www.fairchildsemi.com Application Note AN-4150 Design Guidelines for Flyback Converters Using FSQ-series Fairchild Power Switch (FPS™) 1. Introduction size and weight, while simultaneously increasing efficiency, productivity, and system reliability. The FSQ-series employs an advanced control technique that allows converter to operate with narrow frequency variation, while keeping the quasi-resonant operation. When the converter operates in discontinuous conduction mode (DCM), the controller finds the valley of the drain voltage and turns on the MOSFET at the minimum drain voltage. Meanwhile, the converter can operate with fixed frequency when operating in continuous conduction mode (CCM), which allows converter design as simple as conventional PWM converters. Compared to conventional hard-switched converters with fixed switching frequencies, the quasi-resonant converter (QRC) topology is a very attractive alternative for power supply designers. The increasing popularity of the QRC approach is based on its ability to reduce electromagnetic interference (EMI) while increasing power conversion efficiency. The FSQ-series FPS™ (Fairchild Power Switch) is an integrated Pulse Width Modulation (PWM) controller and Sense FET specifically designed for quasi-resonant off-line Switch Mode Power Supplies (SMPS) with minimal external components. Figure 1 shows the internal block diagram of the FSQ-series. Compared with discrete MOSFET and PWM controller solution, it can reduce total cost, component count, This application note presents practical design considerations of a flyback converter employing the FSQ-series FPS™. It covers designing the transformer, output filter, and sync network; selecting the components; and closing the feedback loop. Sync Vstr Vcc 4 5 2 Drain 6 7 8 + OSC - 0.7V/0.2V + + VCC Vref Idelay FB 3 0.35/0.55V VBurst Vref VCC good - 8V/12V IFB PWM 3R R SoftStart S Q R Q Gate driver LEB 200ns LPF RC=80ns AOCP 1 TSD 6V VSD Sync Vovp S 2.5μs time delay R Q Q VOCP GND (1.1V) 6V Vcc good Figure 1. Block Diagram of FSQ-Series © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 www.fairchildsemi.com AN-4150 APPLICATION NOTE 2. Operation principle of Quasiresonant flyback converter Quasi resonant flyback converter topology can be derived from a conventional square wave, pulse-width-modulated (PWM) flyback converter without adding additional components. Figure 2 shows the simplified circuit diagram of a quasi-resonant flyback converter and its typical waveforms. The basic operation principles are: + - + VIN Vo×Np/Ns - D Lm VO - + I ds Coss During the MOSFET ON time (tON), input voltage (VIN) is applied across the primary-side inductor (Lm). Then, MOSFET current (Ids) increases linearly from zero to the peak value (Ipk). During this time, the energy is drawn from the input and stored in the inductor as much as Lm×Ipk2/2. + Vds - Ids (MOSFET Drain-to-Source Current) Ipk When the MOSFET is turned off, the energy stored in the inductor forces the rectifier diode (D) to turn on. During the diode ON time (tD), the output voltage (Vo) is applied across the secondary-side inductor and the diode current (ID) decreases linearly from the peak value (Ipk×Np/Ns) to zero. At the end of tD, all the energy stored in the inductor has been delivered to the output. During this period, the output voltage is reflected to the primary side as Vo×Np/Ns. The sum of input voltage (VIN) and the reflected output voltage (Vo×Np/Ns) is imposed on the MOSFET. ID (Diode Current) Ipk×Np/Ns Vds (MOSFET Drain-to-Source Voltage) VIN +Vo×Np/Ns Vo×Np/Ns When the diode current reaches zero, the drain-tosource voltage (Vds) begins to oscillate by the resonance between the primary-side inductor (Lm) and the MOSFET output capacitor (Coss) with an amplitude of Vo×Np/Ns on the offset of VIN, as depicted in Figure 2. Quasi-resonant switching is achieved by turning on the MOSFET when Vds reaches its minimum value. Doing this reduces the MOSFET turn-on switching loss caused by the capacitance loading between the drain and source of MOSFET. If the transformer is designed so that the resonance amplitude is larger than VIN by increasing the turns ratio, Np/Ns, "Zero-Voltage-Switching (ZVS)" of the MOSFET is achieved. VIN Vo×Np/Ns VIN -Vo×Np/Ns tON tD tS Figure 2. Typical Waveform of Quasi-Resonant Flyback Converter 3. Control Method of FSQ-Series Other than turning on the MOSFET with minimum drain-tosource voltage, a quasi-resonant converter provides "soft" switching conditions to the switching devices. The MOSFET turns on at zero current and the diode turns off at zero current. This soft switching not only reduces the switching losses, but also lowers the switching noise caused by diode reverse recovery. To overcome the frequency increase problem at light load, FSQ-series employs an advanced control technique illustrated in Figure 3 with typical switching waveforms. Once the MOSFET is turned on, the next turn-on is prohibited during the blanking time (tB). After the blanking time, the controller finds the valley within the detection time window (tW) and turns on the MOSFET (Case B and C). If no valley is found within tW, the MOSFET is forced to turn on at the end of tW (Case A). Thus, the converter can operate with a fixed frequency when operating in continuous conduction mode (CCM). Meanwhile, when the converter operates in discontinuous conduction mode (DCM), the controller turns on the MOSFET at the valley within