AADesign Guide for ICE2PCSxxApp Application note, Ver 1.0, May 2008 Design Guide for Boost Type CCM PFC with ICE2PCSxx Power Management & Supply N e v e r s t o p t h i n k i n g . Edition 2008-08-01 Published by Infineon Technologies Asia Pacific, 168 Kallang Way, 349253 Singapore, Singapore © Infineon Technologies AP 2005. All Rights Reserved. Attention please! The information herein is given to describe certain components and shall not be considered as a guarantee of characteristics. Terms of delivery and rights to technical change reserved. We hereby disclaim any and all warranties, including but not limited to warranties of non-infringement, regarding circuits, descriptions and charts stated herein. Information For further information on technology, delivery terms and conditions and prices please contact your nearest Infineon Technologies Office (www.infineon.com). Warnings Due to technical requirements components may contain dangerous substances. For information on the types in question please contact your nearest Infineon Technologies Office. Infineon Technologies Components may only be used in life-support devices or systems with the express written approval of Infineon Technologies, if a failure of such components can reasonably be expected to cause the failure of that life-support device or system, or to affect the safety or effectiveness of that device or system. Life support devices or systems are intended to be implanted in the human body, or to support and/or maintain and sustain and/or protect human life. If they fail, it is reasonable to assume that the health of the user or other persons may be endangered. Revision History: Previous Version: Page 2008-08 none Subjects (major changes since last revision) Design Guide for Boost Type CCM PFC with ICE2PCSxx License to Infineon Technologies Asia Pacific Pte Ltd Liu Jianwei Luo Junyang Jeoh Meng Kiat We Listen to Your Comments Any information within this document that you feel is wrong, unclear or missing at all? Your feedback will help us to continuously improve the quality of this document. Please send your proposal (including a reference to this document) to: mailto:[email protected] V1.0 AN-PS0029 ICE2PCSxx Table of Contents Page 1 Introduction ...................................................................................................................................5 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10 2.11 2.12 2.13 Boost PFC design with ICE2PCXX ..............................................................................................7 Target specification .........................................................................................................................7 Bridge rectifier .................................................................................................................................7 Power MOSFET and Gate Drive Circuit .........................................................................................7 Boost Diode.....................................................................................................................................8 Boost inductor .................................................................................................................................9 AC line current filter.......................................................................................................................11 Boost Output Bulk Capacitance ....................................................................................................12 Current Sense Resistor.................................................................................................................12 Output voltage sensing divider......................................................................................................13 Frequency setting (only for ICE2PCS01)......................................................................................13 AC Brown-out Shutdown (only for ICE2PCS02) ...........................................................................14 IC supply .......................................................................................................................................15 PCB layout guide ..........................................................................................................................16 3 3.1 3.2 3.3 3.4 3.5 Voltage loop and current loop compensation..........................................................................17 How to achieve PFC function without sinusoidal reference sensing ............................................18 Current Loop Regulation and Transfer Function...........................................................................19 Voltage Loop Compensation.........................................................................................................22 Design Example ............................................................................................................................28 Vcomp and M1, M2 value at full load condition ............................................................................29 Application Note 4 2008-08-01 Abstract ICE2PCS01/02 are the 2nd generation of Continuous Conduction Mode (CCM) PFC controllers, which employ BiCMOS technology. Its control scheme does not need the direct sine-wave sensing reference signal from the AC mains compared to the conventional PFC solution. Average current control is implemented to achieve the unity power factor. In this application note, the design process for the boost PFC with ICE2PCXX is presented and the design details for a 300W output power PFC with the universal input voltage range of 85~265VAC are included. 