Hello, everyone. My name is Pradeep Shenoy. I am a systems engineer at Texas Instruments in the DC solutions business unit.

And today, I have the pleasure of introducing how to design a high-frequency series capacitor buck converter. This presentation in all of its different segments will introduce a new high-frequency dc/dc converter topology called a series capacitor buck converter. You can see a picture of a prototype on the right of this slide.

First, though, we'll start with some background information, and some limitations of today's conventional buck converters, and what are the challenges to operating them at higher frequency. Then, we'll also introduce the series capacitor buck converter, followed by some sample experimental results to kind of show some of the capabilities and opportunities with this new converter topology enables. And then we'll kind of walk through a design example of a high-frequency series capacitor buck converter.

And the whole goal with this effort is to make voltage regulators smaller. So reducing the size, saving space on your board with this new converter. We've recently introduced the TPS54A20, and that's the product that will be used as sort of a background for some of the design examples here.

Now, many of you are probably familiar with a power delivery system that's in a variety of equipment and systems today. Usually, you'll have some sort of power source-- maybe you're connected to the AC grid. And you'll have a power supply that provides some intermediate bus voltage-- often, it's 12 volts in many systems. There's a variety of intermediate voltage levels, but 12 volts is very typical.

Now, the loads in these systems are typically microprocessors, maybe FPGAs, memory. You might have some DDR memory in your system, things like that. And you need to convert from this 12-volt intermediate bus down to the load. And these load voltages are typically low-- maybe 1.2 volts, 1 volt, 1.5, depending on what the load is.

And what is very common today is using a buck converter in these applications that steps down the voltage from 12 to 1. And point-of-load voltage regulators are very common in pretty much most systems today. Now, one thing I'd like to point out is you can see that the voltage conversion ratio in these applications is pretty high, typically. So 12 to 1.2 volts-- that's a 10:1 voltage conversion ratio. And that's going to be key as we go along in this presentation.

Now, these point-of-load voltage regulators can take up considerable size and real estate on a PCB. And so what this new converter topology will enable you to do is increase switching frequency. And why do you want to increase switching frequency?

Well, the main number one reason is you want smaller size. So the picture on the left shows a conventional buck converter. And you can see in that picture that the inductor is the largest single component in that converter. It takes up the most space in terms of area as well as height.

On the right is an example design of a high-frequency series capacitor buck converter. And you can see that it is considerably smaller. The area is about half the size, and the converter volume is significantly reduced.

You can't see this as clearly in this picture-- but the height is also reduced. The picture on the left is about 4 or 5 millimeters tall, whereas the picture on the right is about 1.2 millimeters tall. But what's interesting is you can see that the converter volume of the converter on the right-- it's less than just the inductor volume with the converter on the left. So significantly smaller size.

Another benefit that you get when you operate at high-frequency is faster response. Whenever your load changes, this high-frequency converter can respond to changes in load very quickly. Also, another benefit that you get from reduced size is just reduced bill of materials cost. So you have fewer output capacitors, maybe smaller inductors, and that helps save a lot as well.

Now, here's specifically looking at the inductor size reduction. So let's take a look at that. On the left are two different inductors that are designed for a 10-amp output application using conventional buck converter topology.

Today, maybe they're designed for around 500 kilohertz operation. And the 1k pricing for those, if you buy online through a distributor, is maybe in the range of $2 to $3-- 1k unit pricing. Whereas when you go to high-frequency, you can use much smaller inductors.

In this example, around 2 by 1.2 millimeter inductors-- they're designed for a converter, let's say, operating in the 2 to 5 megahertz range. And these newer inductors-- there's nothing special or magical about them. They're mainly just smaller.

Because you're operating at a higher frequency, you can get away with a smaller converter. And the cost of them in 1k unit pricing is around $0.20 to $0.30. So high-frequency operation, much smaller inductors, and reduced cost.