tW. Accordingly, the switching frequency is limited between 55kHz and 67kHz, as shown in Figure 3 and 4. This allows converter design as simple as in conventional PWM converters. The major drawback of applying a quasi-resonant converter topology is that it causes the switching frequency to increase as the load decreases and/or input voltage increases. As the load decreases and/or input voltage increases, the MOSFET ON time (tON) diminishes and, therefore, the switching frequency increases. This results in severe switching losses, as well as intermittent switching and audible noise. Due to these problems, the conventional quasi-resonant converter topology has limitations for applications with wide input and load ranges. © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 ID Np:Ns www.fairchildsemi.com 2 AN-4150 APPLICATION NOTE 4. Step-by-step Design Procedure ID Ids A Vds tB=15μs This section provides a step-by-step design process, illustrated in the design flow chart of the Figure 5. Figure 6 shows the basic schematic of quasi-resonant flyback converter using FSQ-series, which also serves as a reference circuit for the design process described. Ids tW=3μs 1. Determine the system specifications (Vlinemin, Vlinemax, fL , Po , Eff ) 2. Determine DC link capacitor (C DC) and calculate DC link voltage range ID Ids ID Ids 3. Determine the reflected output voltage (V RO) B Vds 4. Determine the transformer primary side inductance (L m) tB=15μs 5. Choose proper FPS considering input power and Idspeak tW=3μs Ids ID 6. Determine the proper core and the minimum primary turns (Npmin) ID Ids 7. Determine the number of turns for each output and Vcc auxiliary circuit C Vds 8. Determine the wire diameter for each winding tB=15μs tW=3μs Y Is the winding window area (Aw) enough ? tsmax=18μs N Figure 3. Switching Waveforms of FSQ-Series for Different Input Voltages Is it possible to change the core ? Y N When the resonant period is 2μs fs 67kHz C B 59kHz 55kHz CCM 1 17μs 10. Determine the output capacitors 1 18μs A Constant frequency 9. Choose the secondary side rectifier diodes 1 15μs 11. Design the Snubber network Variable frequency within limited range 12. Design the synchronization network DCM 13. Design the feedback control circuit Design finished Vin Figure 4. Frequency Variation as Input Voltage Varies © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 Figure 5. Flow Chart of Design Procedure www.fairchildsemi.com 3 AN-4150 APPLICATION NOTE DR(n) LP(n) VO(n) NS(n) CO(n) Np LP2 DR2 CDC AC IN FSQ-Series Vstr CP(n) VO2 CP2 NS2 C O2 Drain DR Sync PWM LP1 VO1 1 GND NS1 VFB VCC CB Dzc Rcc Ca Da DSY Na Rd RSY1 RSY2 CP1 CO1 Rbias H11A817A R1 CSY CF RF KA431 R2 RSY3 Figure 6. Basic Quasi-Resonant Converter (QRC) Using FSQ-Series [STEP-1] Define the System Specifications [STEP-2] Determine DC Link Capacitor (CDC) Value and Calculate the DC Link Voltage Range When designing a power supply the following specifications should be determined first: In offline SMPS applications, a crude DC voltage (VDC) is obtained first on the DC link capacitor (CDC) by rectifying the AC mains. Then, the crude DC voltage is converted into pure DC outputs. Typically, the DC link capacitor is selected as 2-3µF per watt of input power for universal input range (85~265Vrms) and 1µF per watt of input power for European input range (195~265Vrms). With the DC link capacitor selected, the minimum DC link voltage is obtained as: Line voltage range (Vlinemin and Vlinemax). Line frequency (fL). Maximum output power (Po). Estimated efficiency (Eff): The power conversion efficiency must be estimated to calculate the maximum input power. If no reference data is available, set Eff = 0.7~0.75 for low-voltage output applications and Eff = 0.8~0.85 for highvoltage output applications. With the estimated efficiency, the maximum input power is given by: P in P = ------oE ff V DC 2 ⋅ ( V line P in ⋅ ( 1 – D ch ) ) – -----------------------------------C DC ⋅ f L min 2 (EQ 3) where CDC is the DC link capacitor value; Dch is the duty cycle ratio for CDC to be charged as defined in Figure 7, which is typically about 0.2; Pin, Vlinemin and fL are specified in STEP-1. (EQ 2) The maximum DC link voltage is given as: where Po(n) is the maximum output power for the n-th output. For single output SMPS, KL(1)=1. It is assumed that Vo1 is the reference output that is regulated by the feedback control in normal operation, as shown in Figure 6. © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 = (EQ 1) For multiple output SMPS, the load occupying factor for each output is defined as: Po ( n ) K L ( n ) = -----------Po min V DC max = 2V line max (EQ 4) where Vlinemax is specified in STEP-1. www.fairchildsemi.com 4 AN-4150 APPLICATION NOTE [STEP-4] Determine Inductance (Lm) Minimum DC link voltage DC link voltage t2 [STEP-3] Determine the Reflected Output Voltage (VRO) Figure 8 shows typical waveforms of the drain voltage of quasi-resonant flyback converter. When the MOSFET is turned off, the DC link voltage (VDC), together with the output voltage reflected to the primary (VRO), is imposed on the MOSFET. The maximum nominal voltage across the MOSFET (Vdsnom) is: nom = V DC max + V RO (EQ 5) The transformer primary side inductance is determined for the minimum input voltage and full-load condition. Once the reflected output voltage (VRO) is determined in STEP-3, the flyback converter can be simplified, as shown in Figure 9, by neglecting the voltage drops