AND8023/D Implementing the NCP1200 in Low-Cost AC/DC Converters Prepared by: Christophe Basso ON Semiconductor http://onsemi.com APPLICATION NOTE External MOSFET connection: by leaving the external MOSFET external to the IC, you can select avalanche proof devices which, in certain cases (e.g. low output powers), let you work without an active clamping network. Also, by controlling the MOSFET gate signal flow, you have an option to slow down the device commutation, therefore reducing the amount of ElectroMagnetic Interference (EMI). SPICE model: a dedicated model to run transient cycle–by–cycle simulations is available but also an averaged version to help you closing the loop. Ready–to–use templates can be downloaded in OrCAD’s PSpice, INTUSOFT’s IsSpice and Spectrum–Software’s µCap from the ON Semiconductor web site, www.onsemi.com, NCP1200 related section. Low external part–count: by integrating the principal electronic blocks in one die, the final NCP1200 implementation reveals an obvious gain in component reduction compared to other offers: – Built–in clock generator without external R–C elements. Operating frequencies at 40 kHz, 60 kHz or 100 kHz. – The optocoupler is directly wired to the feedback pin, the internal IC control taking care of the signal flow. – The 250 ns Leading Edge Blanking (LEB) circuitry also saves an external R–C network. INTRODUCTION The NCP1200 implements a standard current mode architecture where the switch–off time is dictated by the peak current setpoint. This component represents the ideal candidate where low part–count is the key parameter, particularly in low–cost AC/DC adapters, auxiliary supplies etc. Thanks to its proprietary Very High–Voltage Integrated Circuit (VHVIC) technology, ON Semiconductor NCP1200 will please experts as well as non–experts in the Switch–Mode Power Supplies (SMPS) arena as the following features demonstrate: No need of auxiliary winding: the VHVIC technology lets you supply the IC directly from the high–voltage DC rail. We call it Dynamic Self–Supply (DSS). In battery charger applications, you no longer need to design a special primary circuitry to cope with the transient lack of auxiliary voltage (e.g. when Vout is low). Short–circuit protection: by permanently monitoring the feedback line activity, the IC is able to detect the presence of a short–circuit, immediately reducing the output power for a total system protection. Once the short has disappeared, the controller resumes and goes back to normal operation. For given applications (e.g. constant output power supplies), you can easily disconnect this protective feature. Low standby–power: if SMPS naturally exhibit a good efficiency at nominal load, they begin to be less efficient when the output power demand diminishes. By skipping un–needed switching cycles, the NCP1200 drastically reduces the power wasted during light load conditions. In no–load conditions, the NPC1200 allows the total standby power to easily reach next International Energy Agency (IEA) recommendations. No acoustic noise while operating: instead of skipping cycles at high peak currents, the NCP1200 waits until the peak current demand falls below a user–adjustable 1/3rd of the maximum limit. As a result, cycle skipping can take place without having a singing transformer … You can thus select cheap magnetic components free of noise problems. Semiconductor Components Industries, LLC, 2001 April, 2001 – Rev. 3 Dynamic Self–Supply To avoid the use of a dissipative resistor, NCP1200 implements a controlled current source whose technology allows a direct connection to the high–voltage rail (up to 450 VDC). Figure 1a depicts the component arrangement. The current source always operates in the ON or OFF states, either delivering 4 mA or zero. As a result, the internal VCC pin ramps up and down with a 2 V ripple centered around 10.6 V (Figure 1b). 1 Publication Order Number: AND8023/D AND8023/D Pin 8 VCCOFF = 11.4 V 2 4 + - VCC 10.6 V Avg. ON/OFF VCCON = 9.8 V 3 ON Pin 6 + 10 V/12 V C OFF Current Source Output Pulses 10.00M Figure 1a. Internal Implementation of the Dynamic Self–Supply 30.00M 50.00M 70.00M 90.00M Figure 1b. DSS Waveforms During Start–Up and Normal Mode (CVCC = 10 F) As one can see from Figure 1a, the current source supplies the IC and the Vcc capacitor. Thanks to the circuitry nature, the current source duty–cycle will automatically adjust depending on the IC average current consumption. Because this is a controlled source, this current stays constant whatever the high–voltage rail excursion (VHV): from 100 VDC up to 370 VDC. The IC contribution to the total SMPS power budget is therefore: VHV . Ipin8. The internal IC consumption is made of the internal electronic blocks (clock, comparators, driver etc.) but also depends on the MOSFET’s gate charge, Qg. If we select a MOSFET like the MTD1N60E, Qg equals 11 nC (max) and with a maximum switching frequency of 48 kHz, the average current necessary to drive the MOSFET (excluding the driver efficiency and neglecting various voltage drops) is: Fsw Qg 530 A Fsw = maximum switching frequency (Hertz) Qg = MOSFET’s gate charge (Coulomb) If we now add this number to the normal IC consumption, we reach a typical total of 1.2 mA. The total IC power contribution alone is therefore: 330 V . 1.2 mA = 396 mW. Decreasing the Standby Power Below stands the future stand–by power recommendations issued by International Energy Agency (IEA). It concerns stand–by power when there is no output power demand: ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ Rated Input Power Phase 1, January 2001 Phase 2, January 2003 Phase 3, January 2005 0.3 W and 15 W 1.0 W 0.75 W 0.30 W 15 W and 50 W 1.0 W 0.75 W 0.50 W 50 W and 75 W 1.0 W 0.75 W 0.75 W through pin 8 (actually the total NCP1200 consumption + a few losses) and integrate over a half–sine only. This leads to the final formula: A typical 4 W universal mains AC/DC wall adapter using the NCP1200 will exhibit a no–load power consumption of less than 380 mW @ Vin = 230 VAC. If a lower power consumption is required, you have several options to achieve it: 1. Use a MOSFET with lower gate charge Qg, but as we saw the driver’s contribution is small. 2. Connect pin 8 through a diode to one of the mains input to apply a rectified half–wave on pin 8. Since we have either 4.0 mA or 0 synchronized with this half–sine wave, we should integrate the V–I product over a cycle to obtain the final average power. However, depending on the VCC capacitor, we may have some half–sine portions where the current source if OFF: the VCC capacitor level has not reached UVLOLow and the current source is left opened. To simplify the curves, we can assume a constant current flowing Pavg 2 * Vmainspeak * lavg (Pin 8) . Our previous number drops to 250 mW. If you carefully look at Figure 2a, you will notice that the reverse voltage is sustained by the diode bridge. The maximum anode voltage of the Dstart diode is also clamped at the high–voltage rail. Therefore, a standard fast diode like the 1N4148 can safely be used in this option. However, because of the lack of synchronization between the DSS and the mains, it is necessary to equilibrate the diode voltage when both diode and DSS are inactive. This can be done by wiring a 220 kΩ in parallel with the diode. Otherwise, a standard 1N4937 can also do the job without any resistor in parallel… http://onsemi.com 2 AND8023/D 220 k Aux D4 15 V 400 mW 1N4148 C3 4.7 F 350 V + NCP1200 1 HV 8 Adj 2 FB NC 7 3 CS VCC 6 D5 1N4148 NCP1200 1 Adj HV 8 2 FB 7 3 CS VCC 6 4 Gnd Drv 5 4 Gnd Drv 5 Figure 2a. A simple diode naturally reduces the average voltage on pin 8. M2 MTD1N60E R3 1K + 10 µF C6 Figure 2b depicts the VCC voltage obtained when using this method. Despite the lack of synchronization between the mains and the DSS, the average VCC level is not affected. Figure 3. By wiring an auxiliary winding, you further decrease the standby power. The typical operating voltage can be set at 12 V, with an overvoltage protection at 15 V. As a benefit, if the optocoupler fails, the SMPS turns into primary regulation mode and prevents any output voltage runaway. Typical measurements using this method gave a standby power of 84 mW at 230 VAC. VCC 2 V/div Skip Cycle Mode The NCP1200 automatically skips switching cycles when the output power demand drops below a given level. This is accomplished by monitoring the FB pin (pin 2). In normal operation, pin 2 imposes a peak current accordingly to the load value. If the load demand decreases, the internal loop asks for less peak current. When this setpoint reaches a determined level, the IC prevents the current from decreasing further down and starts to blank the output pulses: the IC enters the so–called skip cycle mode, also named controlled burst operation. Because this operation takes place at low peak currents, you will not hear any acoustic noise in your transformer. Figure 4a depicts how the IC implements the cycle skipping while Figure 4b shows typical switching patterns at different load levels. Vpin8 50 V/div Figure 2b. By wiring a diode in series with pin 8, you do not affect the average VCC level. 