 |
|
| If you can't view the Datasheet, Please click here to try to view without PDF Reader . |
|
|
|
| If you can't view the Datasheet, Please click here to try to view without PDF Reader . |
|
|
| Datasheet File OCR Text: |
| APPLICATION NOTE DESIGNING WITH THE L296 MONOLITHIC POWER SWITCHING REGULATOR A cost-effectivereplacement for costly hybrids, the L296 Power Switching Regulator delivers 4A at an output voltage of 5.1V to 40V and includes many popular supply features. This comprehensive application guide explains how the device operates and how it is used. Typical application circuits are also presented. The SGS THOMSON L296 is the first monolithic switching regulator in plastic package which includes the power section. Moreover, the circuit includes all the functions which make it specially suited for microprocessor supply. Before the introduction of L296, which realizes the step down configuration, this function was implemented with discrete power components driven be integrated PWM regulator circuits (giving a maximum outputcurrent of 300 to 400mA) or with hybrid circuits. Both of thesesolutionsarecharacterizedby a low efficiency of the power transistor. For this reason it is generally necessary to operate at frequenAN244/1288 cies in the 20kHz to 40kHz range. Of the two alternativesdiscrete solutionsareusuallyless expensive because they do not include as many functions as the L296. With the new L296 regulator the driving problem of the power control stage has been eliminated. Besides a higher overall efficiency, it is therefore also possibleto operatedirectlyat frequenciesas highas 100kHz. At 200kHz the device still operates (further reducing the cost of the L and C external components) when a reductionof a few percent in efficiency is acceptable. 1/43 APPLICATION NOTE The device deliversa maximum current of 4 A to the load, at an output voltage adjustable from 5.1 to 40V ; the maximum operating input voltage is 46V. The high voltage and the high current capabilities of the device are a result of the special technology used and the special care taken in designingthe power transistor.Essentialrequirementsfora goodpower transistor are high gain and high current levels, low saturationvoltage and good second breakdown robustness.To achieve high gain at high current levels, the power transistor has to be designedto maximize the emitter's perimeter/area ratio. In the L296 power transistor, realized with a high voltage (50V) process, current densities in the magnitude order of 10mA/Mil2 are achieved. In its most complete configuration, in which all the available functions are being used, a significant reductionof theexternalcomponentcountis achieved compared with discrete component solution. The L296 is mounted in a MULTIWATT(R) plastic package with 15 pins, minimizing the cost per watt and allowing a low thermal resistance of 3C/W between junction and package and of 35C/W betweenjunctionand ambient.This thermalresistance (inclucing the contact resistance) is comparable to that of the more costly metal TO-3 packages. THE STEP-DOWN CONFIGURATION Fig. 1 shows the simplified block diagramof the circuit realizing the step-down configuration. This circuit operates as follows : Q1 acts as a switch at the frequency f and the ON and OFF times are suitably controlled by the pulse width modulator circuit. When Q1 is saturated, energy is absorbed from the input which is transferred to the output through L. The emitter voltage of Q1, VE, is Vi-Vsat when Q is ON and -VF ( with VF the forward voltage across the D diode as indicated) when Q1 is OFF. During this second phase the current circulates again through L and D. Consequently a rectangular shaped voltage appears on the emitter of Q1 and this is then filtered by the L-C-D network and converted into a continuous mean value across the capacitor C and therefore across the load. The current through L consists of a continuous component, ILOAD, and a triangular-shaped component super-imposed on it, IL, due to the voltage across L. Figure 1 : The Basic Step-down Switching Regulator Configuration. 2/43 APPLICATION NOTE Figure 2 : Principal Circuit Waveforms of the figure 1 Circuit. 3/43 APPLICATION NOTE Fig. 2 shows the behaviour of the most significant waveforms, in different points of the circuit, which help to understandbetter the operationof the power section of the switching regulator. For the sake of simplicity, the series resistance of the coil has been neglected.Fig. 2a shows the behaviour of the emitter voltage (which is practically the voltage across the recirculation diode), where the power saturation and the forward VF drop across the diode era taken into account. The ON and OFF times are established by the following expression : TON Vo = (Vi - Vsat) TON + TOFF Fig. 2b shows the current across the switching transistor. The currentshape is trapezoidaland the operation is in continuous mode. At this stage, the phenomenadue to the catchdiode,that we consider as dynamically ideal, are neglected. Fig. 2c shows the current circulating in the recirculation diode. The sum of the currents circulating in the power and in the diode is the current circulating in the coil as + shown in fig. 2e. In balancedconditionsthe IL current increase occuring during TON has to be equal to the IL- decrease occurring during TOFF. The mean value of IL correspondsto the charge current. The current ripple is given by the following formula : (Vi - Vsat) - V IL+ = IL- = TON = L Vo + VF = TOFF L It is a good rule to respect to IoMIN IL/2 relationship, that implies good operation in continuous mode. When this is not done, the regulator starts operating in discontinuousmode. This operation is still safe but variations of the switching frequency may occur and the output regulation decreases. Fig. 2d shows the behaviour of the voltage across coil L. In balancedconditions, the mean value of the voltageacross the coil is zero. Fig. 2f showsthe current flowing through the capacitor, which is the difference between IL and ILOAD. In balancedconditions, the mean current is equalto zero, and IC = IL. The current IC through the capacitor gives rise to the voltage ripple. This ripple consists of two components: a capacitive component, VC, and a resistive component, VESR, due to the ESR equivalent series resistance of the capacitor. Fig. 2g shows the capacitive component VC of the voltage ripple, which is the integral of a triangular-shaped current as a function of time. Moreover, it should be observed that v C (t) is in quadrature with iC(t) and therefore with the voltage VESR. The quantity of charge Q+ supplied to the capacitor is given by the area enclosed by the ABC triangle in fig. 2f : 1 T IL . . Q = 2 2 2 which therefore gives : IL Q VC = = C 8fc Fig. 2h shows the voltage ripple VESR due to the resistive componentof the capacitor. This component is VESR (t)= iC (t) ESR. Fig. 2i shows the overall ripple Vo, which is the sum of the two previous components. As the frequencyincreases (> 20kHz), which is required to reduce both the cost and the sizes of L and C, the VESR component becomes dominant. Often it is necessary to use capacitors with greater capacitance(or morecapacitorsconnectedin parallel to limit the value of ESR within the required level. We will now examine the stepdown configuration in more detail, referring to fig. 1 and taking the behaviour shown in fig. 2 into account. Starting from the initial conditions, where Q = ON, vC =Vo andiL = iD =0, usingKirckoff secondprinciple we may write the following expression : Vi = vL + vC (Vsat is neglected against Vi). diL diL Vi = L + vC = L + Vo (1) dt dt which gives : (Vi - Vo) diL = dt L (2) The current through the inductance is given by : (Vi - Vo) t (3) IL = L When Vi, Vo, and L are constant, IL varies linearly with t. Therefore, it follows that : (Vi - Vo) TON IL+ = (4) L WhenQ is OFF the current throughthe coil has reached its maximumvalue,Ipeak andbecauseit cannot very instantaneously, the voltage across the coil is inverted and the diode D becomes forward