in MOSFET and diode. The design rules are a bit different for CCM and DCM. max where VDC is as specified in Equation 4. As shown in Figure 8, the capacitive switching loss of the MOSFET can be reduced by increasing VRO. However, this increases the voltage stress on the MOSFET. Therefore, VRO should be determined by a trade-off between the voltage margin of the MOSFET and the efficiency. It is typical to set VRO as 60~90V so that Vdsnorm is 430~460V (65~70% of MOSFET rated voltage). + VDC CCM at full load and minimum input voltage condition, the maximum duty ratio is given by: V RO D max = ------------------------------------min V RO + V DC - FPS With Dmax, the primary-side inductance (Lm) of the transformer is obtained as: Drain + GND Coss + Vds - min Lm VRO Vds VDC max VRO Vdsnom (EQ 7) DCM Design: When designing the converter to operate in DCM at full load and minimum input voltage condition, the maximum duty ratio should be chosen as smaller than the value obtained in Equation 6, as shown in Figure 9: VRO V RO D max < ------------------------------------min V RO + V DC 0V Figure 8. Typical Waveform of MOSFET Drain Voltage for Quasi-Resonant Converter © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 2 ( V DC ⋅ D max ) = --------------------------------------------2Pin f s K RF where VDCmin is specified in Equation 3, Pin is specified in STEP-1, fs is the free-running switching frequency of the FPS device, and KRF is the ripple factor, shown in Figure 9. The ripple factor is closely related to the transformer size and the RMS value of the MOSFET current. It is typical to set KRF = 0.5-0.7 for the universal input range. VRO nom (EQ 6) where VDCmin and VRO are specified in Equation 3 and STEP-3, respectively. VO VRO - CCM Design: When designing a converter to operate in + Lm Primary-Side In respect of EMI, DCM operation is preferred since the MOSFET is turned on at the minimum drain voltage and the secondary-side diode is softly turned off when operating in DCM. The transformer size can be reduced when using DCM because the average energy storage is low compared to CCM. However, DCM inherently causes higher RMS current, which increases the conduction loss of the MOSFET and the current stress on the output capacitors. When considering efficiency as well as magnetic components size, it is typical to design the converter to operate in CCM for low input voltage condition and in DCM for high input voltage condition. Figure 7. DC Link Voltage Waveform V ds Transformer The conventional quasi-resonant converter employs a variable frequency control, which makes the optimum design of the magnetic components difficult. However, FSQseries can operate in both CCM and DCM with near constant switching frequency thanks to the advanced control technique, which allows engineers to use the conventional transformer design procedure of PWM converters. t1 Dch = t1 / t2 = 0.2 the (EQ 8) Since reducing Dmax increases the conduction loss in www.fairchildsemi.com 5 AN-4150 APPLICATION NOTE 1, 3, 6, and 7, respectively, and fs is the FPS free-running switching frequency. MOSFET, too small Dmax should be avoided. Once Dmax is determined, the primary-side inductance (Lm) of the transformer is obtained as: Lm [STEP-5] Choose the Proper FPS Considering Input Power and Peak Drain Current 2 min ( V DC ⋅ D max ) = --------------------------------------------2P in f s (EQ 9) With the resulting maximum peak drain current of the MOSFET (Idspeak) from Equation 10, choose the proper FPS for which the pulse-by-pulse current limit level (ILIM) is higher than Idspeak. Since FPS has ± 12% tolerance of ILIM, there should be some margin in choosing the FPS device. where VDCmin is specified in Equation 3, Pin is specified in STEP-1, and fs is the free-running switching frequency of the FPS device. [STEP-6] Determine the Proper Core and the Minimum primary Turn Lm The initial selection of the core is bound to be crude since there are too many variables. One way to select the proper core is to refer to the manufacture's core selection guide. If there is no reference, use Table 1 as a starting point. The core recommended in Table 1 is typical for the universal input range, 55kHz switching frequency, and single-output application. When the input voltage range is 195-265 VAC or the switching frequency is higher than 55kHz, a smaller core can be used. For an application with multiple outputs, a larger core than recommended in the table should usually be used. VRO Im VDCmin ID Ids K RF = ΔI KRF < 1 ΔI 2IEDC Idspeak IEDC ID Ids With the chosen core, calculate the minimum number of turns for the transformer primary side to avoid the core saturation with the following: Im Dmax = VRO VRO + VDC min NP K RF = 1 Idspeak ΔI ID Im Dmax ≤ VRO VRO +VDC min Figure 9. MOSFET Drain Current and Ripple Factor (KRF) I ds rms = peak ΔI = I EDC + ----2 2 3 ( I EDC ) + I 2 D max ⎛Δ -----⎞ ⎝ 2⎠ -------------3 P in I EDC = ------------------------------------min V DC ⋅ D max V min D DC max Δ I = ----------------------------------- Lm fs (EQ 10) (EQ 14) Aw (EQ 11) (EQ 12) Ae (EQ 13) Figure 10. Window Area and Cross-Sectional Area where Pin, VDCmin, Dmax, and Lm are specified in Equations © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 (turns) If there is no reference data, use Bsat =0.3~0.35 T. Since the MOSFET drain current exceeds Idspeak and reaches ILIM in a transition or fault condition, ILIM is used in Equation 14 instead of Idspeak