1 Introduction The Pin layout of ICE2PCS01 and ICE2PCS02 is shown in Figure 1. GND 1 8 GATE ICOMP 2 7 VCC ISENSE 3 6 VSENSE FREQ 4 5 VCOMP ICE2PCS01 Figure 1 GND 1 8 GATE ICOMP 2 7 VCC ISENSE 3 6 VSENSE VINS 4 5 VCOMP ICE2PCS02 Pin Layout of ICE2PCS01 and ICE2PCS02 From the layout, it can be seen that most of Pins in ICE2PCS02 are the same as ICE2PCS01 except Pin 4. In ICE2PCS01, Pin 4 is to set the switching frequency. However, for ICE2PCS02, Pin 4 is for AC brown out detection and the switching frequency is fixed by internal oscillator at 65kHz. The typical application circuits of ICE2PCS01 and ICE2PCS02 are shown in Figure 2 and Figure 3 respectively. Application Note 5 2008-08-01 Rectifier EMI Filter D1 L1 COUT R1 VOUT=400VDC T1 R2 RSENSE VIN=85V ...265V AC R3 ISENSE Auxiliary Supply GATE ICE2PCS01 VCC ICOMP VCOMP FREQ R4 C2 RFREQ C1 Figure 2 GND VSENSE C3 Typical application circuit of ICE2PCS01 Rectifier EMI Filter D1 L1 R1 VOUT=400VDC T1 D2 R3 ISENSE GATE R5 R6 Figure 3 GND VSENSE ICE2PCS02 VINS Application Note R2 RSENSE VIN=85V ...265V AC COUT ICOMP C1 C4 Auxiliary Supply VCC VCOMP R4 C2 C3 Typical application circuit of ICE2PCS02 6 2008-08-01 2 Boost PFC design with ICE2PCXX 2.1 Target specification The fundamental electrical data of the circuit are the input voltage range Vin, the output power Pout, the output voltage Vout, the operating switching frequency fSW and the value of the high frequency ripple of the AC line current Iripple. Table 1 shows the relevant values for the system calculated in this Application Note. The efficiency at rated output power Pout is estimated to 91 % over the complete input voltage range. Input voltage 85VAC~265VAC Input frequency 50Hz Output voltage and current 390VDC, 0.76A Output power 300W Efficiency >90% at full load Switching Frequency 65kHz Maximum Ambient temperature around PFC 70ºC Table 1 Design parameter for the proposed design 2.2 Bridge rectifier In order to obtain 300W output power at 85 V minimum AC input voltage, the maximum input RMS current is I in _ RMS = Pout Vin _ min ⋅ η = 300 = 3.92 A 85 ⋅ 90% (1) and the sinusoidal peak value of AC current is I in _ pk = 2 ⋅ I in _ RMS = 2 ⋅ 3.92 = 5.54 A (2) For these values a bridge rectifier with an average current capability of 6A or higher is a good choice. Please note here, that due to a power dissipation of approximately PBR = 2 ⋅ VF ⋅ I in _ RMS = 2 ⋅ 1V ⋅ 3.92 A = 7.84W (3) the rectifier bridge should be connected to an appropriate heatsink. Assuming a maximum junction temperature TJmax of 125°C, a maximum ambient temperature TAmax of 70°C, the thermal junction-to-case RthJC of approximate 2.5 K/W and the thermal case to heatsink RthCHS of approximate 1K/W, the heatsink must have a maximum thermal resistance of RthHS _ BR = TJ max − T A max 125 − 70 − RthJC − RthCHS = − 2.5 − 1 = 3.52 K / W PBR 7.84 2.3 Power MOSFET and Gate Drive Circuit (4) Due to the switch mode operation, the loss is only valid during the on-time of the MOSFET. The duty cycle of the transistor in boost converters operating in CCM at minimum AC input RMS voltage is Don = 1 − Vin _ min Vout Application Note = 1− 85 = 0.782 390 (5) 7 2008-08-01 Since rms-values have the same effect on a system as DC-values, it is possible to calculate a characteristic duty cycle for the rms-value. Therefore, the on-state loss of the MOSFET in CCM-mode at a junctiontemperature of 125°C is 2 Pcond = I in _ RMS ⋅ Don ⋅ Rdson (125C ) (6) the MOSFET switching loss can be estimated as PSW = ( E on + E off ) ⋅ f SW (7) where, Eon and Eoff are the switch-on and switch-off energy loss which can be found in MOSFET datasheet, fSW is the switching frequency. For 300W design, if SPP20N60C3 is used, the conduction loss is Pcond = 3.92 2 ⋅ 0.782 ⋅ 0.42 = 5.05W assuming the switching current is about 6A and gate drive resistance Rg=3.6Ω, then the switching loss is PSW = (0.007mWs + 0.015mWs) * 65kHz = 1.43W the total loss is PMOS _ total = Pcond + PSW = 6.48W (8) the required heatsink for the MOSFET is RthHS _ MOS = TJ max − T A max 125 − 70 − RthJC _ MOS − RthCHS = − 0.6 − 1 = 6.89 K / W PMOS _ total 6.48 (9) RthCHS is the Rth of the insulation pad between MOSFET and heatsink. Gate drive resistance is used to drive MOSFET as fast as possible but also keep dv/dt within EMI specification. In this 300W example, 3.6Ω gate resistor is chosen for SPP20N60C3 MOSFET. Beside gate drive resistance, one 10kΩ resistor is also commonly connected between MOSFET gate and source to discharge gate capacitor. 2.4 Boost Diode The boost diode D1 has big influence on the system’s performance due to the reverse recovery behaviour. So the Ultra-fast diode with very low trr and Qrr is necessary to reduce the switching loss. The new diode technology of silicon carbide (SiC) Schottky shows its outstanding performance with almost no reverse recovery behaviour. The switching loss due to the boost diode can be ignored with SiC Schottky diode. Only conduction loss is calculated as below. Pdiode = VF ⋅ I in _ RMS ⋅ (1 − Don ) = 2V ⋅ 3.92 A ⋅ (1 − 0.782) = 1.71W (10) To decide the current rating of a SiC diode, there is a rule of thumb - the SiC diode can handle output power Pout of 100 W to120 W in a CCM-PFC-system per one rated ampere. For example, the SDT04S60 from Infineon Technologies is rated at a forward current IF = 4 A, so it is capable for a system of Pout = 4*100 W = 400 W system in minimum. Therefore, this diode should be suitable for the proposed design. The required heatsink for boost diode is RthHS _ diode = TJ max − T A max 125 − 70 − RthJC _ diode − RthCHS = − 4.1 − 1 = 27.06 K / W Pdiode 1.71 Application Note 8 (11) 2008-08-01 The SiC boost diodes often have a poor surge current handling capability. Therefore a so called bypass diode is necessary such as the diode D3 as Figure 4. For the proposed system, 1N5408 is suitable. D3 Rectifier D1 L1 R1 COUT T1 R2 RSENSE Figure 4 2.5 inrush current bypass diode Boost inductor The peak current that the inductor must carry is the peak line current at the lowest input voltage plus the high frequency ripple current. The high frequency ripple current peak to peak, IHF, can be related to maximum input power and minmum input voltage as equation below. I HF = k ⋅ 2 ⋅ Pin _ max (12) Vin _ min Where, k must be kept reasonably small, and is usually optimized in the range of 15% to 25% for cost effective design based on the current magnetic component status. If k is too high, the larger AC input filter is required to filter out this ripple noise. If k is too low, the value of the inductance is too large and leads to big size of the magnetic core. For example, we choose k = 22%, then, I HF = 22% ⋅ 2 ⋅ Pin _ max Vin _ min = 1 .2 A The peak current passing through inductor is I L _ pk = I in _ peak + I HF 1 .2 = 5.54 + = 6.14 A 2 2 (13) The boost choke inductance must be Lboost ≥ D ⋅ (1 − D ) ⋅ Vout I HF ⋅ f SW (14) D=0.5 will generate the maximum value for the above equation. Lboost ≥ 0.5 ⋅ (1 − 0.5) ⋅ 390V = 1.25mH 1.2 A ⋅ 65kHz The magnetic core of the boost choke can be either magnetic powder or ferrite material. (1) sendust powder toroid core The required effective magnetic volume of the