High Voltage Switching Power Supply Topologies and Power

Sep 22, 2004 - output voltage, basic non-isolated topologies can be used. .... converter and utilize the power transformer of the forward converter to do some of ...
116KB taille 57 téléchargements 128 vues
High Voltage Switching Power Supply Topologies and Power Management ICs

INTRODUCTION

MODERATOR: [slide 1] Hello and welcome to today's National Semiconductor Online Seminar. I’m Michelle Miller and I will be your host. Before we begin I would like to go over the operation of the seminar interface. Slides will appear in the upper right section of your interface. If you would like the slides to be larger, click the “Enlarge ” button. Slides will automatically advance. At the bottom of your interface is an interactive Web browser set to a Web page containing additional resources for this seminar. Questions may be submitted at any time and the presenter will respond via email. To ask a question, enter your question in the “Ask a Question” entry window, then click “Submit” to send the question to the presenter. Today’s topic is High Voltage Switching Power Supply Topologies and Power Management ICs. Today’s seminar will be given by Senior Applications Engineer, Haachitaba Mweene, of the Power Applications Design Center. Welcome Haachitaba. HAACHITABA: Thank you Michelle. Today I will be addressing appropriate topologies for switching power supplies used in a variety of systems with high input voltages. I will justify the choice of topologies for each application and give examples of power management ICs made by National Semiconductor that ease the practical implementation of the selected topology. SINGLE-ENDED

kkk TOPOLOGIES

[slide 2] Every electrical system needs a power supply to connect an AC or DC voltage at one level to the level that the system requires to function properly. In this case we are dealing with systems where the input voltage is high, and high voltage, for the purposes of this seminar, is defined as Vin being greater than 30 volts, and we’ll be focusing on the range up to 100 volts. which sees applications in a wide variety of areas. Examples include automotive with a 42 volt Vin; Power Over Ethernet, with a Vin between 36 and 56 volts; telecomm applications, with a range between 36 and 100 volts. Depending on the system requirements, the input voltages just mentioned might need to be stepped up, stepped down, isolated, and so forth. The power to be processed will range from below one watt to hundreds and even thousands of watts. The power processing all has to be done efficiently, hence the use of switching power supplies. [slide 3] In systems where no electrical isolation is required between the input and the output voltage, basic non-isolated topologies can be used. These basic topologies are the buck converter, the boost converter, the SEPIC, the buck-boost, the Cuk, and others that I have not put down here. Otherwise, if isolation is required between the input and the output voltage, then isolated topologies are required and some of these topologies are the flyback converter, the isolated SEPIC converter, the standard and clamped mode forward converter, September 22, 2004 National Semiconductor Confidential

1

High Voltage Switching Power Supply Topologies and Power Management ICs the push-pull, the standard and D(1-D) half-bridge, the standard and phase-shifted fullbridge, the cascaded buck with push-pull and so forth. In some cases where isolation is not required, an isolated topology might still be used because it might lead to operation in a range which gives better performance [slide 4] A successful implementation of each topology in a high voltage application is [greatly eased if a well-designed, high voltage, power management IC can be used that integrates logically and effectively the various functions that go to make a practical power supply. The essential features of such an ICs are shown in this slide. It needs an accurate voltage reference, against which the output voltage can be set and maintained; it needs an error amplifier that monitors deviations for the correct output voltage and generates a corrective signal; it requires a ramp generator and slope compensator to compare against the corrective signal from the error amplifier and generate the MOSFET gate drive signals; and it requires a beefy driver for the MOSFET which is capable of sinking and sourcing the large amount of current that the MOSFET needs in order to turn on and off efficiently; and, of course, as we know, switching power supplies requires an oscillator to generate a switching signal. In a high voltage power supply a very important piece is a house-keeping or bias supply voltage in the range of five to 20 volts to power all the functions mentioned above. In a low voltage power supply this bias can often be just the input voltage, but in a high voltage supply the input voltage is too high and the output voltage is not available before start up, thus a voltage to start up the power supply is required. After the power supply has started running from the start up supply, it is easy to use a winding on an inductor or a transformer to generate this bias supply for the rest of the operation. But initially, a way to start up the regulator is needed. [slide 5] National Semiconductor has introduced the LM50xx family of power management ICs that address this product issue in a clever way. Where other IC vendors require the user to externally generate a start-up voltage of about 15 volts and apply it to a low voltage IC, National has implemented the LM50xx ICs in a 100-volt process. This means that the Vstart pin shown in this slide here can be connected directly to the input voltage in the range of 15 to 100 volts to power an internal linear regulator that generates 7.7 volts and starts the IC. Once the supply is running, a winding on a transformer or an inductor is used to generate a voltage in the range of eight to 15 volts, which is applied to the Vcc pin. Once that voltage is applied to the Vcc pin, it shuts off the start up regulator within the IC, saving power and making the operation very efficient. The LM50xx family of ICs also contains all the other functions previously described that are required to implement a power supply successfully. [slide 6] We will now discuss the area of application of the various topologies for high voltage power conversion. In each case we will help members of the design community by identifying a member of the LM50xx family that might be used in the topology that is being described.

