Application Note AN-1070 Class D Audio Amplifier Performance Relationship to MOSFET Parameters By Jorge Cerezo, International Rectifier Table of Contents Page Abstract ................................................................................................…………2 Introduction ..........................................................................................…………2 Key MOSFET Electrical Parameters in Class D Audio Amplifiers ........…………2 Drain Source Breakdown Voltage BVDSS ....................................…………2 Static Drain-to-Source On Resistance RDS(on).............................…………4 Gate Charge Qg ..........................................................................…………5 Body Diode Reverse Recovery Charge, Qrr ...............................…………8 Internal Gate Resistance RG(int)……………………………………...…….….11 MOSFET Package .......................................................................….…….11 Maximum Junction Temperature ................................................…...…….12 International Rectifier Digital Audio MOSFET ........................................…..……13 Conclusions…………………………………………………………………….…….…14 References...........................................................................................…….……14 This application note discusses key parameters to be considered in the selection of MOSFETs for class D amplifier and their relationship with amplifier performance such as efficiency, EMI, and THD. In addition, benefits and advantages of IR Digital Audio MOSFETs in class D audio amplifier are presented. www.irf.com AN-1070 1 Abstract This application note discusses key parameters to be considered in the selection of MOSFETs for class D audio amplifier and their relationship with amplifier’s performance such as efficiency, EMI, and THD. In addition, benefits and advantages of IR Digital Audio MOSFETs in class D audio amplifier are presented. Introduction Class D audio amplifier is a switching amplifier that consists in a pulse width modulator (with switching frequency in order of several hundred kHz), a power bridge circuit and a low pass filter. This type of amplifier has demonstrated to have a very good performance. These include power efficiencies over 90%, THD under 0.01%, and low EMI noise levels that can be achieved with a good amplifier design. Details of class D audio amplifier theory are explained in Application Note “Class D Audio Amplifier Basics” [4]. Key factors to achieve high performance levels in the amplifier are the switches in power bridge circuit. Power losses, delay times, and voltage and current transient spikes should be minimized as much as possible in these switches in order to improve amplifier performance. Therefore, switches with low voltage drop, fast on and off switching times and low parasitic inductance are needed in this amplifier. MOSFET have proved to be the best switch option for this amplifier because of its switching speed. It is a majority carrier device, its switching times are faster in comparison with other devices such as IGBT or BJT [1], resulting in better amplifier efficiency and linearity. Key MOSFET Electrical Parameters in Class D Audio Amplifiers. As in any other application, the MOSFET selection is done in base amplifier specifications. Thus, information such as amplifier output power and load impedance (i.e. 100W into 8Ω), power bridge circuit topology (i.e. Full-Bridge or Half-Bridge), and modulation factor (i. e. 80% - 90%) should be known before the selection process. • Drain-Source Breakdown Voltage, BVDSS. Amplifier operating voltage determines the selection of MOSFET voltage rating. However, other circuit design related factors should be taken in consideration, such as stray resistance, MOSFET switching peak voltages because of stray inductance, and power supply fluctuations. Otherwise, the MOSFET can be in avalanche condition during operation degenerating amplifier