A4401 Automotive Quasi-Resonant Flyback Control IC Features and Benefits Description ▪ Multiple output regulator ▪ 7 to 40 V input supply ▪ Low EMI conducted and radiated emissions ▪ Adaptive quasi-resonant turn on/off control ▪ Minimal number of external components ▪ Enable input which can be driven with respect to the battery voltage This device provides all the necessary control functions to provide the power rails for driving a vacuum fluorescent display (VFD) using minimal external components. The power supply is based on a quasi-resonant, discontinuous flyback converter, operating near the critical conduction boundary. A novel adaptive turn-on control scheme is used to optimize the turn-on and turn-off phase of the MOSFET, to reduce EMI emissions while minimizing switching losses. The converter is self-oscillating, operating at switching frequencies depending on the input voltage, load, and external components. An onboard linear regulator that is powered directly from the battery provides the housekeeping supply, avoiding the need for complex bias supplies. Internal diagnostics provide comprehensive protection against overloads, input undervoltage, and overtemperature conditions. Package: 8-pin narrow SOIC (suffix L) The A4401 is supplied in an 8-pin narrow SOIC package (suffix L), which is lead (Pb) free, with 100% matte-tin leadframe plating. Approximate Scale 1:1 Typical Application +VBAT VFD VIN LX A4401 GD ECU 0V ISS EN 0V VA GND A4401-DS, Rev. 2 COMP A4401 Automotive Quasi-Resonant Flyback Control IC Selection Guide Part Number A4401KLTR-T Packing Tape and reel, 3000 pieces/reel Absolute Maximum Ratings* Characteristic Symbol Rating Units VIN –0.3 to 40 V LX Pin Voltage VLX –0.6 to 60 V ISS Pin Voltage VISS –1 to 1 V EN Pin Voltage VEN –0.3 to 40 V VA Pin Voltage VVA –0.3 to 5 V AEC-Q100-002; all pins 2000 V AEC-Q100-011; all pins; inside 500 V AEC-Q100-011; all pins; corner 750 V VIN Pin Voltage Notes ESD Rating – Human Body Model ESD Rating – Charged Device Model Operating Ambient Temperature TA –40 to 125 ºC Junction Temperature TJ –40 to 150 ºC Storage Temperature Tstg –55 to 150 ºC Range K * With respect to ground Characteristic Package Thermal Resistance Symbol RθJA Test Conditions* 4-layer PCB based on JEDEC standard Value Units 80 ºC/W *Additional thermal information available on Allegro website. Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com 2 A4401 Automotive Quasi-Resonant Flyback Control IC Functional Block Diagram D2 US1J LS1 VBAT 7 to 40V Optional EMI Filter C1 C5 100 nF Anode 84 V 50mA C6 22 F C7 100 nF Grid 58 V 50 mA C8 22 F C9 100 nF Filament 8V 200 mA D3 US1J LS2 L1 C4 22 F C2 D4 STPS160U LP LS3 C3 100 nF R2 VIN 10 kΩ LX Linear GD Q1 ZD1 4.7 V C10 100 nF C11 2.2 nF ISS Adaptive turn-on R3 0.100 Ω control + R Enable EN R1 10 kΩ Optional feedback resistor R4 27 kΩ - Control Logic R5 330 kΩ S - Shutdown Fault VA Gm Amp + UVLO Soft Start TSD R6 4.7 kΩ Ref. COMP GND R7 220 kΩ C12 220 pF R8 1 MΩ C13 6.8 nF Pin-out Diagram EN 1 8 VIN COMP 2 7 LX VA 3 6 GD GND 4 5 ISS Number 1 2 3 4 5 6 7 8 Name EN COMP VA GND ISS GD LX VIN Description Enable input; active high Compensation node for Gm amplifier stage Output voltage feedback input Ground reference connection; connect to negative terminal of battery supply MOSFET, Q1, current sense input MOSFET gate drive output Regulator switching node: MOSFET drive output Supply input to power control circuit, MOSFET driver, and reference voltages Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com 3 A4401 Automotive Quasi-Resonant Flyback Control IC ELECTRICAL CHARACTERISTICS1,2 valid at TJ = –40°C to +150°C, VIN = 7 to 40 V (unless noted otherwise) Characteristics Symbol Test Conditions Min. Typ. Max. Units General VIN Quiescent Current LX Leakage Current IINOFF EN = Low – – 10 μA IINON EN = High, no MOSFET switching – 2.3 3.3 mA ILXLEAK EN = Low, VLX = 40 V – – 1 μA LX Input Bias Current ILX EN = High, VLX = 60 V – – 145 μA Minimum Frequency fSW 25 35 45 kHz Soft Start tSS 5 10 15 ms Gate Drive Drive High Voltage, VIN > 10 V Drive High Voltage, 10 V > VIN > 7 V Rise Time, VIN > 10 V Rise Time, 10 V > VIN > 7 V Fall Time, VIN > 10 V Fall Time, 10 V > VIN > 7 V – 8.4 9.5 V – VIN – 0.5 VIN – 0.25 V CLOAD = 1 nF, 10% to 90% of VGS= 9 V – 60 – ns CLOAD = 1 nF, 10% to 90% of VGS , VIN = 7 V – 90 – ns CLOAD = 1 nF, 90% to 10% of VGS = 0 V – 30 – ns CLOAD = 1 nF, 90% to 10% of VGS , VIN = 7 V – 40 – ns 600 800 1000 mV VGDH tr tf Current Sense Input Maximum Sense Voltage (Current Limit) Sense Input Bias Current Current Sense Blanking VCL – – 10 μA tBLANK IISS VISS = –300 mV to 1 V 100 145 190 ns VREF 1.180 1.205 1.230 V Reference Supply Reference Voltage Tolerance Operational Transconductance Amplifier Output Impedance ZOP 10 – – MΩ Gm Constant3 KGm – 470 – μS Output Source Current ISRC VCOMP = 1.4 V, VA = 1.06 V, TA = 25ºC –30 –25 –20 μA Output Sink Current ISIN VCOMP = 1.4 V, VA= 1.36 V, TA = 25ºC 20 25 30 μA Input Bias Current IBIAS – –100 – nA V Enable Input EN Input Low Voltage VIL – – 0.8 EN Input High Voltage VIH 2.4 – – V VIhys 200 500 – mV VIN = 0 V –10 – 10 μA VIN = 14 V – – 50 μA VIN = 40 V – – 200 μA EN Input Hysteresis IINL EN Input Current IINH Protection VIN Turn-On Threshold VUVON Voltage rising 5.4 – 7 V VIN Turn-Off Threshold VUVOFF Voltage falling 4.9 – 6.5 V Undervoltage Hysteresis VUVhys Overtemperature Shutdown TJSD Overtemperature Hysteresis TJSDhys Temperature increasing – 0.5 – V – 165 – ºC – 15 – ºC 1For input and output current specifications, negative current is defined as coming out of (sourcing) the specified device pin. 2Specifications over operating temperature range are assured by design and characterization. 