CS5127 CS5127 Dual Output Nonsynchronous Buck Controller with Sync Function and Second Channel Enable Features Description circuits. The first lockout releases when VIN reaches 8.4V, while the second lockout ensures that VREF is higher than 3.6V. The outputs are held in a low state until both lockouts have released. The controller is configured to utilize the V2ª control method to achieve the fastest possible transient response and best overall regulation. This dual controller is a cost-effective solution for providing VCORE and VIO power solutions in computing applications using a single controller. The CS5127 will operate over an input voltage range of 9.4V to 20V and is available in a 16 lead wide body surface mount package. The CS5127 is a fixed frequency dual output nonsynchronous buck controller. It contains circuitry for regulating two separate outputs. Each output channel contains a high gain error amplifier, a comparator and latch, and a totem-pole output driver capable of providing DC current of 100mA and peak current in excess of 0.5A. A common oscillator controls switching for both channels, and a sync lead is provided to allow parallel supply operation or shifting of the switching noise spectrum. An on-chip 5V reference is capable of providing as much as 10mA of current for external circuitry. The CS5127 also contains two undervoltage lockout Applications Diagram ■ Nonsynchronous Buck Design ■ V2ª Control Topology ■ 100ns Transient Loop Response ■ Programmable Oscillator Frequency ■ 30ns Typical Gate Rise and 10ns Fall Times (No Load) ■ Frequency Synchronization Input ■ ENABLE Input Controls Channel 2 Gate Driver ■ 5V/10mA Reference Output Package Option 12V, 5V to 2.8V @ 7A and 3.3V @ 7A for 233MHz Pentium¨ Processor with MMXª Technology + C3 + 1mF C1, C2 2 x 680mF 16 Lead SOIC Wide +5V 12V 5V C4, C5 2 x 680mF SYNC + Q1 FMMT2222ACT CT C6 0.1mF C7 330pF VFB1 R2 27k COMP1 Q2 IRL3103S L1 + R4 1540 CS5127 RT 2.8V 5mH C10, C11 2 x 680mF D1 1N5821 + C8 1mF VFB2 GATE1 GATE2 20k VFB2 COMP1 Q3 IRL3103S L2 3.3V 5mH C12, C13 2 x 680mF PGnd D2 1N5821 R6 1k C14 330pF ENABLE VFB1 COMP2 VFFB2 LGnd C9 0.1mF ENABLE VFFB1 VREF RT VIN VREF R5 1270 R10 VIN CT SYNC R1 20k 1 + R9 2k R7 2400 R3 18k R8 1500 COMP2 VFFB1 VFFB2 GATE1 GATE2 LGND PGND R11 C15 100mF C16 100mF C17 330pF 20k V2 is a trademark of Switch Power, Inc. Pentium is a registered trademark and MMX is a trademark of Intel Corporation Cherry Semiconductor Corporation 2000 South County Trail, East Greenwich, RI 02818 Tel: (401)885-3600 Fax: (401)885-5786 Email: [email protected] Web Site: www.cherry-semi.com Rev. 11/3/98 1 A ¨ Company CS5127 Absolute Maximum Ratings Operating Junction Temperature, TJ ..................................................................................................................................... 150¡C Storage Temperature Range, TS ...................................................................................................................................-65 to 150¡C ESD (Human Body Model).........................................................................................................................................................2kV Lead Temperature Soldering: Reflow (SMD styles only).............................................60 sec. max above 183¡C, 230¡C peak Lead Symbol VMAX Lead Name VMIN ISOURCE ISINK SYNC Oscillator Synchronization Input 5.5V -0.3V 5 mA 5 mA CT Oscillator Integrating Capacitor 5.5V -0.3V 1mA 1mA RT Oscillator Charge Current Resistor 5.5V -0.3V 1mA 1mA VFB1, VFB2 Voltage Feedback Inputs 5.5V -0.3V N/A N/A COMP1, COMP2 Error Amplifier Outputs 7.5V -0.3V 2mA 50mA VFFB1, VFFB2 PWM Ramp Inputs 5.5V -0.3V 1mA 1mA GATE1, GATE2 FET Gate Drive Outputs 20V -0.3V DC, -2.0V for t < 50ns 200mA DC, 1A peak (t < 100µs) 200mA DC, 1A peak (t < 100µs) LGnd Reference Ground and IC Substrate 0V 0V 25 mA N/A PGnd Power Ground 0V 0V 1A Peak, 200mA DC N/A ENABLE Channel 2 Enable 5.5V -0.3V 1mA N/A VREF Reference Voltage Output 5.5V -0.3V 150mA (short circuit) 5mA VIN Power Supply Input 20V -0.3V N/A 200mA DC, 1A peak (t < 100µs) Electrical Characteristics: 0¡C < TA < 70¡C; 0¡C < TJ < 125¡C; 9.4V < VIN < 20V; CT = 330 pF; RT = 27k½; unless otherwise stated. PARAMETER ■ Reference Section VREF Output Voltage Line Regulation Load Regulation VREF Variation over Line, Load and Temperature Output Short Circuit Current ■ Oscillator Section Oscillator Frequency Variation over Line and Temperature Maximum Duty Cycle Sync Threshold Sync Bias Current TEST CONDITIONS MIN TYP MAX UNIT Room Temperature, IVREF = 1mA, VIN = 12V 4.9 5.0 5.1 V 1 15 20 26 5.15 mV mV V 30 100 150 mA 175 210 245 kHz 80 0.8 90 1.6 170 430 230 98 2.4 250 750 % V µA 1 mA < IVREF < 10 mA 4.85 VSYNC = 2.4V VSYNC = 5.0V Sync Propagation Delay 2 ns CS5127 Electrical Characteristics: 0¡C < TA < 70¡C; 0¡C < TJ < 125¡C; 9.4V < VIN < 20V; CT = 330 pF; RT = 27k½; unless otherwise stated. PARAMETER ■ Error Amplifiers VFB Reference Voltage Input Bias Current Open Loop Gain Unity Gain Bandwidth PSRR COMP Source Current COMP Sink Current COMP Output Low Voltage ■ PWM Comparators VFFB Bias Current Propagation Delay Common Mode Maximum Input Voltage ■ ENABLE Lead ENABLE High Threshold ENABLE Bias Current ■ Gate Driver Outputs Output Low Saturation Voltage Output High Saturation Voltage Output Voltage under Lockout Output Rise Time Output Fall Time TEST CONDITIONS VCOMP = VVFB VFB = 1.275V f = 120Hz VCOMP = 3V, VVFB = 1.1V VCOMP = 1.2V, VVFB = 1.45V VVFB = 1.45V, ICOMP = 0.3 mA TYP MAX UNIT 1.245 1.300 1.0 0.9 10 0.50 1.275 0.1 85 1.0 80 1.3 16 0.85 V µA dB MHz dB mA mA V 20 250 2.9 2.0 100 3.3 VFFB = 0 VFFB rising to VGATE falling channel 2 enabled VENABLE = 0 3 µA ns V 1.5 2.5 3.5 V 250 400 µA 0.1 0.25 1.5 1.6 0.1 30 10 0.4 2.50 2.0 3.0 0.2 V V V V V ns ns 8.4 7.8 9.4 8.8 V V 0.4 17.5 0.8 25 mA mA 7.4 6.8 VIN = 6V VCT = 0V, no load 2.0 24 1.00 100 IGATE = 20 mA IGATE = 100 mA IGATE = 20 mA IGATE = 100 mA VIN = 6V, IGATE = 1 mA no load no load ■ Undervoltage Lockout Turn On Threshold Turn Off Threshold ■ Supply Current Start Up Current Operating Current MIN CS5127 Package Lead Description PACKAGE LEAD # LEAD SYMBOL FUNCTION 16 Lead SO Wide 1 SYNC A pulse train on this lead will synchronize the oscillator. Sync threshold level is 2.4V. Synchronization frequency should be at least 10% higher than the regular operating frequency. The sync feature is level sensitive. 2 CT The oscillator integrating capacitor is connected to this lead. 3 RT The oscillator charge current setting resistor is connected to this lead. 4 VFB1 5 COMP1 6 VFFB1 This lead connects to the non-inverting input of the channel 1 PWM comparator. 7 GATE1 This lead is the gate driver for the channel 1 FET. It is capable of providing nearly 1A of peak current. 8 LGND This lead provides a ÒquietÓ ground for low power circuitry in the IC. This lead should be shorted to the PGND lead as close as possible to the IC for best operating results. 9 PGND This lead is the power ground. It provides the return path for the FET gate discharge. It should be shorted to the LGND lead as close as possible to the IC for best operating results. 