to prevent core saturation during transition. Once Lm is determined, the maximum peak current and RMS current of the MOSFET in minimum-input and full-load condition are obtained by: I ds L m I LIM 6 = ------------------ × 10 B sat A e where Lm is specified in Equation 7, ILIM is the FPS pulseby-pulse current limit level, Ae is the cross-sectional area of the core in mm2, as shown in Figure 10, and Bsat is the saturation flux density in tesla. Figure 11 shows the typical characteristics of ferrite core from TDK (PC40). Since the saturation flux density (Bsat) decreases as the temperature goes high, the high temperature characteristics should be considered. ±12% tolerance of ILIM should be considered. IEDC Ids min www.fairchildsemi.com 6 AN-4150 APPLICATION NOTE resulting Np is larger than the Npmin obtained from Equation 14. The number of turns for the other output (n-th output) is determined by: Magnetization Curves (typical) Material :PC40 25 °C 500 60 °C ( turns ) (EQ 16) 100 °C 400 Flux density B (mT) Vo ( n ) + VF ( n ) N s ( n ) = --------------------------------⋅ N s1 V o1 + V F1 120 °C The number of turns for Vcc winding is determined as: 300 V cc * + V Fa - ⋅ N s1 N a = --------------------------V o1 + V F1 200 100 800 Magnetic field H (A/m) 1600 Figure 11. Typical B-H Characteristics of Ferrite Core (TDK/PC40) Output Power EI Core 0-10W EI12.5 EI16 EI19 EE8 EE10 EE13 EE16 EPC10 EPC13 EPC17 EI22 EE19 EPC19 EE22 EPC25 EER25.5 EPC30 EER28 10-20W EE Core EPC Core + VF(n) - - EER Core DR(n) Np VRO 20-30W (EQ 17) where Vcc* is the nominal value of the supply voltage of the FPS device and VFa is the forward voltage drop of Da as defined in Figure 12. It is typical to set Vcc* 3~4V below Vcc maximum rating (refer to the datasheet). 0 0 ( turns ) NS(n) + VO(n) - + + VF1 - - VFa + EI25 30-50W EI28 EI30 EE25 50-70W EI35 EE30 + Vcc* EER28L Da DR1 + VO1 NS1 Na - - Table 1. Core Quick selection Table (for Universal Input Range, fs=55kHz and Single Output) Figure 12. Simplified Diagram of the Transformer [STEP-7] Determine the Number of Turns for Each Output Figure 12 shows the simplified diagram of the transformer. First, determine the turns ratio (n) between the primary side and the feedback-controlled secondary side as a reference. NP V RO = ------------------------n = --------N s1 V o1 + V F1 With the determined turns of the primary side, the gap length of the core is obtained as: 2 ⎛ NP 1⎞ G = 0.4 × π A e ⎜ ---------------- – ------⎟ 9 ⎝ 10 L m A L⎠ (EQ 15) where Np and Ns1 are the number of turns for primary side and reference output, respectively, Vo1 is the output voltage and VF1 is the diode (DR1) forward voltage drop of the reference output. ( mm ) (EQ 18) where AL is the AL-value with no gap in nH/turns2; Ae is the cross-sectional area of the core in mm2, as shown in Figure 10; Lm is specified in Equation 7; and Np is the number of turns for the primary-side of the transformer. Then, determine the proper integer for Ns1 so that the © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 www.fairchildsemi.com 7 AN-4150 APPLICATION NOTE [STEP-8] Determine the Wire Diameter for Each Winding Based on the rms Current of Each Output where KL(n), VDCmax, VRO, and Idsrms are specified in Equations 2, 4, STEP-3 and Equation 11, respectively; Dmax is specified in Equation 6; Vo(n) is the output voltage of the nth output; and VF(n) is the diode (DR(n)) forward voltage. The typical voltage and current margins for the rectifier diode are: The rms current of the n-th secondary winding is obtained as: I sec ( n ) rms = I ds rms V RO ⋅ K L ( n ) 1 – D max ----------------------- ⋅ -------------------------------------( V D max o ( n ) + VF ( n ) ) (EQ 19) V RRM > 1.3 ⋅ V D ( n ) I F > 1.5 ⋅ I D ( n ) where VRO and Idsrms are specified in STEP-3 and Equation 11, respectively; Vo(n) is the output voltage of the n-th output; VF(n) is the diode (DR(n)) forward voltage drop; Dmax is specified in Equation 6; and KL(n) is the load-occupying factor for n-th output defined in Equation 2. [STEP-10] Determine the Output Capacitor Considering the Voltage and Current Ripple The ripple current of the n-th output capacitor (Co(n)) is obtained as: I cap ( n ) For high current output, it is better to use parallel windings with multiple strands of thinner wire to minimize skin effect. I 2 (EQ 25) I peak V R K ( Vo ( n ) + VF ( n ) ) (EQ 26) When the power MOSFET is turned off, there is a high voltage spike on the drain due to the transformer leakage inductance. This excessive voltage on the MOSFET may lead to an avalanche breakdown and, eventually, failure of the FPS. Therefore, it is necessary to use an additional network to clamp the voltage. (EQ 21) min © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 ) – Io ( n ) [STEP-11] Design the RCD Snubber max V RO K L ( n ) V DC ------------ ⋅ -------------------------------------V RO ( V o ( n ) + V F ( n ) ) rms 2 If it is impossible to meet the ripple specification with a single output capacitor due to the high ESR of the electrolytic capacitor, additional LC filter stages (post filter) can be used. When using the post filters, be careful not to place the corner frequency too low. Too low a corner frequency may make the system unstable or limit the control bandwidth. It is typical to set the corner frequency of the post filter at around 1020% of the switching frequency. The maximum reverse voltage and the rms current of the rectifier diode (DR(n)) of the