Buck converters have been the workhorse of the voltage regulator industry for decades now. And they have certain fundamental limitations, especially when you want to operate them at high-frequency. And one of the very commonly known limitations is that as you increase operating frequency, your power loss goes up-- specifically, your switching losses. And without going into the details of all the different components of switching losses, you can see that as you increase frequency, your power loss due to switching will go up.

Now, there's another important challenge that applies specifically to high-voltage conversion ratio applications. Let's say, for example, you want to operate your buck converter at 5 megahertz, and that would mean effectively a 200-nanosecond period. Now, if you have a 10:1 voltage conversion ratio, if you recall the 12 volts in, 1.2 volts out, the nominal on-time for the high-side side switch is going to be 1/10 of your period, or 20 nanoseconds.

Now, 20 nanoseconds is a very short time to turn on and off the high-side power MOSFET in a controllable fashion. And many converters on the market today-- their minimum on-times are maybe 100 nanoseconds or more, sometimes. So getting down to these very low high-side on-times is very challenging.

So most of the high-frequency converters on the market today-- and when I say high-frequency, HF, that means 3 to 30 megahertz is the definition of HF. So if you look at the HF converters on the market today, they typically have low voltage conversion ratios. What that means is they'll have maybe a 5-volt input and design for maybe a 1-volt output.

And they typically have low current. I think around 1 amp max is what I've seen typically. So there's clearly a need to find a way to overcome these challenges and limitations of a buck converter.

The serious capacitor buck converter topology is a new topology that enables small high-frequency point-of-load voltage regulators. As you can see in the picture on the left, it's very similar to your regular two-phase buck converter, except that there's an additional capacitor inserted between the high-side and low-side switch of phase A, or between Q1A and Q2A.

Now, this unique topology has a lot of benefits. It essentially merges a switch capacitor circuit with a conventional two-phase buck converter into a single conversion stage that provides high efficiency. Another major benefit is that whenever any of the power switches are turned on and off, they are switching at a reduced voltage from the drain to source-- approximately half of the voltage that would be typically seen by a regular buck converter. So this helps to reduce switching loss.

That series capacitor is also softly charged and discharged by the output inductors, listed LA and LB. And this is unique, because in a conventional switch capacitor circuit, there is usually hard charging of the capacitors, which basically means there's these large current impulses. Whereas in this case, you have a very smooth, constant current source effect that the inductors provide.

A unique aspect of this topology is automatic current balancing. Unlike typical multi-phase buck converters that require current sensing as well as a current sharing control loop, this topology inherently, automatically, in a passive fashion, forces even current sharing between the two phases. No external current sensing is required, no control loops, nothing like that. This significantly simplifies design.

And another benefit is that the on-time of the high-side switches are doubled, or the duty ratio is doubled. For the same voltage conversion application of a regular buck converter, the on-time would be doubled in this topology.

Now, there's always some drawbacks. And with this topology, it's basically that you cannot turn on both high-side switches-- Q1A and Q1B-- at the same time. What that effectively does, or what that means is, it's a 50% duty cycle limitation. Now, a 50% duty cycle limitation combined with the fact that that series capacitor voltage is half the input voltage nominally, and you have an inherent sort of 2:1 step down effect because of that, and the switch nodes voltages are at half the input voltage, you're theoretically limited to a input voltage that is four times greater than your output voltage. Practically speaking, it's really more like five times your input voltage.

So what does that mean? If you have a 12-volt input, and you want 5-volt output, this topology won't work for you. But if you have a high-voltage conversion ratio application where your output is 5 times less than your input voltage, it will work just great. 12 volts in, 1.8 volts out-- perfect. 12 volts in, 1 volt out-- you got it.

Another limitation is that you cannot do phase shedding and adding that you would typically do in a multi-phase buck converter. But there are other light load operating modes that you can do that can save losses and improve your efficiency at light load.

So how does this converter work? I'm glad you asked. What I'm going to step through next is the four intervals of operation of this converter.

It's very similar to your regular two-phase buck converter that's interleaved by 180 degrees. On the top left is a picture of the schematic of the converter with the different switches open or closed. The bottom left is the series capacitor voltage differentially. In the top right is the inductor currents and the series capacitor current. And the bottom left graph shows the switch node voltages.