core, Ve, is Application Note 9 2008-08-01 Ve ≥ µ r µ 0 Lboost ( I L _ pk Bmax ) 2 = 125 ⋅ 1.257 e − 6 ⋅ 1.25mH ( 6.14 A 2 ) = 11.6e − 6m 3 = 11.6cm 3 0.8T (15) where, µ r is the relative permeability of the material. It should be noted that µ r changes with different DC magnetizing force H, and so does the inductance. As an example, Figure 5 illustrates the relationship between the Percent Permeability and the DC Magnetizing Force H. µ 0 in (15) is the magnetic field constant which is equal to 1.257e-6; Bmax is the maximum magnetic flux density for the selected magnetic material (for sendust, Bmax is up to 0.8T.) Figure 5 Percent Permeability and DC Magnetizing Force H (from Changsung) Select a core with similar Ve value from the magnetic core datasheet. For example, the core type CS468125 from Chang Sung Corporation is selected. The parameters of CS468125 are Ve=15.584cm3, Ae=1.34cm2, C=11.63cm, µ r =125. The turn number of the boost choke winding is N toroid _ boost = Lboost ⋅ C = 83 µ r µ 0 Ae (16) where, C is the magnetic path length and Ae is the effective magnetic cross section area. To check the actual µ r at low line, maximum power, the DC Magnetizing Force H is calculated H= NI in _ pk C Application Note = 50(Oe) 10 2008-08-01 µr = 125 * 50% = 62.5 according to Figure 5. The actual inductance can be re-calculated as Lboost = N 2 µ r µ 0 Ae = 0.625mH . Hence, the corresponding ripple current will be higher than the C Then previously assumed value. The copper loss of the winding wire can be calculated on Iin_RMS. 2 PL _ boost = I in _ RMS ⋅ RL _ boost (17) Select the proper wire type to fullfil the loss and thermal requirement for the choke. (2) ferrite core To make sure the ferrite core will not go into saturation, the turn number of the boost choke winding with ferrite core is N ferrite _ boost ≥ I L _ pk ⋅ Lboost (18) Bmax ⋅ Amin where, Bmax is up to 0.3T according to ferrite material specification; Amin is the minimum magnetic cross section area. The winding wire copper loss calculation is the same as in the above section of sendust powder toroid core. 2.6 AC line current filter As decribed in section 2.5, there is high frequency ripple current peak to peak IHF passing through boost choke. This ripple will also go into AC line power network. The current filter is necessary to reduce the amplitude of high frequency current component. The filtering circuit consists of a capacitor and an inductor as shown in Figure 6. Rectifier IHF_spec Current Filter IHF Lfilter Cfilter VIN=85V ...265VAC Figure 6 AC line current filter The required Lfilter is I HF L filter ≥ I HF _ spec +1 (19) (2πf SW ) 2 C filter normally there is one EMI X2 capacitor which can act as Cfilter. In this example, if we define IHF_spec as 0.2A peak to peak and asumming X2 capacitance 0.47µF, then L filter 1.2 A +1 0.2 A ≥ = 89 µH (2π ⋅ 65kHz ) 2 ⋅ 0.47 µF Application Note 11 2008-08-01 The leakage inductance of EMI common mode choke can be used for current filter. If the leakage inductance is large enough, no need to add the additional differential mode inductor for filtering. Otherwise, a current filter choke is necessary. The calculation method for the current filter choke is the same as for boost choke. 2.7 Boost Output Bulk Capacitance The bulk capacitance has to fullfil two requirements, output double line frequency ripple and holdup time. (1) output double line frequency ripple limit. The inherent PFC always presents 2*fL ripple. The amplitude of ripple voltage is dependant on output current and bulk capacitance as below. C out ≥ I out π ⋅ 2 * f L ⋅ Vout _ ripple _ pp (20) where, Iout is the PFC output current, Vout_ripple_pp is the output voltage ripple (peak to peak), and fL is the AC line frequency. Please note that ICE2PCXX has enhance dynamic block which is active when Vout exceed ±5% of regulated level. The enchanc dynamic block should be designed to work only during load or line change. During steady state with constant load, the enhance dynamic block should not be triggered, otherwise THD will be deteriorated. That means the target Vout_ripple_pp must be lower than 10% of Vout. For this example, Vout=390VDC, then Vout_ripple_pp must be lower than 39V. if we define Vout_ripple_pp=12V, then C out ≥ I out = 220µF π ⋅ 2 ⋅ f L ⋅ Vout _ ripple _ pp (21) (2) holdup time requirement After the PFC stage, there is commonly a PWM stage to provide isolated DC output for end user. Some applications, especially computing, have the holdup time requirement. It means that PWM stage should be able to provide the isolated output even if AC input voltage become zero for a short holdup time. The common specification for this holdup time is 20ms. If minimum input voltage for PWM stage is defined as 250VDC, then the bulk capacitance will be C out ≥ 2 ⋅ Pout ⋅ t holdup 2 Vout − Vout _ min 2 = 2 ⋅ 300W ⋅ 20ms = 134 µF 390 2 − 250 2 (22) the final Cout capacitance should be higher value calculated from the above two requirements. 2.8 Current Sense Resistor The current sense resistance is calculated based on the IC soft over current control threshold and peak current carried by boost choke. When the Isense signal reaches the soft over control threshold, IC will reduce the internal control voltage and accordingly the duty cycle is reduced in the following cycles. Finally the boost choke current is limited. According to IC datasheet, soft over current control threshold is -0.68V maximum. So the current sense resistor should be Rsense ≤ 0.68V 0.68V = = 0.11Ω 6.14 A I L _ pk Application Note (23) 12 2008-08-01 According to Figure 2 and Figure 3, the transistor current as well as the diode current flows through Rsense. That means, when AC is powered up, a large negative voltage drop at Rsense will be observed when large inrush current in the range of about 150 A to 200 A flows through the resistor. It is therefore necessary to limit the current into Pin 2 (ISENSE) to 1 mA, which is realized with resistor R3. A value of R3 = 220Ω is sufficient for this resistor. 2.9 Output voltage sensing divider The output voltage is set with the voltage divider represented by R1 and R2 in Figure 2 and Figure 3. First, choose the value of the lower resistor R2. Then the value of the upper resistor R1 is R1 = Vout − Vref Vref ⋅ R2 (24) where, Vref is IC internal reference voltage for voltage sensing, 3V typical. If R2=6kΩ, R1 = 390 − 3 ⋅ 10kΩ = 774kΩ 3 It is recommended to take resistor values with a tolerance of 1% for R1 and R2. Due to the voltage stress of R1, it is recommended to split this value into few resistors in series. 