10 GATE2 This lead is the gate driver for the channel 2 FET. See GATE1 lead description for more details. 11 VFFB2 This lead connects to the non-inverting input of the channel 2 PWM comparator. 12 COMP2 13 VFB2 14 ENABLE The regulator controlled by channel 2 may be turned on and off selectively by the user. Pulling the ENABLE lead above 3.5V will turn channel 2 on. Setting the ENABLE lead voltage below 1.5V guarantees that channel 2 is off. 15 VREF This lead is the output of a ± 3% reference. This reference drives most of the on-chip circuitry, but will provide a minimum of 10 mA to external circuitry if needed. The reference is inherently stable and does not require a compensation capacitor, but use of a decoupling capacitor will reduce noise in the IC. 16 VIN This lead is the power supply input to the IC. The maximum input voltage that can be withstood without damage to the IC is 20V. The inverting input of the channel 1 error amplifier is brought out to this lead. The lead is connected to a resistor divider which provides a measure of the output voltage. The input is compared to a 1.275V reference, and channel 1 error amp output is used as the V2ª PWM control voltage. Channel 1 error amp output and PWM comparator input. Channel 2 error amp output and PWM comparator input. Inverting input for the channel 2 error amp. See VFBI for more details. 4 CS5127 Block Diagram COMP1 VFFB1 + PWM Comparator - VFB1 Error Amplifier - Channel 2 Gate Driver + 1.275V GATE1 VREF VIN Undervoltage Lockout VIN Bandgap Voltage Reference Reference Undervoltage Lockout LGND PGND SYNC RT Oscillator CT 1.275V VFB2 Channel 2 Gate Driver + Error Amplifier - GATE2 PWM Comparator - + COMP2 VFFB2 ENABLE Theory of Operation The CS5127 is a dual power supply controller that utilizes the V2ª control method. Two nonsynchronous V2ª buck regulators can be built using a single controller IC. This IC is a perfect choice for efficiently and economically providing core power and I/O power for the latest high-performance CPUs. Both switching regulators employ a fixed frequency architecture driven from a common oscillator circuit. The V2ª control method is illustrated in Figure 1. Both the ramp signal and the error signal are generated by the output voltage. Since the ramp voltage is defined as the output voltage, the ramp signal is affected by any change in the output, regardless of the origin of that change. The ramp signal also contains the DC portion of the output voltage, allowing the control circuit to drive the output switch from 0% to about 90% duty cycle. Changes in line voltage will change the current ramp in the inductor, affecting the ramp signal and causing the V2ª control loop to adjust the duty cycle. Since a change in inductor current changes the ramp signal, the V2ª method has the characteristics and advantages of current mode control for line transient response. Changes in load current will affect the output voltage and thus will also change the ramp signal. A load step will immediately change the state of the comparator output that controls the output switch. In this case, load transient response time is limited by the comparator response time and the transition speed of the switch. Notice that the reaction time of the V2ª loop to a load transient is not dependent on the crossover frequency of the error signal loop. Traditional voltage mode and current mode methods are dependent on the compensation of the error signal loop. The V2ª error signal loop can have a low crossover frequency, since transient response is handled by the ramp signal loop. The ÒslowÓ error signal loop provides DC accuracy. Low frequency roll-off of the error amplifier bandwidth will significantly improve noise immunity. This also improves remote sensing of the output voltage, since switching noise picked up in long feedback traces can be effectively filtered. V2ª line and load regulation are dramatically improved because there are two separate control loops. A voltage V2ª Control Method The V2ª method of control uses a ramp signal generated by the ESR of the output capacitors. This ramp is proportional to the AC current in the inductor and is offset by the DC output voltage. V2ª inherently compensates for variation in both line and load conditions since the ramp signal is generated from the output voltage. This differs from traditional methods such as voltage mode control, where an artificial ramp signal must be generated, and current mode control, where a ramp is generated from inductor current. + PWM Comparator GATE Ramp Signal - COMP Error Amplifier + Error Signal VFFB VFB Reference Voltage Figure 1: V2ª control diagram. 5 CS5127 Theory of Operation: continued mode controller relies on a change in the error signal to indicate a change in the line and/or load conditions. The error signal change causes the error loop to respond with a correction that is dependent on the gain of the error amplifier. A current mode controller has a constant error signal during line transients, since the slope of the ramp signal will change in this case. However, regulation of load transients still requires a change in the error signal. V2ª control maintains a fixed error signal for both line and load variation, since the ramp signal is affected by both. CT Lead Waveform If the sync pulse is longer than the CT lead discharge time, a short Òdead spotÓ will exist during which the output driver is off. Sync Lead Waveform Figure 2a: Sync pulse duration vs. CT lead discharge time. Voltage Mode Control The CS5127 can be operated in voltage mode if necessary. For example, if very small values of output ripple voltage are required, V2ª control may not operate correctly. Details on how to choose the components for voltage mode operation are provided in the section on VFFB component selection. The best way to determine if the pulse width is sufficiently short is to examine the CT lead waveform with an oscilloscope. If Òdead spotsÓ are observed in the CT lead waveform, decreasing the SYNC pulse width should be considered. Alternatively, the SYNC signal may be AC coupled through a small capacitor. In this case, care must be taken to ensure that current pulled out of the IC during the high-to-low transition of the SYNC signal is limited to less than 5mA. Constant Frequency As output line and load conditions change, the V2ª control loop modifies the switch duty cycle to regulate the output voltage. The CS5127 uses a fixed frequency architecture. Both output channels are controlled from a common oscillator. The CS5127 can typically provide a maximum duty cycle of about 90%. SYNC 20k Oscillator 2200p Sync Function It is sometimes desirable to shift the switching noise spectrum to different frequencies. A pulse train applied to the SYNC lead will terminate charging of the CT lead capacitor and pull the CT lead voltage to ground for the duration of the positive pulse level. This reduces the period of oscillation and increases the switching frequency. Synchronization