3. If you permanently force the VCC level above VCCOFF with an auxiliary winding, you will automatically disconnect the internal start–up source and the IC will be fully self–supplied from this winding. Again, the total power drawn from the mains will significantly decrease. However, make sure the auxiliary voltage never exceeds the 16 V limit, particularly in overshoot transients (e.g. the load is suddenly removed). To avoid this trouble and also implement an efficient Over Voltage Protection (OVP), Figure 3 schematic offers a possible solution: http://onsemi.com 3 AND8023/D Clock Max Peak Current S G To Driver Q R 1/4 GAIN FB LEB CS Adj Skip Cycle Current Limit + + - 1.4 V + Figure 4c. The skip cycle takes place at low peak currents which guarantees noise–free operation. Figure 4a. NCP1200 Skip Cycle Implementation The power transfer now depends upon the width of the pulse bunches (Figure 4b). Suppose we have the following component values: P3 Lp, primary inductance = 1 mH Fsw , switching frequency = 48 kHz Ip skip = 300 mA (or 350 mV/Rsense) P2 The theoretical (η = 100%) power transfer is therefore: 1 Lp Ip 2 Fsw 2.16 W 2 (eq. 2) If this IC enters skip cycle mode with a bunch length of 10 ms over a recurrent period of 100 ms, then the total power transfer is: 2.1 . 0.1 = 216 mW. P1 Skip Adjustment By altering the DC voltage present on pin 1, you have the ability to adjust the current level at which skip cycle operation will take place. By default, skip cycle occurs when the peak current demand falls below one third of the maximum peak current. If your design needs to enter standby at a higher or lower current (e.g. for noise reasons), you can alter the internal setpoint by imposing a different level than 1.4 V on pin 1. This can be achieved by using a resistor bridge between VCC and ground. Pin 1 impedance is typically 25 kΩ. Figure 4b. Output pulses at various power levels (X = 5 s/div) P3 < P2 < P1 When FB is above the skip cycle threshold (1.4 V by default), the peak current cannot exceed 1 V/Rsense. When the IC enters the skip cycle mode, the peak current cannot go below Vpin1/4, 350 mV/Rsense by default (Figure 4c). The user still has the flexibility to alter this 1.4 V by either shunting pin 1 to ground through a resistor or raising it through a resistor up to the desired level. http://onsemi.com 4 AND8023/D Overload Operation on normal output load conditions and the maximum peak current allowed by the system. The time–out used by this IC works with the VCC decoupling capacitor: as soon as the VCC decreases from the VCCOFF (typically 11.4 V) the device internally watches for an overload current situation. If this condition is still present when the VCCON is reached, the controller stops the driving pulses, prevents the self–supply current source to restart and puts all the circuitry in standby, consuming as little as 350 µA typical (ICC3 parameter). As a result, the VCC level slowly discharges toward 0. When this level crosses 6.3 V typical, the controller enters a new startup phase by turning the current source on: VCC rises toward 11.4 V and again delivers output pulses at the VCCOFF crossing point. If the fault condition has been removed before V CCON approaches, then the IC continues its normal operation. Otherwise, a new fault cycle takes place. Figure 5 shows the evolution of the signals in presence of a fault. In applications where the output current is purposely not controlled (e.g. wall adapters delivering raw DC level), it is mandatory to implement a true short–circuit protection. A short–circuit actually forces the output voltage to be at a low level, preventing a bias current to circulate in the optocoupler LED. As a result, the FB pin level is pulled up to 4.8V, as internally imposed by the IC. The peak current setpoint goes to the maximum and the supply delivers a rather high power with all the associated effects. Please note that this can also happen in case of feedback loss, e.g. a broken optocoupler. To account for this situation, the NCP1200 hosts a dedicated overload detection scheme. Once activated, this circuitry imposes to deliver pulses in a burst manner with a low duty–cycle. The system recovers when the fault condition disappears. During the start–up phase, the peak current is pushed to the maximum until the output voltage reaches its target and the feedback loop takes over. This period of time depends VCC Regulation Occurs Here 11.4 V Latch–off Phase 9.8 V 6.3V Time Drv Driver Pulses Driver Pulses Time Internal Fault Flag Fault is Relaxed Time Fault Occurs Here Startup Phase Figure 5. If the fault is relaxed during the VCC natural fall down sequence, the IC automatically resumes. If the fault still persists when VCC reached UVLOL, then the controller enters burst mode until recovery. ∗ Always make sure that the output power you are looking for asks for an FB level less than the maximum value to avoid a false overload circuitry trigger. the overload protection activates at roughly twice this value, especially if you implement a low–gain single zener feedback loop. You need to be sure that the output components sustain the corresponding current. Fortunately, because of temperature, the transformer core permeability will decrease as well as the primary inductance. This means less maximum power at higher temperatures. ∗ Because of component dispersions (NCP1200 frequency span, Rsense tolerance etc.), the overload protection will not be active as soon as you exceed the nominal power by a small amount. If you shoot for a given Pout, it is likely that http://onsemi.com 5 AND8023/D Deactivating the Overload Detection the power (except if you have the adequate place on your PCB). However, using the solution of the series diode or the self–supply through the auxiliary winding does not cause any problem with this frequency version. By wiring a 20 kΩ resistor from FB to ground, you permanently deactivate the overload protection circuitry. This is a very useful feature, especially if you need to build a constant output power converter. Calculating the VCC Capacitor for Overload As the above section describes, the fall down sequence depends upon the VCC level: how long does it take for the VCC line to go from 11.4 V to 9.8 V? The required time depends on the start–up sequence of your system, i.e. when you first apply the power to the IC. The corresponding transient fault duration due to the output capacitor charging must be less than the time needed to discharge from 11.4 V to 10 V, otherwise the supply will not properly start. The test consists in either simulating or measuring in the lab how much time the system takes to reach the regulation at full load. Let’s suppose that this time corresponds to 6ms. Therefore a VCC fall time of 10 ms could be well appropriated in order to not trigger the overload detection circuitry. If the corresponding IC consumption, including the MOSFET drive, establishes at 1.5 mA, we can calculate the required capacitor using the following formula: Power Dissipation The NCP1200 is directly supplied from the DC rail through the internal DSS circuitry. The current flowing through the DSS is therefore the direct image of the NCP1200 current consumption. The total power dissipation can be evaluated using: (VHVDC 11 V) ICC2. If we operate the device on a 250 VAC rail, the maximum rectified voltage can go up to 350 VDC. As a result, the worse case dissipation occurs on the 100 kHz version which will dissipate 340 . 