The difference between MOSFET and IGBT
Latest update time：2023-04-03 18:20
Reads：
Due to the structure of MOSFET, it can usually achieve a large current, up to KA, but its voltage resistance is not as strong as IGBT.
IGBT can produce a lot of power, current and voltage, but the frequency is not too high. At present, the hard switching speed of IGBT can reach 100KHZ, which is already good. However, compared to the operating frequency of MOSFET, it is still a drop in the bucket. MOSFET can work to hundreds of KHZ, upper MHZ, or even dozens of MHZ.
In terms of its application: According to its characteristics, MOSFET is used in switching power supplies, ballasts, high-frequency induction heating; high-frequency inverter welding machines; communication power supplies and other high-frequency power supply fields; IGBT is mainly used in welding machines, inverters, frequency conversion devices, electroplating electrolytic power supplies, ultrasonic induction heating and other fields.
The performance of switching power supplies (SMPS) depends largely on the selection of power semiconductor devices, namely switching transistors and rectifiers.
Although there is no foolproof solution to the problem of choosing IGBT or MOSFET, it can still serve as a reference to compare the performance of IGBT and MOSFET in specific SMPS applications and determine the range of key parameters.
This article will explore parameters such as switching losses in hard-switching and soft-switching ZVS (Zero Voltage Switching) topologies and examine the three main power switching losses - conduction losses, conduction losses and turn-off - in relation to circuit and device characteristics. Describe losses. In addition, by taking an example to illustrate that the recovery characteristics of the diode is the main factor that determines the turn-on switching loss of MOSFET or IGBT, the impact of the recovery performance of the diode on the hard-switching topology is discussed.
conduction loss
Except for the longer voltage drop time of IGBT, the conduction characteristics of IGBT and power MOSFET are very similar. It can be seen from the basic IGBT equivalent circuit (see Figure 1) that the time required to fully adjust the minority carriers in the collector base region of the PNP BJT causes the on-voltage tail to appear.
This delay causes a saturation-like effect, preventing the collector/emitter voltage from immediately falling to its VCE(sat) value. This effect also causes the VCE voltage to rise at the moment when the load current is transferred from the anti-parallel diode of the combined package to the collector of the IGBT in the ZVS case. The Eon energy consumption listed in the IGBT product specification is the time integral of the product of Icollector and VCE for each conversion cycle, in joules, and includes other losses related to quasi-saturation. It is divided into two Eon energy parameters, Eon1 and Eon2. Eon1 is the power loss that does not include the energy loss related to the hard-switching diode recovery loss; Eon2 includes the hard-switching conduction energy loss related to the diode recovery, which can be measured by restoring the same diode as the diode packaged with the IGBT combination. Typical The Eon2 test circuit is shown in Figure 2. The IGBT measures Eon by switching on and off with two pulses. The first pulse will increase the inductor current to achieve the required test current, and then the second pulse will measure the Eon loss recovered by the test current in the diode.
In the case of hard switch conduction, the gate drive voltage and impedance and the recovery characteristics of the rectifier diode determine the Eon switching losses. For traditional CCM boost PFC circuits, the recovery characteristics of the boost diode are extremely important in the control of Eon (conduction) energy consumption. In addition to selecting a boost diode with minimum Trr and QRR, it is also important to ensure that the diode has soft recovery characteristics. The degree of softening, or the tb/ta ratio, has a considerable impact on the electrical noise and voltage spikes generated by switching devices. For some high-speed diodes, the current drop rate (di/dt) from IRM (REC) within time tb is very high, so high voltage spikes will be generated in the parasitic inductance of the circuit. These voltage spikes can cause electromagnetic interference (EMI) and can cause excessive reverse voltages on the diodes.
In hard switching circuits, such as full-bridge and half-bridge topologies, a fast recovery transistor or MOSFET body diode is packaged with the IGBT. When the corresponding switch transistor is turned on, the diode has current flowing through it, so the recovery characteristics of the diode determine the Eon loss. Therefore, it is very important to choose a MOSFET with fast body diode recovery characteristics. Unfortunately, the recovery characteristics of a MOSFET's parasitic or body diode are slower than the discrete diodes currently used by the industry. Therefore, for hard-switching MOSFET applications, the body diode is often the limiting factor in determining the SMPS operating frequency.
Generally speaking, the selection of IGBT package diodes should match their applications, with slower ultrafast diodes with lower forward conduction losses packaged together with slower low VCE (sat) motor drive IGBTs. Conversely, soft-recovery ultrafast diodes can be packaged in combination with high-frequency SMPS2 switch-mode IGBTs.