performance. Thus, the minimum BVDSS is chosen by the desired amplifier output power and load impedance, power bridge circuit topology, modulation factor, and an additional factor related with circuit issues (frequently used 10% - 50%) as follows: 2 * POUT * RLOAD BVDSS min (full-bridge conf.) = www.irf.com + M AN-1070 Additional factor due to circuit related issues 2 2* BVDSS min (half-bridge conf.) = 2 * POUT * RLOAD M + Additional factor due to circuit related issues The following table is an example of minimum MOSFET voltage rating selection for different class D audio amplifier conditions: Table 1. Minimum MOSFET voltage rating for different class D audio amplifier conditions. It is important to choose the lowest BVDSS possible because this parameter is related to others such as RDS(on). Higher BVDSS results in higher RDS(on) and higher MOSFET power losses. Therefore, this should be taken into consideration for “factor due to circuit related issues” selection. In summary, the voltage rating of the MOSFET should be selected according to amplifier operating voltage. It should also be large enough to avoid avalanche condition during operation. www.irf.com AN-1070 3 • Static Drain-to-Source On-Resistance, RDS(on). Amplifier efficiency is related to MOSFET total power losses. These power losses are the result of MOSFET conduction, switching, and Gate charge losses. Furthermore, the MOSFET’s junction temperature TJ and heatsink size depend on this power losses amount. High power losses increase TJ, and therefore, heatsink size. MOSFET conduction losses are directly related to RDS(on) parameter. RDS(on) is the Drain-Source resistance, typically specified on datasheet at 25°C with VGS = 10V for standard Gate MOSFETs [5]. RDS(on) along with Drain current define MOSFET conduction losses during amplifier operation, and can be calculated as follows: PCONDUCTION = (ID RMS)2 * RDS(on) RDS(on) is temperature-dependent, increasing when TJ increases. Care must be taken during the thermal design, regarding this, to avoid thermal runaway. In addition, it is important that the maximum MOSFET junction temperature TJ max, should not be higher than specified in the datasheet during all amplifier operating conditions. Then, the maximum MOSFET conduction losses calculation should be done at maximum amplifier operating conditions, using RDS(on) at TJ max, typically specified in the datasheet, and maximum ID RMS current. Therefore, lower RDS(ON) results in lower MOSFET conduction losses and consequently better amplifier efficiency. This is illustrated in Figure 1. Lower RDS(ON) ⇒ Lower PCONDUCTION ⇒ Better efficiency Pout vs. Efficiency Different RDS(on) Comparison Similar switching losses Half-Bridge Configuration, Vbus = +/- 35V, fswitching = 395KHz, finput = 1KHz, Rload = 4Ω 100% 98% 10% THD 96% 1% THD 94% RDS(on) = 51mOhm Efficiency 92% 90% RDS(on) = 95mOhm 88% 86% 84% 82% 80% 0 20 40 60 80 100 120 140 160 180 Pout (W) Figure 1. Amplifier efficiency using different MOSFET RDS(on) and similar switching losses. www.irf.com AN-1070 4 • Gate Charge, Qg. The MOSFET Gate charge Qg is the charge required by the Gate to fully turn-on the MOSFET. This parameter is temperature-independent and is directly related to the MOSFET’s speed. Lower Qg results in faster switching speeds and lower Gate losses; consequently, lower switching losses and better efficiency is achieved. MOSFET switching losses are defined as: PTOTAL SWITCHING = PSWITCHING + PGATE Switching losses are the result of MOSFET turn-on and turn-off switching times illustrated in Figure 2. Switching losses are calculated by multiplying switching energy Esw, with the amplifier’s PWM switching frequency fsw. PSWITCHING = Esw * fsw And, switching energy Esw is obtained by: Esw = ∫ t VDS(t) * ID(t) dt 0 Where, t is the length of switching pulse. ton toff MOSFET Turn -On MOSFET Turn -Off Figure 2. MOSFET turn-on and turn-off waveforms. Acceptable