must always be done at a frequency higher than the typical oscillator frequency. Using a lower frequency will lead to erratic operation and poor regulation. The SYNC pulse train frequency should be at least 10 % higher than the unsynchronized oscillator frequency. Synchronizing the oscillator will also decrease the maximum duty cycle. If the nominal oscillator frequency is 200kHz, increasing the oscillator frequency by 10% (to 220kHz) will decrease the maximum duty cycle from a typical of 90% to about 89%. Increasing the frequency by 25% (to 250kHz) will change the maximum duty cycle to about 87%. A 50% increase (to 300kHz) gives a maximum duty cycle of about 85%. The width of the SYNC pulse should be slightly shorter than the duration of the falling edge of the CT lead waveform (see Figure 2a) so the SYNC pulse doesnÕt interfere with the oscillator function. Figure 2b: Capacitive coupling of the SYNC signal. The external diode is used to clamp the IC substrate diode if ISYNC is greater than 5mA during the negative portion of the input waveform. Overcurrent Protection The CS5127 has no on-board current limit circuitry. An example current limit circuit is provided in the Additional Application Circuits section of this data sheet. 6 CS5127 Applications Information variables the designer must consider. Inductance values between 1µH and 50µH are suitable for use with the CS5127. Low values within this range minimize the component size and improve transient response, but larger values reduce ripple current. Choosing the inductor value requires the designer to make some choices early in the design. Output current, output voltage and the input voltage range should be known in order to make a good choice. The input voltage range is bracketed by the maximum and minimum expected values of VIN. Most computer applications use a fairly well-regulated supply with a typical output voltage tolerance on the order of ±5%. The values of VIN(MAX) and VIN(MIN) are used to calculate peak current and minimum inductance value, respectively. However, if the supply is well-regulated, these calculations may both be made using the typical input voltage value with very little error. Current in the inductor while operating in the continuous current mode (CCM) is defined as the load current plus the inductor ripple current: Selection of Feedback Lead Divider Resistor Values The feedback (VFB) leads are connected to external resistor dividers to set the output voltage. The on-chip error amplifier is referenced to 1.275V, and the resistor divider values are determined by selecting the desired output voltage and the value of the divider resistor connected between the VFB lead and ground. Resistor R1 is chosen first based on a design trade-off of system efficiency vs. output voltage accuracy. Low values of divider resistance consume more current which decreases system efficiency. However, the VFB lead has a 1µA maximum bias current which can introduce errors in the output voltage if large resistance values are used. The approximate value of current sinking through the resistor divider is given by 1.275V IV(FB) = R1 The output voltage error that can be expected due to the bias current is given by (1E - 6) ´ R1 Error Percentage = ´ 100% 1.275 IL = IOUT + IRIPPLE The ripple current waveform is triangular, and the current is a function of the voltage across the inductor, the switch on-time and the inductor value. Switch on-time is the duty cycle divided by the operating frequency, and duty cycle can be defined as the ratio of VOUT to VIN, such that where R1 is given in ohms. For example, setting R1 = 5K yields an output voltage error of 0.39% while setting the feedback divider current at 255µA. Larger currents will result in reduced error. IRIPPLE = Output Driver The peak current can be described as the load current plus half of the ripple current. Peak current must be less than the maximum rated switch current. This limits the maximum load current that can be provided. It is also important that the inductor can deliver the peak current without saturating. VOUT + 1.275V - R2 VFB IOUT(MAX) = ISWITCH(MAX) - R1 GATE COMP R2 can be sized according to the following formula once the desired output voltage and the value of R1 have been determined: ( (VIN(MAX) - VOUT)VOUT 2f ´ L ´ VIN(MAX) Since the peak inductor current must be less than or equal to the peak switch current, the minimum value of inductance can be calculated: Figure 3: Feedback resistor divider. R2 = R1 (VIN - VOUT)VOUT f ´ L ´ VIN VOUT -1 1.275 LMIN = ) (VIN(MIN) - VOUT)VOUT f ´ VIN(MIN) ´ ISWITCH(MAX) Load Current Transient Response The theoretical limit on load current transient response is a function of the inductor value, the load transient and the voltage across the inductor. In conventionally-controlled regulators, the actual limit is the time required by the control loop. Conventional current-mode and voltage-mode control loops adjust the switch duty cycle over many oscillator periods, often requiring tens or even hundreds of Selecting the Inductor There are many factors to consider when choosing the inductor. Maximum load current, core losses, winding losses, output voltage ripple, short circuit current, saturation, component height, EMI/EMC and cost are all 7 CS5127 Applications Information: continued microseconds to return to a steady-state. V2ª control uses the ripple voltage from the output capacitor and a ÒfastÓ control loop to respond to load transients, with the result that the transient response of the CS5127 is very close to the theoretical limit. Response times are defined below. tRESPONSE(INCREASING) = Selecting the Output Capacitor Output capacitors are chosen primarily on the value of equivalent series resistance, because this is what determines how much output ripple voltage will be present. Most polarized capacitors appear resistive at the typical oscillator frequencies of the CS5127. As a rule of thumb, physically larger capacitors have lower ESR. The capacitorÕs value in µF is not of great importance, and values from a few tens of µF to several hundreds of µF will work well. Tantalum capacitors serve very well as output capacitors, despite their bad reputation for spectacular failure due to excessive inrush current. This is not usually an issue for output capacitors, because the failure is not associated with discharge surges. Ripple current in the output capacitor is usually small enough that the ripple current rating is not an issue. The ripple current waveform is triangular, and the formula to calculate the ripple current value is: L(ÆIOUT) (VIN - VOUT) ´ 0.85 tRESPONSE(DECREASING) = L(ÆIOUT) VOUT Note that the response time to a load decrease is limited only by the inductor value. Other Inductor Selection Concerns IRIPPLE = Inductor current rating is an important consideration. If the regulated output is subject to short circuit or overcurrent conditions, the inductor must be sized to handle the fault without damage. Sizing the inductor to handle fault conditions within the maximum DC current rating helps to ensure