In addition to selecting the correct diode, designers can control Eon losses by adjusting the gate drive conduction source impedance. Lowering the driving source impedance will improve the conduction di/dt of IGBT or MOSFET and reduce Eon loss. There is a trade-off between Eon losses and EMI, as higher di/dt results in increased voltage spikes, radiated and conducted EMI. To select the correct gate drive impedance to meet the turn-on di/dt requirements, it may be necessary to perform internal circuit testing and verification, and then the approximate value can be determined based on the MOSFET transfer curve (see Figure 3).
Assume that at turn-on, the FET current rises to 10A. According to the 25°C curve in Figure 3, in order to reach the 10A value, the gate voltage must switch from 5.2V to 6.7V, and the average GFS is 10A/(6.7V- 5.2V)=6.7mΩ.
Equation 1 to obtain the gate drive impedance for the required turn-on di/dt
Apply the average GFS value to equation 1 to get the gate drive voltage Vdrive=10V, the required di/dt=600A/μs, the typical value of FCP11N60 VGS(avg)=6V, Ciss=1200pF; then the conduction can be calculated Gate drive impedance is 37Ω. Since the transient GFS value is a slope in the curve in Figure 3, it will change during Eon, which means that di/dt will also change. The exponentially decaying gate drive current Vdrive and falling Ciss as a function of VGS also enter into the equation, with the overall effect of a surprisingly linear current rise.
Similarly, similar gate drive on-resistance calculations can be performed for IGBT. VGE(avg) and GFS can be determined through the conversion characteristic curve of IGBT, and the CIES value under VGE(avg) should be used instead of Ciss. The calculated IGBT turn-on gate drive impedance is 100Ω, which is higher than the previous 37Ω, indicating that the IGBT GFS is higher and CIES is lower. The key point here is that in order to switch from MOSFET to IGBT, the gate drive circuit must be adjusted.
Be cautious about conduction losses
When comparing devices rated for 600V, IGBTs generally have less conduction losses than 600V MOSFETs of the same chip size. This comparison should be made at a time when collector and drain current densities are clearly sensed, and at a specified worst-case operating junction temperature. For example, the FGP20N6S2 SMPS2 IGBT and FCP11N60 SuperFET both have RθJC values of 1°C/W. Figure 4 shows the relationship between conduction loss and DC current at a junction temperature of 125°C. The curve in the figure shows that the conduction loss of MOSFET is greater after the DC current is greater than 2.92A.
However, the DC conduction loss comparison in Figure 4 is not suitable for most applications. Meanwhile, Figure 5 shows the comparison curve of conduction loss in CCM (continuous current mode), boost PFC circuit, junction temperature of 125°C, AC input voltage Vac of 85V and DC output voltage of 400 Vdc operating mode. In the figure, the MOSFET-IGBT curve intersection point is 2.65A RMS. For PFC circuits, when the AC input current is greater than 2.65A RMS, the MOSFET has large conduction losses. The 2.65A PFC AC input current is equal to 2.29A RMS in the MOSFET calculated from Equation 2. MOSFET conduction loss, I2R, can be calculated using the current defined in Equation 2 and the MOSFET's RDS(on) at 125°C. This conduction loss can be further refined by taking into account the variation of RDS(on) with drain current. This relationship is shown in Figure 6.
An IEEE article titled "How to incorporate the dependence of a power MOSFET's RDS(on) on drain current transients into conduction loss calculations for high-frequency three-phase PWM inverters" describes how to determine the drain current Effect on conduction losses. As a function of ID, changes in RDS(on) have little impact on most SMPS topologies. For example, in the PFC circuit, when the peak current ID of the FCP11N60 MOSFET is 11A - twice the 5.5A (the test condition of RDS(on) in the specification book), the effective value of RDS(on) and the conduction loss will increase by 5 %.
In high pulse current topologies where the MOSFET conducts extremely small duty cycles, the characteristics shown in Figure 6 should be considered. If the FCP11N60 MOSFET is operated in a circuit with a drain current of 20A pulses with a duty cycle of 7.5% (i.e. 5.5A RMS), the effective RDS(on) will be greater than that at 5.5A (test current in the specification) 0.32 ohm is 25% larger.
Equation 2 RMS current in CCM PFC circuit
In Formula 2, Iacrms is the RMS input current of the PFC circuit; Vac is the RMS input voltage of the PFC circuit; Vout is the DC output voltage.
In a practical application, calculating the conduction losses of an IGBT in a PFC-like circuit would be more complex, since each switching cycle occurs on a different IC. The VCE (sat) of IGBT cannot be represented by an impedance. A relatively simple and direct method is to express it as an impedance RFCE connected in series with a fixed VFCE voltage, VCE (ICE) = ICE × RFCE + VFCE. The conduction loss can then be calculated as the average collector current times VFCE, plus the square of the RMS collector current, multiplied by the impedance RFCE.