2.10 Frequency setting (only for ICE2PCS01) The frequency of the ICE2PCS01 is adjustable in the range of 50 kHz up to 250 kHz. The external resistor RFREQ according to Figure 7 programs a current which controls the oscillator. Figure 7 Application Note Resistor-frequency characteristic 13 2008-08-01 2.11 AC Brown-out Shutdown (only for ICE2PCS02) Brown-out occurs when the input voltage VAC falls below the minimum input voltage of the design (i.e. 85V for universal input voltage range) and the VCC has not entered into the VCCUVLO level yet. For a system without input brown out protection (IBOP), the boost converter will increasingly draw a higher current from the mains at a given output power which may exceed the maximum design values of the input current and lead to over heat of MOSFET and boost diode. ICE2PCS02 provides a new IBOP feature whereby it senses directly the input voltage for Input Brown-Out condition via an external resistor/capacitor/diode network as shown in Figure 8. This network provides a filtered value of VIN which turns the IC on when the voltage at Pin 4 (VINS) is more than 1.5V. The IC enters into the standby mode and gate is off when VINS goes below 0.7V. The hysteresis prevents the system to oscillate between normal and standby mode. Figure 8 Block diagram of voltage loop Because of the high input impedence of comparator of C4 and C5, R5 can be high ohmic resistance to reduce the loss. From the datasheet, the bias current on VINS Pin is 1µA maximum. In order to have the design consistence, the current passing through R5 and R6 has to be much higher than this bias current, for example 6µA. Then R6 is: R6 = 0.7V = 117 kΩ 6uA (25) R6 is selected 120KΩ. R5 is selcted by R5 = 2 ⋅ V AC _ on − 1.5V 1.5V ⋅ R6 (26) where, VAC_on is the minimum AC input voltage (RMS) to start PFC, for example 70VAC. R5 = 2 ⋅ 70V − 1.5V ⋅ 120kΩ = 7.8MΩ 1.5V Due to the voltage stress of R5, it is recommended to split this value into few resistors in series. C4 is used to modulate the ripple at the VINS pin. The timing diagram of VINS pin when IC enters brown-out shutdown is shown in Figure 9. Application Note 14 2008-08-01 Figure 9 Timing diagram of VINS Pin when IC enters brown-out shutdown If the bottom level of the ripple voltage touches 0.7V, PFC is in standby mode and gate is off. The ripple voltage defines PFC brown out off threshold of AC input voltage (RMS), VAC_off. C4 can be obtained from the following equation. Assuming V INS _ AVE = R6 ⋅ V AC _ off , where, VAC_off is the maximum AC input voltage R5 + R6 (RMS) to switch off PFC, for example 65VAC. − R6 (2 ⋅ ⋅ V AC _ off − 0.7) ⋅ e R5 + R6 t disch arg e R6C4 = 0.7V (27) assuming tdischarge is equal to half cycle time of line frequency, ie. R6 2⋅ V AC _ off − 0.7V R5 + R6 C 4 = 2 f L R6 ln 0.7V 1 , then 2 fL −1 120kΩ 2⋅ 65V − 0.7V 7 . 8 M 120 k Ω + Ω C 4 = 2 ⋅ 50 Hz ⋅ 120kΩ ln 0.7V 2.12 t disch arg e = (28) −1 = 140nF IC supply The IC supply voltage operating range is 11~26V. There are two stages during IC turned on. First Vcc capacitor is charged from 0V to 7V, the IC internal regulator block starts to reset voltage at all external pins. The reset process will take about 10us. And then when Vcc voltage is charged to Vcc_on threshold, IC starts the soft start with gate switching. In the case of Vcc decoupling capacitance is too low such as 0.1uF, Vcc voltage may be charged up too fast and the time interval from Vcc=7V to Vcc_on is less than the reset time. Then the IC will not go through a proper soft start as the voltages at IC pins are not yet properly reset. To avoid such a problem, the delay circuitry is needed. Application Note 15 2008-08-01 Q1 AUX supply input IC Vcc Cvcc 0.1uF R1 Cdelay 10k 0.47uF R2 10k Power on control Q2 Figure 10 Vcc supply circuitry Figure 10 is a typical circuitry to supply PFC controller. Q2 is NPN transistor and controlled by external “Power on” signal. When “Power on” signal is “high”, Q2 is turned on provides base current for Q1. Q1 is turned on accordingly to supply auxiliary power to IC Vcc. The reset delay time is adjustable by changing the RC time constant of R1, R2 and Cdelay. The recommended values are shown in Figure 10 as 10kΩ, 10kΩ and 0.47uF respectively. The same reset process also happens during IC power down when Vcc is discharged from Vcc_off to 7V. The reset time for power down is around 200us. Because IC is in power down mode with very low current consumption, typically 300uA only, the required Vcc capacitance for power down reset can be calculated as: CVCC ≥ I power _ down _ max ⋅ t reset Vcc _ off _ min − Vreset = 650 µA ⋅ 200 µs = 38.2nF 10.4V − 7V (29) So the common Vcc decoupling capacitance 0.1uF is enough for reset delay requirement. 2.13 PCB layout guide In order to avoid crosstalk on the board between power and signal path, and to keep the IC GND pin as “clean” from noise as possible, the PCB layout for GND must be taken care of properly. Below are some suggestions for GND connection and Figure 11 below illustrates as a good example. (1) Star connection rule for main power stage GND: the PCB tracks of MOSFET source, output load GND, IC auxiliary supply GND and shunt resistor are separated and connected together at bulk capacitor negative Pin. (2) Star connection rule for small signal IC GND: the IC external components which need to be connected to the small signal GND bus highlighted in red color. Such GND bus is connected to IC GND Pin. (3) Connection between main power stage GND and small signal IC GND: in Figure 11, a single PCB track in pink color directly connect IC GND pin to power stage star connection point - bulk capacitor negative. This is to ensure that the voltage between IC Isense Pin and IC GND Pin does not observe the switching rectangular noise current. The dark green and blue tracks denote for flowing paths of high frequency rectangular switching current. (4) Vcc decoupling capacitor Cvcc: the decoupling capacitor need to be placed close to IC Vcc and GND Pins as much as possible. The GND track of Cvcc (green color in Figure 11) should be connected at the point on the single PCB track connecting between IC GND Pin and power GND point so that the large gate charging current will not pass through the small signal GND bus. (5) Vsense capacitor Cvsense: to reduce noise in Vsense Pin, small