3Guaranteed by design. Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com 4 A4401 Automotive Quasi-Resonant Flyback Control IC Functional Description Basic Operation A peak current-mode control scheme is used to regulate one of the converter outputs, which will typically be the highest output voltage. The regulated output voltage is potentially divided down and fed into a Gm stage, where the resulting error signal acts as the control reference. This reference signal is compared against the signal that is produced by the inductor magnetization current flowing through the sense resistor. As shown in figure 1, at the beginning of a switching cycle, the external MOSFET, Q1, is turned on. After the sense resistor signal reaches the control reference amplitude, the PWM comparator resets the synchronous rectification (SR) latch and turns off the MOSFET. When the MOSFET is turned off, the voltage on the LX node rises until the voltage clamps at the battery voltage, VBAT , plus the reflected output voltage, VOUT(RFL). The secondary rectification diodes are forward biased and the energy stored in the coupled inductor is released to the output circuits. During this period, the current through the inductor decreases lin+V VMOSFET ( VLX ) Coupled inductor goes discontinuous; resonant ring occurs VOUT(RFL) VOUT(RFL) VBAT 0 MOSFET turns on MOSFET turns off IMOSFET +I Current released from coupled inductor into output circuit ½ resonant period Figure 1. External MOSFET voltage and current Current builds up in primary winding of coupled inductor early. As the current falls to 0 A, a resonance is set up between the primary magnetizing inductance and any capacitance appearing between the drain and ground. A damped voltage ringing occurs, which resonates around the battery voltage, VBAT. As the resonant ring swings negative, the adaptive turn-on circuit monitors to detect the point at which the voltage reaches a minimum. At this point the MOSFET is commanded on, thereby minimizing the turn-on losses. Also, the relatively slow resonant dV/dt helps to reduce EMI. In most applications, the converter will be operated with a battery input voltage of 13.5 V. To optimize the performance of the regulator at this voltage, the magnetics can be designed to force 0 V across the MOSFET at turn-on. This minimizes switching losses and perhaps more importantly reduces EMI caused by voltage ringing due to the drain to ground capacitor resonating with the primary inductance. The voltage resonance at the MOSFET turn-off can be reduced by a simple low-loss R-C snubber, as described in the Electromagnetic Interference section. If a small enough load is applied to the outputs, and the output of the Gm stage falls below a certain level, the converter will enter a burst mode of operation. Burst mode reduces switching losses while maintaining regulation of the outputs. During startup, assuming the battery voltage is above the turn-on threshold and the EN input is enabled, the controller turns on. A soft start circuit controls the reference voltage, limiting the amount of current drawn on the input and the amount of charge transferred to the output, preventing voltage overshoot. During the initial phase of the soft start, very little or no voltage is present on the output. This means that there will be no resonant phase and the converter will operate in continuous-conduction mode. The converter effectively operates in constant-current mode until regulation is achieved. Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com 5 A4401 Automotive Quasi-Resonant Flyback Control IC In the event of an overload, the current demand signal produced by the Gm amplifier restricts the output current by introducing pulse-by-pulse current limiting. Regulation Voltage The feedback resistors, R5 and R6, determine the voltage of the output rail to which they are connected, according to the following formula: VOUT = VREF ×(R5 + R6) , R6 (1) To determine the sense resistor value, assume that the minimum sense voltage before current limiting occurs is 600 mV. A reasonable maximum voltage to select during normal operation would be 500 mV. Then, the resistor value can be found as follows: RSENSE = In applications where the main control output (anode or grid) can run at relatively light loads (relative to the filament load), it may be necessary to “mix” the feedback signal. This involves adding an additional feedback from the filament output to the VA input. Note that this only applies to DC filament outputs. R3 Current Sense Resistor Selection To