switching losses estimations can be done by using the amplifier’s specifications and the MOSFET’s datasheet parameters as follow: PSWITCHING = [0.5 * ID * Vbus * (tr + tf) * fsw] + [0.5 * Coss * Vbus2 * fsw] + [K * 0.5 * Qrr * Vbus * fsw] Where Vbus is amplifier’s bus voltage, tr and tf are MOSFET rise and fall times, Coss is MOSFET output capacitance, Qrr is MOSFET body diode reverse recovery charge, and K is the factor due to MOSFET TJ and specific amplifier’s conditions such as IF and dIF/dt. www.irf.com AN-1070 5 Gate losses can be estimated: PGATE = 2 * Qg * Vdriver * fsw Where, Vdriver is voltage of the Gate driver. Also, amplifier linearity is affected by switching timing errors such as MOSFET turn-on and turn-off delay times. This implies that the amplifier linearity is also affected by higher Qg. However, these timing errors caused by MOSFET switching are not as significant when compared to dead time and, can be significantly reduced by selecting the right dead time value [4]. This is illustrated in Figures 3 and 4. Pout vs. THD + N Comparison of different Qg and same dead time Full Bridge Configuration, Vbus = 40V, fswitching = 800KHz, Rload = 8Ω THD + N (%) 100.0 Dead Time = 12ns 10.0 MOSFET Qg = 13nC 1.0 MOSFET Qg = 8nC 0.1 1 10 100 Pout (W) Figure 3. Amplifier THD+N at different MOSFET Qg and same dead time. Therefore, MOSFET Qg is related to amplifier efficiency and linearity. However, it affects amplifier efficiency more significantly than linearity. Since linearity can be improved with dead time optimization, Qg should be lower mainly to achieve lower MOSFET switching losses and increase amplifier efficiency. This is illustrated in Figure 5. Lower Qg ⇒ Lower PSWITCHING ⇒ Better efficiency www.irf.com AN-1070 6 Pout vs. THD + N Comparison of different dead time and same Qg Full Bridge Configuration, Vbus = 40V, fswitching = 800KHz, Rload = 8Ω THD + N (%) 100.0 MOSFET Qg = 13nC 10.0 Dead Time = 12ns 1.0 Dead Time = 4ns 0.1 1 10 100 Pout (W) Figure 4. Amplifier THD+N at different dead time and same MOSFET Qg. Pout vs. Efficiency Different Qg comparison, similar R DS(on) Half-Bridge configuration, Vbus = +/- 35V, fswitching = 395KHz, Rload = 4Ω 100% 98% 1% THD 96% 10% THD Qg = 9nC Efficiency 94% 92% Qg = 13nC 90% 88% 86% 84% 82% 80% 0 20 40 60 80 100 120 140 160 180 Pout (W) Figure 5. Amplifier efficiency at different MOSFET Qg and similar RDS(ON). www.irf.com AN-1070 7 • Body Diode Reverse Recovery Charge, Qrr. Intrinsically by design, the MOSFET structure has a built in reverse diode, and its reverse recovery characteristics are related to amplifier performance as well. Reverse recovery charge Qrr, is defined as the area under Irr during trr. This is illustrated in Figure 6. Qrr is mainly determined by IF and dIF/dt. It is a temperature dependent parameter, increasing when TJ increases. Qrr affects amplifier efficiency and EMI performance. The relationship with efficiency is due to the power bridge circuit configuration. During operation, the reverse recovery current generated by the MOSFET’s body diode after commutation current, flows also in the complementary MOSFET of bridge circuit, causing an increase in turn-on switching losses due to the current increment. This is illustrated in figure 7. Details of these switching events are explained in Application Note “Using HEXFET III in PWM Inverters for Motor Drives and UPS Systems” [2]. IF Figure 6. Typical MOSFET body diode reverse recovery waveforms. However, similar to amplifier linearity, the selection of right dead time can improve efficiency due to Qrr reduction. Reduction of dead time causes the commutation current to flow most of the time into the MOSFET channel, reducing MOSFET body diode current pulse width, and hence, minority carrier charge and Qrr. Still, smaller dead time may cause shoot through current, as shown in Figure 8. This is a risky condition for power bridge MOSFETs and also it