biased to allow the recirculation of the current through the load. 4/43 APPLICATION NOTE WhenQ switches OFF, the followingsituationis present : vC(t) = Vo, iL (t) = iD (t) = Ipeak And the equation associated to the following loop may be written : diL VF + L + vC = 0 (5) dt where : vC = Vo dIL = - (VF + Vo)/L dt It follows therefore that : VF + Vo iL (t) = - t L all theothersystemlosses.Theexpressionof the efficiency becomes therefore the following : Po = (12) Po + Psat + P D + PL + Pq + psw DC LOSSES Psat : saturation losses of the power transistor Q. These losses increase as Vi decreases. Vo TON Psat = Vsat . Io = Vsat Io (13) T Vi TON Vo where = and Vsat is the power T Vi transistor saturation at current Io. PD : losses due to the recirculation diode. These losses increase as Vi increases, as in this case the ON time of the diode is greater. Vo Vi - Vo PD = VF Io = VF Io (1 - ) (14) Vi Vi where VF is the forward voltage of the recirculation diode at current Io. PL : losses due to the series resistance RS of the coil (15) PL = RS Io2 Pq : losses due to the stand-by current and to the power driving current : TON Pq = Vi I'3q + Vi I''3q (16) T (6) (7) The negative sign may be interpretatedwith the fact thatthe current isnow decreasing.AssumingthatVF may be neglected against Vo, during the OFF time the following behaviour occurs : Vo IL = t (8) L therefore : Vo IL - = L TOFF (9) But, because IL+ = IL- if follows that : (Vi - Vo) TON Vo TOFF = L L which allows us to calculate Vo : TON TON Vo = Vi = Vi TON + TOFF T (10) where T is the switching period. Expression (10) links the output voltageVo to the input voltageVi andto the duty cycle. The relation-ship between the currents is the following : TON IiDC = IoDC . T EFFICIENCY The system efficiency is expressed by the following formula : Po %= 100 Pi where Po = VoIo (with Io = ILOAD) is the output power to the load and Pi is the input power absorbedby the system. Pi is given by Po, plus where being : TON Vo = it follows that : T Vi Pq = Vi I'3q + Vo I''3q in which : I'3q = I3q at 0 % duty cycle I''3q = I3q(100 % d.c.) - I3q (0 % d.c.) SWITCHING LOSSES Psw : switching losses of the power transistor : tr + tf Psw = Vi Io 2T The switching losses of the recirculation diode are neglected (which are anyway negligible) as it is assumed that diode is used with recovery time much smaller than the rise time of the power transistor. We can neglect losses in the coil (it is assumed that IL is very small compared to Io) and in the output capacitor, which is assumed to show a low ESR. 5/43 APPLICATION NOTE Calculation of the inductance value, L Calculation TON and TOFF through (4) and (9) respectively it follows that : + IL . L I- . L L TON = TOFF Vi - Vo Vo But because : TON + TOFF = T it follows that : and IL+ = IL- = IL, Finally, calculating C it follows that : (Vi - Vo) Vo C= 8 Vi VC f2 L where : (20) IL . L IL . L + =T Vi - Vo Vo Calculating L, the previous relation becomes : (Vi - Vo) Vo L= T (18) Vi IL Fixing the current ripple in the coil required by the design (for instance 30% of Io), and introducing the frequency instead of the period, it follows that : (Vi - Vo) Vo L= where L is in Henry and f in Hz Vi . 0.3 . Io . f Calculation of the output capacitor C From the output node in fig. 3 it may be seen that the current throughthe outputcapacitoris given by : ic (t) = iL (t) - Io Figure 3 : Equivalent Circuit Showing Recirculation when Q1 is Turned Off. L is in Henrys C is in Farads f is in Hz Finally, the following expression should be true : VCmax ESRmax = (21) IL It may happen that to satisfy relation (21) a capacitance value much greater than the value calculated through (20) must be used. TRANSIENT RESPONSE Sudden variations of the load current give rise to overvoltages and undervoltages on the output voltage. Since ic = C (dvc/dt) (22), where dvc = Vo, the instantaneous variation of the load current Io is suppliedduringthe transientbythe outputcapacitor. During the transient, also current through the coil tends to change its value. Moreover, the following is true : diL vL = L (23) where diL = Io. dt vL = Vi - Vo for a load increase vL = Vo for a load decrease Calculating dt from (22) and (23) and equalizing, it follows that : dvc diL L =C vL ic Calculating dvc and equalizing it to Vo, it follows that : LI o2 Vo = (24) for + Io C(Vi - Vo) LI o2 Vo = (25) for - Io CVo From the behaviour shown in fig. 2 it may be calculated that the charge current of the output capacitor, within a period, is IL/4, which is supplied for a time T/2. It follows therefore that : IL T IL T IL VC = = = (19) 4C 2 8C 8fC but, remembering expression (4) : Vo (Vi - Vo) TON IL+ = and T ON = T L Vi therefore equation (19) becomes : (Vi - Vo) Vo VC = 8 Vi f2 L C 6/43 From these two expressions the dependence of overshoots and undershootson the L and C values maybe observed.To minimize Vo it is thereforenecessary to reduce the inductance value L and to increase the capacitance value C. Should other auxiliary functionsbe requiredin thecircuit like reset or crowbar protectionsand very variable loads may be present, it is worthwhile to take special care for minimizing these overshoots, which could cause spurious operation of the crowbar, and the undershoot, which could trigger the reset function. APPLICATION NOTE DEVICE DESCRIPTION Fig. 4 shows the package in which the device is mounted and the pin function assignments. The internalstructureof the deviceis shownin fig. 5. Each block will now be examined. Power supply The deviceis providedwith an internalstabilized power supply that, besides supplying the reference Figure 4 : Pin Assignments of the L296. voltage of 5.1V for the whole system, also supplied the internal analog blocks. Special features of the voltage reference are its accuracy, temperaturestability and high line rejection. Through zenze-zap trimming, the voltage is within 2% limits. Figure 5 : Block Diagram of the L296. In Addition to the Basic Regulation Loop the Device includes Functions such as Reset, Crowbar and Current Limiting. 7/43 APPLICATION NOTE OSCILLATOR The oscillator block generatesthe saw-tooth waveformthat setsthe switchingfrequencyofthe system. This signal, compared with the output voltageof the erroramplifier, generatesthe PWM signalto be sent to the power output stage. The saw-tooth, whose amplitude is between 1.2V and 3.2V, is generated by charging rapidly the Cosc capacitorwhich then discharges across the Rosc resistance. As shown in fig. 6, the oscillator is realized by a comparator(with grounded compatible input) with hysteresis whose thresholdsare 1.2V and 3.2V respectively. The Cosc capacitor and the Rosc resis-tance are connected to the non-inverting input of the comparator which set the oscillating frequency is fixed. When the voltage on pin 11 is less than 3.2V, the switch S1 is closed and the current generator charges the Cosc capacitor rapidly ; in this phase S2 is also closed. As soon as 3.2V is reached the comparator output drives S2 open (thereforeopeningS1, too); the invertinginput voltage is reduced to about 1.2V and the capacitor Figure 6 : Internal Schematic of the Oscillator. startsto dischargeitself across the Rosc resistor (the Ibias effect is neglected). When the voltage reaches 1.2V, S2 and S1 close again and a new cycle starts. The generated waveform is shown in fig. 7. To achieve a good accuracy of the switching frequency it is essential to have a charging time of the capacitor which is much smaller than the discharging time. In this way, the oscillation frequency only dependson the externalcomponentsCosc andRosc. For thisreasonthe capacitorcharging current(when S1 is ON) is typically around 10mA. For example, with a 2.2nF capacitor to switch from 1.2V to 3.2V about 400ns is required, which is negligible compared to the 10s period that occurs when the operation is performed at 100kHz. The diagrams shown in fig. 8 allow the calculation of the Rosc value (R1 in fig. 8) with Cosc as a parameter (C3 in fig. 8) when the oscillation frequency required for operation has been previously fixed. 