determine the resistance value, the maximum peak current needs to be determined. First determine the average input current, IAV , as follows: IAV = POUT(max) H% × VIN(max) , (2) where POUT is the output power. Then, the peak current through the sense resistor: IPK = 2 × IAV D(max) , (3) where D is limited to 0.7, or can be precisely found as described in the Magnetics Design section. Note that a D of 0.7 is chosen in order to achieve 0 V switching with a VBAT of 13.5 V. . (4) The power losses in the resistor can be found by first determining the rms current through it: ⎛ D(max) ⎞½ IRMS = IPK × ⎜⎜ ⎟⎟ ⎝ 3 ⎠ where R6 should be approximately 5 kΩ. The internal 1.2 V reference has a ±2% worst case tolerance, plus there is an input bias current, IBIAS, on the feedback node, VA, that has a small influence. This current flows into the ground referenced resistor, R6, creating a small voltage offset. 500 mV IPK Then, the losses in sense resistor are: PRDS = I ²RMS × RSENSE . . (5) (6) The power rating of the resistor can be selected based on the power dissipation. When selecting a resistor it is worth noting that the maximum power rating is valid up to 70°C and derates linearly to 0 W at a temperature of typically between 120°C and 140°C. Check the resistor manufacturer guidelines. Note that is imperative that this resistor be a low inductance type; avoid wire wound. Standard surface mount devices are usually acceptable. Soft Start When power is initially applied, assuming the input voltage turn-on threshold is reached, and the EN input is enabled, the controller is initiated and the MOSFET, Q1, is turned on for the first switching cycle. Initially, while the output volts are rising towards the target regulation point set by the soft start circuit, the MOSFET will run at current limit. During a soft start cycle, the reference voltage is ramped from 0 to 1.2 V in 32 steps over a period of 10 ms under the control of a DAC. This forces the output of the amplifier to vary between 0.8 and 1.5 V, which in turn reduces the effects of inrush current and voltage overshoot on the outputs. Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com 6 A4401 Automotive Quasi-Resonant Flyback Control IC If there is a special requirement for larger output capacitors, and the onboard soft start is insufficient, an external soft start can be introduced. This can be implemented by “pulling” down the amplifier output (COMP pin) and then “releasing” it gradually over the duration of the new soft start period. The pull-down circuit has to be capable of sinking at least 30 μA. how long the driver takes to remove the Miller (gate to drain) charge. The Miller charge, QGD , is quoted in the MOSFET datasheet. The driver capability can be found from data in the Electrical Characteristics table, as follows. At turn-off, the driver can shift a charge of (7 V – 0.5 V) × 1 nF, in 40 ns. The drive current required to do this is: Q1 MOSFET Selection When selecting the RDS, it is important to consider its value at minimum battery voltage, as it tends to increase with low VGS values. Below a battery voltage, VBAT, of 10 V, the actual drive amplitude, VGS, is VBAT minus 500 mV (worst case). MOSFET suppliers usually quote the variation of RDS with VGS amplitude. Another factor that influences the “real” RDS is the operating temperature. At a temperature of 140°C, this figure is increased by approximately 1.8. Again, manufacturers will provide this information. Worst case losses will occur at minimum battery voltage and maximum load. These can be considered in terms of static losses, switching turn-off losses, and switching turn-on losses: Static Losses The rms current, IRMS, that flows in the MOSFET is identical to the current that flows in the sense resistor previously calculated. Therefore: ¨ PSTATIC = ©© IPK × ª ² ⎛ D(max) ⎞½·¸ ⎜⎜ ⎟⎟ ¸ × RDS ⎝ 3 ⎠ ¹ . 6.5 V × 1 nF = 163 mA 40 ns (8) Then, the time, tLOSS, taken to shift the Miller charge can be found: tLOSS = QGD IDRIVE . (9) The turn-off switching loss can now be estimated: PTURNOFF = IPK × VDS × tLOSS × fSW(min) 2 , (10) where fSW(min) is specified in the Electrical Characteristics table and the peak current, IPK , is identical to the current that flows in the sense resistor, the calculation of which is shown in the Current Sense Resistor Selection section. VDS is calculated in the next section. +V MOSFET turned-off commanded VGS ( V ) In general, the higher the RDS and the smaller the package, the lower the cost of the MOSFET, Q1. On the basis of selecting a MOSFET to minimize cost, it is important to understand the power losses associated with it. IDRIVE = 0 +I, +V (7) Switching Turn-Off Losses Assume that the turn-off threshold, VTH, is similar to where the Miller “plateau” effect takes place. Figure 2 illustrates how an approximation can be made in terms of the turn-off losses. The duration of the tloss region is determined by VDS Loss Region tLOSS Time Figure 2. MOSFET, Q1, loss approximations Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com 7 A4401 Automotive Quasi-Resonant Flyback Control IC Switching Turn-On Losses The turn-on