September 22, 2004 National Semiconductor Confidential

2

High Voltage Switching Power Supply Topologies and Power Management ICs In applications where the requirement is step down with no isolation, for example, if you have an input voltage of 60 volts and an output voltage of five volts, and there is no need for isolation, then the best topology is a simple buck converter. It is, as I have said, very simple and efficient. Apart from the switch and the diode it just requires one inductor and one capacitor and it works very well, except for cases where the ratio of the input voltage to the output voltage is very large. In those circumstances, you end up operating with a duty ratio which is very small. For example, if the input voltage is 60 volts, as we said, and the output voltage is 1.5 volts, then we will be operating with a duty ratio of 2.5 percent. What that would do is that it would lead to the use of an excessively large inductor, L1, and the rms current in the transistor Q1 would be large in relation to the average current. In situations like that, even though you do not need isolation, you would be better off to use the forward converter and utilize the power transformer of the forward converter to do some of the step down. For example, if you use a 10 to 1 step down transformer in the forward converter then you can increase the duty ratio by a factor of 10 to 25, and now you are operating Q1 with a duty ratio of 25 percent, which leads to better operation. [slide 7] In situations where you just need simple voltage step up and you don’t need isolation and you don’t need inversion of the voltage, that is to say the input voltage and the output voltage have the same polarity, then the topology to use is a boost converter. For example, you might have a system where you have 80 volts in and you need to generate 150 volts out at the same polarity, then you use a boost converter. The boost converter, like the Buck converter, is very simple. It just has one inductor and one capacitor, and it is also very efficient. The voltage stresses across the MOSFET and the diode in the boost converter are low, they are just simply the output voltage. (Whereas in the buck converter, the voltage stresses are just equal to the input voltage.) In theory, from looking at the formula for the transfer function for the boost converter, Vout=Vin/(1-D), according to to which if “D” goes to 1, you can get an infinitely large output voltage, in reality, the losses in the circuit make this impossible and in practice you can only get step up to about eight or ten times the input voltage before you cannot go any higher. So, this is the limitation of the boost converter. If, for example, you wanted to go from 80 volts to 800 or 1,000 volts then you would not use a simple boost converter, but you would use a topology, with a transformer and, the transformer would help you with some of the step up. [slide 8] In a lot of applications where the input voltage varies over a very wide range, the output voltage will have an intermediate value which is higher than the lowest input voltage, but lower than the highest input voltage. For example, your input voltage might be 20 volts to 75 volts and your output voltage might be 30 volts. In such a case you can’t use a buck converter because a buck converter can only drop the voltage down, and you can’t use a boost converter because a boost converter can only boost the voltage up. You have to use a topology that does both bucking and boosting and one such topology, the simplest one, is the buck-boost converter. And you would use the buck-boost converter in this case if your output voltage were inverted, that is to say, had the opposite polarity from the input voltage. The Buck-boost converter is simple and efficient, it has just one inductor and one capacitor, but it has the disadvantage that it has high voltage stresses on the diode and the MOSFET. The stresses are (Vin+Vout) on each of the components, and this topology also has a lot of ripple current in the output capacitor, so you would need to use an expensive capacitor with a high ripple current rating. September 22, 2004 National Semiconductor Confidential