degenerates the amplifier performance [4]. Therefore, optimized dead time can help to reduce Qrr and improve efficiency and linearity as well. This is illustrated in Figure 9. Qrr is also related with amplifier EMI performance. High recovery current dIrec/dt (i. e. faster tb) generates high dVrec/dt, and this results in large high-frequency current and voltage ringing transients in the www.irf.com AN-1070 8 MOSFET due to stray inductances and capacitances in the amplifier circuit, increasing radiated and conducted EMI noise. Therefore, a smaller and soft recovery is fundamental to avoid these transients to improve EMI performance. Details of this EMI and reverse recovery relationship are explained in White Paper “Ultra-fast Recovery Diodes Meet Today Requirements for High Frequency Operations and Power Ratings in SMPS Applications” [3]. Hence, smaller and soft reverse recovery improves amplifier efficiency and EMI performance, due to reduction in MOSFET switching losses and current-voltage transient ringing. Smaller and soft Qrr ⇒ Lower PSWITCHING ⇒ Better efficiency and EMI performance Q Drive current Q Commutation current a) Amplifier’s drive and commutation current Drive current Commutation current b) Load current (Iload = Idrive + c) High side Q1 MOSFET Pon VDS Peak current due to Qrr of complementary MOSFET's Body Diode ID d) MOSFET turn-on during drive current Figure 7. MOSFET turn-on waveforms during drive current. www.irf.com AN-1070 9 Qst as a function of Overlap Tim e & Rg Vbus = 60V, Id = 2A, Vgs = 12V (Overlap tim e m easured from 50% Vgs high side fall to 10% Vgs low side rise) 120 Low Side Side VGS VGS High Side VGS 100 Qst (nC) 80 Shoot Through Current (2A/div) Rg=1Ohms 60 Rg=5Ohms Rg=10Ohms Device Qg = 37nC 40 Qst = ∫ Ist(t)dt 20 0 -10 -5 0 5 10 15 20 Overlap time (ns) Figure 8. Example of shoot through charge [Qst = ∫ Ist(t)dt] as a function of dead time. Pout vs. Efficiency Comparison of different dead time Full Bridge Configuration, Vbus = 40V, fswitching = 800KHz, Rload = 8Ω 95.0% 10% THD Efficiency 90.0% Dead Time 12ns Dead time 4ns 85.0% Icommutation Icommutation 80.0% Qrr Qrr 75.0% 0 10 20 30 40 50 60 70 80 Pout (W) Figure 9. Amplifier efficiency at different dead times. www.irf.com AN-1070 10 • Internal Gate Resistance, RG(int). Internal Gate Resistance RG(int) is a temperature dependent parameter, increasing when temperature increases. This parameter affects MOSFET on and off switching times. Higher RG(int) increases total Gate resistance, decreases Gate current (as shown in Figure 10), increases switching times, and thus, MOSFET switching losses. Furthermore, a large variation of RG(int) affects dead time control. Therefore, RG(int) parameter distribution should be taken in consideration for amplifier performance tolerances. D IG RG(ext) G RG(int) RGtotal = RG(ext) + RG(int) S Figure 10. Total Gate resistance, RGtotal = RG(ext) + RG(int). • MOSFET Package. The selection of MOSFET package is of particular importance because it can significantly affects the amplifier performance and cost. Packages characteristics such as dimensions, power dissipation capability, current capability, internal inductance and resistance, electrical isolation and mounting process can all be significant to define PCB and heatsink size, assembly process, and MOSFET electrical parameters. It is well known the importance of package thermal resistance RθJC in MOSFET performance. Mainly, because lower RθJC reduces MOSFET junction temperature TJ during operation, improving the MOSFET reliability and performance. Pmax = ΔTJ / RθJC max Again, stray inductances and capacitances in amplifier circuit affect EMI performance. Internal package inductance can make a significant difference in EMI noise generation. Figure 11 shows EMI noise comparison of two packages with same MOSFET die and different internal inductance, DirectFETTM MOSFET (<1nH) vs. TO-220 (~ 12nH). It can be observed