losses are determined by the amount of energy the resonant capacitor, C11, has to discharge into the MOSFET. At low battery voltage, the resonant swing should force the volts across the capacitor to only a few volts, making this loss negligible. Total Losses The total MOSFET power loss can be estimated as follows: PTOTAL = PSTATIC + PTURNOFF . (11) The thermal resistance, RθJA, can be determined by two methods. One is by estimating a maximum junction temperature, TJ(max). The other is to test for the operating junction temperature, using the given device package mounted on a printed circuit board with copper trace area connected to the device. RθJA is then calculated as follows: RθJA = T J – TA PTOTAL . (12) The drain-to-source rating, VDS , is the sum of the maximum input voltage, VBAT (max), plus the reflected output voltage. Adequate margin should also be added, to allow for tolerancing effects and parasitic voltage ring. It can be calculated as follows: ⎛ NP ⎞ VDS = ⎜⎜VOUT × ⎟ + VBAT (max) NOUT ⎟⎠ ⎝ . (13) D2, D3, and D4 Output Diodes Selection For the low voltage outputs such as the filament supply, it is recommended that a Schottky diode be used. For the higher voltage rails, ultrafast rectifier diodes are recommended. For each output, estimate the maximum reverse voltage, VRRM, and the maximum average current that the diode is subjected to. The VRRM rating should exceed at least 20% of the maximum VDIODE voltage, calculated as follows: VDIODE = VBAT (max) × NS1 + VOUT NP . (14) The maximum average current through the diode is simply the maximum load current. The diode should be rated to handle this current with some margin. In addition, the diode should be able to handle the power dissipation. The majority of the power loss is simply the static loss: PSTATIC = Vf × ILOAD . (15) The forward voltage drop, Vf , can be found from the diode characteristics at maximum load. C1 Input Capacitor Selection In the interests of cost and performance, it is recommended that ultralow impedance electrolytic capacitors be used. The ratio of the source impedance to the impedance of the input capacitor will determine how much of the input current is drawn from the input capacitor. For example, if the source impedance is relatively high, then the input capacitor would have to supply the triangular current that flows through the primary winding, the MOSFET Q1, and the current sense resistor. This rms current was worked out in the Current Sense Resistor Selection section. Electrolytic capacitors experience heating effects caused by the rms current flowing through the ESR of the device. The maximum rms current is normally quoted at 100 kHz and 105°C. Frequency correction factors for the ripple current are provided when the operating frequencies are less than 100 kHz. A 50 VDC rating should be adequate for most applications. C4, C6, and C8 Output Capacitors Selection The overall equivalent capacitance on the output should not be less 22 μF (see Control Loop section Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com 8 A4401 Automotive Quasi-Resonant Flyback Control IC for more information). The current that flows in and out of the capacitor is similar to the current that flows through the corresponding rectifier diode. Worst case power dissipation due to the ESR will occur at VBAT(min) and maximum load. The duty cycle under these conditions is a maximum of 0.7. The rms current in the capacitor can be found as follows: ⎛ 0.7 ⎞½ IRMS = 4 × ILOAD × ⎜⎜ ⎟⎟ . ⎝ 3 ⎠ (16) power stage effectively contains one pole formed by the output capacitors and loads. In terms of “closing the loop,” the optimal shaping components are shown in the Functional Block diagram, connected to the COMP pin. The control loop is optimized for an equivalent output capacitor of 22 μF. Larger capacitor values can be used, however, those will tend to reduce the bandwidth of the control loop. Smaller values should not be used, as they may cause instability issues. When selecting a suitable capacitor, the rms current rating should have reasonable margin with respect to the above value. In addition, the current rating should be derated to take into account the frequency correction at values of less than 100 kHz. Magnetics Design The impedance of the output capacitor will affect the amount of voltage ripple and noise that appears on the output. The impedance is composed of two components: ESR and reactance, XC. Even with a modest amount of capacitance on the output, the ESR will tend to dominate the overall impedance. • Minimum switching frequency, fSW(min). This occurs at VBAT(min) and maximum load. Note, it is recommended that the lowest possible switching frequency be used, in order to minimize switching losses and to shift the differential harmonics down the frequency spectrum. This is a compromise with the sizing of the magnetics and filtering components. The ESR can be estimated by multiplying the capacitance reactance by the dissipation factor, DF (tan δ): ESR = 1 2× × fSW × COUT × DF . (17) Note that DF is normally quoted at 25°C, however, it is usually fairly constant as the temperature