n-th output are obtained as: rms ( ID ( n ) where Co(n) is the capacitance; Rc(n) is the effective series resistance (ESR) of the n-th output capacitor; KL(n), VRO, and Idspeak are specified in Equation 2, STEP-3, and Equation 10, respectively; Dmax is specified in Equation 6; Io(n) and Vo(n) are the load current and output voltage of the n-th output, respectively; and VF(n) is the diode (DR(n)) forward voltage. [STEP-9] Choose the Rectifier Diode in the Secondary Side Based on the Voltage and Current Ratings. = I ds D Co ( n ) fs If the required window (Awr) is larger than the actual window area (Aw), go back to STEP-6 and increase the core. If it is impossible to change the core due to cost or size constraints and the converter is designed for CCM and the winding window (Aw) is slightly insufficient, go back to STEP-4 and reduce Lm by increasing the ripple factor (KRF). The minimum number of turns for the primary (Npmin) of Equation 14 decreases, which results in the reduced required winding window area (Awr). rms = o ( n ) max ds RO C ( n ) L ( n ) Δ V o ( n ) = ------------------------ + ---------------------------------------------------------- (EQ 20) where Ac is the actual conductor area and KF is the fill factor. Typically the fill factor is 0.2~0.25 for single-output application and 0.15~0.2 for multiple outputs application. ID (n) rms where Io(n) is the load current of the n-th output and ID(n)rms is specified in Equation 22. The ripple current should be smaller than the ripple current specification of the capacitor. The voltage ripple on the n-th output is given by: Verify that if the winding window area of the core, Aw is enough to accommodate the wires (refer to Figure 10). The required winding window area (Awr) is given by: V DC ⋅ ( Vo ( n ) + VF ( n ) ) V D ( n ) = V o ( n ) + ---------------------------------------------------------------V RO (EQ 24) where VRRM is the maximum reverse voltage and IF is the average forward current of the diode. The current density is typically 5A/mm2 when the wire is greater than 1m long. When the wire is short, with a small number of turns, a current density of 6-10 A/mm2 is also acceptable. Avoid using wire with a diameter larger than 1mm to avoid severe eddy current losses and to make winding easier. A wr = A c ⁄ K F (EQ 23) rms (EQ 22) www.fairchildsemi.com 8 AN-4150 APPLICATION NOTE The RCD snubber circuit and MOSFET drain voltage waveform are shown in Figures 13 and 14, respectively. The RCD snubber network absorbs the current in the leakage inductance by turning on the snubber diode (Dsn) once the MOSFET drain voltage exceeds the voltage of node X, as depicted in Figure 13. In the analysis of snubber network, it is assumed that the snubber capacitor is large enough that its voltage does not change significantly during one switching cycle. The capacitor used in the snubber should be ceramic or a material that offers low ESR. Electrolytic or tantalum capacitors are unacceptable for these reasons. R sn V DC + C DC C sn X - VX V sn + reasonable. The snubber capacitor voltage (Vsn) of Equation 27 is for the minimum input voltage and full-load condition. When the converter is designed to operate in CCM under this condition, the peak drain current, together with the snubber capacitor voltage, decrease as the input voltage increases, as shown in Figure 14. The peak drain current at the maximum input voltage and full load condition (Ids2peak) is obtained as I ds2 = 2 ⋅ P in --------------fs ⋅ Lm (EQ 29) where Pin, and Lm are specified in Equations 1 and 7, respectively, and fs is the FPS free-running switching frequency. Np The snubber capacitor voltage under maximum input voltage and full load condition is obtained as: V RO + peak 2 2 D sn FPS peak V sn2 V RO + ( V RO ) + 2R sn L lk f s ( I ds2 ) = -----------------------------------------------------------------------------------------------------2 (EQ 30) L lk Drain where fs is the FPS free-running switching frequency, Llk is the primary-side leakage inductance, VRO is the reflected output voltage, and Rsn is the snubber resistor. + V ds GND - Idspeak Figure 13. Circuit Diagram of the Snubber Network Ids2peak The first step in designing the snubber circuit is to determine the snubber capacitor voltage (Vsn) at the minimum input voltage and full-load condition. Once Vsn is determined, the power dissipated in the snubber network at the minimum input voltage and full-load condition is obtained as: Ids2peak < Idspeak ==> Vsn2 < Vsn 2 ( V sn ) V sn peak 2 1 P sn = ---------------- = --- f s L lK ( I ds ) -------------------------R sn V sn – V RO 2 Vsn2 VRO (EQ 27) Vsn peak is specified in Equation 10, fs is the FPS freewhere Ids running switching frequency, Llk is the leakage inductance, Vsn is the snubber capacitor voltage at the minimum input voltage and full-load condition, VRO is the reflected output voltage, and Rsn is the snubber resistor. Vsn should be larger than VRO and it is typical to set Vsn to be 2~2.5 times VRO. Too small a Vsn results in a severe loss in the snubber network, as shown in Equation 27. The leakage inductance is measured at the switching frequency on the primary winding with all other windings shorted. VRO VDC Minimum input voltage & full load V C sn R sn f s From Equation 30, the maximum voltage stress on the internal MOSFET is given by: V ds (EQ 28) max = V DC max + V sn2 (EQ 31) where VDCmax is specified in Equation 4. where fs is the FPS free-running