capacitor up to 0.1uF can be added between Vsense Pin and small signal GND bus. Application Note 16 2008-08-01 Rectifier EMI Filter L1 R1 T1 VOUT =400VDC C OUT R2 R SENSE VIN=85V ...265V AC R3 ISENSE GATE Cvcc VSENSE ICOMP FREQ VCOMP RFREQ C1 Figure 11 3 Cvsense ICEXPCS01 VCC Auxiliary Supply GND R4 C2 C3 Good PCB layout illustration Voltage loop and current loop compensation This section provides a model and a tool for evaluating and improving the control loop characteristics of ICE2PCS02-based PFC pre-regulators in boost topology. The goal is not only to ensure a narrow bandwidth in order to achieve a high Power Factor, but also to have enough phase margin so as to make sure the system is stable over a large range of operating conditions. The design example is demonstrated as well. Traditional diode rectifiers used in front of the electronic equipment draw pulsed current from the utility line, which deteriorates the line voltage, produce radiated and conducted electromagnetic interference, leads to poor utilization of the capacity of the power sources. In compliance with IEC 61000-3-2 harmonic regulation, active power factor correction (PFC) circuit is getting more and more attention in recent years. For low power up to 200W, discontinuous conduction mode (DCM) PFC is popular due to its lower cost. Furthermore, there is only one control loop, i.e. voltage loop, in its transferring control blocks. The design is easy and simple for DCM operation. However, due to its inherent high current ripple, DCM is seldom to be used for high power applications. In high power applications, continuous conduction mode (CCM) PFC is more attractive. V, I OUT I IL IIN DCM operation Figure 12 Application Note CCM operation DCM and CCM PFC principle 17 2008-08-01 3.1 How to achieve PFC function without sinusoidal reference sensing 3.1.1 Boost converter modeling Figure 13 shows the inductor current waveform for boost converter operating in continuous conduction mode. iL diL I0 ton toff TSW Figure 13 inductor current waveform of boost converter operating in CCM mode assuming Vin is boost converter input DC voltage, Vout is the boost converter output voltage, L is the boost choke inductance, ton is the on time duration in one switching cycle, toff is the off time duration in one switching cycle, doff is the off time duty cycle and Tsw is the time duration in one switching cycle. During “on” interval, diL Vin = L dt (30) During “off” interval, diL Vin − Vout = L dt (31) And then the boost inductor current variation after one switching cycle is: diL = V −V ⋅ d Vin V −V ⋅ ton + in out ⋅ toff = in out off ⋅ TSW L L L (32) The instant boost inductor current after n switching cycle is: iL _ n = iL _ n −1 + 3.1.2 Vin _ n − Vout _ n ⋅ d off _ n L ⋅ TSW (33) PFC IC control principle with boost topology PFC IC control block is inserted in boost converter as shown in Figure 14. Application Note 18 2008-08-01 Vin iL _ n Boost converter Vin _ n − Vout _ n ⋅ d off _ n = iL _ n−1 + ⋅ TSW L doff iL IC PWM modulation doff=K*iL Figure 14 PFC current loop principle IC senses boost inductor average current, and calculate the off duty cycle to be proportional to inductor current, and then send such off duty cycle back to boost converter. The negative feedback loop can be seen from Figure 14. A small disturb increasing on iL will result in a little bit increasing on off duty cycle. The increasing off duty cycle will lead to decreasing of iL after processing by boost converter. In the stead state, (34) Vin = Vout ⋅ d off = Vout ⋅ K ⋅ iL Where, K is the modulation gain defined by IC. It can be seen that boost inductor current shape follows AC input voltage and it is how PFC function to be achieved. In the following sections, detail mathematical analysis of current loop and voltage loop will be described and the transfer function for each block is given in order to design IC external compensation network components. 3.2 Current Loop Regulation and Transfer Function The detail block diagram of current loop for ICE2PCS02 is shown in the Figure 15. The boost converter stage Kboost is elaborated in S-plane. Vin Vout M2 PWM Comparator Kc(S) Doff - X + iL 1/sL Boost Converter Power Stage Kboost(s) Vicomp Current Averaging Kave(S) M1 Figure 15 3.2.1 Block diagram of current loop Current Averaging Circuit IC sense the boost inductor current via shunt resistor Rsense as shown in Figure 2. The sensing signal is sent to Isense Pin. As the voltage in Isense Pin is negative signal together with switching ripple, IC need to do signal averaging and convert the polarity to positive for following PWM modulation blocks. The output of averaging block is Vicomp voltage at Icomp Pin. the block diagram of current averaging block is shown in Figure 16. Application Note 19 2008-08-01 Figure 16 current averaging block diagram The transfer function of averaging circuit block can be derived as below. K AVE ( s ) = Vicomp iL K 1 Rsense M1 = K 1C icomp 1+ s ⋅ M 1 g OTA 2 (35) where, K1 is a ratio between R501 and R7 which is equal to 4, Cicomp is the capacitor at Icomp Pin, gOTA2 is the trans-conductance of the error amplifier of OTA2 for current averaging, typical 1.0mS as shown in Datasheet, M1 is the variable controlled by voltage loop. The function of the averaging circuit is to filter out the switching current ripple. So the corner frequency of the averaging circuit fAVE must be lower than the switching frequency fSW. Then, C icomp ≥ 3.2.2 g OTA2 M 1 K 1 ⋅ 2πf AVE (36) PWM comparator block The averaged Vicomp signal is sent to PWM comparator block and compared with internal triangular ramp signal to derive duty cycle. The timing diagram of this block is shown in Figure 17. Application Note 20 2008-08-01 Ramp Vicomp PWM Comparator To PWM logic and gate driver block C1 Vicomp Gate drive From protection logic Vramp=M2*Kfq Oscillator Tosc Figure 17 The block diagram and timing sequence of PWM comparator block The operating principle is explained as following. Gate output is in “low” state in the beginning of the each cycle. Gate output is turned to “high” at the intersection of the triangular ramp signal and Vicomp signal. Gate output is turned to “low” by oscillator synchronous signal. Based on the operating principle, the transfer function of KC(s) is: K C (s) = d off Vicomp = 1 K FQ M 2 (37) Where, KFQ is a design constant which is equal to 9.183, M2 is the variable controlled by voltage loop. 3.2.3 Boost converter stage The transfer function of boost converter stage KBoost(s) can be obtain via State-Space Averaging method. Combining equation (30) and (31) by state –space averaging, Vin − Vout d off V − Vout di L Vin d on + in = d off = dt L L L (38) Make Laplace transformation for equation (38) with assuming Vin and Vout are constant for current loop analysis, i L ( s ) = (Vin − Vout d off ( s )) 1 sL (39) The equation (39) has been described in current loop block diagram in Figure 15. Although Vin is not physically sensed by circuit, the input sinusoidal signal is presented in transfer functions only if boost topology is applied. 