3

High Voltage Switching Power Supply Topologies and Power Management ICs

Because the output voltage of the buck-boost converter is negative, that is, it has the opposite polarity to the input voltage, you cannot directly connect components in the usual way to the output of this converter to the power management IC to observe the output voltage. You need to put a level shifting circuit to help you to do that, but such circuits are quite straight forward to implement and we can always help with them [slide 9] If you are in a situation like the one that we just described, for the buckboost converter, where the output voltage, lies somewhere within the range of the input voltage, that is to say it is higher than the lowest input voltage, but lower than the highest input voltage, but you do not need inversion, then the output voltage has the same polarity as the input voltage, and you would use a non-isolated SEPIC converter. And one good thing about the SEPIC converter is, as you can see, that the MOSFET is ground referenced so it is very easy to drive; and because the output voltage has the same polarity as the input voltage, the feedback accessing the output voltage is very easy to do. The main disadvantage of the SEPIC converter is that instead of having just two energy source elements, like the topologies that we have looked at already that is to say, just one inductor and one capacitor, it has two inductors, L1 and L2, and two capacitors, C1 and Cout. So the disadvantage of the SEPIC converter is it complexity, but it is a very widely used topology. [slides 10] In a lot of cases, you want to create a multi-output power supply from one input voltage; and for low power systems the flyback converter is the best topology for that. [slide 11] The flyback converter is an isolated version of the buck-boost converter, but because there is isolation between the primary and the secondary, you can arrange the transformer so that relative to the primary side, the voltage on the secondary side can be up or down, and it also allows you to generate both positive and negative output voltages. And furthermore, just by having multiple windings, each one with its own diode and its own capacitor, you can have different output voltages. For example, in this slide here, if the winding associated with D1 has, let’s say, a number of turns of “N,” and the winding associated with D2 also has “N,” turns, then the output voltage, Vout2, would be twice Vout1, so basically the output voltage is just regulated by the turns ratios of the respective windings. The flyback converter is very widely used for low to medium power applications, that is to say, from below one watt to about 150 watts. It is used for both up and down conversion and as I have just said, it is very ideal for both single output and multi-output use, and the outputs can all be one polarity or they can be of different polarities. So it is probably the most versatile topology of all. But, it has its limitations, and one limitation is that it has high voltage stressors on the FETs and on the diode. The stresses on the FET, just like in the buck-boost converter, are equal to the input voltage plus the output voltage, where the output voltage is transformed by the [turns ratio of the transformer, and the voltage stresses on the diode are also equal to the output voltage plus the transformed input voltages, so they are high. Also, when the MOSFET turns off you have a lot of leakage inductance energy stored in the transformer which needs to be dissipated somewhere and it tends to be a problem to dissipate this energy.. This energy leads to high voltage spikes and it leads to inefficient operation. And finally, because the flyback converter is a single ended topology, that is to September 22, 2004 National Semiconductor Confidential