8/43 APPLICATION NOTE Figure 7a : Oscillator Waveform at Pin 11 with f = 100Khz (Rosc = 4.3K, Cosc = 2.2nF). Fig. 8 shows two suggested values for the Cosc capacitance. Excessively low capacitance value may give rise to an inaccuracyof the upperthreshold due to the switching delays of the comparator. This inaccuracy in caused by an excessively short rise time of the voltage. A capacitance value too high gives rise to a charging time which is too compared to the discharging time. An additional inaccuracy cause would be therefore present for the switching frequency, now due to spread of the charge current. The oscillation frequency is given by the following formula : 1 (26) fosc = Rosc Cosc PWM (see fig. 9) Figure 7b : Oscillator Waveform at Pin 11 with f = 50Khz (Rosc = 9.1K, Cosc = 2.2nF). The PWM signal is generated on the comparator output ; the triangular-shaped waveform and the continuous signal coming from the output of the transconductance error amplifier are sent to its inputs. The PWM signal is then transferred to the driving stage of the output power transistor. SOFT START (see fig. 9) Soft start is an essentialfunctionfor correct start-up, to prevent stresses and possible breakdown from occurring in the powertransistor and to obtaina monotonically increasing output voltage. In particular, the L296, as it does not have any duty cycle limitation and due to the type of current limitation does not allow the output to be forced to a steady state without the aid of the soft-start facility. Soft-start operates at the start-up of the system, after the inhibit has been activated, after an interventionof the current limitation andafterthe intervention of the thermal protection. The soft-start function is realized through a capacitor connected to pin 5 which is charged at constant current ( 100A) up to a value of about VREF. During the charging time, throughPNP transistorQ58, the voltage on pin 9 is forced to increase with the same rising speed as on pin 5. Starting from the discharged capacitor condition (pin 5 voltage = 0V) the power transistor is in the OFF condition, as the voltage on pin 9 is smaller than the minimum level of the ramp voltage.As the capacitoris charged,the PWM signal begins to be generated as soon as the error amplifier outputvoltage crosses the ramp ; the power stage starts to switch with steadily increasing duty cycle. This behaviour is shown in fig. 10. As soon as the steadycondition is reachedthe duty cycle sets itself to the right value due to the effect of the feedback network while the soft-start capacitor completesits charging to a value very close to VREF. 9/43 Figure 8 : Nomogram for the Choice of Oscillator Components. APPLICATION NOTE The soft-start effect is determined, apart from the switch-on time, when thecurrentlimitationoperates, due to either an overload or a short circuit, to keep the mean value of the current absorbed by the power supply low. Moreover from fig. 11 it may be observed that since the voltage on pin 9 can decrease under the minimum ramp level and increase over the maximum level no limitationshave been provided on theduty cycle, which thereforemay vary between 0 and 100%. Figure 9 : Partial Internal Schematic Showing PWM and Soft Start Blocks. Figure 10 : Soft Start Waveforms. When power is applied, or after an inhibit, the L296's output current rises slowly under control of the soft start circuit. 10/43 APPLICATION NOTE Figure 11 : Waveform for Calculation of Duty Cycle and Soft Start Time. CALCULATING THE DUTY CYCLE AND SOFT-START TIME Assume, for simplicity, that the rising edge of the rampis instantaneous; Vr isthe outputvoltageof the error amplifier and Vc the ramp voltage (see fig. 11). The PWM comparator block switches when Vr = Vc ; therefore : t - Rosc Cosc Vr = Vc = E e Consequently: t = Rosc Cosc In E Vr where Css is the soft-start capacitor and I5so is the charging current. Considering as the soft-start time the time required for the soft-start capacitor to charge from (1.2 V - 0.5V) to V r - 0.5V, gives : Css (Vr - 1.2) tss = Isso substituting Vr from (27) gives : Vo - ( 1- ) Vi Vr = E e substituting into (28) gives : Vo ( V -1) Css i tss = (E e - 1.2) Isso SYNCHRONIZATION The synchronization function is available on pin 7, this function allows the device to be switched at an externally generated frequency (leaving pin 11 open), or to mutually synchronize several devices, using one of themas masterand the othersas slave (fig. 12). This allows several devices to be operated at the same frequency,avoidingundesirableintermodulation phenomena.The numberif mutuallysynchronizable devices is obviously much greater than the three devices shown in the figure. It is anyway diffi- The time obtained from this expression is the TOFF timeof the powertransistor.Theduty cycle d isgiven by : E TON T - Rosc Cosc In Vr d= = = T T (27) E Vo = 1 - In = Vr Vi Consequently, starting with the capacitor discharged, the outputof the regulator will be at the nominal level when the voltage at the terminal of the capacitor (which is charged by a constant current) has reached Vr - 0.5V. Css (Vr - 0.5V) tstart-up = I5so 11/43 APPLICATION NOTE cult to establish an exact maximum number of devices, as it depends on different conditions. The first consideration concerns the accuracy which must be achieved and maintained on the oscillation frequency.Since thebias current on pin 7 is anoutput current, thesum of all the biascurrents must be much smaller than the capacitor discharge current in close proximity to thelower dischargethreshold.Therefore, assuming Cosc = 2.2nF and Rosc = 4.3K, it follows that : 1.2V = 280A 4.3K Assuming that a 10% variation may be accepted, it follows therefore that the number of synchronizable devices is given by : 28A N= Ibias max This means that if the overall Ibias is too high it may modify the discharging time of the capacitor. The second consideration concerns the layout design. In the presence of a great number of devices to be synchronized, the lenght of the paths may become significant and therefore the distributed inductance introduced along the paths may begin to modify the triangular shaped waveform, particularly the rising edgewhichis very steep.This effectwould affectthe devicesthat arephysicallylocatedmoredistantfrom the master device. Theamplitude of the saw-toothto be externallyconnected must be with in 0.5V and 3.5V, values also representingthe maximum swing of the error amplifier output. CURRENT LIMITATION The current limitation function has been realized in a rather innovative way to avoid overload condition during the short circuit operation.In fact, while for all the other devices a constantcurrent limitation is implemented by acting on the duty cycle (therefore, in short circuit conditions an output current is equal to themaximum limitation current),the new control approach allows operation in short circuit conditions with a mean current much smaller than the allowed 4A value.Operationof the currentlimiter will now be described. Refer to the block diagram, fig. 13. The current which is delivered from the output transistor to the load flows through the current sensing resistor RS. When the voltage drop on RS is equal to the offset voltage of the current comparator, the comparator generates a set pulse for the flip-flop, with a delay of about1sec. The purpose of this delay is to avoid triggering of the protection circuit on thecurrent peak that occurs during the recirculation phase.Therefore,the outputQ goeslow and the power stageis immediately switched off, while the output Q goes high and acts directly on the soft-start capacitor dischargng the soft-start capacitor at a constant current (about 50A). When the voltage on pin 5 reaches 0.4V the comparator triggers, supplying a reset pulse to the flipflop ; from now on, the power stage is enable and thesoft-startphase starts again.When the limitation cause, either overload or short circuit, is still present thecycle repeatsagain. Thewaveformof the output current on pin 2 is shown in fig. 14. Figure 12 : In multiple supplies several l296's can be synchronized as shown here. 12/43 APPLICATION NOTE From fig. 14 it may be observedhow this current limitation technique allows the short circuit operation with a very low output current value. It is possible to reduce the maximum current value by actingon pin 4. Onthispin a voltageof about3.3V is present ; by connecting a resistance a constant current, given by 3.3/R, is sent to ground. This current reduces the offset voltage of the current comparator, therefore anticipating its triggering threshold. Figure 13 : Partial Schematic Showing the Current Limiter Circuit. Figure 14a : Current Limiter Waveforms. 