is increased. The peak to peak voltage can then be found: . Vpk-pk = ESR × ILOAD × 4 The following are the known variables: • Maximum output power, POUT . • Minimum battery voltage, VBAT(min). • Maximum duty cycle. This should typically not exceed 0.7, in order to avoid excessive losses in the secondary output circuits. • Efficiency of converter, η% , at VBAT(min). This will typically be between 80% and 85%. (18) The voltage rating should be chosen to provide at least a 20% margin above the maximum output voltage. Control Loop As the converter operates in discontinuous mode, the inductor does not feature in the power stage. The • Resonant capacitor, C11, connected between the LX pin and ground. The resonant half period of 1 μs may be achieved with the parasitic capacitance that exists in the circuit. Fine tuning can be performed by adding additional capacitance. • Known magnetic core set. Material selection should be based on performance at elevated temperature, and include consideration of flux density, losses, Curie temperature, and so forth. Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com 9 A4401 Automotive Quasi-Resonant Flyback Control IC • Magnetics sizing can be determined using techniques such as area product or geometric product, or by following manufacturers guidelines, usually in the form of nomograms. To simplify the design process, the resonant switching transition is ignored and the calculations are based on the switcher operating at the boundary condition of continuous/discontinuous conduction. This is a reasonable assumption as the resonant period forms a small percentage of the overall period at minimum line voltage and maximum load (where the magnetics are designed because of worst case conditions). The objective of the design is to achieve 0 V switching when VBAT = 13.5 V (note that this voltage can be adjusted for any value). At VBAT voltages of less than 13.5 V, 0 V switching is still achieved. In addition, the LX voltage is prevented from swinging negative as the MOSFET, Q1, is commanded on as soon as 0 V is reached. To ensure that the reflected output voltage forces the LX node to 0 V, the following condition must be met: n= VOUT 13.5 , (19) Then, maximum peak current can be found: IPEAK(max) = D(max) = VOUT (VBAT(min) × n ) + VOUT . (20) The primary magnetizing inductance can be determined from the following formula, which is derived by equating the input energy to the output energy times the efficiency: LPRI = H × ( VBAT(min) × D(max)) ² 2 × fSW(min) × POUT . (21) . (22) The worst case operating flux density, BOP , can be found from the ferrite core manufacturers datasheet, by taking the saturation flux density, BSAT , at elevated temperatures and subtracting a margin, approximately 15% , to allow for operation in current limit mode or during startup. After an appropriate magnetic core set has been selected, the number of turns required on the primary winding can be found: NP = VBAT(min) × D(max) fSW(min) × BOP × Ae . (23) where Ae is the magnetics cross-sectional area in m2. The number of secondary turns can be derived through the turns ratio found previously: NS1 = n × NP . (24) If there is more than one output, the additional secondary windings are simply scaled from the main secondary as shown in the following formula: where n is the step-up turns ratio from primary to controlled output. Worst case conditions in terms of core saturation occur when VBAT is at a minimum and duty cycle, D, a maximum: VBAT(min) × D(max) fSW(min) × LPRI NS2 = NS1× VOUT2 VOUT1 , (25) where: NS2 is the quantity of turns in the additional windings, VOUT1 is the output voltage from main winding, and VOUT2 is the output voltage from any additional windings. The total air gap, lg, can be found. First, an approximate air gap, lg(approx), is found before flux fringing is taken into account. Given: μO = 4 × 10 -6 , (26) 10 Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com A4401 then: Automotive Quasi-Resonant Flyback Control IC μ × Ae × NP² Ig(approx) = O LP . (27) Because of flux fringing effects, the above air gap should be modified, according to the following formulas. Given: Ig(approx) ⎛ 2×G ⎞ . F= ⎟⎟ × ln ⎜⎜ A½ e ⎝ Ig(approx) ⎠ (28) where lg(approx) is the previously calculated approximate air gap, and G is the bobbin width (the effective winding width). . The voltage amplitude across the filament winding is: VFIL = VBAT × NSF N + VOUT1 × SF NP NS1 , (30) where: NSF is the number of turns on the filament winding, NS1 is the number of turns on the main controlled output winding, and VOUT1 is the output voltage of the main controlled output. then the total air gap can now be found: Ig = Ig(approx) × (1 + F) ing in other converter topologies because, during the MOSFET off-time, the voltage is regulated. (29) Note that most ferrite core manufacturers provide a limited number of air gap sizes. It is therefore recommended to select a standard size. Size is indicated in terms of the Al factor, which is expressed in L/N2 units. The Al factor can be derived from the above two formulae. To minimize flux leakage effects, it is recommended