4

High Voltage Switching Power Supply Topologies and Power Management ICs say, power is transferred through the transformer only when the FET is on, the transformer has a DC flux in it. That is, the transformer has a DC magnetizing current, which means is that the core of the transformer is not utilized efficiently, so the transformer is bigger it needs to be. And also the transformer needs to store a lot of energy, so it is be big and bulky and the topology has a lot of ringing. [slides 12] Another topology, which finds a lot of application, is the forward topology, which is nothing but an isolated buck converter. I should look back a little bit now and say that the flyback technology is an isolated version of the buck-boost converter. So the forward technology is an isolated buck converter and it is used in much the same power range as the flyback converter, that is, in low power to medium power applications. Its main advantage over the flyback converter is that because it has an output inductor, the inductor makes sure that high input ripple currents do not flow in the output capacitor. And what that means is that for high voltage applications where voltage capacitors expensive you can use cheap electrolytic capacitors, which don’t have a good ripple current rating, but they will work well because the output inductor is taking care of the ripple current. [slides 13] As I said, the applications for this are low to medium power applications and it is a simple topology compared to bridges, which we will be discussing later. The MOSFET is very easy to drive because it is ground referenced and the topology can be used for both step down or step up, simply by choosing the appropriate turns ratio. If you choose a turns ratio such that the primary turns are greater than the secondary turns, then you are doing step down. If you choose the turn ratio of the transformer such that the secondary turns are greater in number than the primary pin turns, then you are doing step up. So for any input voltage that you have, you can have any output voltage that you like. The disadvantages of the forward converter are that just like the flyback converter, it has high voltage stresses, the transformer also has a DC flux in it, which makes it inefficient, and also in contrast to the flyback converter, where the transformer is resets itself, the forward converter does not have a mechanism to reset itself. You have to add a reset winding to it and so the transformer is slightly more complicated than for the flyback converter. [slide 14] A topology which is related to the forward converter is – (actually I’ve still got the forward converter to treat), is the single-ended clamped mode forward converter. And essentially on the secondary side, that is to say, through the transformer, from the transformer secondary to the output it is exactly the same in operation as the forward converter. The main difference is that -, instead of needing a reset winding, this topology resets the transformer through a blocking capacitor, Cb, and a MOSFET Q2. The two MOSFETs, Q1 and Q2, are driven in a complementary manner. When Q1 is on, Q2 is off and when Q2 is on, Q1 is off and in the small dead time between the two gate drives there is no cross-conduction, the FETs do not both conduct at the same time. This topology is used in exactly the same applications as the forward converter that I discussed before. [slides 15] There is another version of this topology called the double-ended clamped mode forward converter. And in the double-ended clamped mode forward converter there are two diodes on the output and power is transferred through the transformer all the time; when the switch Q1 is on and when it is off. The way that the transformer is reset is exactly the same as in the single ended version of the topology, but the output voltage of this September 22, 2004 National Semiconductor Confidential

5

High Voltage Switching Power Supply Topologies and Power Management ICs topology for the same turns ratio is twice that of the single-ended one. And in contrast to the regular forward converter and the single-ended clamped mode forward converter, because of the presence of diode diodeD2, it does not have an interval when the output voltage at the junction on the left-hand side of the output inductor is zero. So, that voltage is always some non-zero value. [slides 16] What that means is that on the secondary side of the topology you always have a voltage which corresponds to the on-time of one of the diodes, therefore the output diodes can very easily be replaced by synchronous rectifiers that are self-driven, which allows you to have very high efficiency in a very simple way. And also, both the single-ended and double-ended clamped mode forward converters have very clean MOSFET waveforms because if you go through the analysis, you can easily see that you can have zero voltage switching of the MOSFETs. Therefore this topology is used in a lot of applications where electrical noise has to be kept to a minimum. The disadvantages of this topology are that instead of having just one FET to drive as in the standard forward converter you have two FETs to drive; and then secondly, if you go through the analysis, you’ll find that it’s not easy to make everything work well over a very wide input voltage range, so it is good for an input voltage range that is narrow. [slide 17] These topologies that I have just spoken about, the isolated topologies, the flyback converter and the various versions of the forward converter are suitable for use at low to medium power. Now we will go to topologies that are suitable for very high power. But before we go to those I need to mention two products from National that are of special interest. One product is the LM5070 Power Over Ethernet controller. This is an IC that National is coming out with that is designed to satisfy all the requirements of the 802.3af spec for power over Ethernet applications. It has signature detection, it does identification, and has current limiting that ensures that you do not violate a maximum power, around 13 watt limit that is allowed for POE applications. Basically it is just one of the LM50xx family ICs, but with extra circuitry put in to make sure that it satisfies the requirements for Power Over Ethernet applications. This IC will greatly simplify the design of all topologies that have a low side MOSFET, that is to say, boost converters, SEPIC converters, flyback converters, forward converters, and so forth in PoE applications [slide 18] Another product that I need to speak about is the LM5021 ac to dc controller that National Semiconductor is coming out with. And basically where this differs from other members of the 50xx family is that the other members can run off a voltage only up to 100 watts, beyond that you can’t run them. The LM5021 has been designed in such a way that you can run it from arbitrary high voltages basically from 15 volts up to 1 kilovolt or 2 kilovolts if you like. You don’t have to put in any external start-up regulators, you connect a very high value resistor and a small capacitor to the start up pin, the IC starts and once it starts, it then runs from a voltage that is generated by the power supply. This IC has been designed to be very cheap and it is aimed for the ac to dc market. We expect it to find a lot of use in universal AC power supplies with input voltages from 100 volts to 400 volts DC and it will implement all the topologies that have a low-side FET, like flyback, forward, SEPIC and so forth. September 22, 2004 National Semiconductor Confidential