switching frequency. In general, 5~10% ripple of the selected capacitor voltage is © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 Maximum input voltage & full load Figure 14. MOSFET Drain Voltage and Snubber Capacitor Voltage The snubber resistor with proper rated wattage should be chosen based on the power loss. The maximum ripple of the snubber capacitor voltage is obtained as: sn Δ V sn = ----------------------- VDC max min www.fairchildsemi.com 9 AN-4150 APPLICATION NOTE Verify that Vdsmax is below 90% of the rated voltage of the MOSFET (BVdss), as shown in Figure 15. The voltage rating of the snubber diode should be higher than BVdss. Usually, an ultra fast diode with 1A current rating is used for the snubber network. Np FSQ-Series In the snubber design in this section, neither the lossy discharge of the inductor, nor stray capacitance, is considered. In the actual converter, the loss in the snubber network is generally less than the designed value. Drain CO + Ids V ds - - 0.7/0.2V Sync GND VCC Na Rcc Voltage Margin > 10% of BVdss Da RSY1 Ca Effect of stray inductance (5-10V) Vsn2 Vo1 Lm Sync comparator + BVdss Ns1 RSY2 Vsync VRO DSY CSY RSY3 VDC max Figure. 16 Synchronization Circuit 0V Figure 15. MOSFET Drain Voltage and Snubber Capacitor Voltage Vds TR π LmCeo = 4 2 [STEP-12] Design the Synchronization Network The optimum MOSFET turn-on point is indirectly detected by monitoring the Vcc winding voltage, as shown in Figures 16 and 17. The output of the sync-detect comparator (CO) becomes high when the sync voltage (Vsync) rises above 0.7V and becomes low when the Vsync drops below 0.2V. The MOSFET is turned on at the falling edge of the syncdetect comparator output (CO). VOVP Vsyncpk 0.2V TQ To synchronize the Vsync with the MOSFET drain voltage, the sync capacitor (CSY) should be chosen so that TQ is same as a quarter of the resonance period (TR/4), as shown in Figure 17. TR /4 and TQ are given as: TR π ⋅ L m ⋅ C eo ------ = --------------------------------4 2 Vsyns 0.7 V RC time delay internal delay (200ns) CO Gate (EQ 32) Figure. 17 Synchronization Waveforms R SY1 ⋅ ( R SY2 + R SY3 ) T Q = ---------------------------------------------------------- ⋅ C SY + 200ns (EQ 33) R SY1 + R SY2 + R SY3 The peak value of the sync signal is determined by the voltage divider network RSY1, RSY2, and RSY3 as where Lm is the primary-side inductance of the transformer, Ceo is the effective MOSFET output capacitance, and 200ns is the internal delay time. © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 V pk sync R SY3 Na = --------------------------------------------------------------- ⋅ ------------ ⋅ ( V + V ) 01 F1 R SY1 + R SY2 + R N S1 SY3 (EQ 34) www.fairchildsemi.com 10 AN-4150 APPLICATION NOTE where Na and Ns1 are the numbers of the turns for Vcc winding and Vo1, respectively, and VF1 is the forward voltage drop of D1. Choose the voltage divider RSY1, RSY2, and RSY3 so that the peak value of sync voltage (Vsyncpk) is lower than the OVP threshold voltage (6V) to avoid triggering OVP in normal operation. It is typical to set Vsyncpk to be 4~5V. vˆ o1 G vc = -------vˆ FB (EQ 36) K ⋅ R L V DC ( N p ⁄ N s1 ) ( 1 + s ⁄ w z ) ( 1 – s ⁄ w rz ) = ----------------------------------------------------- ⋅ ---------------------------------------------------------1 + s ⁄ wp 2V RO + v DC where VDC is the DC input voltage; RL is the effective total load resistance of the controlled output, defined as Vo12/Po; Np and Ns1 are specified in STEP-7; VRO is specified in STEP-3; Vo1 is the reference output voltage; Po is specified in STEP-1; and K is specified in Equation 35. The pole and zeros of Equation 36 are defined as: [STEP-13] Design the Feedback Loop Since FSQ-series employs current-mode control, the feedback loop can be simply implemented with a one-pole and one-zero compensation circuit, as shown in Figure 18. In the feedback circuit analysis, it is assumed that the current transfer ratio (CTR) of the opto-coupler is 100%. 2 RL ( 1 – D ) 1 (1 + D) w z = -------------------- , w rz = ---------------------------------------- and w p = ------------------2 R c1 C o1 R L C o1 DL m ( N s1 ⁄ N p ) The current control factor of FPS, K is defined as: (EQ 37) I pk I LIM K = --------- = ----------------V FB V FBsat where Lm is specified in Equation 7, D is the duty cycle of the FPS, Co1 is the reference output capacitor, and RC1 is the ESR of Co1. (EQ 35) where Ipk is the peak drain current and VFB is the feedback voltage, respectively, for a given operating condition; ILIM is the current limit of the FPS; and VFBsat is the feedback saturation voltage, which is typically 2.5V. When the converter has more than one output, the low frequency control-to-output transfer function is proportional to the parallel combination of all load resistance, adjusted by the square of the turns ratio. Therefore, the effective load resistance is used in Equation 36 instead of the actual load resistance of Vo1. To express the small signal AC transfer functions, the small signal variations of feedback voltage (vFB) and controlled output voltage (vo1) are introduced as vˆ FB and vˆ o1 . vo1' FPS vFB RB CB RD Notice that there is a right half plane (RHP) zero (wrz) in the control-to-output transfer function of Equation 36. Because the RHP zero reduces the phase by 90°, the crossover frequency should be placed below the RHP zero. vo1 ibias Figure 19 shows the variation of a CCM flyback converter control-to-output