1.8 mA @ Tj = 25°C = 612 mW (however this 1.8 mA number will drop at higher operating temperatures). A DIP8 package offers a junction–to–ambient thermal resistance RθJ–A of 100°C/W. The maximum power dissipation can thus be computed knowing the maximum operating ambient temperature (e.g. 70°C) together with the maximum allowable junction temperature (125°C): t V C, with ∆V = 2 V (eq. 3). Then for a wanted ∆t i Tj max TA max P max 550 mW. As we can see, we do R J A of 10 ms, C equals 8 µF or 10 µF for a standard value. When an overload condition occurs, the IC blocks its internal circuitry and its consumption drops to 350 µA typical. This happens at VCC = 10 V and it remains stuck until VCC reaches 6.3 V: we are in latch–off phase. Again, using the calculated 9.8 µF and 350 µA current consumption, this latch–off phase lasts: 109 ms. not reach the worse consumption budget imposed by the 100 kHz version. Two solutions exist to cure this trouble. The first one consists in adding some copper area around the NCP1200 DIP8 footprint. By adding a min–pad area of 80mm of 35µ copper (1 oz.), RθJ–A drops to about 75°C/W which allows the use of the 100 kHz version. The other solutions are a) add a series diode with pin 8 (as suggested in the above lines) to drop the maximum input voltage down to 225 V (707/π) and thus dissipates less than 410 mW b) implement a self–supply through an auxiliary winding to permanently disconnect the self–supply. SO–8 package offers a worse RθJ–A compared to that of the DIP8 package: 178°C/W. Again, adding some copper area around the PCB footprint will help decreasing this number: 12mm x 12mm to drop RθJ–A down to 100°C/W with 35µ copper thickness (1 oz.) or 6.5mm x 6.5mm with 70µ copper thickness (2 oz.). As one can see, we do not recommend using the SO–8 package for the 100 kHz version with DSS active as the IC may not be able to sustain Protecting the Power MOSFET If the leakage inductance is kept low, an avalanche rugged MOSFET such as the MTD1N60E can withstand accidental avalanche energy, e.g. during a high–voltage spike superimposed over the mains, without the help of a clamping network. However, if this leakage path permanently forces a drain–source voltage above the MOSFET BVdss (e.g. 600 V), a clamping network is mandatory and must be built around a passive RC network or a Transient Voltage Suppressor (TVS). Figure 6a depicts the phenomenon while the below lines details the calculation steps: http://onsemi.com 6 AND8023/D RC network or TVS BV dss HV rail HV rail You need 1 4 5 3 TVS Ultra–fast Figure 6a. Care must be taken to ensure a safe operation of the MOSFET 2. Transient Voltage Suppressor 1. RC Network Despite the low–cost offered by the above RC solution, the clamping level unfortunately varies with the peak current. If you need a very precise clamping level, you must implement a zener diode or a TVS. There are little technology differences behind a standard zener diode and a TVS. However, the die area is far bigger for a transient suppressor than that of zener. A 5 W zener diode like the 1N5388B will accept 180 W peak power if it lasts less than 8.3 ms. If the peak current in the worse case (e.g. when the PWM circuit maximum current limit works) multiplied by the nominal zener voltage exceeds these 180 W, then the diode will be destroyed when the supply experiences overloads. A transient suppressor like the P6KE200 still dissipates 5 W of continuous power but is able to accept surges up to 600 W @ 1 ms. If the peak power is really high, then turn to a 1.5 KE200 which accepts up to 1.5 kW @ 1 ms. The RC network will permanently impose a fixed clamping level that will oppose to the leakage voltage. As a result, the drain will be clamp to VHVrail + Vclamp. You normally select a clamping level between 40 to 80 volts above the reflected output voltage when the supply is heavily loaded. To calculate the component values, the following formula will help you: Rclamp 2 Vclamp (Vclamp (Vout Vf sec) N) Lleak Ip 2 F SW (eq. 4) Cclamp Vclamp Vripple F SW Rclamp (eq. 5) The power dissipated by Rclamp can also be expressed by: P Rclamp 1 Lleak Ip 2 F SW 2 Select the zener or TVS clamping level between 40 to 80 volts above the reflected output voltage when the supply is heavily loaded. Vclamp (Vout Vf sec) N Vclamp 1 (Vout Vf sec) N 3. Snubber Network with: Vclamp: the desired clamping level. Ip: the maximum peak current (e.g. during overload) Vout + Vf: the regulated output voltage level + the secondary diode voltage drop Lleak: the primary leakage inductance N: the Ns:Np conversion ratio Fsw: the switching frequency Vripple: the clamping ripple, could be around 20 V Another option lies in implementing a snubber network which will damp the leakage oscillations but also provide more capacitance at the MOSFET’s turn–off. The peak voltage at which the leakage forces the drain is calculated by: V max Ip Lleak (eq. 6) where Clump Clump represents the total parasitic capacitance seen at the MOSFET opening. Depending on the output power, you can either wire a simple capacitor across the MOSFET or an R–C network, as shown by Figure 6b. http://onsemi.com 7 AND8023/D Csnubb To lower the dissipated heat, you can wire a capacitor C Csnubb in series with R: C power is low, you can directly wire this capacitor between the MOSFET drain and ground (not between drain–source to avoid substrate injection). Unfortunately, you discharge this capacitor in the MOSFET every time it turns on... Further tweaking is thus necessary to tune the dissipated power versus standby power. Rsnubb RS 1 (eq. 8). If the output F ring R RS Figure 6b. If the output power is low, you can wire a simple capacitor MOSFET’s drain and ground. ON Semiconductor Protection Devices SMPS protection clearly needs fast switching components to ensure a reliable operation in the event of dangerous transients. ON Semiconductor portfolio offers a comprehensive list of semiconductors dedicated to protection: fast diodes, zeners, TVS etc. Below is a small list of typical component you can select to protect the MOSFET in your application. The complete list of TVS devices can be found at the following URL: www.onsemi.com or by ordering the selection guide TVSPROMO1299/D by sending an email to: [email protected]: To calculate the values of this RC snubber, you need to measure the ringing frequency imposed by the leakage inductance and all the stray capacitances. Make sure you use a low capacitance probe, otherwise you might affect the observed frequency. Once you have it, calculate the impedance of the leakage inductance at the ringing frequency: Z 2 F ring L leak (eq. 7). Wiring a resistor R whose value equals Z should solve the problem but at the expense of a large power dissipation. Clipping Elements ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ Reference Nominal Voltage (V) Average Power (W) Maximum Peak Power 1N5953B 150 1.5 98 W @ 1 ms 1N5955B 180 1.5 98 W @ 1 ms 1N5383B 150 5.0 180 W @ 8.3 ms 1N5386B 180 5.0 180 W @ 8.3 ms 1N5388B 200 5.0 180 W @ 8.3 ms P6KE150A 150 5.0 600 W @ 1 ms P6KE180A 180 5.0 600 W @ 1 ms P6KE200A 200 5.0 600 W @ 1 ms 1.5KE150A 150 5.0 1.5 kW @ 1 ms 1.5KE180A 180 5.0 1.5 kW @ 1 ms 1.5KE200A 200 5.0 1.5 kW @ 1 ms Reference VRRM Ton (typical) IF max MUR160 600 V 50 ns 3A MUR1100E 1000 V 25 ns 3A 1N4937 600 V 200 ns 1A MSR860* 600 V 100 ns 8A MSRB860–1* 600 V 100 ns 8A Fast Diodes *Soft recovery diodes http://onsemi.com 8 AND8023/D Calculating the Component Values for a Typical Application To simplify the calculation, we will neglect the charging period and thus consider a total discharge time Suppose that we would like to build a simple AC/DC wall adapter with an NCP1200 delivering a raw DC voltage with the following specs: equal to evaluate the equivalent current (Iload) drawn by the charge at the lowest input line condition. Let’s us adopt a 40% ripple level, or a 50 V drop from the corresponding Vinpeak. To evaluate the equivalent load current (which discharges Cbulk between the peaks), we divide the input power by the average rectified voltage: Pout = 3.5 W Target efficiency: η = 75% → Pin = 4.66 W Vout = 6 V, Rload = 10.3 Ω, Iout = 580 mA VACin = 100 VAC to 250 VAC The first step lies in calculating the peak rectified value obtained from this line range: Vin peak Vin 2 2 Vf with : Iload @ 100 VAC input voltage. Thanks to Figure 7b information, we can evaluate the capacitor value which allows the drop from Vpeak down to Vavg – (Vripple/2) to s t a y w i t h i n o u r 5 0 V t a r g e t : dV C i load dt: → Vinpeak = 140 V at low line → Vinpeak = 352 V at high line Cbulk The bulk capacitor supplies the high–frequency current pulses to the SMPS, while it is refueled at low rate by the mains. Figure 7a shows a typical rectifying configuration. However, the bulk capacitor should be dimensioned to provide enough energy between the line peaks. The worse case appears at the lowest line level combined with the maximum output current. Figure 7b depicts the voltage waveform observed over Cbulk in Figure 7a example. Vripple 2 High–voltage DC Rail Range From the above formula, we can extract the ripple amplitude at any line level: Vripple Vpeak 1 Vout Bridge + + out Pout Cbulk Fline Vpeak 2 (eq. 12) 4.67/V(out) B1 Current V HVavg Vin 2 2 Vf Vripple 2 (eq. 13) → 115 VDC @ 100 VAC with a 50 Vpp ripple (initial target) → 343 VDC @ 250 VAC with 19 Vpp ripple (calculated) 5 Cbulk 10 µ N 1 Let us now calculate the final average DC value delivered to the SMPS at high but also low line: Iout Vline Pout 2 Fline Vripple Vpeak = (eq. 11) 9.3 F or 10 F for a normalized value. Bulk Capacitor 2 Pout Pin (eq. 10) = 46 mA Vripple Vrect avg Vpeak 2 (eq. 9) Vin: the AC input voltage Vf: a forward drop of a bridge diode, 0.7 V @ Id 1 + 1 . From the design characteristics, we can 2 Fline To confirm the validity of our calculation, a simple SPICE simulation can be run where the load is replaced by a current source. However, thanks to the feedback loop of the converter, the input current will vary depending on the line level. To avoid tweaking the current source at different line levels, Figure 7a’s B1 element computes the input – Figure 7a. A Full–Wave Rectification Configuration Vripple current according to the required power: Iload Vinpeak P SMPS Vbulk exactly as the final SMPS would do. Simulations gave a 29 Vpp ripple at low line (VHVavg = 126 V), turning into 12 Vpp at high line (VHVavg = 342 V). The calculations gave less pessimistic results simply because we have neglected the capacitor re–charge time which reduces its discharge time. VHVavg Diode Bridge Selection 10 ms To select the right rectifiers, it is necessary to know the RMS current flowing through its internal diodes. Prior to reach this final result, we need to evaluate the diode conduction time. From Figure 8a, we can see that the diode starts to conduct when VACin reaches Vmin and stops when reaching Vinpeak: Figure 7b. A Typical Ripple Voltage over the Bulk Capacitor http://onsemi.com 9 AND8023/D a surge current (when you plug the SMPS in the AC outlet) less than the diode maximum peak current (Ifsm). Thanks to these numbers, we compute the apparent power at low line: 86 mA x 100 V = 8.6 VA which compared to our 4.66 Watts of active power (neglecting the input diode bridge and Cbulk losses) gives a power factor tc Vinpeak Vmin Vbulk of: PF W 0.54 (eq. 19) conform to what we could V.A expect from this kind of offline power supply. With a reverse voltage of 400 V, you can select either a 1 A/400 V bridge or 4 x 1N4007 diodes. Transformer Calculation 300 mA Transformer calculation can be done in several manners: a) you evaluate ALL the transformer parameters, electrical but also physical ones, including wire type, bobbin stack–up etc. b) you only evaluate the electrical data and leave the rest of the process to a transformer manufacturer. We will adhere to the latest option by providing you with a list of potential transformer manufacturers you can use for prototyping and manufacturing. However, as you will discover, designing a transformer for SMPS is an iterative process: once you freeze some numbers, it is likely that they finally appear either over or under estimated. As a result, you re–start with new values and see if they finally fit your needs. To help you speed–up the transformer design, an Excel design–aid sheet is available from the ON Semiconductor web site, www.onsemi.com/NCP1200. Let’s start the process with the turn ratio calculation: VAC in Idiode X–2 ms/div Figure 8a. When VACin reaches Vpeak, the diode stops conducting. From a mathematical point of view, we can calculate the time VACin takes to reach Vmin: V ACin sin( t) Vmin. Since Vpeak is reached at the input sinusoid top (or one fourth of the input period), then the diode conducting time tc is simply: sin 1 1 tc 4 Fline V V min in2 AC 360 Fline = 2 . 1 m s @ Vi n = 100 VAC (eq. 14). During these 2.1 ms, Vbulk is the seat of a rising voltage equal to Vripple or 29 Vpp. This corresponds to a brought charge Q of: Qbulk Vripple Cbulk = 290 µC (eq. 15). From Figure 8a, we can calculate the amount of charge Q drawn from the input by integrating the input current over i Turn Ratio and Output Diode Selection The primary/secondary turn ratio affects several parameters: The drain plateau voltage during the OFF time: the lowest plateau gives room for the leakage inductance spike before reaching the MOSFET’s BVdss: tc the diode conduction time: Qin diode(t).dt Vplateau (eq. 16). The The secondary diode Peak Inverse Voltage (PIV) is linked to the turn ratio and the regulated output voltage by: 0 expression of idiode(t) is: Ipeak tc t (eq. 17). After tc PIV Ns VinDC max Vout (eq. 21). If you lower the Np proper integration, it comes: Qin 1 Ipeak tc . If we 2 plateau voltage, you will increase the reverse voltage the secondary diode must sustain. The amp–turns equation Np . αIp = Ns . Is should satisfy the average output current demand with now equate Qbulk and Qin and solve for Ipeak, it comes: Ipeak Qbulk 2 (eq. 17) or 280 mA peak, as confirmed tc by the simulation. We can now evaluate the RMS c u r r e n t flowing through the diodes: Irms = Fline (idiode(t)) dt Ipeak tc3 2 Fline Iout avg tc 2 Np (Vout Vf) VinDC max (eq. 20). Ns Ip toff Fsw Np 2 Ns (eq. 22). The α parameter illustrates the energy diverted by the leakage inductance at the switch opening (take 0.95 for low leakage designs). With these numbers in mind, you can tweak the turn ratio according to the MOSFET BVdss and the diode. Below are given ON Semiconductor references for Schottky diodes: = 0 (eq. 18) 86 mA @ VAC = 100. A 400 V/1A diode bridge or four 1N4007 can thus be selected for the rectifying function. A small resistor is however put in series to ensure http://onsemi.com 10 AND8023/D ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁ Reference VRRM (V) Io (A) Case MBRM120LT3 20 1.0 PowerMite MBRM130LT3 30 1.0 PowerMite MBRA130LT3 30 1.0 SMA MBRA140LT3 40 1.0 SMA MBRS120LT3 20 1.0 SMB MBRS130LT3 30 1.0 SMB MBRS140LT3 40 1.0 SMB MBRS190T3 90 1.0 SMB MBRS1100T3 100 1.0 SMB MBRS320T3 20 3.0 SMC MBRS330T3 30 3.0 SMC MBRS340T3 40 3.0 SMC MBRS360T3 60 3.0 SMC 1N5817 20 1.0 Axial 1N5818 30 1.0 Axial 1N5819 40 1.0 Axial *Please see brochure BR1487/D for thermal and package details. If we select an MBRA140LT3 (VRRM = 40 V), then the PIV should be selected around 35 V at high line: current to build up with a slope of VinDC . Because the N PIV Vout (eq. 23). If we select Np:Ns = 1: 0.08, VinDC max NCP1200 operates in current–mode, the ON time expires when the current setpoint Ip has been reached: then PIV = 34 V at 350 VDC input voltage which is okay with the selected diode. The plateau voltage at the drain will establish around 430 VDC: it leaves up to 170 V for the leakage spike. The average diode Idavg current is the converter’s DC output current which is 580 mA, in line with our 1 A MBRA140LT3. The repetitive peak current seen by the ton Lp Lp Ip (eq. 28). In DCM, we must satisfy equation VinDC 27. That is to say, the OFF time lasts until the core is fully reset or the secondary current has come back to zero: toff Lp Ip (eq. 29), with N the turn ratio N (Vout Vf) Np (eq. 24). The diode Ns between the secondary and the primary and Vf the secondary rectifier forward drop at a given current (≈1 V). In a FLYBACK SMPS, the input power flow is evaluated R M S c u r r e n t I drms c a n b e e v a l u a t e d u s i n g : using the formula: Pin 1 Lp Ip 2 Fsw (eq. 30) with Id rms Ipeak sec Lp the primary inductance, Ip the primary current at the end of the ON time and Fsw the converter’s switching frequency. Pin is linked to Pout by the efficiency using: diode is: Ipeak sec Ip Fsw3 toff 2 (eq. 25). Finally, the total conduction losses of the diode can be assessed through the following equation: Pdiode avg Vf Id avg Rd Id rms 2 (eq. 26) with Vf the forward drop at Id = Ipeaksec and Rd, the dynamic Pin Pout (eq. 31). Combining equations 27, 28, 29, 