13/43 APPLICATION NOTE Figure 14b : Load Current in Short Circuit Conditions (Vi = 40V, L = 300H, f = 100K). the limits required to supplythe microprocessor correctly. The reset function is realized through the use of 3 pins : the reset input pin 12, the reset delay pin 13 and the reset outputpin 14. Whenthe voltageon pin 12 is smaller than 5V the comparator output is high and the reset capacitor is not charged because the transistor Q is saturated and the voltage on pin 14 is at low level, since Q2 is saturated, too. When the voltage on pin 12 goes above 5V, the transistor Q switches OFF and the capacitor can start to charge through a current generator of about 100A. When the voltage on pin 13 goes above 4.5V the output of the related comparator switches low and the pin 14 goes high. As the output consists of an open collector transistor, a pull-up external resis-tance is required. In contrast, when the reset input voltage goes below5V, lessa hysteresisvoltageof about100mV, the comparatortriggers again and instantaneously sets the voltage on pin 14 low, therefore forcing to saturation the Q1 transistor, that starts the rapid discharge of the capacitor. Obviously, the reset delay is again present when the voltage on pin 13 is allowed to go under 4.5V. To achieve switching operations without uncertainties the two comparators have been provided with an hysteresis of about 100mV. In every operating condition the reset switching is guaranteed with a minimum reset input of 4.75V,the value requiredfor correct operation of the microprocessor even in the presence of the minimum VREF value. Normally pin 12 is used connected to pin 10. When it is connected to the output, the function may be more properly called "reset" ; on the other hand, when it is connectedthroughresistive divider, to the input voltage, the function is called "power fail". Fig. 16 and fig. 17 show the two possible usages. The"power-fail" functionis used to predict,with a given advance, the drop of the regulator output voltage, due to main failures, which is enough to save the data being processed into protected memory areas. Fig. 18 summarizes the reset function operation. Figure 14c : Current at Pin 2 when the Output is Short Circuited. RESET The reset function is of great importance when the device is used to supplymicroprocessors, logic devices, and so on. This function differentiates the L296 device from all previous devices. The block diagram of the function is shown in fig. 15. A reset signalis generatedwhen the outputvoltageis within 14/43 APPLICATION NOTE Figure 15 : Partial Schematic Showing Reset Circuit. Figure 16 : For Power - On reset the reset block is connected as shown here. Figure 17 : To obtain a power fail signal, the reset block is connected like this. 15/43 APPLICATION NOTE Figure 18 : Waveform of the Reset Circuit. CROWBAR This protection function is realized by a completely independent block, using pin 1 as input and pin 15 as output. It is used to prevent dangerous over voltagesfrom occurringwhen the outputexceeds 20% of rated value. Pin 15 is able to output a 100mAcurrent to besent to the gate of a SCRwhich,triggering, short circuits either output or the input. When connected to the input, as the SCR is triggered a fuse in series connectedto power supply is blown and to bring the system back to operationmanual intervention is requested. Figs. 19, 20 and 21 show the different configurations. When the voltage on pin 1 exceeds by about 20% the VREF value the output stage is activated, which sends a current to the SCR gate, after a delay of about 5sec to make the system insensitive to low duration spikes. When activated, the output stage delivers about100mA ; when not activated,it drains about 5mA and shows a low impedanceto the SCR gate to avoid incorrect triggering due to random noise. If the crowbar function is not used connect pin 1 to ground. Figure 19 : Connection of Crowbar Circuit at Output for 5.1V Output Applications. 16/43 APPLICATION NOTE Figure 20 : Connection of Crowbar Circuit at Output for Output Voltages above 5.1V. Figure 21 : Connection of Crowbar Circuit to Protect Input. When triggered, the scr blows the fuse. INHIBIT The inhibit input (pin 6) is TTL compatible and is activated when the voltage exceeds 2V and deactivated when the voltage goes under 0.8V. As may be seen in the block diagram, the inhibit acts on the power transistor, instantaneously switching it off and also acts on the soft-start, discharging its capacitor. When the function is unused, pin 6 must be grounded. THERMAL PROTECTION The thermal protection function operates when the junction temperaturereaches 150C ; it acts directly on the powerstage,immediately switchingit off, and on the soft-start capacitor, discharging it. The thermal protection is provided with hysteresis and, therefore, after an intervention has occurred, it is necessary to wait for the junction temperatureto decrease of about 30C below the intervention threshold. APPLICATIONS Though the L296 is designed for step-downregulator configurationsit may be usedin a variety of other applications. We will now examine these possibilitiesand showhow the capabilitiesof the device may be extended. In fig. 22 the complete typical application is shown, where all the functions available on the device are being used. This circuit delivers to the load a maximum current of 4A and a voltage which is established by the voltagedivider constitutedby R7 andR8 resistances.The following tableis helpfulfor a quick calculation of some standard output voltages : Resistor Value for Standard Output Voltages Vo 12 15 18 24 V V V V R8 4.7 4.7 4.7 4.7 k k k k R7 6.2 k 9.1 k 12 k 18 k To obtain Vo = VREF thepin 10 is directly connected to the output, therefore eliminating both R7 and R8. The switching frequency is 100kHz. 17/43 APPLICATION NOTE Figure 22 : Schematic, PCB Layout and Suggested Component Values for the Evaluation Circuit used to characterizethe L296. This is a typical stepdown application which exercises all the device's functions. C7, C8 : EKR (ROE) SUGGESTED INDUCTOR (L1) Core Type Magnetics 58930 A2MPP Thomson GUP 20 x 16 x 7 No Turns 43 65 Wire Gauge 1.0 mm. 0.8 mm. 1 mm. Air Gap SUGGESTED INDUCTOR (L1) (continued) Core Type Siemens EC 35/17/10 (B6633 & - G0500 - x 127) No Turns 40 Wire Gauge 2 x 0.8 mm. VOGT 250 H Toroidal Coil, Part Number 5730501800 18/43 APPLICATION NOTE Figure 23 : Oscilloscope PhotographsShowing Main Waveform of the Figure 22 Circuit. 