transfer function for different input voltages. This figure shows the system poles and zeros, together with the DC gain change, for different input voltages. The gain is highest at the high input voltage condition and the RHP zero is lowest at the low input voltage condition. Rbias iD 1:1 CF RF R1 KA431 Figure 20 shows the variation of a CCM flyback converter control-to-output transfer function for different loads. This figure shows that the low frequency gain does not change for different loads and the RHP zero is lowest at the full-load condition. R2 Ipk MOSFET current For DCM operation, the control-to-output transfer function of the flyback converter, using current-mode control, is given by: Figure 18. Control Block Diagram V o1 ( 1 + s ⁄ w z ) vˆ o1 --------- ---------------------------G vc = -------ˆv - = V FB ⋅ ( 1 + s ⁄ w p ) FB For CCM operation, the control-to-output transfer function of the flyback converter, using current-mode control, is given by: where (EQ 38) 1 w z = ------------------- , wp = 2 ⁄ R L C o1, R c1 C o1 Vo1 is the reference output voltage, VFB is the feedback voltage for a given condition, RL is the effective total © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 www.fairchildsemi.com 11 AN-4150 APPLICATION NOTE resistance of the controlled output, Co1 is the controlled output capacitance, and Rc1 is the ESR of Co1. Figure 21 shows the variation of the control-to-output transfer function of a flyback converter in DCM for different loads. Contrary to the flyback converter in CCM, there is no RHP zero and the DC gain does not change as the input voltage varies. As can be seen, the overall gain, except for the DC gain, is highest at the full-load condition. The feedback compensation network transfer function of Figure 18 is obtained as: ˆ v FB w i 1 + s ⁄ w zc - ⋅ ---------------------------------ˆ - = - ---s 1 + 1 ⁄ w pc v o1 (EQ 39) RB 1 1 - ;w =--------------------------------- ; w pc =--------------- ; where w i = ---------------------R 1 R D C F zc ( R F + R 1 )C F RB CB Figure 21. DCM Flyback Converter Control-to-Output TransferFunction Variation for Different Loads RB is the internal feedback bias resistor of FPS, which is typically 2.8kΩ; and R1, RD, RF, CF and CB are shown in Figure 18. When the input voltage and the load current vary over a wide range, it is not easy to determine the worst case for the feedback-loop design. The gain, together with zeros and poles, varies according to the operating condition. Even though the converter is designed to operate in CCM or at the boundary of DCM and CCM in the minimum input voltage and full-load condition, the converter enters into DCM, changing the system transfer functions as the load current decreases and/or input voltage increases. One simple and practical solution to this problem is designing the feedback loop for low input voltage and fullload condition with enough phase and gain margin. When the converter operates in CCM, the RHP zero is lowest in low input voltage and full-load condition. The gain increases about 6dB as the operating condition is changed from the lowest input voltage to the highest input voltage condition under universal input condition. When the operating mode changes from CCM to DCM, the RHP zero disappears, making the system stable. Therefore, by designing the feedback loop with more than 45° of phase margin in low input voltage and full load condition, the stability over the operating ranges can be guaranteed. Figure 19. CCM Flyback Converter Control-to-Output Transfer Function Variation for Different Input Voltages The procedure to design the feedback loop is as follows: Determine the crossover frequency (fc). For CCM mode flyback, set fc below 1/3 of right half plane (RHP) zero to minimize the effect of the RHP zero. For DCM mode, fc can be placed at a higher frequency, since there is no RHP zero. When an additional LC filter is employed, the crossover frequency should be placed below 1/3 of the corner frequency of the LC filter, since it introduces a -180° phase drop. Never place the crossover frequency beyond the corner frequency of the LC filter. If the crossover Figure 20. CCM Flyback Converter Control-to-Output Transfer Function Variation for Different Loads © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 www.fairchildsemi.com 12 AN-4150 APPLICATION NOTE frequency is too close to the corner frequency, the controller should be designed to have a phase margin greater than 90° when ignoring the effect of the post filter. The resistors Rbias and RD, used together with opto-coupler H11A817A and shunt regulator KA431, should be designed to provide proper operating current for the KA431 and to guarantee the full swing of the feedback voltage for the FPS device chosen. In general, the minimum cathode voltage and current for the KA431 are 2.5V and 1mA, respectively. Therefore, Rbias and RD should be designed to satisfy the following conditions: Determine the DC gain of the compensator (wi/wzc) to cancel the control-to-output gain at fc. Place a compensator zero (fzc) around fc/3. Place a compensator pole (fpc) above 3fc. Loop gain T 40 dB fzc 20 dB 0 dB Control to output (EQ 42) V OP ------------- > 1mA R bias (EQ 43) Compensator fpc fp V o1 – V OP – 2.5 ---------------------------------------- > I FB RD where Vo1 is the reference output voltage; VOP is optodiode forward voltage drop, which is typically 1V; and IFB is the feedback current of FPS, which is typically 