that the air gap should be located on the center limb. If, however, a distributed air gap is used, the air gap figure should be divided by two. Some applications require an AC filament output. Typically this may be a center tapped winding with the center tap held at some bias voltage. During the MOSFET off-time, the output voltage of the control winding is simply reflected through the turns ratio of the magnetics. During the on-time of the MOSFET, the windings are driven as a forward converter because there is no rectifying diode to isolate this action. The voltage during this interval is simply the battery voltage transformed by the turns ratio of the filament winding and the primary winding. This means that as the battery voltage varies, there will be a variation in the filament voltage. However, this variation will be less than those aris- It is probably desirable to optimize the filament voltage at nominal battery conditions; for example, at VBAT = 13.5 V. Due to the low voltage out, the number of integer turn combinations is limited. So, the filament voltage may not be exact. The turns range may only be 2 or 3. The magnetic wire sizing for each winding is determined by the ampere-turns ratio as a proportion of the total ampere-turns of all the windings. The amount of bobbin area available for the windings is influenced by the amount of insulation required, the winding construction technique, and the packing density of the circular wire. A conservative utilization factor is 0.5, that is, 50% of the bobbin window area filled with copper. The rms current of each winding has to be determined. The worst case condition is at minimum input voltage and maximum load. The primary winding current is identical to the current flowing in the current sense resistor (see the Current Sense Resistor Selection section). The current in each of the other output windings can be found as follows. Given: IPK = 2 × IOUT D'(max) , (31) 11 Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com A4401 Automotive Quasi-Resonant Flyback Control IC where IOUT is the maximum load current, and D'(max) is the duty cycle, limited to 0.3. Then the rms current in the winding is: ⎛ D' ⎞½ ⎟⎟ IRMS = IPK × ⎜⎜ ⎝ 3 ⎠ , (32) Because the number of turns has already been worked out, the ampere-turns factor can now be determined. After all of the ampere-turns are known for each winding, the bobbin window can be apportioned to each winding. It is recommended that the current density in each winding should be kept below 5 A per mm2. Another consideration when selecting the wire gauge is the skin depth (depth within which the current flows), especially at higher frequencies. Skin depth can be calculated as: δ= 75 f ½ SW , (33) For example, if 45 kHz were the minimum frequency at minimum input voltage and maximum load, then to ensure maximum wire utilization for the first four switching harmonics, the switching frequency would be 180 kHz. The conduction depth at 180 kHz would equal 0.18 mm, therefore, the wire diameter should not exceed 0.36 mm. For any particular winding, if the current rating of the wire is insufficient even though the wire meets the skin depth criteria, multiple wires wound in parallel will be necessary. It is recommended to locate the start and finish of each winding as close as possible on the bobbin. This minimizes the “loop area” and reduces the effects of noise pick-up. When winding the high voltage windings, such as for the anode or grid, it is advisable to insert a layer of polyester insulating tape between each layer as well as between adjacent windings. C11 Resonant Capacitor Selection The resonance that occurs when the MOSFET, Q1, turns off is formed by the interaction of the primary magnetizing inductance and the capacitance between the LX node (the drain terminal of the MOSFET) and ground. The design is optimized for a half resonant period of 1 μs. This means the resonant capacitor value can be found from the following formula: ⎛T CRES = ⎜⎜ R ⎝ ⎞² 1 ⎟⎟ × ⎠ LPRI , (34) where TR is a half resonant period of 1 μs. It is advisable to measure the half resonant period in the application, as the parasitic capacitance between the LX node and ground can be substantial and may even be sufficient to meet the requirements with very little additional capacitance. PCB Layout Guidelines The layout can be considered as two blocks: primary and secondary: Primary Block To minimize parasitic noise appearing on the ground return, and at the LX and ISS nodes, as well as to maximize the effectiveness of the EMI LPRI VBAT LX node A4401 CIN MOSFET Q1 ISS node RSENSE Minimize this loop area Figure 3. Main power loop 12 Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com A4401 Automotive Quasi-Resonant Flyback Control IC filtering, it is imperative that the “power loop” formed by the input filter, primary winding, MOSFET, Q1, and sense resistor is as short and “tight” as possible. This means components should be placed as close together as possible, and the loop area should be minimized, in order to reduce the effects of magnetic pickup and noise generation at higher frequencies. A