100F/40V capacitors have been connected in parallel. The behaviourof the impedance as a function of frequency is shown in fig. 24. Alsothe selectionof the catchdioderequires special care. The best choice is a Schottkydiode which minimizes the losses because of its smaller forward voltage drop and greater switching frequency rate. A possible limitation comes from the backward voltage, that generally reaches 40V max. When the full input voltage range of the device is requiredin thisapplicationit ispossibleto usesuperfast siodes with 35 to 50ns rated recovery time, where no more problemsonthe backwardvoltageoccur(onthe otherhand,they showa greaterforwardvoltage).The use of slower diodes, with trr = 100ns or more is not recommended ; The photographsin fig. 25 show the effectsonthe powercurrent andon thevoltageonpin 2,due to thediodesshowing differentspeeds.Diodes showing trr greaterthan 35-50ns will reduce the overall efficiency of the system, increasing the powerdissipated by the device. The third component requiring care is the inductor. Fig. 22a shows the part numbers of some types used for testing. Besides havingthe required inductance value, the coil has to show a very high saturation current. Therefore, a correct dimensioning requires a saturation current above the maximum value of I2L, the current limit threshold. To achieve high saturation with ferrite cores an air gap between the two core halves must be provided ; the air gap causes a leakage flux which is radiated in the surroundingspace. To better limit this phenomenon "pot cores" may be used, whose geometry is such to better limit the flux radiated to the outside. Using toroidal cores, for instanceof Magnetic 58930A2moly-permalloy kind,boththerequirementsofhigh saturationand low leakage flux are satisfied. The saturation is softer that the saturation shown by the ferrite materials. The air gap is not concentrated in one area,butis finely distributedalongthe wholecore ; this gives the low leakage flux value. Careful selection of the external componentstherefore allows the realization of a powersupply system whose benefits are significant when compared to a system withthe same performancebut realized with the linear technique. The oscilloscope photographs of the main waveforms are shown in fig. 23. The outputvoltageripple Vo dependson the current ripple in the coil and on the performance of the output capacitor at the switching frequency (100kHz). A capacitor suitable for this kind of application must have a low ESR and be able to accept a high current ripple, at the working frequency.For this applicationthe RoedersteinEKR series capacitorshave been selected,desi-gnedfor high frequency applications (> 200kHz) and manufactured to show low ESR value and to accept high current ripples. To minimize the effects of ESR, two 19/43 APPLICATION NOTE Figure 24 : Typical Impedance/Frequencycurves for EKR Capacitors. Figure 25 : Oscilloscope Photographs Showing the Waveform obtained with Diodes having Different trr Values. LOW COST APPLICATION AND PREREGULATOR Fig. 26 shows the low cost application of a 4A and Vo = 5.1V power supply. A minimum amount of essential external components is required, which are necessary for correct operation. It is impossible to save other components, specially the soft-start capacitor. Without soft-start, the system cannot reach the steady state and there is also a serious risk of damaging the device. This application is very well suited not only as a lowcostpowersupply,but alsoas pre-regulatorfor postregulators distributed in different circuit points, or even on different boards (fig. 27). The post-regulators may be selectedamongthe low-drop types, like L4805 and L387 for example, still obtaining a high efficiency, combined with an excellent regulation. The use of L387 device allows us to usealso the reset function, useful to power a microprocessor. 20/43 APPLICATION NOTE SWITCHING vs LINEAR Switching regulators are more efficient than linear types so the transformer and heatsink can be smaller and cheaper. But how much can you gain ? We can estimate the savings by comparing equivalent linear and switching regulators. For example, suppose that we want a 4 A/5 V supply. Linear For a goodlinearregulator the minimum dropoutwill be at least 5V at 4A. The minimum input voltage is given by : 1 Vripple Vi min = Vo + Vdrop + 2 where : Vripple 4 x 8 x 10 - 3 Io t1 = C 10 x 10 - 3 = 3.2 V (a good approximation is 8ms for t1 at mains frequency of 50Hz and 10.000F for C, the filter capacitor after the bridge). Therefore Vimin 1.6V. Since operationmustbe guaranteedevenwhen the mains voltagefalls 20%,the nominal voltageon load at the terminals of the regulator must be : 10.6 Vi min Vnom = = = 13.25V 0.8 0.8 To allow even a small margin we have to choose : Vnom = 14V The power that the series element must dissipate is therefore : Pd = (Vnom Vo) Io = 36W and a heatsink will be necessary with a thermal resistance of : Rth heats. = 0.8C/W and the transformer must supply a power of : Pdiss = 14 x 4 = 56W It must therefore be dimensioned for : 56 PD = = 62VA 0.9 Switching (L296) Assuming the same nominal voltage (14V), the L296 data sheet indicates that the power dissipated in this case is only 7W. And this power is dissipated in two elements ; the L296 itself and the recirculation diode. It follows that the transformermust be roughly30VA and the heatsink thermal resistanceabout 11C/W. +Linear Transformer Heat sink 62 VA 0.8 /W +Switching 30 VA 11 /W Thiscomparison shows thatthe L296switching regulatorallowsa saving of roughly 50% onthe cost ofthe transformerand an impressive 80-90%on the cost of the heatsink. Considering also the extra functions integratedbytheL296thetotalcostofactiveandpassive componentsis roughly the same for both types. Finally, it is important to note that a lower power dissipation means that the ambient temperature in the regulatorenclosure can be lower - particularly when the circuit is enclosed in a box - with all the advantages cooler operation brings. 21/43 APPLICATION NOTE Figure 26 : A Minimal Component Count 5.1V / 4A Supply. Figure 27 : The L296 may also be used as a preregulatorin distributed supply systems. (*) L2 and C2 are necessary to reduce the switching frequency spikes. 22/43 APPLICATION NOTE If for some reason it is necessaryto use higher supply voltages the switching technique,and hence the L296, becomes even more advantageous. POWER SUPPLY COMPLETE WITH TRANSFORMER Fig. 28 shows a power supply complete of transformer, bridge and filter, with regulation on the output voltage from 5.1V to 15V. As alreadystatedabove,the outputcapacitors have to show some speciale features, like low ESR and high current ripple, to obtain low voltage ripple values and high reliability. The input filter capacitors must not be neglected because they have to show excellent features, too, having to supply a pulsed current, required by the device at the switching frequency. The current ripple is rather high, greater than the load current. For this application, two parallel connected3300F/50VEYF (ROE)capacitors have been used. POWER SUPPLY WITH MAINS SWITCHING PREREGULATOR When it is desirable to eliminate the 50/60Hz transformer - in portable or volume-limited equipment-a mains preregulatorcan be addedto reducethe input voltage to a level acceptable for the L296. In this case the pre-regulator circuit is connectedto the primary of the transformer which now operates at the switching frequency and is therefore smaller and lighter. Using a UC3840 which includes the feed-forward functionit is possible to compensatemains variation within wide limits. The secondary voltage is therefore only affected by load variations. Using one or more L296s as postregulators,feedback to the primary is no longer necessary, reduces the complexity and cost of the transformer which needs only a single secondary winding. Fig. 28A shows a multi-output supply with a mains preregulator. Figure 28 : A Typical Variable Supply showing the Mains Transformer. Vo = 5.1 to 15V Io = 4A max. (min. load current = 100mA) ripple 20mV load regulation (1A to 4A) = 10mV (Vo = 5.1V) line regulation (200V 15% and to Io = 3A) = 15mV (Vo = 5.1V) 23/43 APPLICATION NOTE Figure 28A: A Multiple Output Supply using a Switching Preregulator rather than a Mains Trans- POWER SUPPLY WITH 0 - 30V ADJUSTABLE VOLTAGE When output voltages lower than 5V are required, the circuit shown in fig. 29 may be used. Calibration is performed by groundingthe P1 slider. Acting on P2, the current which flows through the 10kW resistor is fixed approximately 2.5mA to obtain an output voltage of 30V. The equivalent circuit is shown in fig. 30. Acting now on the slider of P1, the current flowing throughthe divider may be varied. The new equivalent circuit is shown in fig. 31. Reducing the current flowing, also the voltage drop across the 10k resistance is reduced, together with VO. When the current reaches zero, it follows that VO = VREF. When the voltageon the slider of P1 exceeds VREF, the current start to flow in opposite direction and VO begins to decrease below 5V. When I 1 x 10k = VREF it follows that VO = 0. DUAL OUTPUT REGULATOR The application shown in fig. 32 is specially interesting because it provides two output voltages. The first voltage, the main one, is directly controlled by the feedbackcircuit. The secondvoltageis obtained through an auxiliary winding. It often happens, when microprocessors, logic devices etc., have to be power supplied, that a main 5V output and an auxiliary +12V or -12V output are required, the latter with lower current requirements (100 to 200mA)and a stabilization level not excessively high.As the auxiliary power supply isobtained througha completely separatedwinding, it is possible to obtain either a positive or negative voltage (compared to the main voltage or also a completely isolated voltage. With Vi variable between 20V and 40V, VO = 5.1V and IO = 2.5A, the auxiliary 12V/0.2Avoltage is within a 2% tolerance. 