3.2.4 Open loop transfer function gain for current loop The open loop gain of current regulation loop is: K 1 RsenseVout K FQ M 1 M 2 L V GC ( s ) = K AVE ( s ) K C ( s ) out = K 1Cicomp sL s (1 + s ⋅ ) M 1 g OTA2 Application Note (40) 21 2008-08-01 The selected Cicomp must also meet the requirement that the cross over frequency of the current loop fC is much lower than the switching frequency fSW. 3.2.5 Steady state solution of IL Solving the current loop in Figure 15, i L ( s ) = (Vin − Vout d off ( s )) 1 1 = (Vin − Vout K C ( s ) K AVE ( s )i L ( s )) sL sL Vin Vin sL sL i L ( s) = = Vout K C ( s ) K AVE ( s ) 1 + GC ( s ) 1+ sL (41) For AC line frequency which is much lower than fC, then Gc ( s ) >> 1 K FQ M 1 M 2Vin Vin Vin K 1 RsenseVout sL i L ( s) = ≈ sL = K1C icomp 1 + GC ( s ) GC ( s ) 1+ s ⋅ M 1 g OTA2 (42) For AC line frequency which is also much lower than fAVE, s ⋅ K 1C icomp M 1 g OTA 2 << 1 , then the steady state IL can be derived as IL = K FQ M 1 M 2Vin (43) K1 RsenseVout from the above steady state solution of IL, it can be seen that the choke current IL is always following input voltage Vin. This is how PFC function is achieved. 3.3 Voltage Loop Compensation The control loop block diagram for ICE2PCS02 based CCM PFC is shown in Figure 18 and Figure 19. There are four blocks in the loop. IC PWM Modulator G2(s) has been discussed in above Section 3. the rest of them are Error Amplifier G1(s), nonlinear block GNON(s), boost converter output stage G3(s) and Feedback Sensing G4(s). Vin Vcomp_DC 0V Vref + - Error Amplifier G1(s) Vcomp Nonlinear block GNON(s) M1M2 PWM Modulator G2(s) iL Boost converter output Stage G3(s) Vout 400V Vsense 5V Feedback G4(s) Figure 18 Application Note Large signal modeling of voltage loop 22 2008-08-01 PWM modulation G2(s) + ∆Vsense Voltage loop Error Amplifier G1(s) ∆Vcomp Nonlinear GNON(s) ∆ ( M 1M 2 ) Output Stage G3(s) ∆I L _ rms I L _ rms M 1M 2 + - Vinrms Vout _ AVE ∆I out Output Stage 1 sCout ∆Vout I L _ rms Vout _ AVE Feedback Figure 19 3.3.1 Small signal modeling of voltage loop Boost converter output stage G3(s) Boost converter output stage is described as influencing of variation on iL to bulk output voltage Vout. The transfer function of power stage, G3(s), is separated to two stages as: G3 ( s ) = ∆Vout ∆Vout ∆I out = ⋅ ∆I L _ rms ∆I out ∆I L _ rms (44) where Vout is the DC output voltage, Iout the DC output current and IL_rms is the boost inductor current. 3.3.1.1 ∆Vout / ∆Iout Under the above assumption, the power stage can be modeled as illustrated in Figure 20: a controlled current source (with a shunt resistor Re) that drives the output bulk capacitor Cout and the load resistance Rout (= Vout / Iout). The zero due to the ESR associated with Cout is far beyond the crossover frequency thus it is neglected. Iout Figure 20 Re Cout Rout Vout Power stage modeling A few algebraic manipulations would show that the shunt resistor Re always equals the DC load resistance Rout, thus it changes depending on the power delivered by the system. There are two kinds of load in the application. Two cases will give a different result in case of resistive load or constant power load. For purely resistive load, the AC load resistance equals Ro. In case of constant power load like additional isolated PWM DC/DC converter, the AC load resistance is equal to -Ro (if the DC bus decreases, the current demanded of the PFC increases. hence the negative sign is shown.). As a result, the parallel combination with Re tends to infinity and the two resistances cancel. The current source drives only the output capacitor. The result is summarized as below: Application Note 23 2008-08-01 ∆Vout ∆I out Rout R C 2(1 + s ⋅ out out ) = 2 1 sC out Resistive Load (45) Constant Power Load In this application note, the calculation is only carried out for constant power load situation ∆Iout / ∆IL_rms 3.3.1.2 The current source Iout can be characterized with the following considerations as shown in Figure 21. The low frequency component of the boost diode current is found by averaging the discharge portion of the inductor current over a given switching cycle. The low frequency current, averaged over a mains half-cycle yields the DC output current Iout: IL_PK iL idiode IOUT Figure 21 I out = 1 π π∫ 0 The simplification and characterization for Iout / IL_rms (1 − Don ) I L _ PK Sinαdα = 2Vinrms I L _ rms πVout _ AVE ∫ π 0 ( Sinα ) 2 dα = Vinrms I L _ rms Vout _ AVE (46) So, V ∆I out = inrms ∆I L _ rms Vout _ AVE (47) where, Don is the switch duty cycle; α is the instantaneous phase angle of the mains voltage, Vinrms is the input RMS voltage value, IL_PK is choke current sinewave peak value and Vout_AVE is the averaging bulk DC output voltage. In case of constant power load, the transfer function of G3(s) is: G3 ( s ) = 3.3.2 ∆Vout ∆Vout ∆I out V 1 = ⋅ = inrms ⋅ ∆I L _ rms ∆I out ∆I L _ rms Vout _ AVE sC out (48) Small signal transfer function of ∆Vout/∆(M1M2) for voltage loop analysis There is a internal feedback from Vout to G2(s). this inner loop has to be solved to obtain the transfer function of ∆Vout/∆(M1M2). Rewrite the equation (43) at input voltage RMS point: I L _ rms = K FQ M 1 M 2Vinrms (49) K1 RsenseVout Application Note 24 2008-08-01 making a perturbation on IL_rms, (M1M2), Vout, then ∆I L _ rms = I L _ rms M 1M 2 ∆( M 1 M 2 ) − I L _ rms Vout _ AVE ∆Vout (50) replacing ∆IL_rms by ∆Vout/G3(s) according to voltage loop block diagram, I L _ rms I L _ rms ∆Vout = ∆( M 1 M 2 ) − ∆Vout G3 ( s ) M 1 M 2 Vout _ AVE (51) then the transfer function of dVout/dVcomp is Vout _ AVE G23 ( s ) = ∆Vout M 1M 2 = ∆( M 1 M 2 ) Vout _ AVE 2 C out = s +1 I L _ rmsVinrms 1 With f 23 = 3 2π K 1 RsenseVout _ AVE C out K FQ M 1 M 2Vinrms Vout _ AVE M 1M 2 (52) 3 K 1 RsenseVout _ AVE C out K FQ M 1 M 2Vinrms 2 s +1 , 2 