ground plane is not necessary, however, a good star ground connection should be formed between the input filter capacitor and the sense resistor. Circuit traces should be as wide as possible, in order to minimize leakage inductance. Figure 3 illustrates the main power loop. A local ground plane around the A4401 can be used to minimize ground bounce issues. This can be done with a simple connection from the ground pin of the A4401 to the star ground, being careful to avoid any connection between this local ground plane and the power loop. The compensation components connected to the COMP pin, the filter capacitor connected to VIN, the feedback resistors connected to VA, and the resonant capacitor connected to LX should be located as close to their respective pins as possible. In addition, the length of the traces between those components and ground should be as short as possible. Secondary Block Each output power circuit should be laid out with identical principles relative to the primary side power circuit, that is, with each secondary winding, rectifying diode, and output capacitor positioned close together and forming a tight loop. The high voltage output circuits should be kept well away from all the control circuitry. It is recommended that good connections be made between each of the local output grounds and the star ground. In addition, all of the output grounds should be connected together via wide traces and not ground planes. Figure 4 illustrates one of the output loops. The feedback resistor connected to the regulated output rail should be located as close to the VA pin as possible. The PCB trace that connects the top of this resistor to the output rail should not be located near any of the output power loops or ground connections. Electromagnetic Interference COUT LSEC Minimize this loop area Figure 4. One of three output power loops Some of the previous sections provide information on reducing EMI in terms of input filter capacitance selection, magnetics design (to achieve zero voltage switching at nominal input voltage), and board layout. This section provides some additional advice on reducing EMI. Effects of Magnetics Design on 0 V Switching Figure 5 illustrates a converter running at full power with VBAT = 13.5 V. The upper trace is the voltage across LX and the lower trace is the voltage across the sense resistor (essentially the current through the primary winding and the MOSFET, Q1). The magnetic set in this case was designed to achieve 0 V switching at a VBAT of 7 V. 13 Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com A4401 Automotive Quasi-Resonant Flyback Control IC It can be seen that at MOSFET turn-on there is a considerable sinusoidal ringing on the voltage sense, in the region of 3 MHz. In addition, there is a considerable ringing during the MOSFET turn-off, in the region of 6 MHz. These ringing effects are caused by the primary-to-secondary leakage inductance interacting with parasitic capacitance. These noise sources will probably have an impact on the conducted emissions and should therefore be suppressed. By rescaling the magnetics so that 0 V switching is achieved with VBAT = 13.5 V, the turn-on ringing is almost completely damped, as shown in Figure 6. Because the resonant capacitor, C11, had discharged to 0 V before switching, there was effectively little energy left. The turn-off noise is also improved dramatically as well. It can be seen that the ringing amplitude is lower, dampens more quickly, and the ringing frequency is higher making it easier to snub (see the Radiated Emissions on the LX Node subsection). The reason for this effect is that a lower step-up ratio is required to achieve the output voltage. This improves the magnetic coupling coefficient and therefore the leakage inductance is reduced. Running the system with a lower turns ratio means that the duty cycle is greater. The advantage of this is that the peak current is lower, resulting in lower turn-off switching losses and reduced harmonics of the input current, thus reducing the conducted emissions. Radiated Emissions on LX Node A potential source of radiated emissions is turning off the MOSFET. A resonance is set up between the leakage inductance of the magnetics and the parasitic capacitance on the LX node. Assuming the magnetics are well designed in terms of reducing leakage inductance, this resonance should occur in the region of tens of megahertz. Another potential source of radiated emissions is turning off the output rectifying diodes. The following procedure can be applied to address both issues. MOSFET Turn-On MOSFET Turn-On VLX MOSFET Turn-Off MOSFET Turn-Off VLX VRSENSE VRSENSE Figure 5. Voltages with 0 V switching of Q1 at VBAT of 7 V Figure 6. Voltages with 0 V switching of Q1 at VBAT of 13.5 V 14 Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com A4401 Automotive Quasi-Resonant Flyback Control IC A simple low-loss R-C snubber can be deployed to dampen these oscillations. To do so, follow these simple steps to determine the component values required: 1. Measure the voltage resonant frequency, fRES , on the LX