6

High Voltage Switching Power Supply Topologies and Power Management ICs

HIGH POWER TOPOLOGIES

[slides 19] Now, let us move on to high power topologies. All the topologies that have been discussed so far for isolated conversion have been so-called single-ended topologies, where there is just one MOSFET on the primary side and the power in the transformer is processed during only one half of the switching cycle. What that means is that there is DC flux in the transformer and the transformer tends to be big and the voltage stresses tends to be high and these topologies are not suitable for high power. If you want to do high power conversion, then you can use topologies like push-pull converters and bridge converters that I’m going to describe here. The push-pull converter, -and this is clear if you if you go through the analysis,- is just a pair of inter-leaved forward converters with a transformer, which is center tapped on both the primary side and the secondary side. The input voltage is connected to the primary center tap and two MOSFETs are connected to the outer ends of the transformer. The FET are controlled in such a way that first one FET is on, then both FETs are off, then the other FET is on, and then both are off and so on.. If you go through the analysis you will find that one FET causes the flux in the transformer to increase in one direction and the other FET causes it to increase in the opposite direction, or to decrease by the exact same amount. So, the transformer is self-balancing, it is self-resetting. And then on the secondary side, the push-pull converter is double-ended to get power processed through the transformer during both halves of the switching cycle. Because of this, the transformer is used very efficiently indeed, and for the same size of transformer you can actually get twice the power out of the push-pull converter that you would get out of the forward converter. Push-pull converters are used in applications ranging from low power, maybe 50 watts up to hundreds or thousands of watts, even hundreds of thousands of watts, and they are used both for step down and step up, just by picking the right turns ratio for the transformer. [slides 20] The advantage of the push-pull converter is that the two MOSFETs are ground referenced and as I have explained, the transformer is utilized very well because there is no DC flux in it. The disadvantages about the push-pull converter, and these disadvantages are really with respect to the bridges that I will describe just after this, are that the stresses on the MOSFETs are high. Each MOSFET sees a voltage which is equal to two times the input voltage, and the primary of the transformer needs to be center tapped. [slide 21] If you do not want to use a push-pull converter, then you can use a bridge converter and there are several variants of the bridge converter. Some are called half-bridges and others are full-bridges. Let us start with the standard half bridge. In a standard half bridge, the secondary side of the full half bridge looks exactly like the secondary side of a push-pull converter, but on the primary side it only has one winding. And it has two FETs connected to one side of the transformer across the input of the power supply and it has two capacitors, also connected across the power supply, going to the other side of the transformer. The MOSFETs are driven in exactly the same way as in a push-pull converter. That is to say, one MOSFET is on, then both MOSFETs are off, then the other MOSFET is September 22, 2004 National Semiconductor Confidential