Vout _ AVE G23 ( s ) = 3.3.3 ∆Vout M 1M 2 = s ∆( M 1 M 2 ) 1+ 2πf 23 (53) Nonlinear block GNON(s) The Vcomp voltage is sent to nonlinear gain block. The output of nonlinear is two internal variables, M1 and M2. The two variables are used to define boost choke current amplitude IL as in equation (43). The characteristic of nonlinear gain block is shown in Table 2 and Figure 22. The small signal gain between ∆(M1*M2) and ∆Vcomp can be derived as well at different operating point. Vcomp 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 Application Note M1 4.686E-02 4.685E-02 4.665E-02 4.685E-02 4.823E-02 8.153E-02 1.261E-01 1.901E-01 2.747E-01 3.768E-01 4.884E-01 5.992E-01 6.992E-01 7.816E-01 8.443E-01 8.888E-01 9.184E-01 9.339E-01 M2 4.964E-04 7.072E-04 1.199E-03 3.292E-03 3.224E-02 1.075E-01 1.921E-01 2.796E-01 3.686E-01 4.590E-01 5.523E-01 6.539E-01 7.794E-01 9.669E-01 1.287E+00 1.802E+00 2.442E+00 2.911E+00 25 M1*M2 2.326E-05 3.313E-05 5.595E-05 1.542E-04 1.555E-03 8.766E-03 2.423E-02 5.316E-02 1.013E-01 1.729E-01 2.697E-01 3.918E-01 5.449E-01 7.557E-01 1.087E+00 1.601E+00 2.243E+00 2.719E+00 2008-08-01 4.50 4.75 5.00 9.350E-01 9.351E-01 9.351E-01 2.911E+00 2.911E+00 2.911E+00 2.722E+00 2.722E+00 2.722E+00 Table 2 nonlinear block characteristic data 3.5 3.0 2.5 M1 2.0 M2 M1*M2 1.5 1.0 0.5 0.0 0 1 2 3 4 5 Vcomp Figure 22 3.3.4 The characteristics of nonlinear block Error Amplifier compensation G1(s) The circuit of error amplifier compensation circuit is shown in Figure 23. The sensing voltage Vsense is compared to internal reference voltage 3V typical. The difference between Vsense and internal reference is sent to transconductance error amplifier and converted to a current source to charge or discharge the RC components in Vcomp Pin. Figure 23 Error Amplifier compensation G1(s) The transfer function is: Application Note 26 2008-08-01 G1 ( s) = ∆Vcomp ∆Vsense = ∆Vcomp ∆I OTA1 ⋅ = ∆I OTA1 ∆Vsense 1 + sR4 C 2 ⋅ g OTA1 R4 C 2 C 3 (C 2 + C 3 ) s(1 + s ) C 2 + C3 (54) where, gOTA1 is the trans-conductance of OTA1, 42uS typically for ICE2PCS02. With f CZ = G1 ( s ) = 1 and f CP = 2πR4 C 2 1 , R4 C 2 C 3 2π C 2 + C3 g OTA1 (1 + s ) 2πf CZ (C 2 + C3 ) s(1 + s ) 2πf CP (55) The pole and zero are to regulate the overall voltage loop with the cross-over frequency below 100Hz and create the phase margin for the loop stability. 3.3.5 Feedback G4(s) The Feedback block is a simple voltage divider to monitor the bulk capacitor output voltage. The circuit is shown in Figure 24. Vout G4 ( s ) = ∆Vsense R2 = R1 + R2 ∆Vout R1 (56) Vsense R2 Figure 24 3.3.6 bulk voltage sensing divider Overall Open Loop Transfer Function GV(s) With combining all of the blocks above, the overall open loop gain for voltage loop is equal to: GV ( s ) = G1 ( s )G NON ( s )G23 ( s )G4 ( s ) (57) Due to PF requirement, inherent PFC dynamic voltage loop compensation is always implemented with low bandwidth in order not to make the response for 2*fL ripple. For example, for 50Hz AC line input, PFC voltage loop bandwidth is normally set below 20Hz. The compensation circuit R4, C2 and C3 are used to optimize the loop gain and phase margin. 3.3.7 Enhance dynamic response Application Note 27 2008-08-01 As mentioned in Section 4.6, the inherent low bandwidth of voltage loop in PFC application will lead to slow response in case of sudden load step and result in large output overshoot or drop. Enhance dynamic response feature is integrated in ICE2PCS02 to have a fast response in the case of load step. The voltage loop with including enhance dynamic response block is shown in Figure 25. Vin Vcomp_DC 0V Vref + - Error Amplifier G1(s) Vcomp + +/- Nonlinear block GNON(s) M1M2 PWM Modulator G2(s) iL Boost converter output Stage G3(s) Vout 400V Vsense 5V Enhance dynamic Feedback G4(s) Figure 25 voltage loop block diagram including enhance dynamic response When Vsense voltage variation is within -5% to +5% of nominal value, there is no function of enhance dynamic response block. However, when Vsense variation is out of such +/-5% range, enhance block will add offset voltage on top of Vcomp voltage to influence the current amplitude. The timing diagram of enhance dynamic response operation is shown in Figure 26 with sudden load jump situation. It can be seen that during enhance dynamic operation, the high current of boost choke is delivered for fast response. Within half sinusoidal period, when Vsense operating around the boundary of -5% threshold, the first part of boost choke current follows high amplitude profile due to enhance mode offset and the rest of boost choke current come back to low amplitude profile without enhance mode offset. When Vsense voltage is pulled back within +/-5% range, enhance dynamic offset disappear and boost choke current waveform will stay as perfect sinusoidal shape. Normal enhance Normal Vin Iin Vcomp 0 Pin Pin_ave 0 Ichg Ichg_ave 0 Vout Vout_ave nominal - 5% Figure 26 timing diagram for enhance dynamic operation 3.4 Design Example Assuming a 300W application with universal input AC voltage 85~265VAC, constant power load efficiency=90% Application Note 28 2008-08-01 Vout=400VDC Cout=220uF/450V fSW=125kHz Rsense=0.1ohm Boost choke inductance L=1.2mH (please note that the inductance may change at different choke current) Vsense divider: R1=390kohm*2=780kohm, R2=6kohm 3.5 Vcomp and M1, M2 value at full load condition (1) 85VAC: RMS AC input current under full load: I L _ rms _ 85 = Pout 300 = = 3.92 A η ⋅ Vinrms _ 85 0.9 ⋅ 85 (58) From equation (43), With K FQ = 4.34 and K 1 = 4 from the ICE2PCS02 Datasheet, M 1M 2 85VAC = I L _ rms _ 85 K1RsenseVout K FQVinrms _ 85 = 3.92 ⋅ 4 ⋅ 0.1⋅ 400 = 1.70 4.34 ⋅ 85 (59) From table 2 and Figure 22, it can be obtained Vcomp M1 M2 M1*M2 3.75 8.888E-01 1.802E+00 1.601E+00 4.00 9.184E-01 2.442E+00 2.243E+00 With Linear approximation: Vcomp _ 85 = Vcomp _ 1 + Vcomp _ 85 = 3.75 + M 1M 2 Vcomp _ 2 − M 1M 2 Vcomp _ 1 M1 _ 2 − M1 _ 1 Vcomp _ 2 − Vcomp _ 1 ⋅ (Vcomp _ 2 − Vcomp _ 1 ) (60) ⋅ (Vcomp _ 85 − Vcomp _ 1 ) (61) 0.918 − 0.889 = 0.889 + ⋅ (3.79 − 3.75) = 0.894 4 − 3.75 M 2 85VAC = M 2 _ 1 + M 2 85VAC Vcomp _ 1 1.70 − 1.601 ⋅ (4 − 3.75) = 3.79V 2.243 − 1.601 M 1 85VAC = M 1 _ 1 + M 1 85VAC M 1M 2 85VAC − M 1M 2 M 2 _ 2 − M 2 _1 Vcomp _ 2 − Vcomp _ 1 ⋅ (Vcomp _ 85 − Vcomp _ 1 ) 2.442 − 1.802 = 1.802 + ⋅ (3.79 − 3.75) = 1.91 4 − 3.75 (62) The small signal gain of nonlinear block is G NON ( s ) 85VAC = M 1 M 2 Vcomp _ 2 − M 1 M 2 Vcomp _ 