in this comparison that the DirectFET MOSFET shows better EMI performance. Approximately 9dB lower noise than TO-220, even when the DirecFET MOSFET rise and fall times are approximately three times faster than TO-220. In summary, the right package selection improves amplifier reliability, performance and cost www.irf.com AN-1070 11 Test Conditions: Half-Bridge Configuration, No Heatsink, Vbus = +/- 35V, fsw = 395kHz, Pout = 12.5W, Rload = 4 Ohms, Input = 250mVrms Sinusoidal @ 1kHz Note: Not shielded test room. Both Packages are tested with the same MOSFET die TO-220 DirectFET a) DirectFET package EMI noise. b) TO-220 package EMI noise. Frequency (MHz) Frequency (MHz) c) DirectFET VDS rise and fall times d) TO-220 VDS rise and fall times Figure 11. Amplifier conducted EMI performance comparison. DirectFET vs. TO-220 • Maximum Junction Temperature, TJ. The maximum Junction Temperature TJ max is a parameter that is not directly related with the amplifier performance. However, special highlight is done because it is significant to define heatsink size. MOSFET devices with higher TJ max, allow handling higher power losses, and hence, allow smaller heatsink reducing amplifier size and cost. www.irf.com AN-1070 12 International Rectifier Digital Audio MOSFET. IR Digital Audio MOSFETs are devices specifically designed for class D audio amplifier applications. Its parameters are optimized to improve amplifier performance and this goal is achieved by selecting the right die size for specific amplifier specifications. As mentioned before, RDS(on) and Qg are key parameters that determine MOSFET power losses. These parameters are related to MOSFET die size and there is a tradeoff between them. A larger MOSFET die size means lower RDS(on) and higher Qg, and vice versa. Bigger Die Size ⇒ Lower RDS(ON) & Higher Qg ⇒ Lower PCONDUCTION & Higher PSWITCHING Smaller Die Size ⇒ Higher RDS(ON) & Lower Qg ⇒ Higher PCONDUCTION & Lower PSWITCHING Therefore, the optimal die size will results in lower MOSFET power losses. An example of this is illustrated in Figure 12. Optimal MOSFET die size ⇒ Optimal MOSFET Parameters 3.5 3.0 Total Losses = Conduction Losses + Switching Losses Power Losses (W) Total Losses 2.5 Optimal MOSFET Die Size 2.0 1.5 1.0 Switching Losses Conduction Losses 0.5 0.0 1 10 Normalized Size NormalizedDie Die Size Figure 12. Example of IR Digital Audio MOSFET die size selection. Furthermore, IR Digital Audio MOSFETs have a guaranteed maximum RG(int), low Qrr, a 175°C maximum TJ, and are assembled in most efficient packages such as DirectFET MOSFET. These features combine www.irf.com AN-1070 13 to make this MOSFET a highly efficient, robust, and reliable device for class D audio amplifier applications. Conclusions There are key MOSFET parameters such as BVDSS, RDS(on), Qg, Qrr, RG(int), TJ max and packaging that are related with Class D audio amplifier performance, size and/or cost. The right selection of these parameters can improve the amplifier performance and reduce its size and/or cost. IR Digital Audio MOSFETs are devices in which those key parameters have been optimized to achieve the best overall amplifier performance. References 1. Barkhordarian V. “Power Mosfet Basics”. International Rectifier Application Note. <http://www.irf.com>. 2. Grant D. “Using HEXFET III in PWM Inverters for Motor Drives and UPS Systems” International Rectifier Application Note. <http://www.irf.com>. 3. Guerra A., Andoh K. and Fimiani S. “Ultra-fast Recovery Diodes Meet Today Requirements for High Frequency Operations and Power Ratings in SMPS Applications”. International Rectifier White Paper. <http://www.irf.com>. 4. Honda J. and Adams J. “Class D Audio Amplifiers Basics”. International Rectifier Application Note. <http://www.irf.com>. 5. Pearce R., Brown S. and Grant D. “Measuring HEXFET® Characteristics”. International Rectifier Application Note. <http://www.irf.com>. www.irf.com AN-1070 14