7

High Voltage Switching Power Supply Topologies and Power Management ICs on, then both MOSFETs are off, so that is exactly like a push-pull converter. But the difference is that one of the FETs has a high-side drive, in is particular case, Q1 and therefore you need a high-side driver, you need an IC that is capable of driving a MOSFET which is switching at say, 100 volts. So that is a disadvantage of the half-bridge converter. [slides 22] The voltage at the junction of the capacitor is equal to half of the input voltage, Vin/2. Therefore, the voltage applied across the primary of the transformer is just half of Vin, compared to Vin in a push-pull converter and so to get the same output voltage the turns ratio of the half-bridge converter has to be only half of that of the push-pull converter. Another variant of the bridge is the full-bridge and in the full-bridge the two capacitors, C1 and C2, are simply replaced by MOSFETs and in the standard full-bridge the MOSFETs are switched in diagonal pairs. Going to the slide, Q1 and Q4 are switched on, then all the FETs are off, then Q2 and Q3 are switched on, then all the FETs are off, and then we go back to the beginning. The secondary side is the same as for the push-pull and the half-bridge converter. The turns ratio of the full-bridge converter is twice the turns ratio of the half-bridge converter. You have two FETs that need high-side gate drives, so you need a high-side driver in order to operate them properly. However, if you go through the calculations and look at the total silicon area that is required for a full-bridge converter and for a half-bridge converter and you also look at the RMS current in the two topologies, you will find that they are exactly equal. Not only that, the voltage stresses on the FETs in both cases are equal to Vin and the voltage stresses on the output rectifier are the same and actually they are the same as in the push-pull converter as well. [slide 23] Then there is a variant of the full bridge called the phase-shifted full bridge, which works in a clever manner. Looking at the schematic, it looks exactly like the schematic for a standard full bridge. However, the difference is in the way that the FETs are driven. [slide 24] Referring to the schematic for the phase-shifted full bridge, the way the FETs are driven are such that each top and bottom leg of the bridge, that is to say, each left side of the bridge or right side of the bridge, has the FET driven with a duty ratio of exactly 50 percent, with just a small dead time between them to make sure that there is no cross conduction. So, we have Q1 at 50 percent, then Q2 at 50 percent, then Q1 at 50 percent, then Q2 at 50 percent. This is for the right-hand side of the bridge. The same situation is obtained for the left-hand side of the bridge. You have Q3 with a 50 percent duty ratio, then Q4 with a 50 percent duty ratio, then Q3 and Q4 again with a duty ratio very close to 50 percent except for a small dead time between them. In order now to exchange the output voltage of the system, you take the right leg, that is Q1 and Q2, and you slide the gate drive in relation to the left leg, which contains Q3 and Q4. And so, as you slide the Q1/Q2 leg versus the Q3/Q4 leg, you end up with a voltage across the transformer whose duration varies as shown here, with duty ratio changes. And when this voltage is rectified by the diode you will get the waveform shown in the last graph and the average value of this waveform is the output voltage of this system.

September 22, 2004 National Semiconductor Confidential

8

High Voltage Switching Power Supply Topologies and Power Management ICs So, why the phase-shifted bridge? The phase-shifted bridge is preferred because actually during the time when you switch the FET in the standard half and standard full bridge you get a lot of voltage ringing, which leads to a lot of noise. If you go through the analysis for a phase-shifted bridge, it does not have any ringing so it is very quiet electrically, and it is used in very high power systems where radiated and conducted EMI would otherwise cause a lot of problems. [slide 25] One thing that needs to be mentioned is that all the push-pull, the halfbridge, and the standard full bridge, and phase-shifted full bridges are all just isolated versions of the buck converter. And in fact, if you look at the waveforms on the secondary side of these topologies, the secondary side actually does not know whether it is being driven by a push-pull, a standard half-bridge or a standard full bridge. So what that means is that when you are selecting a topology to use and you are trying to decide which of these topologies to pick, you don’t have to worry about the secondary side, you can decide on the basis of other things. [slide 26] All these, the standard half-bridges and the standard full bridges have the advantage that their voltage stresses on the MOSFETs are very low, they are just equal to Vin. Remember that in the push-pull converter it was 2Vin, it was two times Vin. And they have excellent transformer utilization because there is no DC flux, just as in the push-pull converter. And, of course, the disadvantage is that in both half-bridges and full bridges some of the FETs need high-side drivers. [slides 27] There is a very interesting variant on the half-bridge that [is used sometimes, and this is the D(1-D) or complementary duty ratio half-bridge. And in this half-bridge instead of Q1 and Q2 being driven with signals of the same duty ratio, that is to say, -if you remember what we said about the way the bridge switches, Q1 is on, then both are off, then Q2 is on, then both are off, then Q1 is on again; - in a D(1-D) half-bridge Q1 is on for a period DT and then Q2 is on for a period (1-D)T. And the only time when both of them are off is just a small time period of maybe 50 to 100 nanoseconds to avoid cross conduction. [slide 28] Now, what this does is that, if you go through the analysis again, sorry to keep saying that, is that if you look at the secondary side waveform there is never a time when the voltage on the left-hand side of the inductor is zero. There is always a voltage on the secondary of the transformer and this means that, again, just like in the case of the clamped-mode forward converter, you always have a voltage that you can use to drive synchronous rectifiers on the output side. So it makes synchronous rectification very easy and there is no ringing in the circuit, it is electrically very quiet and therefore it is used in very high power systems where people are worried about conducted and radiated EMI. The only problem with the topology is that you need to restrict the input voltage to a very narrow range. If the input voltage is very wide, then you start seeing very large voltages on some of the rectifiers and you start seeing very large RMS currents in some of the FETs on the input. [slides 29] And finally, there is a topology for which National designed a special IC, the LM5041, and this is a cascaded buck-current fed push-pull topology. Basically this topology has a front-end which consists of just a straight forward buck converter, which is regulated. Just to put some numbers in the equation, the input side of the buck converter might be, September 22, 2004 National Semiconductor Confidential