30, 31 and solving for Lp we obtain the critical inductance value above which we would go into Continuous Conduction Mode (CCM) at the lowest input voltage resistance dVf @Id = Ipeaksec. dId VinDC Once Ip have been evaluated, you will need to confirm the agreement with the diode maximum rating specs. avg Primary Inductance and Peak Current Vripple and maximum output power: 2 N (Vout Vf) VinDC Lp N (Vout Vf) VinDC When the SMPS operates in the Discontinuous Conduction Mode (DCM), the sum of the ON and OFF 2 Pout Fsw 2 times equal the switching period: ton toff 1 (eq. 32). Finally, the peak current corresponding to this inductance value can be computed from equations 30, 31: (eq. 27). From the FLYBACK equations, we can easily calculate ton and toff. During the ON time, the converter applies VinDC to the primary inductance which forces the Ip Fsw (eq. 33). With our example data in 2Lp Pout Fsw http://onsemi.com 11 AND8023/D Ip2 mind, the calculation gives a primary inductance of 6.4 mH at Fsw = 40 kHz (nominal NCP1200 switching frequency). The corresponding Ip is 190 mA. In equation 30, two unknowns exist: Lp and Ip. The above calculations show how to select Lp from a conduction mode point of view (always stay in DCM), which leads to a final operating peak current. However, from equation 6, we can see that a high peak current leads to a large inductance spike when the switch opens. We should then try to work at a low primary peak current to avoid this potentially lethal voltage kick. However, working at a low peak current means a higher primary inductance to satisfy equation 30 which, in turn, induces a larger leakage energy (the leakage inductance is proportional to the square of the primary turns)…We thus need a balance between the low primary current requirements combined with our low leakage inductance needs. Let’s go over the whole process and see how we can tweak the parameters, keeping in mind that the lowest turn number also ensures the lowest manufacturing cost… Id You have that You expect this Vg Prop. Delay Figure 9a. The NCP1200 propagation delay leads to a slightly higher peak current than expected. The data we have are: Lp = 6.4 mH ±10%, Fsw = 40 kHz ±15%, Rsense = 3.9 Ω ±5% and Vsensemax = 0.9 ±5%. From these values, we can combine worse case together to see a) are we always able to deliver our 3.5 W? b) what is the maximum output power capability?: VinDC min Vsense min Pout min 1 Lp min t prop 2 Rsense max Lp min Step 1: Evaluate the critical inductance value with eq. 32. Ip is calculated via eq. 33 → Lp = 6.4 mH, Ip = 190 mA. Fsw min 3.3 W Step 2: The NCP1200 limits the peak current to: Ip1 (eq. 37) Vsense max VinDC max Pout max 1 Lp max t prop 2 Rsense min Lp max 0.9 V Rsense (eq. 34), 0.9 V varying ±10%. We should then select a sense resistor from the E24 series, also affected by a ±5% tolerance, which authorizes the peak current to grow up to the number calculated by eq. 33. The E24 shunt series looks like the following for medium output powers: …0.56 Ω, 0.68 Ω, 0.82 Ω, 1 Ω, 1.2 Ω, 1.5 Ω, 1.8 Ω, 2.2 Ω, 2.7 Ω, 3.3 Ω, 3.9 Ω, 4.7 Ω… As you can see, it will be easier to pick up an available reference from the E24 series then refine the inductance value to reach our goal. This option avoids to freeze the inductance without having the necessary flexibility on the shunt. From step 1 and equation 34, Rshunt = 4.24 Ω. If we select 4.7 Ω, we will have difficulties to reach our peak current with the previous inductance (remember that we cannot increase it otherwise we go into CCM). Another problem would be the FB permanently pushed to the maximum current demand and triggering the overload mode. Therefore, let’s select 3.9 Ω from the E24 series and later on diminish the inductance value to fit eq. 30. Now, we need to apply worse cases of tolerance variations to see how they affect the final output power result. To refine the calculation, we will include the NCP1200 overcurrent propagation delay which is 120 ns typical. Propagation delay corresponds to the actual time taken by the internal chain to really pull the MOSFET’s gate to ground when the peak current setpoint has been reached. During this time, the primary current keeps Fsw max 6.3 W 2 2 (eq. 38) From the above numbers, we can see that we cannot deliver the target power in the worse case. Let’s reduce the Rsense value to the next reference, 3.3 Ω: (you can use the spreadsheet to speed–up this process) Poutmin = 4.6 W Poutmax = 8.8 W Rsense = 3.3 Ω, Lp = 6.4 mH From these results, we can consider that 4.6 W is more than we need. By reducing the inductance value to 5.5 mH, the numbers are updated to: Poutmin = 4 W Poutmax = 7.6 W Rsense = 3.3 Ω, Lp = 5.5 mH With this case, the 4 W output power offers a guardband above 14% (4 W compared to 3.5 W) and it will thus leave some dynamic on the FB level without triggering the overvoltage protection circuitry. Step 3: This final step should give us the type of transformer we can use for our application. In FLYBACK converters, we need a gap to expand the storage capability of a given un–gapped core. The gap presence imposes a decrease in the remnant flux density Br and allows a larger field excursion before reaching the saturation (a gap does not change the saturation level Bsat nor the coercitive field values Hc). However, the gap is usually made by grinding growing up with a slope of VinDC (eq. 36). Figure 9a Lp details this principle. http://onsemi.com 12 AND8023/D 3. You now need to calculate the specific inductance value the center leg of the selected core (E or RM etc.). This operation can be difficult, especially on small cores used for powers as the one we are looking for. For this reason, most of the manufacturers do not recommend gaps above 0.6 mm. Large gaps also generate fringing flux which can lead to disturbing ElectroMagnetic Interference (EMI) leaks. Two methods can be used for the design of the transformer depending on the information given by the core manufacturer: by applying: AL 4. Evaluate the desired gap length lg (in mm) through the following formula: lg Numerical Application 1. From eq. 34 and Lp value, evaluate the maximum required storage energy capability Lpmax.Ip2max. 2. Depending on your application constraints (dimensions, weight etc.) select a given core geometry (E, RM etc.). 3. From the core manufacturer handbook, look at the graph which gives L.I versus AL curves. Draw a horizontal line which corresponds to the above required L.I2 number for the selected geometry. Any core references appearing below this curve can be used. However, in an attempt to minimize the leakage inductance (less primary turns), try to select the largest AL . 4. Plot a vertical line which passes through the intersection point step 3 created. This gives you the AL. 5. From the AL versus air–gap length curve, extract the corresponding air–gap and check if it is machinable (< 600 µm?). 6. From the AL, calculate the primary turn number with: ALLp For our 3.5 W charger operating a 40 kHz, we have selected an E16 core in B2 material (Thomson–LCC) offering the following parameters: Ae = 0.0000198m Bsat = 300mT @ 100°C µa > 1000 (T° > 100°C, B = 330mT) lm = 0.0348 meter To lower the turn numbers (for leakage but also for winding time reasons), we have purposely decreased the previously calculated Lp value down to 2.7 mH which gives an operating peak current of 290 mA and a sense resistor of 2.7 Ω. Applying method 1 leads to: AL = 60 nH/turn2 Np = 212 turns Ns = 17 turns (eq. 39) The manufacturer will need the primary and secondary RMS currents calculated at the lowest line level: 7. Finally, check that Bs . N . Ae > Lpmax . Ipmax with Bs the core saturation limit at 100°C and Ae the effective core area. If it fails, reduce the AL value by keeping the air–gap within reasonable limits and check that the updated turn numbers are still in agreement with the bobbin capability (the transformer manufacturer will normally tell you about it). 