and isolated from each other. The same effect of MOSFET is called "ORing" FET, because they are essentially based on "OR" logic to connect multiple power output.


Figure 1: Topology used for N +1 parallel redundant power control MOSFET.
Applications in the ORing FET, MOSFET is the role of switching devices, but because of server-class applications work uninterrupted power supply, the switch is in fact always in a conduction state. The switching function is only to play in the startup and shutdown, and when the power fails.
Compared to the core applications in order to switch designers, ORing FET application designers must focus clearly different characteristics of the MOSFET. To the server, for example, during normal operation, MOSFET is equivalent to a conductor. Therefore, ORing FET application designers are most concerned about is the minimum conduction losses.
Low RDS (ON) can be reduced to the minimum BOM and PCB size
In general, MOSFET manufacturers use RDS (ON) parameters to define the on-resistance; for ORing FET applications, RDS (ON) is the most important device characteristics. Data Manual defines RDS (ON) and gate (or drive) voltage VGS and the current flowing through the switch, but for the full gate drive, RDS (ON) is a relatively static parameters. For example, Fairchild FDMS7650 data sheet requirements for 10V gate drive, the maximum RDS (ON) of 0.99 mΩ.
If the designer attempts to develop the smallest, lowest-cost power, low on-resistance is double important. In power supply design, each power supply is often required multiple ORing MOSFET in parallel, you need multiple devices to transmit the current to the load. In many cases, designers must parallel MOSFET, to reduce RDS (ON).
Need to remember that in the DC circuit, the equivalent parallel resistance of the load impedance is less than the impedance of each load individually. For example, two parallel 2Ω resistor is equivalent to a 1Ω resistor. Therefore, in general, a low RDS (ON) values ​​MOSFET, with a large rated current, it allows designers to use the power MOSFET in the number of minimum.
In addition to RDS (ON), the MOSFET's selection process there are several parameters of the power MOSFET design staff is very important. In many cases, designers should pay close attention to the safety data sheet on the work area (SOA) curve, the curve also describes the drain current and drain-source voltage. Basically, SOA defines a safe working MOSFET power supply voltage and current. In the ORing FET applications, the primary question is: "full-state" under the FET current transmission capacity. SOA curve is actually no need to drain current value can be obtained. Then FDMS7650, for example, the device is rated for 36A, it is ideal for server applications typically used in DC-DC power supply.
If the design is hot swappable capabilities, SOA curve may be better to play a role. In this case, MOSFET conduction needs some work. SOA curve defines the different current and voltage during the pulse limits.
Note that the current rating just mentioned, it is worthwhile to consider the thermal parameters, because the conduction of the MOSFET is always very easy to heat. In addition, increasing the junction temperature rise can lead to RDS (ON) increases. MOSFET data sheet provides a thermal impedance parameters, defined as the junction MOSFET package thermal capability. RθJC The simplest definition is from junction to case thermal resistance. Fine words, the actual measurement of its representatives from the junction (for a vertical MOSFET, that is near the upper surface of the die) to the outer surface of the package thermal resistance, are described in the data sheet. The use of PowerQFN package shell is defined as the center piece of this large drain. Therefore, RθJC die and packaging system defines the thermal effect. RθJA defined from the die surface to ambient thermal resistance, and generally marked by a footnote to the relationship with the PCB design, including the number of layers and thickness of copper.
All in all, RθJC in power beyond the control of the design team, because it is used in device packaging technology decisions. Advanced thermal-enhanced package, such as Fairchild's Power 56, the RθJC specifications 1 and 2 oC / W between, FDMS7650 specifications for the 1.2 oC / W. Design team to change the PCB design can RθJA. Ultimately, a robust thermal design helps to improve system reliability, extending the system MTTF (MTBF).
The MOSFET switching power supply
Now let us consider the switching power supply applications, and how such applications need a different angle to look at the data sheet. By definition, this application requires MOSFET turns on and off periodically. Meanwhile, dozens of topology can be used for switching power supply, consider a simple example here. DC-DC power supplies commonly used relies on two basic buck converter MOSFET to perform switching functions (Figure 2), these switches alternately in the inductor to store energy, then releasing the energy to the load. Currently, designers often choose to hundreds of kHz frequencies above 1 MHz as well, because the higher the frequency, magnetic components can be smaller and lighter. 图2:用于开关电源应用的MOSFET对。(DC-DC控制器)