the coil doesnÕt overheat. Not only does this prevent damage to the inductor, but it reduces unwanted heat generated by the system and makes thermal management easier. Selecting an open core inductor will minimize cost, but EMI/EMC performance may be degraded. This is a tough choice, since there are no guidelines to ensure these components will not prove troublesome. Core materials influence the saturation current and saturation characteristics of the inductor. For example, a slightly undersized inductor with a powdered iron core may provide satisfactory operation because powdered iron cores have a ÒsoftÓ saturation curve compared to other core materials. Small physical size, low core losses and high temperature operation will also increase cost. Finally, consider whether an alternate supplier is an important consideration. All of these factors can increase the cost of the inductor. and output ripple voltage due to inductor ripple current is given by: VRIPPLE(ESR) = (VIN - VOUT) ´ VOUT ´ ESR f ´ L ´ VIN A load step will produce an instantaneous change in output voltage defined by the magnitude of the load step, capacitor ESR and ESL. DI DVO = (DIO ´ ESD) + DT ESL A good practice is to first choose the output capacitor to accommodate voltage transient requirements and then to choose the inductor value to provide an adequate ripple voltage. Increasing a capacitorÕs value typically reduces its ESR, but there is a limit to how much improvement can be had. In most applications, placing several smaller capacitors in parallel will result in acceptable ESR while maintaining a small PC board footprint. A warning is necessary at this point. The V2ª topology relies on the presence of some amount of output ripple voltage being present to provide the input signal for the ÒfastÓ control loop, and it is important that some ripple voltage be present at the lightest load condition in normal operation to avoid subharmonic oscillation. Externally generated slope compensation can be added to ensure proper operation. Operating in Discontinuous Current Mode For light load designs, the CS5127 will operate in discontinuous current mode (DCM). In this regime, external components can be smaller, since high power dissipation is not an issue. In discontinuous mode, maximum output current is defined as: IOUT(MAX) = (VIN - VOUT)VOUT f ´ L ´ VIN (IPK)2 f ´ L(VIN) 2VOUT ´ (VIN(MAX) - VOUT) Selecting the VFFB Lead Components where IPK is the maximum current allowed in the switch FET. The VFFB lead is tied to the PWM comparatorÕs non-inverting input, and provides the connection for the externally-generated artificial ramp signal that is required whenever duty cycle is greater than 50%. 8 CS5127 Applications Information: continued The DC voltage for the VFFB pin is usually provided from the output voltage through an RC filter if VOUT is less than 3V. If VOUT is greater than 2.9V, a resistor divider from VOUT is recommended for proper circuit bias due to the common mode input range limitations of the PWM comparator. In most cases, the FB pin resistor divider can be used for this purpose with very little error, but a separate divider is recommended if high accuracy is required. The filter network is typically composed of a 1K resistor (RFFB) and a 330 pF capacitor (CFFB). This filter gives a 330 ns time constant which is sufficient to remove switching noise from the DC voltage. Note that in cases where a resistor divider provides the ramp signal, the resistor between VOUT and the VFFB pin serves as RFFB. An artificial ramp signal is generated using an NPN transistor (Q1), a small coupling capacitor (CC) and a second resistor (RR). The NPN transistor collector is connected either to the external 5V supply or to the ICÕs 5V on-chip reference. The transistorÕs base is connected to the CT pin, and the ramp on the CT pin is used to provide the artificial ramp. The transistorÕs emitter is connected to the coupling capacitor. The capacitor value should provide a low impedance at the switching frequency. A 0.1 µF capacitor represents 6.4 ohms at 250 kHz. A resistor is placed in series between this capacitor and the VFFB pin to set the amplitude of the ramp signal. if DC voltage is provided from the output, or (RESR) (VOUT)(R1) VRAMP = 2000 (L OUT) (R1 + R2) if DC voltage is provided from a resistor divider as in figure 5. where RESR is the equivalent series resistance in ohms of the total output capacitance, VOUT is the output voltage in volts and LOUT is the inductor value in Henries. The result is VRAMP given in millivolts per oscillator period. This value is the optimum amplitude for the artificial ramp. Note that COMP pin voltage changes and output ripple voltage must be added to the ramp amplitude for proper operation. Once the total ramp signal has been determined, the value of the ramp resistor (RR) can be determined. The ramp resistor and filter resistor RFFB create a resistor divider between the output voltage and the artificial ramp voltage. We can assume the output does not change, and that the maximum input voltage to the divider is equal to the DC output voltage plus the CT pin voltage swing of 2.1V. The ramp amplitude on the filter capacitor is then the divider output voltage: GATE VOUT R2 VRAMP = + (2.1V) (RFFB) (RR + RFFB) VFB CT 5V Q1 CT Rearranging, we have VFFB RR CC RE RR = RFFB RFFB CFFB R1 ( ) 2.1V -1 VRAMP Selecting the Catch Diode Figure 4: Artificial ramp components CC, CFFB, RR and RFFB must be provided for each channel if duty cycle for that channel exceeds 50%. Q1 and RE are common to both channels. DC voltage is shown supplied to VFFB through the VFB resistor divider. The schottky ÒcatchÓ diode must be capable of handling the peak inductor current and must withstand a reverse voltage at least equal to the value of VIN. Since the catch diode only conducts during switch off-time, the average current through the catch diode is defined as: The amount of artificial ramp is dependent on oscillator frequency, output voltage, output capacitor equivalent series resistance (ESR), and inductor value. It also assumes very small voltage fluctuations on the COMP pin. If the added ramp is too small, it will not be sufficient to prevent subharmonic oscillation. If the ramp is too large, V2ª control will be defeated, and loop regulation will enter voltage mode control. DC regulation will be adequate, but transient response will be degraded. However, this may be desirable in cases where very low values of output ripple voltage are desired. The artificial ramp amplitude can be calculated as follows: ICATCH = IOUT ( VIN - VOUT VIN ) Minimizing the diode on-voltage will improve efficiency. Selecting Oscillator Components RT and CT The on-chip oscillator frequency is set by two external components. RT sets the oscillator charge current. It is connected to a voltage reference approximately equal to 2.5V. The current generated in this fashion charges the CT capacitor between threshold levels of 1.5V and 3.6V. CT capacitor discharge is done by a saturating NPN, and the (RESR) (VOUT) VRAMP = 2000 (L OUT) 9 CS5127 Applications Information: continued should conduct all the ripple current. RMS ripple current can be as large as half the load current, and can be calculated as: discharge time is typically less than 10% of the charge time. External components CT and RT allow the switching frequency to be set by the user in the range between 10kHz and 500kHz. CT can be chosen first based on size and cost constraints. For proper operation over temperature, the value of RT should be chosen within the range from 20k½ to 40k½. Any type of one-eighth watt resistor will be adequate. Larger values of RT will decrease the maximum duty cycle slightly. This occurs because the sink current on the CT lead has an exponential relationship to the charge current. Higher charge currents will discharge the CT lead capacitor more quickly than lower currents, and a shorter discharge time will result in a higher maximum duty cycle. Once the oscillator frequency and a value of CT have been selected, the necessary value of RT can be calculated as follows: IRIPPLE(RMS) = IOUT ÆV = IRIPPLE(RMS) ´ ESR (fOSC)(CT) The type of capacitor is also an important consideration. Aluminum electrolytic capacitors are inexpensive, but they typically have low ripple current ratings. Choosing larger values of capacitance will increase the ripple current rating, but physical size will increase as well. Size constraints may eliminate aluminum electrolytics fro consideration. Aluminum electrolytics typically have shorter operating life because the electrolyte evaporates during operation. Tantalum electrolytic capacitors have been associated with failure from inrush current, and manufacturers of these components recommended derating the capacitor voltage by a ratio 2:1 in surge applications. Some manufacturers have product lines specifically tested to withstand high inrush current. AVX TPS capacitors are one such product. Ceramic capacitors perform well, but they are also large and fairly expensive. where fOSC is the oscillator frequency in hertz, CT is given in farads, and the value of RT is given in ohms. ESR effects are negligible since the charge and discharge currents are fairly small, and any type of capacitor is adequate for CT. Selecting the Compensation Capacitor As previously noted, the error amplifier does not contribute greatly to transient response, but it does influence noise immunity. The fast feedback loop input is compared against the COMP pin voltage. The DC bias to the VFFB pin may be provided directly from the output voltage, or through a resistor divider if output voltage is greater than 2.9V. The desired percentage value of DC accuracy translates directly to the VFFB pin, and the minimum COMP pin capacitor value can be calculated: CCOMP = VIN2 Peak current requirement, load transients, ambient operating temperature and product reliability requirements all play a role in choosing this component. Capacitor ESR and the maximum load current step will determine the maximum transient variation of the supply voltage during normal operation. The drop in the supply voltage due to load transient response is given as: 1.88 RT = VOUT(VIN - VOUT) Startup (16mA)(TOSC) (VFFBDC Bias Voltage)(tolerance) At startup, output switching does not occur until two undervoltage lockouts release. The first lockout monitors the VIN lead voltage. No internal IC activity occurs until VIN lead voltage exceeds the VIN turn-on threshold. This threshold is typically 8.4V. Once this condition is met, the on-chip reference turns on. As the reference voltage begins to rise, a second undervoltage lockout disables switching until VREF lead voltage is about 3.5V. The GATE leads are held in a low state until both lockouts are released. As switching begins, the VFB lead voltage is lower than the output voltage. This causes the error amplifier to source current to the COMP lead capacitor. The COMP lead voltage will begin to rise. As the COMP lead voltage begins to rise, it sets the threshold level at which the rising VFFB lead voltage will trip the PWM comparator and terminate switch conduction. This process results in a soft start interval. The DC bias voltage on VFFB will determine the final COMP voltage after startup, and the soft start time can be approximately calculated as: If fOSC = 200kHz, VFFB DC bias voltage is 2.8V and tolerance is 0.1%, CCOMP = 28.6µF. This is the minimum value of COMP pin capacitance that should be used. It is a good practice to guard band the tolerance used in the calculation. Larger values of capacitance will improve noise immunity, and a 100µF capacitor will work well in most applications. The type of capacitor is not critical, since the amplifier output sink current of 16mA into a fairly large value or wide range of ESR will typically result in a very small DC output voltage error. The COMP pin capacitor also determines the length of the soft start interval. Selecting the Input Bypass Capacitor The input bypass capacitors minimize the ripple current in the input supply, help to minimize EMI, and provide a charge reservoir to improve transient response. The capacitor ripple current rating places the biggest constraint on component selection. The input bypass capacitor network TSOFT START = 10 VFFB ´ CCOMP ICOMP(SOURCE) CS5127 Applications Information: continued where TSOFT START is given in seconds if CCOMP is given in farads, ICOMP(SOURCE) in amperes, and VFFB in volts. Note that a design trade off will be made in choosing the value of the COMP lead capacitor. Larger values of capacitance will result in better regulation and improved noise immunity, but the soft start interval will be longer and capacitor price may increase. VOUT VIN L RL C CONTROL LOGIC RA R RC RB PWM VFFB VR COMP VCONTROL R1 C2 R2 VFB EA C1 1.275V Figure 6: Voltage mode control equivalent circuit with two pole, one zero compensation network. VIN is the switch supply voltage, R represents the load, RL is the combined resistance of the FET RDS (on) and the inductor DC resistance, L is the inductor value, C is the output capacitance, RC is the output capacitor ESR, RA and RB are the feedback resistors and VR is the peak to peak amplitude of the artificial ramp signal at the VFFB pin. C1, C2, R1 and R2 are the components of the compensation network. Based on the application circuit from page 1, values for the 2.8V output equivalent circuit are: Figure 5: Measured performance of the CS5127 at start up. CCOMP =100µF, ICOMP(SOURCE)=1.3mA, VFFB = 2.8V, TSOFTSTART = 0.22s. Normal Operation VIN = R= RL = C= RC = RA = RB = L= During normal operation, the gate driver switching duty cycle will remain approximately constant as the V2ª control loop maintains the regulated output voltage under steady state conditions. Changes in supply line or output load conditions will result in changes in duty cycle to maintain regulation. Voltage Mode Control Voltage Mode Operation There are two methods by which a user can operate the CS5127 in voltage mode. The first method is simple, but the transient response is typically very poor. This method uses the same components as V2ª operation, but by increasing the amplitude of the artificial ramp signal, V2ª control is defeated and the controller operates in voltage mode. Calculate RR using the formula above and divide the value obtained by 10. This should provide an adequately large artificial ramp signal and cause operation under voltage mode control. There may be some dependence on board layout, and further optimization of the value for RR may be done empirically if required. 