Now, let's start with the scenario where the high-side switch phase A, labeled Q1A, is on. Now, for this case, I'm assuming the input voltage is, let's say, 12 volts. You can see in the graph in the bottom left that the series capacitor voltage is approximately 6 volts, or half the input voltage, with some ripple across it-- around maybe 150 millivolts of ripple.

Now, when Q1A is on, there is current flowing through Q1A, through the series capacitor, through inductor LA to the output. And in this situation, the series capacitor voltage is increasing, because the series capacitor is essentially being charged by inductor A. You can see that in the top right, where the inductor A current in blue is increasing. And the red dotted line which represents the series capacitor current is exactly the same as the inductor current, and it's positive.

The switch node voltage of phase A, labeled VSWA-- it's shown as 6 volts as well, or half the input voltage. The phase B inductor is connected to ground through Q2B, just like any regular buck converter, when its low-side switches on. And its inductor current is sloping down, its switch node voltage is 0.

Now, interval 2 is very simple. It is essentially the same as a regular two-phase buck converter, where you have both low-side switches on at the same time, both inductor currents are sloping down. In this case, there are no currents going into or out of the series capacitor, as you can see in the top right. And as a result, the voltage across a series capacitor is not changing, as you can see in the bottom left. In the bottom right, you can see that both switch node voltages are 0.

Interval 3 is where things get interesting. In this case, the series capacitor CT is essentially acting as an input capacitor to phase B. You can see that the minus, or the negative side of the series capacitor, is connected to ground through Q2A, the low-side switch of phase A.

And the plus side is connected to the switch node of B through its high-side switch, or Q1B. And the inductor of phase B, labeled LB-- it is discharging that series capacitor by a certain amount-- in this case, 115 millivolts in the end. And the current that's coming out of that series capacitor is the same as the current that's flowing through that phase B inductor.

Now, because the current is coming out of the capacitor, its voltage is dropping, as you can see in the figure in the bottom left. And the switch node voltage of phase B is also basically half the input voltage, or around 6 volts, as you can see in the figure in the bottom right. And the switch node voltage for phase A is 0 volts, and that's why the inductor current in phase A is sloping down. And because the switch node voltage of phase B is 6 volts, that's why its inductor current increases.

Interval 4 is basically the same thing as interval 2-- we just include it for completeness' sake. Both low-side switches are on, both inductor currents are sloping down. The series capacitor current is 0. That's why the series capacitor voltage doesn't change, and both switch node voltages are 0. This is just the same as you'd have in a regular multi-phase buck converter.

The series capacitor buck converter has many benefits. One of the major benefits is a reduction in switching loss. Now, there's a very detailed model that you could create of switching loss in these converters.

But let's just highlight the major benefits. Now, because the switches are turning on and off at half the input voltage, or half of what would be seen in a regular buck converter, the switching loss is reduced. In the figure on the right, you can see a comparison of the energy lost per switching cycle due to the parasitic output capacitance of, in example, MOSFET.

In this case, if you were to switch it on and off at 12 volts, you'd have around 30 nanojoules of energy lost every switching cycle. But if you're switching at 6 volts, you have around 10 nanojoules, or this represents about a 67% decrease in switching loss. Now, that's just one component of switching loss.

You also get a reduction in switching loss due to the reduction of the voltage current overlap during the transition. So the reduction in these loss components enables high-frequency operation of the converter without having excess loss or heating in the converter. Another benefit that you get is a small amount of reduction in inductor current ripple. It can be up to 33%, depending on your voltage conversion ratio.

Now, this compares a conventional buck converter with a series capacitor buck converter for the same values of inductance, VIN, VOUT, and switching frequency. So all else being equal, the current ripple in the series capacitor buck converter will be reduced.

If you have a 10:1 voltage conversion ratio in that application, it'd be around 11% reduction. Where this is beneficial is it really helps to reduce inductor core loss. Another way of looking at it is if you wanted to design the converter for the same ripple in the inductor current, you could have a slightly lower required inductance for the same voltage conversion ratio conditions as well as switching frequency.