The example in Figure 5 only considers the conduction loss of the CCM PFC circuit, which assumes that the design goal is to maintain a worst-case conduction loss of less than 15W. Taking the FCP11N60 MOSFET as an example, the circuit is limited to 5.8A, while the FGP20N6S2 IGBT can operate at an AC input current of 9.8A. It can conduct more than 70% of the power of a MOSFET.
Although IGBTs have smaller conduction losses, most 600V IGBTs are PT (penetration) type devices. PT devices have NTC (negative temperature coefficient) characteristics and cannot be shunted in parallel. Perhaps, these devices can be connected in parallel with limited effectiveness through matching devices VCE(sat), VGE(TH) (gate emitter threshold voltage) and mechanical packaging, so that the temperatures of the IGBT chips can maintain consistent changes. In contrast, MOSFETs have a PTC (Positive Temperature Coefficient) that provides good current shunting.
Turn-off losses—the problem isn’t over yet
In hard-switched, clamped inductive circuits, MOSFETs have much lower turn-off losses than IGBTs due to the IGBT tail current, which is related to clearing the minority carriers of the PNP BJT in Figure 1. Figure 7 shows the collector current ICE as a function of junction temperature Tj, Eoff, whose curves are provided in most IGBT datasheets. These curves are based on clamped inductive circuits with the same test voltage and include tail current energy losses.
Figure 2 shows a typical test circuit for measuring IGBT Eoff. Its test voltage, VDD in Figure 2, varies between manufacturers and the BVCES of individual devices. VDD under these test conditions should be considered when comparing devices because testing and operating at lower VDD clamping voltages will result in lower Eoff energy consumption.
Reducing the gate drive turn-off resistance has minimal impact on reducing IGBT Eoff losses. As shown in Figure 1, when the equivalent majority carrier MOSFET is turned off, there is still a storage time delay td(off)I in the IGBT minority carrier BJT. However, reducing the Eoff drive impedance will reduce the risk of current injection into the gate drive loop caused by the Miller capacitance CRES and the dv/dt of the turn-off VCE, and avoid re-biasing the device into a conductive state, thereby causing multiple Eoff switch action.
ZVS and ZCS topologies have advantages in reducing turn-off losses of MOSFETs and IGBTs. However, the operating advantages of ZVS are not that great in IGBTs, because when the collector voltage rises to a potential value that allows excess storage charge to be dissipated, the tail impact current Eoff will be triggered. ZCS topology can improve the maximum IGBT Eoff performance. The correct gate driving sequence can prevent the IGBT gate signal from being cleared before the second collector current zero-crossing point, thereby significantly reducing the IGBT ZCS Eoff.
The Eoff energy dissipation of a MOSFET is a function of its Miller capacitance Crss, gate drive speed, gate drive turn-off source impedance, and parasitic inductance in the source power circuit path. The circuit parasitic inductance Lx (shown in Figure 8) creates a potential that increases turn-off losses by limiting the current speed drop. At turn-off, the current decrease rate di/dt is determined by Lx and VGS(th). If Lx=5nH, VGS(th)=4V, the maximum current decreasing speed is VGS(th)/Lx=800A/μs.
Summarize
There is no one-size-fits-all solution when selecting a power switching device. Circuit topology, operating frequency, ambient temperature, and physical size all play a role in making the best choice.
In ZVS and ZCS applications with minimal Eon losses, MOSFETs are able to operate at higher frequencies due to their faster switching speed and less turn-off losses.
For hard-switching applications, the recovery characteristics of the MOSFET parasitic diode can be a disadvantage. Conversely, excellent soft-recovery diodes can be used with higher-speed SMPS devices because the diodes within the IGBT combination package are matched to the specific application.
postscript
There is no essential difference between MOSFE and IGBT. The question people often ask, "Which is better, MOSFET or IGBT?" is wrong in itself. As for why we sometimes use MOSFETs and sometimes use IGBTs instead of MOSFETs, we cannot simply distinguish between good and bad. We need to use a dialectical approach to consider this issue.
*Disclaimer: This article is original by the author. The content of the article is the personal opinion of the author. The reprinting by Semiconductor Industry Watch is only to convey a different point of view. It does not mean that Semiconductor Industry Watch agrees or supports the view. If you have any objections, please contact Semiconductor Industry Watch.
Today is the 3359th content shared by "Semiconductor Industry Observation" with you. Welcome to pay attention.
Recommended reading
★ Diamond semiconductor, getting closer
★ What kind of chip is Jim Keller working on?
★ The "legacy" left to us by Gordon Moore
Semiconductor Industry Watch
" Semiconductor's First Vertical Media "
Real-time professional original depth
Identify the QR code , reply to the keywords below, and read more
Wafers | Integrated circuits | Equipment | Automotive chips | Storage | TSMC | AI | Packaging
Reply
Submit an article
and read "How to Become a Member of "Semiconductor Industry Watch""
Reply Search and you can easily find other articles you are interested in!
Featured Posts
京公网安备 11010802033920号