8. With the help of the spreadsheet, you have the choice to tweak Lp in order to decrease the turn number to finally reduce the leakage inductance. Isec rms Ipeak sec Fsw3 toff or 2.7 A (eq. 42) The current primary Iprim rms Ipeak prim is obtained with: Fsw3 ton or 184 mA (eq. 43) MOSFET Selection The transformer being designed, we can look at the MOSFET type. From equation 20, the plateau voltage (during the toff duration) rises up to 438 V. By selecting a 600 V device, we ensure enough headroom for the leakage spike. If this spike is finally too energetic, we can either clamp it with an adequate network or choose a 800 V MOSFET. From equation 43, a 1 A MOSFET is suitable and can be selected from the below devices: Second Method 1. Depending on your application constraints (dimensions, weight etc.) select a given core geometry (E, RM etc.) 2. From the calculated Lp value, evaluate the minimum turn number with: N o a N 2 Ae Lp lm 1000 (eq. 41) with Lp a µo the air permeability (4.π.10–7), µa the amplitude permeability (the core permeability at high flux excursions), N the turn number, lm the mean magnetic path length (m). First Method N Lp N2 Lp max Ip max (eq. 40) Bs Ae http://onsemi.com 13 AND8023/D ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ Reference BVdss (V) RDS(ON) () Peak Current (A) Package NTB10N60 600 0.75 – TO–220 NTB10N60 600 0.75 – D2PAK NTD4N60 600 2.3 – DPAK NTP6N60 600 1.1 – TO–220 NTP6N60 600 1.1 – D2PAK MTP4N80E 800 3.0 4.0 TO–220 MTD1N80E 800 12 1.0 DPAK MTD1N60E 600 8.0 1.0 DPAK MTB3N60E 600 2.2 3.0 D2PAK MTP2N60E 600 3.8 2.0 TO–220 MTP3N60E 600 2.2 3.0 TO–220 MTP6N60E 600 1.2 6.0 TO–220 Let’s pick–up an MTD1N60E which features the following specs: needed to push enough Coulomb in the gate to properly bias the MOSFET. However, the time during which the drain falls corresponds to the charging time of the Miller capacitance or the VGS “plateau” duration. This plateau, which occurs at 6 V for the MTD1N60E is not the VGSth parameter (≈3 V). From the MOSFET’s gate–charge chart, we can extract the amount of necessary Coulomb Qg2 to cross–over the plateau area: 3.2 nC. The time needed to push or extract this charge with a given gate resistor is: BVdss = 600 V RDS(ON) = 13 Ω @ Tj = 100°C Case = DPAK RθJA = 71 °C/W on min pad area Max junction temperature = 150°C Qg = 11 nC max. (VGS = 10 V) Coss = 40 pF tson If we operate at an ambient of 70°C, the maximum power the component can dissipate on free–air is given by: Pmax (eq. 49), with VCC the lowest supply level delivered by the DSS (9.2 V). To minimize the EMI problems, the NCP1200 output impedance is asymmetrical: Rsource = 40 Ω (turn–on) while Rsink = 12 Ω (turn–off), as already accounted for in eq. 47 and 48. With these values, we obtain tson = 45 ns and tsoff = 6.4 ns. These numbers are rather low because of the small parasitic elements constituting the MTD1N60E. At the switch opening, the MOSFET current immediately drops thanks to a small tsoff. But the leakage inductance forces Ip to circulate through all the stray capacitance (Clump) and pulls up the drain with a slope of Tj max T A = 1.1 W (eq. 45). The total MOSFET R JA losses are a combination of the conduction losses and the switching losses. From equation 43, we can compute the MOSFET conduction losses: Pcond RDS (ON) Id 2RMS = 440 mW (eq. 46). Switching losses appear because of the overlap between drain current and drain–source voltage during transitions. We have losses during the switch closing (Pswon) but also during the switch opening (Pswoff). In DCM, the turn–on losses are mainly created by the recurrent discharge of the stray parasitic capacitors in the MOSFET. The MOSFET contributes to these losses via its Coss capacitor. Unfortunately, Coss does not linearly vary with Vds during the transition and an accurate calculation would require the use of an equivalent capacitor, resulting from the Coss integral over the switching period. To simplify the evaluation, we will stick to the classical Coss given in the data–sheet, 40 pF max and neglect the other stray capacitance (e.g. from the transformer or a snubber): P swon Qg2 Rsource Qg2 Rsink (eq. 48) and tsoff Vcc Vplateau Vplateau Ip (eq. 50). Because the MOSFET is fully open when Clump Vds rises, there are very few associated losses. In the case Vds (rising) and Ip (decreasing) would intersect together in their middle, the final Pswoff losses could be computed by: P swoff Vds max Ip max ts off Fsw max (eq. 51) with Vds 6 the highest drain–source level (600 V in worse case and without a clamping network). After computation, we obtain 14 mW a slightly optimistic number. The total power dissipation is thus: 440 mW + 175 mW + 14 mW = 630 mW, neglecting the gate losses. This number is below eq. 45 result and offers a comfortable theoretical security margin. We will confront these results with practical measurements on the board to ensure reliable operation. Coss (Vds plateau) 2 Fsw max 2 (eq. 47) with Vdsplateau given by eq. 20. We obtain 175 mW. To estimate this ON transition duration, we can calculate the time ts http://onsemi.com 14 AND8023/D Output Capacitor Selection is fully reset (toff caption). First, let’s calculate this toff duration knowing that the secondary downslope is: To evaluate the output capacitor value Cout, we need to define a given, acceptable level of ripple. The ripple finds its root in two different sources: 1. The natural integration of the secondary current (amputed by the DC current delivered to the load) through the capacitance gives birth to a capacitive voltage of Vcap 1 Cout i (t) S sec peak current Isec equal to N . Ip = 0.29 . 12.5 = 3.6 A, toff will last 8.9 µs for a 3.5 W nominal output power (IoutDC = 500 mA). With this value in mind, we can calculate the integrated voltage over Cout when subject to a ramping down current starting from (Isec – IoutDC) and lasting 8.9 dt. This integral is Vcap 1 Cout µs: 0 (Isec Iout DC) dt toff (Isec Iout DC). Thanks to 2C this formula, we can extract the Cout value by: Cout toff (Isec Iout DC) or 490 µF with Vcap = 30 mV 2 Vcap ripple. The simulation show higher ripple numbers because of an output power of 4.5 W. Unfortunately, the ESR contribution is by far the higher contributor to the output ripple. This ohmic loss will generate a thin voltage spike equal to: R ESR . (Isec–IoutDC). In our case, a 100 mΩ ohmic loss produces up to 300 mV and adds up with the capacitor voltage ripple. The ESR also affects the capacitor dissipation by: PCout = I2capRMS . RESR. You can smooth the ESR spike by inserting a secondary LC filter, as proposed by the various sketches of this application note. Thanks to simulation, Figure 9b is able to show these components separately and how they combine together: Total Secondary ripple, 200 Mv/div Iout, DC tofftoff t toff valid during the secondary current decay until DCM is reached (Isec = 0). The remaining portion of time includes the dead–time (if any) plus the switch ON time. 2. The total parasitic contribution of Equivalent Series Resistor (ESR) and Inductor (ESL) that produce a voltage spike at every switch openings. The ESR contribution only is usually culprit for the majority of the ripple amplitude. Q brought to Cout (Vout Vf) N 2 or 405 mA/µs. With a secondary Lp Icap, 2 A/div On Duration Simulating the NCP1200 To ease your design phase, we have developed a transient and an AC averaged SPICE model for the NCP1200. Ready–to–use templates are available to download at www.onsemi.com in these three versions: INTUSOFT’s IsSpice4, OrCAD’s PSpice and Spectrum–Software’s Cap. Describing these models is beyond the scope of this application note. However, we will show some typical waveforms you could quickly get with the help of these tools. Figure 10a portrays a typical IsSpice transient simulation using the NCP1200 in our 3.5 W charger application. Vcap, 20 mV/div VESR, 200 mV/div Figure 9b. The final ripple is made of the ESR and the output capacitance. By looking at Figure 9b’s INTUSOFT’s IsSpice plot, we can evaluate the output capacitor voltage rise until the core http://onsemi.com 15 AND8023/D Isec X1 XFMR RATIO = 0.08 R3 200 m 12 Iout R16 10 m 31 1 R4 100 m Iripple L2 2.7 mH + L1 2.2 µH 4 3 Iprim Vsec D1 MBR140P 17 R17 300 m 6 32 C1 470 µF IC = 5.9 IHV + Vout Vout Rload 8 C2 10 µF 16 L5 80 µH VInput 126 Vadj 15 1 2 3 4 NCP1200 X2 NCP1200 Fs = 40 kHz + + IDrain 8 Drv 7 6 5 R15 470 VDrain 5 VCC 19 10 X7 MTD1N60E 13 X4 MOC8101 C3 470 pF 18 R5 100 m D3 1N752 Vsense 23 CVCC 10 µF IC= 12.1 Rsense 2.8 VFB Figure 10a. A typical transient analysis of the NCP1200 with a dedicated SPICE model. C3 is purposely wired between drain–source to highlight the potential substrate injection problems. The model includes the start–up source, various propagation delays, the short–circuit protection and the driver dual impedance