Figure 2: MOSFET for switching power supply applications right. (DC-DC controller)
Clearly, the power supply design is very complicated and there is no simple formula can be used to evaluate the MOSFET. But we might consider some of the key parameters, and why is it important of these parameters. Traditionally, many power supply designers have adopted a comprehensive quality factor (gate charge QG ×-resistance RDS (ON)) to evaluate the MOSFET or the right to conduct grading.
Gate charge and on-resistance is important, because both the efficiency of the power supply has a direct impact. Impact on the efficiency loss can be divided into two types - conduction and switching losses.
Gate charge is to produce the main reason for switching losses. Gate charge unit Naku Lun (nc), is the MOSFET gate charge and discharge the energy required. Gate charge and on-resistance RDS (ON) in semiconductor design and manufacturing processes are interrelated, in general, the device's low gate charge value, the higher on-resistance parameters.
Switching power MOSFET in the second most important parameters include the output capacitance, threshold voltage, gate resistance and the avalanche energy.
Topology will change some special MOSFET parameters related to different quality, for example, the conventional synchronous buck converter and resonant converter to compare. Resonant converter only in the VDS (drain-source voltage) or ID (drain current) than zero before the MOSFET switch, which can minimize the switching losses. These technologies are becoming soft-switching or zero voltage switching (ZVS) or zero current switching (ZCS) technology. Since switching losses are minimized, RDS (ON) in such topologies become more important.
Low output capacitance (COSS) value of these two types of converters are good. Resonant converter in the resonant circuit mainly by the transformer leakage inductance and COSS decision. In addition, the two MOSFET turn-off dead time, the resonant circuit must COSS fully discharged. Therefore, the resonant topology is valued lower COSS. Consider Figure 3 shows the COSS and VDS Fairchild FDMS7650 diagram.  

Figure 3: FDMS7650 of COSS and VDS diagram.
Low output capacitance also help traditional step-down converter (sometimes called hard-switch converter), but for different reasons. Because each cycle is stored in the hard switching of the output capacitor energy will be lost, whereas the energy in the resonant converter in repeated cycles. Thus, low output capacitance synchronous buck regulator for low-side switch is especially important.
MOSFET motor control applications
Motor control applications come in handy is another power MOSFET applications, then the choice of the most important benchmark may be different from another other. Unlike modern switching power supplies, motor control circuit is not high frequency switch. A typical half-bridge control circuit 2 MOSFET (full-bridge is used 4), but the two MOSFET's turn-off time (dead time) equal. For such applications, the reverse recovery time (trr) is very important. In the control inductive loads (such as motor windings), the control circuit to the bridge circuit of the MOSFET switch to the off state, then the bridge circuit to another switch through the MOSFET body diode conduction current temporary reverse. Thus, the current re-circulation, continues to motor power. When the first MOSFET on again, the other MOSFET diode stored charge must be removed, discharged through the first MOSFET, which is an energy loss, so trr shorter, this loss is smaller.
Therefore, if the team needs power circuit design using MOSFET, before the start of the assessment process, be careful the opponent's application full consideration. According to their needs rather than the manufacturer's particular specifications to boast of the parameters for prioritization.
Added: the use of IC and package design was minimal RDS (ON) specifications
In the MOSFET selection process to assess the parameters of the designers usually by careful analysis of the relevant specifications to understand their own in the end what is required. But sometimes in-depth understanding of how IC manufacturers provide operating characteristics is necessary. With RDS (ON), for example, you might usually expect the specifications only with the device design and semiconductor manufacturing process related. But in fact, the package design of the on-resistance RDS (ON) minimization has a huge impact.
Package of RDS (ON) is a huge role because the parameter depends on the conduction losses, while the package will no doubt affect the conduction losses. Consider the body of this article referred to Fairchild FDMS7650 and 1mΩ RDS. The device can be low RDS (ON) value, about half can be attributed in the package design. The package uses a solid copper clip technology to replace commonly used aluminum or gold wire bonding to connect the source and the lead frame. This approach minimizes the package resistance, and reduce the source inductance, source inductance is the main reason for switching devices generate ringing.
Author: Mike Speed
Fairchild Semiconductor