Now, a really unique benefit of this converter topology is the automatic current sharing. And without going into all the math and details, the high-level explanation for this, is that this series capacitor forms an average current feedback mechanism inherent to the topology.

Now, think of it like this. if one of the inductor currents is higher than the other one, then the average serious capacitor voltage is going to go up or down by a certain amount. And that, in turn, is going to impact the switch node voltage of each of the phases in such a way that it forces the inductor currents to go back into even sharing.

And we've tested this. It's very robust variations in inductance as well as the DC resistance of the inductors. Because, at the end of the day, it comes down to charge balance.

The picture on the bottom left-- you can see that this kind of depicts a scenario where you'd have very large differences in inductance. And therefore, if one inductor was a lot smaller than the other one, its peak to peak ripple would be a lot larger. But it turns out that the area Q1 and Q2 are the same, so it doesn't really matter what the inductance is. It just matters that the on-times are essentially the same.

We've tested this experimentally as well. On the top right, you can see how the inductor current shares versus output currents. And in this case, we tested with two different inductance values of approximately 100 nanohenries on phase A and 200 nanohenries on phase B. And you can see that the currents share perfectly.

The bottom left shows an image of basically that same case, where you can see that the peak to peak ripple in the inductors is quite different. This demonstrates that there are different inductance values, but that the average current of both of them is the same. If you're interested in learning more, you can look at the paper that's cited at the bottom of the slide.

Now, this converter topology that we're introducing has a unique high-frequency controller added onto it. The core of it is an adaptive constant on-time controller, which provides very fast transient response, and is internally compensated. That basically makes designing the converter much, much easier.

There's a unique added feature to this controller, where we enable fixed-frequency operation in steady state. Now, the way that that works is there's a phase lock loop that's added on to your conventional constant on-time controller. So a conventional constant on-time controller-- you look at your feedback, and your area amplifier will trigger it on-time. And you just kind of trigger on-time pulses whenever the output demands it.

But what we've added onto the on-time generator is this phase lock loop that compares the switching signals to some reference clock. It could be the internal oscillator or an external sync clock And it'll slowly adapt or adjust the on-time such that in steady state, the converters fix frequency.

There's also the potential options to do light load efficiency improvements. Let's look at some comparisons of actual hardware results of the series capacitor buck converter and a conventional converter. As you can see in the graph on the left, the current density is significantly higher with the TPS54A20, or the series capacitor buck converter. It's around 3 to 7 times smaller than anything else that's on the market today. It's even, as you can see, much higher density than a variety of research publications.

One of the major benefits to increasing the density of the series capacitor buck converter is that it has a significantly lower profile. So, as you can see on the right-hand side, a conventional buck converter is around almost 5 millimeters high. Whereas a series cap buck, as designed using the TPS54A20, is a mere 1.2 millimeters tall.

And this is key, because it enables unique applications and placements of this high-current 10-amp voltage regulator. For example, you could put it on the backside of a board. Or maybe on a card, where there's some height limitation.

Now, the efficiency of this converter is still high, even though we're operating at high-frequency. The graph on the left shows a comparison of measured efficiency of two converters-- the 54A20 and the 54020. Both of these are designed for 12 volts in, 10 amp out applications.

But the difference is, the series cap buck converter's operating around 2 megahertz per phase, whereas the regular buck converter's at around 500 kilohertz. In both these applications, it's 12 volts in, 1.2 volts out. There was no airflow, just operated at room temperature.

And you can see that over the load range, there is higher efficiency in the series capacitor buck converter, even though it's switching at four times the frequency. And just as a note, the inductors in this comparison were selected for the equivalent DCR, so it would be a fair comparison.

The low transient response of the series capacitor buck converter is extremely good. It has very fast transient response with very good characteristics. In this example, we have a load step up on the left, and a load step down on the right. And you can see the output voltage, the inductor currents, and the load current shown.