concept. Typical waveforms are shown on Figure 10b and 10c. Thanks to short simulation times, the NCP1200 SPICE model will help you confirming the theoretical results you have already calculated. http://onsemi.com 16 AND8023/D Vsense 200 mV/div Vdrain 100 V/div Vdrain 100 V/div Gate current contribution Vgate 5 V/div Vsense 500 mV/div Possible substrate injection Figure 10b. Steady–State Drain–Source and Sense Voltages Figure 10c. Zoomed portion of MOSFET switch–off time when the snubber capacitor is wired between drain and source. The Final Application Schematic By feeding the various components with the choices you have made (e.g. transformer turn ratio), you can immediately verify that you do not exceed the safety limits. Such verifications could end–up into smoke on a real prototype test … L1 330 µH + C1 4.7 µF SMD 400 V Universal Input Figure 11a depicts the complete application schematic used in our NCP1200 AC/DC demoboard. Thanks to the weak leakage inductance exhibited by the transformer (≈80 µH), a simple capacitor in parallel with the MOSFET allows a safe operation up to 250 VAC, but to a slight detriment of switching losses (eq. 47). I2 4.7 µH D2 MBRA140LT3 + C2 4.7 µF 400 V 1 Adj HV 8 2 FB 7 3 CS1 VCC 6 + + C6 470 µF/10 V M1 MTD1N60E C4 470 pF 1 kV Thru Holes D3 MMSZ5V1 10 R1 1W C3 10 µF 16 V + 1WSMD R4 2.7 IC1 SFH6156–2 R6 470 C5 2.2 nF Y Type Figure 11a. The 3.5 W AC/DC adapter available as a demoboard http://onsemi.com 17 C7 10 µF/10 V Ground T1 4 Gnd Drv 5 6 V @ 600 mA AND8023/D The PCB exhibits compact dimensions (64mm x 35mm) and includes a simplified EMI filter (L1). In case the common mode noise induced by M1 switching too fast exceeds the standard limits, you still have the option to Normal Mode slow it down a little bit by inserting a resistor between pin 5 and its gate. Below are some typical oscilloscope shots taken from the board: Vdrain 100 V/div VCCH VCCL Short Circuit Latch–Off Phase Start and Stop Vsense 500 V/div 0V Figure 11b. The NCP1200 DSS behavior in normal and short circuit conditions. Figure 11c. Typical 3 W operation that can be compared to Figure 10b. Vsense 200 V/div Vdrain 100 V/div Vdrain 100 V/div Possible Substrate Injection Vsense 200 V/div Gate 5 V/div Gate 5 V/div Figure 11c. This shot confirms our turn–on 45 ns calculation and shows the absence of immediate drain current. The capacitive peak is blanked with the NCP1200 LEB. Figure 11d. As predicted by the calculations, the turn–off time is extremely fast and does not engender losses. The switch is immediately closed and the current is capacitive. This plot can be compared to Figure 10c. Figure 11d represents the turn–off stage when the snubber capacitor is wired between drain and source. When the leakage starts to ring, the induced dV/dt forces a capacitive current to circulate inside the MOSFET Coss and the snubber. This current unfortunately creates a negative ringing over Rsense and can potentially inject into the NCP1200 substrate. This situation is highly undesirable as the injected electrons can go anywhere, possibly engendering an erratic behavior of the IC. To prevent this situation, simply wire the snubber between the drain and ground, as indicated by Figure 6b and confirmed by Figure 11e. In application where you do not install any snubber, the problem should not appear. http://onsemi.com 18 AND8023/D Vsense 500 V/div Vdrain 100 V/div, X = 500 µs/div Vdrain 100 V/div Vdrain 100 V/div, X = 5 µs/div Zoomed Current 50 ns/div Figure 11e. Wiring the snubber between drain and source avoids the substrate injection. Figure 11g. Standby operation, upper curve is Vdrain with X = 500 s/div. However, in some higher power applications, the case can arise where the spike is present without the snubber in place. In that later case, we advise to wire a small low–pass RC filter between the sense resistor and pin 3. Typical values are R = 1 kΩ C = 220 pF. Another options lies in connecting a Schottky diode between pin 3 and ground (cathode to pin 3). Depending on the layout, the FB pin can pick–up some noise which leads to spurious oscillations. To cure this problem, wire a 1 nF capacitor between pins 2 and 4. VCC, 5 V/div Vout 100 V/div Vgate, 2 V/div Figure 11h. Typical Operating Curves in Short–Circuit Conditions Figure 11f. Demoboard transient response when banged from 10 mA to 800 mA. Typical measurements on the 3.5 W AC/DC adapter: ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁÁÁ ÁÁÁÁÁÁÁ Output Power DC Input Level Input Power 120 134 mW 0 – 120 732 mW 480 mW 65% 120 1.44 W 1.0 W 69.5% 120 4.08 W 3.3 W 80.6% 325 339 mW 0 – 325 1.06 W 480 mW 45% 325 1.84 W 1.0 W 54% 325 4.3 W 3.11 W 72% http://onsemi.com 19 AND8023/D At low line, the DSS contribution is low at about 100 mW. It represents nearly all the power consumption and does not significantly affect the efficiency. At higher line levels, because of the DSS constant current nature, its contribution increases and degrades the efficiency at weak output power levels. To greatly improve these numbers, you can apply Figure 2 trick or work with an auxiliary winding (Figure 3): it will save more than 200 mW. With the prototype under test, we measure an overload activation at a power level of 7 W. your SMPS into a precise Constant Voltage–Constant Current (CV–CC) generator. For less precise requirements, Figure 12a depicts a configuration where a second loop has been added. This loop operates in parallel with the standard zener one and deviates the LED bias current to regulate the current (Q1 active). Of course, the current reference being the transistor Vbe, it is likely to change over temperature (–2.2 mV/°C)… However, this design can be selected where ± 10% current precision is enough. More precise designs can be made through a TL431 for better output precisions but also by combining the bipolar with a CTP to compensate the current loop temperature drift. Figure 12a shows a typical configuration plugged into our NCP1200 averaged SPICE model. Thanks to this model, we can test for the open–loop stability by drawing a Bode plot of both loops (I or V) and transient test the validity of the CV–CC behavior: Battery Charger Configuration Certain applications require the control of the output current, e.g. battery chargers. To precisely monitor the output current flow, dedicated circuits already exist from ON Semiconductor such as the MC33341. This IC has been tailored to directly drive an SMPS optocoupler and turn Iout NCP1200 IN Averaged OUT 120 CTRL 1.37 11 + VIn 120 V 0 FB GND LoL 1 kH R4 100 m R17 300 m 6.81 12 CoL 1 kF 7 6.80 C1 470 µF 13 + Vstim 0.547 out 1 Vout 6 C2 10 µF AC = 1 5 out 2 FB 1.37 R8 470 1 X4 MOC8101 R9 68 6.14 6.12 15 10 R3 47 C4 100 nF 6.80 ∆ X1 NCP1200_AV Fs = 40 k L = 2.7 mh RI = 2.7 X1 XFMR D1 R1 RATIO = 0.08 MBR140P out 1 L1 out 2 10 m 2.2 µH + 89.6 1 2 6.81 6.80 7.17 6.81 D3 1N752 6.14 17 Q1 2N2222 0.222 19 R5 100 R6 68 Rshunt 2.2 Figure 12a. An Averaged SPICE Simulation in a Battery Charger Application http://onsemi.com 20 Rload 25 AND8023/D By reflecting the bias points to the schematic, the simulator (IsSpice4 in this example), indicates which loop is active: the voltage loop in this case (Q1’s Vbe is 203 mV and is off). By sweeping Vstim, Figure 12b and 12c reveal the Bode plot of the whole configuration where both loops have been represented. VFB (dB) VFB (dB) No comp. With comp. Pm = –57° Pm = 42° Pm = 90° VFB (°) VFB (°) 0 0 10 dB/div 45°/div 10 dB/div 45°/div Figure 12b. The voltage sweep does not reveal any problem with a 90° PM... Figure 12c. However, the current loop is unstable with a large bandwidth. As the below curves show, there is no problem when the voltage loop operates. The high gain finds one of its root in the NCP1200 internal collector resistance which equals 6.4 kΩ. But as you can see, it offers a good theoretical bandwidth associated with a safe phase margin. If we now activate the current loop, Figure 12c details a strong instability because of the bipolar presence. To cure this problem in the simplest manner, we can wire a 100 nF capacitor between FB and ground to roll–off the gain at lower frequencies. Updated Bode plots confirm the stability gained by this capacitor addition. This instability was observed on the evaluation board and also eradicated by the 100 nF capacitor. By decreasing LoL and CoL elements down to 1p, we actually close the feedback loop. By sweeping the output load through a variable resistor, we plot the CC–CV curve as shown by Figure 12d. Vout in Y 1 V/div CV Operation CC Operation Iout in X 100 mA /div Figure 12d. The simulated CC–CV output shape as delivered by Figure 12a. Please note that these average simulation circuits work with the demo versions of INTUSOFT, OrCAD and Spectrum–Software. The complete battery charger schematic appears on Figure 13 and is available as another NCP1200 demoboard. http://onsemi.com 21 AND8023/D SMB D2 MBRA140LT3 L1 330 µH + + C1 C2 4.7 µF 4.7 µF 400 V 400 V 6 V @ 600 mA + 1 Adj HV 8 2 FB 7 3 CS1 VCC 6 Universal Input I2 4.7 µH M1 MTD1N60E C6 470 µF 10 V T1 + C7 10 µF/10 V Ground C4 470 pF/1 kV Thru Holes 4 Gnd Drv 5 R10 470 R5 68 10 R1 1W C8 100 nF C3 10 µF 16 V + SMD R4 2.7 IC1 SFH6156–2 D3 MMSZ5V1 R6 47 Q1 2N2222 R7 68 R8 100 R9 2.2 SMD C5 2.2 nF Y Type Figure 12e. The Complete Battery Charger Demonstration Board Different scheme will allow various bandwidth and response times. In that case, the averaged NCP1200 SPICE model appears as an invaluable tool to help you to place poles/zeroes at the correct location. 