5V 0.4½ 0.02½ 1320µF 0.025½ 1540½ 1270½ 5µH A resistor change is necessary to increase the artificial ramp magnitude to VFFB1. Changing R10 from 20k to 2k will give a peak to peak amplitude of about 2V. Thus, VR = 2V. The transfer function from VCONTROL to VOUT is VOUT VCONTROL = R ´ VIN ´ (sCRC + 1) s2LC (R + RC) + s[L + RLC(R + RC) + RCRC] + R + RC 1 VR Voltage mode control may be refined by removing the COMP pin capacitor and adding a two pole, one zero compensation network. Consider the system block diagram shown in figure 6. Using the component values provided, this reduces to 11 ´ CS5127 Applications Information: continued divider. This factor will further reduce the overall system gain. 1 + s(3.3E-5) s2(2.772E-9) + s(2.902E-5) + 0.42 By adding the two pole, one zero compensation network shown in figure 6, we can maximize the DC gain and push out the crossover frequency. The transfer function for the compensation network is The zero frequency due to the output capacitor ESR is given as 1 = 4.8 kHz. (2¹CRC) VCONTROL s C1(R1 + R2) + 1 = -s C2 R1(s C1 R2 + 1) VFB The double pole frequency of the power output stage is This can be rewritten in terms of pole and zero frequencies and a gain constant A. 1 R + R1 = = 1.95 kHz. LC(R + RC) (2¹) VCONTROL s/(2¹fZ + 1) = VFB -A s ((s/2fP) + 1) The ESR zero approximately cancels one of the poles, and the total phase shift is limited to 90. Bode plots are provided below. fZ = where fP = 20 1 and A = R1 C2 2¹ C1R2 Note that, due to the first s term in the denominator, a pole is located at f = 0. This will provide the maximum DC gain. The optimum performance can be obtained by choosing fZ equal to the output double pole frequency and setting fP to approximately half of the switching frequency. Gain factors can be chosen somewhat arbitrarily. 0 Gain, (dB) 1 (2¹ C1 (R1 + R2)) -20 -40 -60.0 1 102 10 103 105 104 Values between 1E-6½F and 20E-6½F are practical. We then have a set of equations that can be solved for component values: 107 106 Frequency (Hz) Figure 7: Bode plot of gain response for VOUT/VCONTROL. C1 R1 = Phase, (degree) 90 0 1 2¹ [ 1 fZ - 1 fP ] , C1 R2 = 1 2¹fP , C2 = A R1 Since there are only three equations, we must arbitrarily choose one of the components. One option is to set the value of R1 fairly large. This provides a high impedance path between the VFB pin and the COMP pin. -90 For our design, we have fZ = the double pole frequency = 1.95 kHz and fP = fOSC/2 = 100kHz. LetÕs arbitrarily choose R1 = 4.7K. Then we solve the first equation for C1 and obtain C1 = 17nF. Use a standard value of 22 nF. -180 -270.0 1 10 102 103 104 105 106 107 Frequency (Hz) We next solve for R2. With C1 =22 nF, R2= 72½. Use a standard value of 75½. We can choose a gain factor from somewhere in the middle of our range and solve for C2. If A = 10E-6½F, we have C2 = 2.1 nF. Use a standard value of 2.2 nF. Figure 8: Bode plot of phase response for VOUT/VCONTROL. This uncompensated system is stable, but the low gain will result in poor DC accuracy, and the low cutoff frequency will result in poor transient response. Note that we have not yet included the gain factor from the feedback resistor 12 CS5127 Applications Information: continued Now that we have the compensation components chosen, we can put together a transfer function for the entire control loop. The transfer function is the product of the VOUT to VCONTROL transfer function, the gain of the feedback resistor divider and the negative inverse of the compensation loop transfer function. That is, Entering the loop transfer function in a mathematics program or a spreadsheet and evaluating the performance from resulting Bode plots may help to further optimize the compensation network component values. Compensation may be further optimized by using a two poleÐtwo zero compensation network as shown below. TLOOP = - (TVC-VO ´ TDIVIDER ´ TCOMPENSATION) C2 or [ R ´ VIN ´ (sCRC + 1) s2LC (R + RC) + s[L + RLC(R + RC) + RCRC] + R + RC [ ][ ´ R3 C1 TLOOP = 1 RB ´ VR RA + RB ][ ´ sC1 (R1 + R2)+ 1 sC2 R1 (sC1 R2 + 1) ] C3 R1 R2 From VOUT VFB COMP ] Figure 11: Two poleÐtwo zero compensation network. The two zeros are placed close to the resonant frequency of the LC output circuit. That is, Bode plots for this transfer function are shown below. 1 2¹ 100 1 2¹ C1 R2 Å 1 2¹ R3 C3 The two poles are placed near half the switching frequency, or 60 Gain, (dB) LC Å 20 fSW 2 -20 Å 1 2¹ C1 R1 Å 1 2¹ R3 C2 -60.0 Channel 2 ENABLE Feature -100.0 1 10 102 103 104 105 106 107 The ENABLE lead controls operation of channel 2. Channel 2 operates normally if the ENABLE lead voltage is greater than 3.5V. Setting the ENABLE lead voltage below 1.5V will guarantee that channel 2 is disabled. In this case, the GATE2 lead will be held low and no switching will occur. This feature can be used to selectively power up or power down circuitry that may not always need to be on. For example, in a laptop computer, channel 1 could power the microprocessor while channel 2 controlled the disk drive. Channel 2 could be turned off if the drive was not in use. Frequency (Hz) Figure 9: Bode plot of gain response for compensated voltage mode system. Phase, (degree) 90 0 -90 Thermal Management for Semiconductor Components -180 -270.0 1 10 102 103 104 105 106 Semiconductor components will deteriorate in high temperature environments. It is necessary to limit the junction temperature of control ICs, power MOSFETs and diodes in order to maintain high levels of reliability. Most semiconductor devices have a maximum junction temperature of 125¡C, and manufacturers recommend operating their products at lower temperatures if at all possible. Power dissipation in a semiconductor device results in the generation of heat in the pin junctions at the surface of the 107 Frequency (Hz) Figure 10: Bode plot of phase response for compensated voltage mode system. 