24/43 APPLICATION NOTE Figure 29: Variable 0-30V supply illustrating how output voltages below 5.1V are obtained. Figure 30: When setting up the figure 29 circuit the slider of P1 is grounded, giving the equivalent circuit shown here, and P2 adjusted to give an output voltage of 30V. Figure 31: Partial Schematic showing Output Voltage Adjustment of Figure 29. 25/43 APPLICATION NOTE Figure 32: Dual output regulator showing how an additional winding can be added to the inductor to generate a secondary output. PERSONAL COMPUTER POWER SUPPLY Using two mutually synchronized devices it is possible to obtain a four output power supply suitable for power a microprocessor system. V01 = 5.1V/4A V02 = 12V/2.5A (up to 4A) V03 = -5V/0.2A V04 = -12V/0.2A The schematic diagram is shown in fig. 33. The 5V output is also provided with the reset function, that is available also for the 12V output. 26/43 APPLICATION NOTE Figure 33: Microcomputer Supply with 5V, -5V, 12V and -12V Outputs. 27/43 APPLICATION NOTE Figure 34: Battery charger circuit illustrating how the device is used to regulate the output current. The feedbackis direct,no other externalcomponent is used and no calibration is therefore required. An outputisobtainedwith the accuracyof the reference voltage (2%). For the 12V output, by using a resistive divider with 1% resistance an output is obtained whose spread is within 4%. The two devices are mutually synchronized not to give rise to intermodulation which could generate unpleasant noise and, at the same time, a further component saving is achieved. The crowbar function is implemented on both 5V and 12V outputs, using a single SCR connected to the input. The latter, by discharging to ground the electrolytic filter capacitors, blows the fuse connected in series with the devices power supply. In this way, should a faulty be present on either of the main outputs,the supply is switched off for whole system. To inhibit both the devices with a single input signal, it is possible to connect the two inhibit inputs (pin 6) together;the5k resistanceis used when theinhibit input is left open. If this input is not used it must be grounded. As may be noted in the diagram, to obtain the two auxiliary voltages is very simple and cost-effective. It is suggested that the diodes are fast types (trr < 50ns); should slowerdiodes be requiredsome more turns have to be added to the auxiliary winding. BATTERY CHARGER When the device has to be used as current generator it is necessaryto avoid the internal current limiter is operated fig. 34 shows the circuit realizing constant current limitation. In this way it is possible to obtain a 6V, 12V and 24V battery charger. For each of these voltages a max. current of 4A is available, which is large enough for batteries up to 4045Ah (for 12V type). With reference to the electric diagram through the 2k potentiometer the max. outputcurrent is set, while throughthe R1 - R2 output divider the voltage is set. (R1 may be replaced by either a potentiometer or a 3 position switch, to directly obtain the three 6V, 12V and 24V voltages). HIGHER INPUT VOLTAGE Since a maximum input voltage of 46V (operating value) may be applied to the device the diagram shown in fig. 35 may be used when it is necessary to exceed this limit. This system is particularly useful when operating at low output voltages.In this casea mean current IiDC which has a low value when compared to IO is obtained. In fact, since Vo = Vi (TON/T) and Vo Io = Vi IiDC (assuming the device has an ideal efficiency), it follows that IiDC = IO (TON/T). 28/43 APPLICATION NOTE Assuming to be: Vo = 5V Io = 4A and V3 37V it follows that: 5 TON/T = Vo / Vi = = 0.135 37 IiDC = 4 x 0.135 = 0.54A. With input voltage50V and Io = 4A,the externaltransistor dissipates about 7W. High good efficiency is still achieved, around 74%; in the real case, considering also the device losses, an efficiency around 62% is achieved. During output short circuits the externaltransistor is not overloaded because in this condition IiDC reduces to values lower than 100mA. It is not possible to realize this application with series post-regulator because the efficiency would be unacceptablelow. MOTOR CONTROL The L296 is also suitable for use in motor controls applications. Fig. 36 shows how to use the device to drive a motor with a maximum power of about 100W and provided with a tachometergeneratorfor a good speed control. Figure 35: The maximum input voltage can be raised above 46V by adding a transistor as shown here. Figure 36: With a tacho dynamo supplying feedback the L296 can be used as a motor speed controller. 29/43 APPLICATION NOTE HIGHER CURRENT REGULATORS It is possible to increase the output current to the load above 4A through the use of an external power transistor. Fig 37 shows a suitable circuit. The frequency is around 40kHz to prevent the device from loosing excessive power due to switching on the external power. The circuits shown in fig. 38 and fig. 39 show how current limitation may be realized in two different ways:througha sensingresistorconnectedin series with the collector of the external power transistor or through a current transformer. In the first case, the sensing resistor is a low value resistor able to withstand the maximum load current required. The VCE of the power transistor is higher than its VCEsat; when the resistor is connected in series to the collector VCE is reduced; consequently since the overall dissipated power is constant, the power dissipated by the sensing resistor is subtracted from that dissipated by the power transistor. The values indicated in figs. 38 and 39 realize adjustable current limitation for load currents around 10A. Figure 37: The output current may be increased by adding a power transistor as shown in this circuit. 30/43 APPLICATION NOTE Figure 38: This circuit shows how current limiting for the external transistor is obtained with a sensing resistor. Figure 39: A small transformer is used in this example for current limiting. 31/43 APPLICATION NOTE Figure 40: A step-up Converter using a Power MOS Transistors. STEP-UP CONVERTER With the L296it isalsoeasyto realize a step-upconverter, by using a MOS power transistor. Fig. 40 shows the electric diagram of the step-upconverter. The frequency is 100kHz, operation is in discontinuous mode and the device internal current limiter is used. Thereforeno other external protectionis required. The input voltage could be a 12V car battery, from which an output voltage of 35V may be obtained. Lower output voltage of 35V may be obtained. Lower outputvoltage valuesmay be obtained by reducing the value of R7. DESCRIPTION OF OPERATION Fig. 41 shows the diagramof the circuit realizing the step-up configuration. When the transistor Q1 is ON, the inductance L charges itself with a current given by: Vi iL = t L The peak current in the coil is: Vi Ipeak = TON L In this configuration, unlike the step-down configuration, the peak current is not strictly related to the load current. The energystoredin thecoil is successively discharged across the load when the transi- Figure 41: Basic Schematic for Step-up Configurations. stor switches OFF. To calculate the Io load current, the following procedure may be used: 12 L I peak = Vo Io T 2 Io = L I2peak Vi TON = 2 Vo T 2 L Vo T For a greater output power to be available, the internallimitation must be replacedby an externalcircuit to protect the externalpower devicesand to limit the current peak to a convenientvalue. A dual comparator(LM393) with hysteresis is used to avoid uncertaintes when the current limitation operates. The electric diagram is shown in fig. 42. 