1V / div D1 MBR140P C2 10 µF Vout L1 330 µH 1 C1 470 µF GND 300 mA / div R15 270 16 Figure 12f. The Resulting I–V Curve of the NCP1200 Charger Demoboard X4 MOC8101 Figure 12f depicts the resulting CV–CC obtained when sweeping the charger’s output through a MOSFET and displaying the values with an X–Y oscilloscope. Vout R2 14 k 15 C5 100 nF X3 TL431 14 Precise Secondary Regulation Any TL431–based regulation scheme can be employed with NCP1200. Figure 13 gives in example the feedback section only when using a TL431 reference voltage. The TL431 is usually compensated by wiring a capacitor between the resistor bridge middle–point and the cathode. R7 10 k GND Figure 13. A Precise Regulation Option using a TL431 Reference Voltage http://onsemi.com 22 AND8023/D Primary Regulation Option are depicted in Figure 13a and 13b sketches. To avoid classical peak rectification, a light load e.g. 470 Ω can be connected to load the auxiliary supply. Auxiliary and main power winding ratio should be adjusted to match the desired output voltage. Despite the lack of internal error amplifier, two kind of primary regulation options can be implemented: one in which the short–circuit works as usual and another one where this option is permanently invalidated. Both options HV Rail D6 1N5819 D5 1N4148 + C3 470 µF 10 V Aux C2 10 µF 10 V 6 V @ 600 mA NCP1200 D4 5 V1 1 2 3 4 HV Adj FB CS1 VCC Gnd Drv 8 7 6 5 M2 MTD1N60E R1 470 + Rsense1 C1 10 µF Q1 2N3904 Figure 13a. A primary regulation possibility where the short–circuit protection still operates. HV Rail D6 1N5819 R2 220 D5 1N4148 Aux Vaux + C3 470 µF 10 V C2 10 µF 10 V D4 6 V2 6 V @ 600 mA NCP1200 1 2 3 4 HV Adj FB CS1 VCC Gnd Drv 8 7 6 5 VCC M2 MTD1N60E R1 1k + R1 20 k C1 10 µF Rsense1 Figure 13b. Another primary regulation possibility where the short–circuit protection no longer operates. Transformer Availability E16 version corresponds to a 3.5 W charger operating with an NCP1200P40 (40 kHz) while the 1.8 mH version should be used for a 5 W NCP1200P60 version (60 kHz). T1 transformer for the depicted circuits (charger and AC adaptor) is immediately available from several manufacturers whose details are given below. The 2.7 mH http://onsemi.com 23 AND8023/D Email: [email protected] Eldor Corporation Headquarter Via Plinio 10, 22030 Orsenigo (Como) Italia Tel.: +39–031–636 111 Fax : +39–031–636 280 Email: [email protected] www.eldor.it ref. 1: NCP1200–10 W–UM: 10 W for USB (Lp = 1.8 mH, 60 kHz, 1:0.1, RM8 pot core) Coilcraft 1102 Silver Lake Road Cary, Illinois 60013 USA Tel: (847) 639–6400 Fax: (847) 639–1469 Email: [email protected] http://www.coilcraft.com ref. 1: 2262.0058C: 3.5 W version (Lp = 2.9 mH, Lleak = 80 µH, E16) ref. 2: 2262.0059A: 5 W version (Lp = 1.6 mH, Lleak = 45 µH, E16) ref. 1: Y8844–A: 3.5 W version (Lp = 2.9 mH, Lleak = 65 µH, E16) ref. 2: Y8848–A: 10 W version (Lp = 1.8 mH, Lleak = 45 µH, 1:01, E core) EGSTON GesmbH Grafenbergerstraβe 37 3730 Eggenburg Austria Tel.: +43 (2984) 2226–0 Fax : +43 (2984) 2226–61 Email: [email protected] http://www.egston.com/english/index.htm SPICE Editors As mentioned in the text, NCP1200 ready–to–use SPICE platforms are available to download from the ON Semiconductor Web site in the following formats: O r C A D ’s P S p i c e , I N T U S O F T ’s I s S p i c e a n d Spectrum–Software’s µCap. You will find both transient and AC analysis templates as described by Figures 10a and 12a. Some of these examples can work on the editor’s demo versions, some will require the full version to operate. Editors demo versions can be downloaded at the following locations: www.orcad.com (PSpice), www.intusoft.com (IsSpice) and www.spectrum–soft.com (µCap). ref. 1: F0095001: 3.5 W version (Lp = 2.7 mH, Lleak = 30 µΗ, sandwich configuration, E16) Atelier Special de Bobinage 125 cours Jean Jaures 38130 ECHIROLLES FRANCE Tel.: 33 (0)4 76 23 02 24 Fax: 33 (0)4 76 22 64 89 ON Semiconductor and are trademarks of Semiconductor Components Industries, LLC (SCILLC). SCILLC reserves the right to make changes without further notice to any products herein. SCILLC makes no warranty, representation or guarantee regarding the suitability of its products for any particular purpose, nor does SCILLC assume any liability arising out of the application or use of any product or circuit, and specifically disclaims any and all liability, including without limitation special, consequential or incidental damages. “Typical” parameters which may be provided in SCILLC data sheets and/or specifications can and do vary in different applications and actual performance may vary over time. All operating parameters, including “Typicals” must be validated for each customer application by customer’s technical experts. SCILLC does not convey any license under its patent rights nor the rights of others. SCILLC products are not designed, intended, or authorized for use as components in systems intended for surgical implant into the body, or other applications intended to support or sustain life, or for any other application in which the failure of the SCILLC product could create a situation where personal injury or death may occur. Should Buyer purchase or use SCILLC products for any such unintended or unauthorized application, Buyer shall indemnify and hold SCILLC and its officers, employees, subsidiaries, affiliates, and distributors harmless against all claims, costs, damages, and expenses, and reasonable attorney fees arising out of, directly or indirectly, any claim of personal injury or death associated with such unintended or unauthorized use, even if such claim alleges that SCILLC was negligent regarding the design or manufacture of the part. SCILLC is an Equal Opportunity/Affirmative Action Employer. PUBLICATION ORDERING INFORMATION NORTH AMERICA Literature Fulfillment: Literature Distribution Center for ON Semiconductor P.O. Box 5163, Denver, Colorado 80217 USA Phone: 303–675–2175 or 800–344–3860 Toll Free USA/Canada Fax: 303–675–2176 or 800–344–3867 Toll Free USA/Canada Email: [email protected] Fax Response Line: 303–675–2167 or 800–344–3810 Toll Free USA/Canada N. American Technical Support: 800–282–9855 Toll Free USA/Canada EUROPE: LDC for ON Semiconductor – European Support German Phone: (+1) 303–308–7140 (Mon–Fri 2:30pm to 7:00pm CET) Email: ONlit–[email protected] French Phone: (+1) 303–308–7141 (Mon–Fri 2:00pm to 7:00pm CET) Email: ONlit–[email protected] English Phone: (+1) 303–308–7142 (Mon–Fri 12:00pm to 5:00pm GMT) Email: [email protected] CENTRAL/SOUTH AMERICA: Spanish Phone: 303–308–7143 (Mon–Fri 8:00am to 5:00pm MST) Email: ONlit–[email protected] Toll–Free from Mexico: Dial 01–800–288–2872 for Access – then Dial 866–297–9322 ASIA/PACIFIC: LDC for ON Semiconductor – Asia Support Phone: 1–303–675–2121 (Tue–Fri 9:00am to 1:00pm, Hong Kong Time) Toll Free from Hong Kong & Singapore: 001–800–4422–3781 Email: ONlit–[email protected] JAPAN: ON Semiconductor, Japan Customer Focus Center 4–32–1 Nishi–Gotanda, Shinagawa–ku, Tokyo, Japan 141–0031 Phone: 81–3–5740–2700 Email: [email protected] ON Semiconductor Website: http://onsemi.com EUROPEAN TOLL–FREE ACCESS*: 00–800–4422–3781 *Available from Germany, France, Italy, UK, Ireland For additional information, please contact your local Sales Representative. http://onsemi.com 24 AND8023/D