13 CS5127 Applications Information: continued IC. This heat is transferred to the surface of the IC package, but a thermal gradient exists due to the thermal properties of the package molding compound. The magnitude of this thermal gradient is denoted in manufacturerÕs data sheets as QJA , or junction-to-air thermal resistance. The on-chip junction temperature can be calculated if QJA , the air temperature at the ICÕs surface and the on-chip power dissipation are known: TJ - TA PD = Q + Q + Q JC CS SA where PD is on-chip power dissipation in watts, TJ is junction temperature in ¡C, TA is ambient temperature in¡C, and thermal impedance QJC , QCS , and QSA are in¡C/W. All these quantities can be calculated or obtained from data sheets. The choice of a heatsink is based on the value of QSA required such that the calculated power dissipation does not cause junction temperature to exceed the manufacturerÕs maximum specification. TJ = TA + (QJA ´ P) TJ and TA are given in degrees centigrade, P is IC power dissipation in watts and QJA is thermal resistance in degrees centigrade per watt. Junction temperature should be calculated for all semiconductor devices to ensure they are operated below the manufacturerÕs maximum junction temperature specification. If any componentÕs temperature exceeds the manufacturerÕs maximum specification, some form of heatsink will be required. Heatsinking will improve the thermal performance of any IC. Adding a heatsink will reduce the magnitude of QJA by providing a larger surface area for heat transfer to the surrounding air. Typical heat sinking techniques include the use of commercial heatsinks for devices in TO-220 packages, or printed circuit board techniques such as thermal bias and large copper foil areas for surface mount packages. EMI Management Switching regulators generate noise a consequence of the large values of current being switched on and off in normal operation. Careful attention to layout of the printed circuit board will usually minimize noise problems. Layout guidelines are provided in the next section. However, it may be necessary in some cases to add filter inductors or bypass capacitors to the circuitry to achieve the desired performance. Layout Considerations The following guidelines should be observed in the layout of PC boards for the CS5127: When choosing a heatsink, it is important to break QJA into several different components. 1. Connect the PGND lead to the external ground with a wide metal trace. QJA = QJC + QCS + QSA 2. Connect both LGND and PGND together with a wide trace as close to the IC as possible. where all components of QJA are given in ¡C/W. QJC is the thermal impedance from the junction to the surface of the package case. This parameter is also included in manufacturerÕs data sheets. Its value is dependent on the mold compound and lead frames used in assembly of the semiconductor device in question. QCS is the thermal impedance from the surface of the case to the heatsink. This component of the thermal impedance can be modified by using thermal pads or thermal grease between the case and the heat sink. These materials replace the air gap normally found between heatsink and case with a higher thermal conductivity path. Values of QCS are found in catalogs published by manufacturers of heatsinks and thermal compounds. Finally, QSA is the thermal impedance from the heatsink to ambient temperature. QSA is the important parameter when choosing a heatsink. Smaller values of QSA allow higher power dissipation without exceeding the maximum junction temperature of the semiconductor device. Values of QSA are typically provided in catalogs published by heatsink manufacturers. The basic equation for selecting a heatsink is 3. Make all ground connections to a common ground plane with as few interruptions as possible. Breaks in the ground plane metal should be made parallel to an imaginary line between the supply connections and the load. 4. Connect the ground side of the COMP lead capacitors back to LGND with separate traces. 5. Place the VFFB lead capacitors as close to the VFFB leads as possible. 6. Place the 5V line bypass capacitors as close to the switch FETs as possible. 7. Place the output capacitor network as close to the load as possible. 8. Route the GATE lead signals to the FET gates with a metal trace at least 0.025 inches wide. 9. Use wide straight metal traces to connect between the 5V line and FETs, between FETs and inductors and between inductors and loads to minimize resistance in the high current paths. Avoid sharp turns, loops and long lengths. 14 CS5127 Additional Application Circuits 5V 5V R1 6.3K R3 10K VOUT Q2 FMMT2907ACT R1 24K Q1 Q2 Q3 FMMT2907ACT FMMT2907ACT FMMT2907ACT R2 1k R4 10K OVP OUT TO VFFB1 Q1 FMMT2222ACT ENABLE 1 SIGNAL R2 5K TO COMP1 Q4 FMMT2222ACT R3 1K VOVP = (R1 + R2)(0.65V) R2 Figure 14: An external circuit can be built to provide an enable function for channel 1. The circuit shown above connects to the VFB1 and COMP1 pins as indicated. If the ENABLE1 signal is left floating or is pulled high, channel 1 is enabled. If the ENABLE1 pin is pulled below 1V, Q1 will conduct, and mirror Q3 pulls VFFB1 up at the same time as Q2 and Q4 pull COMP1 low. This will force GATE1 to go low and turn off the switch FET. The circuit above will provide about 1mA of additional drive to the VFFB1 pin components. This additional current must be sufficient to pull VFFB1 up to about 1V in order to guarantee GATE1 is held low. Figure 12: Example external over voltage protection circuit. If VOUT exceeds VOVP, OVP out goes high. Resistor values shown above provide a +10% tolerance for a 3.3V output. 5V VOUT PGOOD OUT R1 18K R3 10K Q2 FMMT2222ACT Q1 FMMT2222ACT R2 5K VPGOOD = (R1 + R2)(0.65V) R2 Figure 13: Example external Power GOOD circuit. PGOOD(OUT) is low until VOUT exceeds VPGOOD. VPGOOD is typically chosen to be 10% below nominal VOUT. Resistor values above provide a -10% tolerance on VOUT =3.3V. 15 CS5127 Additional Application Circuits continued +5V +12V C1 680mF C2 680mF C3 1UF C4 680mF C5 +5V 680mF + + + m1 R1 3.3K 1 2 3 R2 100K R3 24K C6 390pF 5 C7 7pF Q1 IRL3103S C10 680mF 2.8V 4 L1 + R4 1540 6 7 8 5mH CS5127 VIN 16 CT VREF 15 RT ENABLE SYNC VFB1 VFB2 14 +C8 1mF 13 C52 7pF 12 COMP1 COMP2 VFFB1 VFFB2 GATE1 GATE2 LGND PGND 11 Q2 IRL3103S L2 10 09 R7 D2 1N5821 2400 R10 2K R6 2K R7 10 R5 1270 3.3V + D1 1N5821 C11 680mF C13 C12 680UF 680mF 5mH R65 18K C14 0.2mF C16 0.22mF C50 0.22mF Q4 2N3906 R11 2.2K Q7 2N3906 C15 0.01mF R8 1500 R66 10 C17 0.2mF C24 0.01mF R48 100K C15 0.1mF Figure 15: CS5127 12V, 5V input to 2.8V @ 7A and 3.3V @ 7A Voltage Mode Control Application Circuit with External Soft Start. +12V D3 1N5818 C2 680mF C1 680mF D4 1N5818 + + C18 0.1mF 50V Q1 FMMT2222ACT m1 1 2 3 R1 20K C7 330PF Q2 IRL3103S 7 8 L1 5mH + R4 1540 R5 1270 5 6 C10 680mF 2.8V 4 R2 27k C6 0.1mF R12 30 CS5127 SYNC VIN CT VREF RT ENABLE VFB1 VFB2 COMP1 COMP2 VFFB1 VFFB2 GATE1 GATE2 LGND PGND C4 680mF 15 14 C8 1mF C3 10mF + + 13 D5 1N5248 12 C9 0.1mF Q3 IRL3103S 11 10 C13 680mF L2 09 5mH D1 1N5821 C11 680mF C5 680mF 16 D2 1N5821 3.3V + C12 680mF R7 2400 R8 1500 R9 2K R10 20K C14 330PF C16 100mF C15 100mF R6 1K Figure 16: CS5127 12V only to 2.8V @ 7A and 3.3V @ 7A Application Circuit. 