1mA. For example, Rbias < 1kΩ and RD < 1.5kΩ for Vo1=5V. fc frz -20 dB fz -40 dB 1Hz 10Hz 100Hz 1kHz 10kHz Miscellaneous Notes 100kHz Vcc capacitor (Ca): The typical value for Ca is 10-50µF, which is enough for most applications. A smaller capacitor than this may result in an under-voltage lockout of FPS during the startup. Too large a capacitor may increase the start-up time. Figure 22. Compensator design Vcc resistor (Ra): The typical value for Ra is 5-20Ω. In the case of multiple outputs flyback converter, the voltage of the lightly loaded output, such as Vcc, varies as the load currents of other outputs change due to the imperfect coupling of the transformer. Ra reduces the sensitivity of Vcc to other outputs and improves the regulations of Vcc. Determining the feedback circuit component includes some restrictions, such as: The voltage divider network of R1 and R2 should be designed to provide 2.5V to the reference pin of the KA431. The relationship between R1 and R2 is given as: 2.5 ⋅ R 1 R 2 = -----------------------V o1 – 2.5 (EQ 40) where Vo1 is the reference output voltage. The capacitor connected to feedback pin (CB) is related to the shutdown delay time in an overload condition by: t delay = ( V SD – 2.5 ) ⋅ C B ⁄ I delay (EQ 41) where VSD is the shutdown feedback voltage and Idelay is the shutdown delay current. These values are given in the product datasheet. A 10 ~ 50ms delay time is typical for most applications. Because CB also determines the highfrequency pole (wpc) of the compensator transfer function, as shown in Equation 39, too large a CB can limit the control bandwidth by placing wpc at too low a frequency. A typical value for CB is 10-50nF. © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 www.fairchildsemi.com 13 AN-4150 APPLICATION NOTE Design Example Application Device Input Voltage Output Power Output Voltage (Rated Current) DVD player FSQ0365RN 85-265VAC 18.1W 5.1V (1.0A) (60Hz) 3.4V (1.0A) 12V (0.4A) 16V (0.3A) Key Design Notes To maximize the efficiency, the power supply is designed to operate in CCM for minimum input-voltage and full-load condition and in DCM for high input voltage condition. 1. Schematic C209 47pF T101 EER2828 RT101 5D-9 1 R105 100kΩ C104 10nF 1kV R102 56kΩ C103 33μF 400V L201 11 C210 47pF 2 R108 62Ω 12V, 0.4A 3 10 IC101 FSQ0365RN 1 BD101 Bridge Diode 3 Vstr 4 Sync 3 4 C102 100nF,400V C105 47nF 50V C203 470μF 35V C204 470μF 35V L203 C106 C107 100nF 22uF SMD 50V Vcc 2 GND 1 D202 UF4003 12 8 Drain 7 Drain 6 Drain Vfb C202 470μF 35V C201 470μF 35V L202 D101 1N 4007 2 5 16V, 0.3A D201 UF4003 6 R103 5Ω D102 1N 4004 R104 12kΩ 5.1V, 1A D203 SB360 4 C205 1000μF 10V C206 1000μF 10V L204 5 9 ZD101 1N4746A LF101 40mH 3.4V, 1A D204 SB360 D103 1N4148 R106 R107 6.2kΩ 6.2kΩ C110 33pF 50V C208 1000μF 10V C207 1000μF 10V 8 C302 3.3nF C101 100nF 400V R201 510Ω R203 6.2kΩ R202 1kΩ R204 20kΩ C209 100nF IC202 FOD817A TNR F101 FUSE IC201 KA431 R205 6kΩ AC IN © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 www.fairchildsemi.com 14 AN-4150 APPLICATION NOTE 2. Transformer Specifications No Np/2 Pin (s→f) 3→2 Wire 0.25φ Turns ×1 Winding Method 50 Center Solenoid Winding 4 Center Solenoid Winding 2 Center Solenoid Winding 16 Center Solenoid Winding 14 Center Solenoid Winding 18 Center Solenoid Winding 50 Center Solenoid Winding Insulation: Polyester Tape t = 0.050mm, 2 Layers N3.4V 9→8 0.33φ × 2 Insulation: Polyester Tape t = 0.050mm, 2 Layers N5V 6→9 0.33φ × 1 Insulation: Polyester Tape t = 0.050mm, 2 Layers Na 4→5 0.25φ × 1 Insulation: Polyester Tape t = 0.050mm, 2 Layers N12V 10 → 12 0.33φ × 3 Insulation: Polyester Tape t = 0.050mm, 3 Layers N16V 11 → 12 0.33φ × 3 Insulation: Polyester Tape t = 0.050mm, 2 Layers Np/2 2→1 0.25φ × 1 Insulation: Polyester Tape t = 0.050mm, 2 Layers Core: EER2828 (Ae=86.7mm2) Bobbin: EER2828 Electrical Characteristics Pin Specification Remarks Inductance 1-3 1.4mH ± 10% 100kHz, 1V Leakage 1-3 25µH Max Short all other pins © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 www.fairchildsemi.com 15 AN-4150 APPLICATION NOTE Hang-Seok Choi, Ph.D. Power Conversion / Fairchild Semiconductor Phone: +82-32-680-1383 Facsimile : +82-32-680-1317 Email: [email protected] DISCLAIMER FAIRCHILD SEMICONDUCTOR RESERVES THE RIGHT TO MAKE CHANGES WITHOUT FURTHER NOTICE TO ANY PRODUCTS HEREIN TO IMPROVE RELIABILITY, FUNCTION, OR DESIGN. FAIRCHILD DOES NOT ASSUME ANY LIABILITY ARISING OUT OF THE APPLICATION OR USE OF ANY PRODUCT OR CIRCUIT DESCRIBED HEREIN; NEITHER DOES IT CONVEY ANY LICENSE UNDER ITS PATENT RIGHTS, NOR THE RIGHTS OF OTHERS. LIFE SUPPORT POLICY FAIRCHILD’S PRODUCTS ARE NOT AUTHORIZED FOR USE AS CRITICAL COMPONENTS IN LIFE SUPPORT DEVICES OR SYSTEMS WITHOUT THE EXPRESS WRITTEN APPROVAL OF THE PRESIDENT OF FAIRCHILD SEMICONDUCTOR CORPORATION. As used herein: 1.Life support devices or systems are devices or systems which, (a) are intended for surgical implant into the body, or (b) support or sustain life, or (c) whose failure to perform when properly used in accordance with instructions for use provided in the labeling, can be reasonably expected to result in significant injury to the user. © 2006 Fairchild Semiconductor Corporation FSQ-Series Rev. 1.0.0 10/23/06 2.A critical component is any component of a life support device or system whose failure to perform can be reasonably expected to cause the failure of the life support device or system, or to affect its safety or effectiveness. www.fairchildsemi.com 16