9

High Voltage Switching Power Supply Topologies and Power Management ICs let’s say, 36 or 48 volts and it is regulated w so that the output voltage is [fixed at a value of, let’s say 24 volts. So the input buck converter sees all the variation in the input voltage. This voltage is applied to a current fed push-pull converter, which runs at a duty ratio of 50 percent all the time, and doesn’t have any regulation. This push-pull converter can be run in this way because its input and output voltages converter are both fixed The push-pull converter can have a transformer but no output inductor. . In this example if the output voltage were 2.4 volts, then it would have a transformer with a turns ratio of ten to one. This mode of operation separates the two functions of the switching power supply of regulation and isolation. The Buck converter does the regulation part of the job and the push-pull converter does the isolation part. Each one can be designed more optimally for higher efficiency. The buck converter does not see an output voltage of 2.4 volts and its attendant heavy current, but is designed for the 10 times smaller current at 24 volts, making it more efficient.,. On the other hand, the push-pull converter does not see the high voltages that are associated with an input voltage of 48 volts because it is always operating from an input voltage of 24 volts. So this makes this topology very efficient over a very wide input voltage range. [slide 30] When the input voltage is very low, where a lot of topologies have poor efficiency due to too much conduction loss, and when the input voltage is very high, where a lot of topologies have too much loss due to very high switching losses, the conduction losses of the cascaded topology less are than in a single-stage topology and the switching losses are also less than a single-stage topology. So you get very good efficiency over a very wide range of input voltages. For this reason this topology is ideal for very wide input voltage ranges and it is ideal for very high power. Also because the current fed push-pull converter operates with a duty ratio always of 50 percent, there is no time when the secondary side voltage is equal to zero and, synchronous rectification is very easy to implement, and the synchronous rectifiers can self-drive themselves from the secondary of the transformer and you can get very good efficiency [do to this as well. [slide 31] So the preceding has been a presentation of some of the topologies that might be used in a variety of high voltage applications starting from low power to high power and starting from step down to step up and so forth. And National Semiconductor, as I mentioned before, has a family of high voltage ICs that allow you to implement all of the topologies that have been described in this talk, and they are listed here on this slide. If you are interested, you can get information about them from our National Website. SUMMARY

[slide 32] So, in summary, this talk has surveyed the most common used topologies for high voltage power conversion; and I have tried to indicate typical areas in which each topology can be used; and I have tried to indicate the advantages and disadvantages of each topology; and if you are interested in them, I have suggested a National Power Management part that you can look at to help you to implement that topology; and some of these power management ICs that I have discussed here are actually supported by our Web Simulation September 22, 2004 National Semiconductor Confidential

10

High Voltage Switching Power Supply Topologies and Power Management ICs Software, WEBENCH, and the datasheets for all of them can be looked at on National’s Website. And that concludes our presentation, I will turn it over to Michelle. MODERATOR: [slide 33] Thank you, Haachitaba. Thank you, everyone, for joining us for this seminar, High Voltage Switching Power Supply Topologies and Power Management ICs, brought to you by National Semiconductor and IVT. This concludes today’s on-line seminar. When you close your seminar window a survey form will appear. Please fill out and submit the survey form. Your answers to the survey will help us in the development of new products as well as future seminars. Thank you for attending and good day. (END OF PRESENTATION)

September 22, 2004 National Semiconductor Confidential

11