16 C17 330PF R11 20K R3 18K CS5127 Additional Application Circuits continued +5V R69 10K C2 C1 680mF 680mF Q4 2N2907 Q7 2N2907 Q1 FMMT2222ACT R67 15k R73 1K + - 3 C7 330PF 4 5 C6 0.1mF Q2 IRL3103S 2.8V + CS5127 VIN 16 CT VREF 15 RT ENABLE SYNC C4 680mF R77 15K VFB1 14 COMP1 COMP2 12 6 VFFB1 8 VFFB2 11 7 GATE1 GATE2 10 8 LGND PGND 09 C3 1mF + C8 1mF 13 VFB2 R78 15K + C9 0.1mF Q3 IRL3103S C12 C13 680mF 680mF RDROOP L2 5mF D2 1N5821 D1 1N5821 C11 680mF R4 1.54K C25 0.1mF L1 5mH .008 + C5 680mF - m1 R68 15k RDROOP R76 1M + LM2903 U4B 2 C10 680mF R74 1K LM2903 U4A 1 R1 20K R75 1M R72 20K R70 20K C24 0.1mF +12V R71 10K R66 1M R65 1M +5V +12V +12V + .008 R7 2.40K R8 1.50K R11 20K R5 1.27K 3.3V R9 2K 20K R10 R6 1K C14 330PF C17 C15 100mF C16 100mF R3 18K 330PF Figure 17: 200kHz, V2ª, 5V/12V input, 2.8V@ 7A and 3.3V @ 7A outputs with current limit. +12V +5V C1 680mF +5V C2 680mF C4 680mF + C3 1mF C5 680mF + + Q1 FMMT2222ACT m1 1 2 3 R1 20K C7 330PF C6 0.1mF Q2 IRL3103S 7 8 L1 10mH + R4 1540 R5 1270 5 6 C10 680mF 2.8V 4 R2 27K CS5127 SYNC VIN CT VREF RT ENABLE VFB1 VFB2 COMP1 COMP2 VFFB1 VFFB2 GATE1 GATE2 LGND PGND 16 15 14 C8 1mF C9 0.1mF 12 Q3 IRL3103S 11 10 C13 680mF L2 09 D1 1N5821 C11 680mF + 13 10mH R17 2400 + C12 680mF D2 1N5821 3.3V R7 2400 R8 1500 R9 2K R10 20K C14 330pF R6 1K C16 100mF C15 100mF 8 Q4 IRL3103S + 1 - 2.5V 4 C17 330pF R11 20K R3 18K U2A 1/2 LM358 3 2 R18 2400 C18 47mF Figure 18: CS5127 12V, 5V input to 2.8V @7A and 3.3V @ 7A Switching Regulator with External 1A, 2.5V Linear Output for Vclock. 17 CS5127 Typical Performance Characteristics 5.005 18 5.004 Line Regulation (mV) 16 VREF (V) 5.003 5.002 5.001 14 12 10 5.000 4.999 8 0 10 20 30 40 50 60 70 0 10 Temperature (C) 30 40 50 60 70 Temperature (C) Figure 19: VREF vs Temperature, 1mA Load. Figure 22: Line Regulation vs Temperature 9V to 20V. 130 Short Circuit Current (mA) 18 16 Load Regulation (mV) 20 14 12 10 120 110 100 90 80 70 8 0 10 20 30 40 50 60 0 70 10 20 30 40 50 60 70 Temperature(C) Temperature (C) Figure 23: VREF Short Circuit Current vs Temperature. Figure 20: Load Regulation vs Temperature 1mA to 10mA. 4.992 1.00E+06 4.990 5.00E+05 4.998 2.00E+05 Frequency (kHz) VREF (V) 150pF 4.988 4.986 390pF 1.00E+05 680pF 5.00E+04 1.5nF 4.980 2.00E+04 4.978 1.00E+04 3.3nF 0 10 20 30 40 50 60 20 70 22 24 26 28 30 32 34 36 RT (kW) Temperature (C) Figure 21: VREF vs Temperature, 10mA Load. Figure 24: Oscillator Frequency vs RT,CT (VIN = 12V, T = 25C) 18 38 40 CS5127 Typical Performance Characteristics: continued 91.5 1.65 1.60 1.55 Sync Threshhold (V) Maximum Duty Cycle (%) 91.0 90.5 90.0 89.5 1.50 1.45 1.40 1.35 1.30 89.0 1.25 88.5 1.20 0 20 10 30 40 Temperature (C) 60 50 70 10 30 40 Temperature (C) 20 50 70 60 Figure 28: SYNC Threshold vs Temperature. 210 170 208 165 Sync Input Curernt (mA) Oscillator Frequency (kHz) Figure 25: Oscillator Maximum Duty Cycle vs Temperature. 0 206 204 202 160 155 150 145 200 140 0 10 20 30 40 50 60 70 0 10 20 Temperature (C) Figure 26: Oscillator Frequency vs Temperature. CT = 330pF, RT =27k 30 40 Temperature (C) 50 60 70 Figure 29: SYNC Input Current vs Temperature (VSYNC = 2.4V). 95 1.2785 150pF 1.2780 VFB Reference Voltage (V) Duty Cycle (%) 90 85 390pF 80 680pF 75 3.3nF 1.5nF 70 1.2775 1.2770 1.2765 1.2760 1.2755 20 22 24 26 28 30 32 34 36 38 40 0 RT (kW) Figure 27: Oscillator Duty Cycle vs CT, RT (VIN = 12V, T = 25C). 10 20 30 40 Temperature (C) Figure 30: VFB Reference Voltage vs Temperature. 19 50 60 70 CS5127 Typical Performance Characteristics: continued 225 70 60 180 50 40 Gain (dB) 135 30 Phase (degrees) 20 10 0 -10 -20 -30 90 45 0 -45 1.00E +00 1.00E +01 1.00E +02 1.00E +03 1.00E +04 1.00E +05 1.00E +06 1.00E+00 1.00E+01 1.00E+02 1.00E +07 Frequency (Hz) 1.00E+05 1.00E+06 1.00E+07 Figure 34: Error Amplifier Phase vs Frequency. 1.310 450 1.305 Source Current (mA) 430 Input Curernt (mA) 1.00E+04 Frequency (Hz) Figure 31: Error Amplifier Gain vs Frequency. 410 390 370 1.300 1.295 1.290 1.285 1.280 350 0 10 20 30 40 Temperature (C) 50 60 0 70 10 20 40 40 50 60 70 Temperature (C) Figure 35: Error Amplifier Source Current vs Temperature. Figure 32: SYNC Input Current vs Temperature (VSYNC = 5V). 0.90 0.105 0.103 0.85 Output Low Voltage (V) VFB Bias Current (mA) 1.00E+03 0.101 0.099 0.097 0.095 0.80 0.75 0.70 0 10 20 30 40 Temperature (C) 50 60 70 0 10 20 30 40 50 Temperature (C) Figure 36: Error Amplifier Output Low Voltage (500µA) vs Temperature. Figure 33: VFB Bias Current vs Temperature. 20 60 70 CS5127 Typical Performance Characteristics: continued 2.28 3.35 ENABLE Threshold (V) Maximum Common Mode Voltage (V) 3.37 3.33 3.31 3.29 2.26 2.24 2.22 2.20 3.27 0 10 20 30 40 50 60 0 70 10 20 Figure 37: PWM Comparator Maximum Common Mode Input Voltage vs Temperature. 50 60 70 50 60 70 50 60 70 280 270 ENABLE Bias Current (mA) 17.0 16.5 Sink Current (mA) 40 Figure 40: ENABLE Threshold vs Temperature. 17.5 16.0 15.5 15.0 14.5 260 250 240 230 220 210 0 10 20 30 40 50 60 70 0 10 20 Temperature (C) Figure 38: Error Amplifier Sink Current vs Temperature. 30 40 Temperature (C) Figure 41: ENABLE Bias Current vs Temperature. 1.07 215 1.06 GATE Low Voltage (mV) 210 1.05 VFFB Bias (mA) 30 Temperature (C) Temperature (C) 1.04 1.03 1.02 1.01 205 200 195 190 185 1.00 0.99 180 0 10 20 30 40 50 60 70 0 Temperature (C) Figure 39: VFFB Bias Current vs Temperature. 10 20 30 40 Temperature (C) Figure 42: GATE Low Voltage (100mA) vs Temperature. 21 CS5127 1.70 18.0 1.65 17.5 Lockout Voltage (mV) GATE High Voltage (mV) Typical Performance Characteristics: continued 1.60 1.55 1.50 1.45 17.0 16.5 16.0 15.5 1.40 15.0 0 10 20 30 40 Temperature (C) 50 60 70 0 10 20 30 40 50 60 70 Temperature (C) Figure 43: GATE High Voltage (100mA) vs Temperature. Figure 46: GATE Low Voltage (Lockout) vs Temperature. 8.615 48 47 8.613 Start up Threshold (V) GATE Low Voltage (mV) 46 45 44 43 42 41 8.611 8.609 8.607 40 8.605 39 0 10 20 30 40 Temperature (C) 50 60 0 70 10 20 30 40 Temperature (C) 50 60 70 50 60 70 Figure 47: VIN Start-up Threshold vs Temperature. Figure 44: GATE low voltage (20mA) vs Temperature. 400 1.60 1.55 380 Start up Current (mA) GATE High Voltage (V) 1.50 1.45 1.40 1.35 1.30 360 340 320 1.25 300 1.20 0 10 20 30 40 Temperature (C) 50 Figure 45: GATE High Voltage (20mA) vs Temperature. 60 0 70 10 20 30 40 Temperature (C) Figure 48: Start-up Current vs Temperature. 22 CS5127 Typical Performance Characteristics: continued 18.5 7.924 7.922 18.0 IC Supply Current (mA) Shutdown Threshhold (V) 7.92 7.918 7.914 7.912 7.91 17.5 17.0 16.5 16.0 7.908 7.906 15.5 0 10 20 30 40 50 60 0 70 Temperature (C) 10 20 40 30 Temperature (C) 50 60 70 Figure 50: IC Supply Current vs Temperature. No Load on GATE pins. RT = 27k, CT = 330pF Figure 49: VIN Shutdown Threshold vs Temperature. 23 CS5127 Package Specification PACKAGE DIMENSIONS IN mm (INCHES) PACKAGE THERMAL DATA Thermal Data RQJC typ RQJA typ D Lead Count 16 Lead SOIC Wide Metric Max Min 10.50 10.10 English Max Min .413 .398 16 Lead SOIC Wide 23 105 ûC/W ûC/W Surface Mount Wide Body (DW); 300mil wide 7.60 (.299) 7.40 (.291) 10.65 (.419) 10.00 (.394) 0.51 (.020) 0.33 (.013) 1.27 (.050) BSC 2.49 (.098) 2.24 (.088) 1.27 (.050) 0.40 (.016) 2.65 (.104) 2.35 (.093) 0.32 (.013) 0.23 (.009) D REF: JEDEC MS-013 0.30 (.012) 0.10 (.004) Ordering Information Part Number CS5127GDW16 CS5127GDWR16 Rev. 11/3/98 Cherry Semiconductor Corporation reserves the right to make changes to the specifications without notice. Please contact Cherry Semiconductor Corporation for the latest available information. Description 16 Lead SOIC Wide 16 Lead SOIC Wide (tape & reel) 24 © 1999 Cherry Semiconductor Corporation