2 2 32/43 APPLICATION NOTE Figure 42: High power step-up converter showing how the current limiting function is realized externally. 33/43 APPLICATION NOTE LAYOUT CONSIDERATIONS Both for linear and switching power supplies when the current exceeds 1A a careful layout becomes importanttoachieve a goodregulation.Theproblem becomes more evident when designing switching regulators in which pulsed currents are over imposed on dc currents. In drawing the layout, therefore, special care has to be taken to separate ground paths for signal currents and ground paths for load currents, which generally show a much higher value. When operating at high frequencies the path length becomes extremely important. The paths introduce distributed inductances, producing ringing phenomena and radiating noise into the surrounding space. The recirculation diode must be connected close to pin2, to avoid giving rise to dangerous extra negative voltages, due to the distributed inductance. Fig. 43 and fig. 44 respectivelyshow the electricdiagramand the associatedlayout whichhas beenrealized taking these problems into account. Greater care must be taken to follow these rules when two or more mutually synchronized devices are used. Figure 43: Typical application circuit showing how the signal and power grounds are connected. SUGGESTED INDUCTOR (L1) Core Type Magnetics 58930 A2MPP Thomson GUP 20 x 16 x 7 Siemens EC 35/17/10 (B6633 & - G0500 x 127) No. Turns 43 65 40 Wire Gauge 1.0mm. 0.8mm. 2 x 0.8mm. 1mm. Air Gap Resistor Value for Standard Output Voltages Vo 12V 15V 18V 24V R8 4.7k 4.7k 4.7k 4.7k R7 6.2kW 9.1k 12k 18k VOGT 250H Toroidal Coil, Part Number 5730501800 34/43 APPLICATION NOTE Figure 44: A Suitable PCB Layout for the Figure 43 Circuit realized in Accordance with the Suggestions in the Text (1:1 scale). HEATSINK DIMENSIONING The heatsink dissipates the heat produced by the device to prevent the internaltemperature from reacing values which could be dangerous for device operation and reliability. Integratedcircuits in plastic packagemust neverexceed 150C even in worst conditions. This limit has been set because the encapsulating resin has problems of vitrification if subjected to temperatures of more than 150C for long periods or of more than 170C for short periods. In any case the temperature accelerates the ageing process and therefore influences the device life; an increase of 10C can halve the device life. A well designed heatsink shouldkeepthe junctiontemperaturebetween90C and 110C. Fig. 45 shows the structure of a power device. As demonstrated in thermo-dynamics, a thermal circuit can be considered to be an electrical circuit where R1, R2 representthe thermalresistance ofthe elements(expressedin C/W) (seefig. 46). C1,C2 I V are the thermal capacitance (expressed in C/W) is the dissipated power is the temperature difference with respect to the reference (ground) CC Ch Rjc Rh is the thermal capacitance of the die plus that of the tab. is the thermal capacitance of the heatsink is the junction case thermal resistance is the heatsink thermal resistance Figure 45. Figure 46. This circuit can be simplified as shown in fig. 47, where: 35/43 APPLICATION NOTE Figure 47. wise the dissipator can no longer be considered as a concentrated constant and the calculation becomes difficult. This concept is better explained by the graph in fig. 51 which shows the case (and therefore junction) temperature variation as a function of the distance between two dissipating elements with the same type of heatsink and the same dissipated power. Th graph in Fig. 51 refers to the specific case of two elements dissipating the same power, fixed on a rectangular aluminium plate with a ratio of 3 between the two sides. The temperature jump will depend on the total dissipatedpower and on the total dissipatedpower and on the devices geometrical positions. We want to show that there exists an optimal position between the two devices: Figure 49. If we now consider the ground potentialas ambient temperature, we have: Tj = T a + (Rjc + Rh)Pd a) Tj - Ta - Rjc Pd Rh = b) Pd c) Tc = Ta + Rh Pd Thermal contactresistancedependson various factors such as the mounting, contact area and planarity of the heatsink. With no material between thedevice and heatsink the thermal resistance is around 0.5C/W; with silicone grease roughly 0.3C/W and with silicone grease plus a mica insulator about 0.4C/W. See fig. 49. In application where one external transistor is used together, the dissipated power must be calculated for each component.The various junction temperature can be calculated by solving the circuit shown in fig. 50. This applies if the dissipating elementsare fairly close with respect to the dissipator dimensions, other- But since the aim of this section is not that of studing the transistors, the circuit can be furtherreduced as shown in figure 48. Figure 48. Figure 50. 1 side of the plate. 2 Fig. 52 shows the trend of the temperature as a function of the distance between two dissipating eled= 36/43 APPLICATION NOTE Figure 51. Figure 52. 37/43 APPLICATION NOTE APPENDIX A CALCULATING SYSTEM STABILITY This section is intended to help the designer in the calculation of the stability of the whole system. Figure A1 shows the entire control system of the switching regulator. The problem which arises immediately is the transfer function of the PWM block and output stage, which is non-linear. If this function can be considered linear the analysis is greatly simplified. Since the circuit operates at a constant frequency and the internal logic is fairly fast, the error introduced by assuming that this function linear is minimized. Factors which could contribute to the non-linearity are an excessive delay in the output power transistor, ringing and parasitic oscillations generated in the power stage and non-linearity introduced by magnetic part. In the case of the L296,in which the powertransistor is internaland driven by well-controlled and efficient logic, the contributionto non-linearityis furtherreduced. For the assumptionof linearity to be valid the cut-off frequency of the LC filter must be much lower than the switching frequency.In fact, switching operation introduces singularities (poles) at rougly half the switching frequency. Consequently, as long as the LC filter is still dominant, its cut-off frequency must be at least an order of magnitude lower than the switching frequency.This condition is not, however, diffucultto respect.Thecharacteristicsof LC filter affect the output voltage waveforms; is generally much less than an order of magnitude below the switching frequency. Figure A1: The control Loop of the Switching Regulator. GAIN OF THE PWM BLOCK AND OUTPUT STAGE The equation which links Vo to Vi is: TON Vo = Vi T A variation TON in the conduction time of the switching transistor causesa correspondingvariation in the output voltage, Vo, giving: Vi Vo = TON T Indicatingwith Vr the outputvoltage of the error amplifier,and with Vct theamplitudeof the ramp (thedifference between the maximum and minimum values), TON is zero when Vr is at the minimum values), TON is zero when Vr is at the minimum value and equal to T when Vr is at a maximum. Consequently: TON T = Vct Vr The gain is given by: 38/43 Vo Vi = Vr Vct Since Vct isabsolutelyconstantthe gain of the PWM block is directly proportionalto the supplyvoltageVi. Figure A2: Open