1 Vcomp _ 2 − Vcomp _ 1 = 2.243 − 1.601 = 2.568 4 − 3.75 (63) The inherent pole of f23 is Application Note 29 2008-08-01 f 23 85VAC 1 3 K1 RsenseVout _ AVE C out = 2π = 1.54 Hz K FQ ⋅ ( M 1 M 2 ) 85VAC ⋅ Vinrms _ 85 (64) 2 (2) 265VAC RMS AC input current under full load: I L _ rms _ 265 = Pout 300 = = 1.257 A η ⋅ Vinrms _ 265 0.9 ⋅ 265 (65) From equation (43), M 1M 2 265VAC = I L _ rms _ 265 K1RsenseVout K FQVinrms _ 265 = 1.257 ⋅ 4 ⋅ 0.1 ⋅ 400 = 0.175 4.34 ⋅ 265 (66) From table 2 and Figure 22, it can be obtained Vcomp M1 M2 M1*M2 2.25 3.768E-01 4.590E-01 1.729E-01 2.50 4.884E-01 5.523E-01 2.697E-01 With Linear approximation: Vcomp _ 265 = Vcomp _ 1 + Vcomp _ 265 = 2.25 + M 1M 2 Vcomp _ 2 − M 1M 2 Vcomp _ 1 M1 _ 2 − M1 _ 1 Vcomp _ 2 − Vcomp _ 1 ⋅ (Vcomp _ 2 − Vcomp _ 1 ) (67) ⋅ (Vcomp _ 265 − Vcomp _ 1 ) 0.4884 − 0.3768 = 0.3768 + ⋅ (2.266 − 2.25) = 0.386 2.5 − 2.25 M 2 265VAC = M 2 _ 1 + M 2 265VAC Vcomp _ 1 0.175 − 0.1729 ⋅ (2.5 − 2.25) = 2.255V 0.2697 − 0.1729 M 1 265VAC = M 1 _ 1 + M 1 265VAC M 1M 2 265VAC − M 1M 2 M 2 _ 2 − M 2 _1 Vcomp _ 2 − Vcomp _ 1 (68) ⋅ (Vcomp _ 265 − Vcomp _ 1 ) 0.5523 − 0.459 = 0.459 + ⋅ (2.255 − 2.25) = 0.461 2.5 − 2.25 (69) The small signal gain of nonlinear block is G NON ( s ) 265VAC = M 1 M 2 Vcomp _ 2 − M 1 M 2 Vcomp _ 1 Vcomp _ 2 − Vcomp _ 1 = 0.2697 − 0.1729 = 0.3872 2.5 − 2.25 (70) The inherent pole of f23 is f 23 265VAC = 2π Application Note 1 3 K1 RsenseVout _ AVE C out K FQ ⋅ ( M 1 M 2 ) 265VAC ⋅ Vinrms _ 265 = 1.54 Hz (71) 2 30 2008-08-01 3.5.1 Current Averaging Circuit With gOTA2=1.0mS from Datasheet, M1@85VAC, and assuming fAVE=13kHz which is 10 times less than switching frequency 125kHz, then C icomp ≥ g OTA2 M 1 85VAC K 1 ⋅ 2πf AVE = 1.0 E − 3 ⋅ 0.895 = 3nF 4 ⋅ 2π ⋅ 24 E 3 (72) Select Cicomp=3.3nF 3.5.2 Current Loop Regulation Insert M1 and M2 value in equation (40). The amplitude and phase angle of GC(s) is shown in Figure 27 to verify the stability of current loop and the requirement of fC less than switching frequency. 100 85VAC full load 265VAC full load 50 Gain(db) 0 -50 -100 -150 0 10 1 10 2 10 3 4 10 10 5 10 6 10 7 10 f(HZ) Application Note 31 2008-08-01 -90 85VAC full load 265VAC full load -100 -110 Phase Angle -120 -130 -140 -150 -160 -170 -180 0 10 1 10 2 10 3 4 10 10 5 10 6 10 7 10 f(HZ) Figure 27 The bode plot and phase angle for current loop The cross over frequency and phase margin are 3kHz and 75º for 85VAC, and 10kHz and 25º for 265VAC. Application Note 32 2008-08-01 3.5.3 Voltage Loop Regulation From the above sections, it can be obtained: G1 ( s ) = ∆Vcomp ∆Vsense G NON ( s ) = = g OTA1 (1 + s ) 2πf CZ (C 2 + C 3 ) s (1 + s ) 2πf CP (73) ∆( M 1 M 2 ) ∆Vcomp (74) Vout _ AVE G23 ( s ) = G4 ( s ) = ∆Vout M 1M 2 = s ∆( M 1 M 2 ) 1+ 2πf 23 (75) ∆Vsense R2 6.2 = = . = 0.0077 ∆Vout R1 + R2 806.2 (76) The open loop gain for voltage loop is to times all above factors together as: GV ( s ) = G1 ( s )G NON ( s )G23 ( s )G4 ( s ) G1(s) is used to provide enough phase margin and also limit the bandwidth below 20HZ. R4, C2 and C3 can be chosen as required. fCZ normally select to be compensate the pole in G23(s). fCP normally select to be 40~70Hz in order to fast put down the gain amplitude and reject the high frequency interference. In this example f23 is around 1.54Hz at 85VAC/ 265VAC and full load. So the initial target is: fCZ is chosen to be close to 1.5Hz, and fCP is chosen to be 50Hz. C2 and C3 is calculated to obtain Gv(s) cross over frequency around 10Hz. The gain amplitude of GNON*G23*G4 in 85VAC and full load is shown in Figure 28. It can be seen that at f=10Hz, the gain is about 4.52dB. So G1 should provide the gain +4.52dB at f=10Hz. Considering that C2>>C3 due to fcz<fcp and 10Hz>>1Hz=fCZ, then G1 (10 Hz ) = g OTA1 10 Hz 1Hz = +4.52dB C2 ⋅ 2π ⋅ 10 Hz (77) 10 Hz 1Hz = 3.69µF C2 = 4.52 10 20 ⋅ 2π ⋅ 10 Hz 39 ⋅ 10− 6 ⋅ 3.97uF is not common for ceramic type capacitor. So select C2=1uF, then fCZ is recalculated as: Application Note 33 2008-08-01 G1 (10 Hz ) = f CZ = gOTA1 1 + (10 Hz = +4.52dB 10 Hz 2 1µF ⋅ 104.52 20 ⋅ 2π ⋅ 10 Hz −1 39 ⋅ 10 − 6 (78) = 4.30 Hz 1 = 4.30 Hz then 2πR4C2 1 = 37 kΩ 2π ⋅ 4.30 Hz ⋅ C2 select R4=33kΩ, and f CP = C3 = )2 C2 ⋅ 2π ⋅ 10 Hz according to f CZ = R4 = f CZ (79) 1 1 ≈ = 50 Hz R4 C 2 C 3 2πR4 C 3 2π C 2 + C3 1 = 96.5nF 2π ⋅ 50 Hz ⋅ R4 (80) select C3=100nF The gain amplitude and phase angle of overall voltage loop GV(s) at 85VAC and 265VAC in full load condition is shown in Figure 28 and Figure 29. At 85VAC, the cross over frequency fV is around 9.5Hz and the phase margin is about 63º. At 265VAC, the cross over frequency fV is around 14Hz and the phase margin is about 62º. Application Note 34 2008-08-01 60 Gv=G1*Gnon*G23*G4 Gnon*G23*G4 40 20 Gain(db) 0 -20 -40 -60 -80 -100 -120 -1 10 0 10 1 2 10 10 3 4 10 10 f(HZ) -90 Gv -100 -110 Phase Angle -120 -130 -140 -150 -160 -170 -180 -1 10 0 10 1 2 10 10 3 10 4 10 f(HZ) Figure 28 Application Note the bode plot and phase angle for voltage loop at 85VAC and full load 35 2008-08-01 60 40 20 Gain(db) 0 -20 -40 -60 -80 -100 -120 -1 10 0 10 1 2 10 10 3 10 4 10 f(HZ) -90 -100 -110 Phase Angle -120 -130 -140 -150 -160 -170 -180 -1 10 0 10 1 2 10 10 3 10 4 10 f(HZ) Figure 29 Application Note The bode plot and phase angle for voltage loop at 265VAC and full load 36 2008-08-01 References [1] Infineon Technologies: ICE2PCS01 - Standalone Power Factor Correction Controller in Continuous Conduction Mode; Preliminary datasheet; Infineon Technologies; Munich; Germany; Sept. 2007. [2] Infineon Technologies: ICE2PCS02 - Standalone Power Factor Correction (PFC) Controller in Continuous Conduction Mode (CCM) at Fixed Frequency, Preliminary datasheet; Infineon Technologies; Munich; Germany; Sept. 2007. [3] Luo Junyang, Liu Jianwei, Jeoh Meng Kiat, 300W CCM PFC Evaluation Board with ICE2PCS02, CoolMOS™ and SiC Diode thinQ!™, Application note, Infineon Technologies, Munich, Germany, Feb. 2007. [4] Luo Junyang, Liu Jianwei, Jeoh Meng Kiat, ICE2PCSxx, New generation of BiCMOS technology, Application note, Infineon Technologies, Munich, Germany, Feb, 2007 [5] Luo Junyang, Liu Jianwei, Jeoh Meng Kiat, ICE1PCS01 Based Boost Type CCM PFC Design Guide - Control Loop Modeling, Application note, Infineon Technologies, Munich, Germany, May, 2007. [6] Luo Junyang, Liu Jianwei, Jeoh Meng Kiat, ICE1PCS01/02 Boost Type CCM PFC Design with ICE1PCS01. Application note, Infineon Technologies, Munich, Germany, Apr. 2007. Application Note 37 2008-08-01