node. 2. Add an additional capacitance (C11) between LX and ground until the resonant frequency is halved, fRES/2: 1 , fRES = (35) × (LLEAK × CLEAK)½ × fRES = 1 × × (LLEAK × CNEW)½ CNEW = CLEAK + CADD , (36) (37) , where: Figure 6 illustrates a converter running at full power with VBAT = 13.5 V. The upper trace is the voltage across LX and the lower trace is the voltage across the sense resistor, R3. It can be seen that at turn-on, the noise is very low, as the resonant action introduces a very slow voltage slew which reaches 0 V before the MOSFET is turned on. This action ensures minimal noise is produced. However, at turn-off there is a high frequency ringing of 25 MHz which could cause issues in terms of complying with radiated emissions. Figure 7 illustrates the effects of adding an R-C snubber (R = 30 Ω and C = 330 pF in series between the LX pin and GND). It can be seen that, through the addition of the snubber, the turn-off ringing has been considerably damped. This was achieved with almost negligible additional power loss. Conducted Emissions To help reduce the conducted CADD is the additional capacitance added, emissions, an EMI filter may be necessary as shown in the Functional Block diagram. CLEAK is the parasitic capacitance, and LLEAK is the parasitic inductance. Note that the value of the additional capacitance should be in the region of a few hundredths of picofarads. 3. Now to calculate component values that will result in fRES/2, first calculate CLEAK, given: , CLEAK + CADD = 4 × CLEAK CLEAK = CADD 3 . (38) VLX 5. Finally, solve for the characteristic impedance of the parasitic components: ⎞½ ⎟⎟ ⎠ . MOSFET Turn-Off (39) 4. With CLEAK solved, solve for LLEAK ⎛ L RO = ⎜⎜ LEAK C ⎝ LEAK MOSFET Turn-On VRSENSE (40) RO can be selected as the damping resistor value (not shown in functional block diagram). Typically a 1/8 W resistor is adequate. Figure 7. Voltages with 0 V switching of Q1 with an R-C snubber 15 Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com A4401 Automotive Quasi-Resonant Flyback Control IC The capacitance selection has been dealt with in the Input Capacitor Selection section. In terms of selecting an inductance, generally the higher the inductance the better as far as noise rejection of the differential conducted emissions is concerned. However, a large inductance usually means additional cost and increased size, therefore, it is advisable to select an inductance as large as possible within the constraints of board space and cost. The maximum average current that flows through the inductor is similar to the maximum average current that was worked out in the Current Sense Resistor Selection section. Both the rms and saturation current ratings of the inductor selected should be higher than the maximum average current. Care should be taken when selecting inductors for operation at elevated temperatures, because some manufacturers rate their parts using only self-generated heating, giving the impression of operability at higher temperature conditions. 16 Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com A4401 Automotive Quasi-Resonant Flyback Control IC Package L, 8-Pin Narrow SOIC 4.90 ±0.10 4° ±4 8 0.21 ±0.04 3.90 ±0.10 6.00 ±0.20 1.27 8 1.75 +0.43 0.84 –0.44 A 1 0.65 5.60 (1.04) 2 0.25 1 SEATING PLANE GAUGE PLANE 9X SEATING PLANE 0.20 C 1.75 MAX 0.41 ±0.10 1.27 +0.08 0.18 –0.07 C B 2 PCB Layout Reference View For Reference Only (reference JEDEC MS-012 AA) Dimensions in millimeters Dimensions exclusive of mold flash, gate burrs, and dambar protrusions Exact case and lead configuration at supplier discretion within limits shown A Terminal #1 mark area B Reference land pattern layout (reference IPC7351 SOIC127P600X175-9AM); all pads a minimum of 0.20 mm from all adjacent pads; adjust as necessary to meet application process requirements and PCB layout tolerances; when mounting on a multilayer PCB, thermal vias at the exposed thermal pad land can improve thermal dissipation (reference EIA/JEDEC Standard JESD51-5) Copyright ©2007-2008, Allegro MicroSystems, Inc. The products described here are manufactured under one or more U.S. patents or U.S. patents pending. Allegro MicroSystems, Inc. reserves the right to make, from time to time, such departures from the detail specifications as may be required to permit improvements in the performance, reliability, or manufacturability of its products. Before placing an order, the user is cautioned to verify that the information being relied upon is current. Allegro’s products are not to be used in life support devices or systems, if a failure of an Allegro product 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. The information included herein is believed to be accurate and reliable. However, Allegro MicroSystems, Inc. assumes no responsibility for its use; nor for any infringement of patents or other rights of third parties which may result from its use. For the latest version of this document, visit our website: www.allegromicro.com 17 Allegro MicroSystems, Inc. 115 Northeast Cutoff, Box 15036 Worcester, Massachusetts 01615-0036 (508) 853-5000 www.allegromicro.com