Loop Frequency and Phase Response of Error Amplifier APPLICATION NOTE The error amplifier is a transconductanceamplifier (it trasforms a voltage variation at the input into a current variation at the output). It is used in open loop configuration inside the main control loop and its gain and frequency response are determined by a compensationnetworkconnectedbetweenits output and ground. In the applicationa series RC networkis recommended which gives high system gain at low frequency to ensure good precision and mains ripple rejection and a lower gain at high frequencies to ensure stability of the system. Figure A2 shows the gain and phase curves of the uncompensatederror amplifier. The amplifier hasone poleat about7kHzand a phase shift which reaches about -90 at frequencies around 1MHz. The introduction of a series network Rc Cc between the outputand ground modifies the circuit as shown in figure A3. Figure A4 shows the gain and phase curves of the compensatederror amplifier. Figure A3: Compensation Network of the Error Amplifier CALCULATING THE STABILITY For the stability calculation refer to the block diagram shown in figure A5. The transfer functions of the various blocks are rewritten as follows. The simplified transfer function of the compensated error amplifier is: 1 + s Rc Cc 1 gm = 2500 s Cc The DC gain must be considered equal to: Ao = gm Ro PWM block and output stage: Vi GPWM = Vct GEA = gm Zc = LC FILTER: GLC = 1 + s C ESR s LC + s C ESR + 1 2 Where ESRis the equivalentseries resistanceof the outputcapacitor which introducesa zero at high frequencies, indispensablefor system stability. Such a filter introduces two poles at the angular frequency. 1 = LC Refer to the literature for a more detailed analysis. Feedback: consists of the block labelled = 1 when Vo = VREF (and therefore Vo = 5.1V) and R2 = when Vo > VREF R1 + R2 Figure A4: Bode Plot Showing Gain and Phase of Compensated Error Amplifier. Figure A5: Block Diagram Used in Stability Calculation. 39/43 APPLICATION NOTE To analysethe stability we will use a Bode diagram. The values of L and C necessary to obtain the required regulator output performance, once the frequency is fixed, are calculated with the following formulae: L= C= (Vi - Vo) Vo Vi f IL (Vi - Vo) Vo Figure A7: Bode Plot of Complete System Taking into Consideration the Equivalent Series Resistance of the Output Capacitor. 8L f2 IL Since this filter introduces two poles at the angular frequency 1 = LC we place the zero of the Rc Cc network in the same place: 1 z = Rc Cc Taking into accountalso the gain of the PWM block, the Bode plot of figure A6 is obtained. The slope where the curve crosses the axis at 0dB is about 40dB/decadetherefore the circuit is unstable. Taking into account now the zero introducedby the equivalent series resistance (ESR) of the outputcapacitor, we have further condition for dimensioning the Rc Cc network. Knowing the ESR (which is supplied by the manufacturer for the quality components) we can determine the value of Rc so that the axis is crossed at 0dB with a single slope. The zero introduced by the ESR is at the angular frequency: 1 zESR = ESR C The overall Bode diagram is therefore as shown in figure A7. Figure A6: Bode Plot of System Taking Filter and Compensation Network into Account. DC GAIN AND LINE REGULATION Indicating the open-loop gain of the error amplifier with Ao, the overall open-loop gain of the system is: Vi R2 At = Ao Vct R1 +R2 When Vo = VREF, the gain becomes: Vi At = Ao Vct Considering the block diagram of figure A8 and calculating the output variation Vo caused by a variation of Vi, from the literature we obtain: Vo = Vo Ao Vi R2 Vct R1 + R2 This espressionis of generalvalidity. In ourcase the percentage variation of the reference must be added by vector addition. Figure A8: Block Diagram for Calculation of Line Regulation. Vi Vi 40/43 APPLICATION NOTE APPENDIX B REDUCING INTERFERENCE The main disadvantage of the switching technique is the generationof interferencewhich can reach levels which cause malfunctions and interfere with other equipment. For each applicationit is thereforenecessary to study specific means to reduce this interference within the limits allowed by the appropriate standards. Among the main sources of noise are the parasitic inductances and capacitances within the system which are charged and discharged fastly. Parasitic capacitances originate mainly between the device case and the heatsink, the windings of the inductor and the connectionwires. Parasitic inductancesare generally found distributed along the strips of the printed circuit board. Fast switching of the power transistors tendsto cause ringing and oscillations as a result of the parasitic elements. The use of a diode with a fast reverse recovery time (trr) contributes to a reductionin the noise flowing by the current peak generated when the diode is reverse biased. Radiated interference is usually reduced by enclosing the regulator in a metal box. To reduce conducted electromagnetic interference (or radio frequencyinterferences- RFI) to the levels permitted a suitably dimensioned filter is added on the supply line. The best method, generally, to reduce conductednoise is to filter each output terminal of the regulator. The use of a fixed switching frequency allow the use of a filter with a relatively narrowbandwidth. For off-line switchingeregulators this filter is usuallycostly and bulky. In contrast,if the device is supplied from a 50/60Hz transformer the RFI filter problem is gratly reduced. Testshave beencarried out the laboratoriesof Roederstein to determine the dimensions of a mains supply filter which satisfies the VDE 0871/6.78, class B standard. The measurements (see figs. B1 and B2) refer to the application with the L296 supplied with a filtered secondary voltageof about 30V, with Vo = 5.1V and Io = 4A. The switching frequency is 100kHz. Figure B1 shows the results obtainedby introducing on the transformer primary a 0.01F/250V ~ class X capacitor (type ERO F1753-210-124). To reduce interference furtherbelow the limit set by the standards an additional inductive filter must be added on the primary of the transformer. Figure B2 shows the curves obtainedby introducing this inductive filter (type ERO F1753-210-124). Measurements have also been performed beyond 30MHz; the maximum value measured is still well below the limit curve. Figure B1: EMI Measurements with a Capacitor Connected across the Primary Transformer with Screen Grounded (A) 41/43 APPLICATION NOTE Figure B2: EMI results with the addition of an inductive filter on the mains input. 42/43 APPLICATION NOTE Information furnished is believed to be accurate and reliable. However, SGS-THOMSON Microelectronics assumes no responsibility for the consequences of use of such information nor for any infringement of patents or other rights of third parties which may result from its use. No license is granted by implication or otherwise under any patent or patent rights of SGS-THOMSON Microelectronics. Specification mentioned in this publication are subject to change without notice. This publication supersedes and replaces all information previously supplied. SGS-THOMSON Microelectronics products are not authorized for use as critical components in life support devices or systems without express written approval of SGS-THOMSON Microelectronics. (c) 1996 SGS-THOMSON Microelectronics - Printed in Italy - All Rights Reserved SGS-THOMSON Microelectronics GROUP OF COMPANIES Australia - Brazil - Canada - China - France - Germany - Hong Kong - Italy - Japan - Korea - Malaysia - Malta - Morocco The Netherlands - Singapore - Spain - Sweden - Switzerland - Taiwan - Thailand - United Kingdom - U.S.A. 43/43 |
Price & Availability of AN244
 |
|
|
|
|
All Rights Reserved © IC-ON-LINE 2003 - 2022 |
| [Add Bookmark] [Contact Us] [Link exchange] [Privacy policy] |
|
Mirror Sites : [www.datasheet.hk]
[www.maxim4u.com] [www.ic-on-line.cn]
[www.ic-on-line.com] [www.ic-on-line.net]
[www.alldatasheet.com.cn]
[www.gdcy.com]
[www.gdcy.net] |