In this case, it's a 12 volt in, 1 volt out. And the slew rate of the load is extremely fast-- 500 amps per microsecond. And this is basically as fast as we could turn on our power MOSFET load.

And the deviation in the output voltage in this case is less than 25 millivolts. And the recovery time is around 4 microseconds. As you can see, the switching frequency changes during the transient, because we're using a constant on-time controller that either spits out a bunch of pulses during the load step up, or just kind of waits and lets the inductor currents decrease during a load step down.

Now, a key point here to observe is that the inductor currents have very good dynamic current sharing. So even during a low transient, as you can see on the load step up on the left, the average inductor currents are perfectly matched. On the load step down, they're also very well matched. There's a little bit of an offset at the bottom of the inductor currents due to how the controller responds-- it fires one phase, and then it fires the next phase. But you can see that they very quickly align into equal current sharing and correct for this disturbance in a very stable manner.

OK, let's now dive into the design of the series capacitor buck converter. And one of the first places that you typically start is actually figuring out, what are the specifications for the voltage regulator you want to design? What's the input voltage, output voltage? What sort of size or frequency you are trying to target.

So assuming that the voltage conversion ratio-- in this case, we're going to go with roughly 12 volts in, 1.2 volts out. And usually the first step is picking the switching frequency, because that will dictate how you design a lot of the other components. Now, as you can see in the graph on the right, there are some tradeoffs with switching frequency.

Now, as you increase switching frequency, you can help reduce your converter size. But the tradeoff, as you can see on the right, is that as you increase frequency, your efficiency tends to decrease. So making that tradeoff-- it kind of depends on your application, where you want to be. If you are able to handle a little bit lower efficiency, and you'd rather have smaller size, you can do that. If efficiency is the most important thing to you, then you might want to reduce your switching frequency.

And the reason why frequency is so important is because the value of inductance that you need is inversely proportional to switching frequency. So as you increase switching frequency, the inductance required goes down, and also the size of the inductor goes down. You also get some benefits like I mentioned previously about faster transient response. So you have fewer output caps typically needed as well.

Now, inductors also impact efficiency. You can choose a variety of sizes and types of inductors. And typically, a higher inductance will result in a higher peak efficiency. And this is mainly due to the fact that you have maybe a lower core loss and lower RMS currents in your converter.

Whereas, for the same size inductor, if you have a lower inductance value, that will typically mean you have lowered winding resistance, because there may be fewer turns of wire in the inductor. And that will help result in a higher full load efficiency. Because at full load is when most of these conduction losses really come into play, or the DC resistance.

So the graph on the right is a comparison of for 12 volts in, 1.2 volts out, 2 megahertz per phase, using the same size inductors from the same vendor. What efficiency we measured with just different inductance values. So it's the same material, same everything else, just a different inductance value.

And you can see that the higher inductance values typically give higher peak efficiency. But lower full load efficiency and the lower inductance values give higher full load efficiency. So you can kind of tune what sort of efficiency graph you want to get with your inductor selection.

Now, the series capacitor is a unique aspect of this topology. And it shouldn't cause any concern or undue anxiety. The selection of it is fairly simple and straightforward.

There's a couple things you have to keep in mind. Now, the series capacitor, as you may recall, is charged and discharged by the inductors. So if you have a low value of capacitance, then you'll have a larger ripple in the voltage on that series capacitor. If you have a higher value of capacitance, that voltage ripple will decrease, as you can see in the diagram on the top right.

Typically, we recommend selecting the capacitor value to keep the voltage ripple relatively low-- maybe less than 8% in this example. Probably in the 5% to 10% range, because you want that series capacitor to act like a DC voltage source in the converter. That's its job.

So you can see an equation here for calculating what that series capacitance can be. The paper also has more details, if you want to look at that. The tradeoff, really, with the capacitance selection is with the delay during startup.

So the capacitance value calculated, as you can see, is relatively low-- around 1 microfarad. Typically, we see around 1 or 2 microfarads being what you get or what you need in this application. So it's relatively small. And you might be tempted to increase your capacitance to a higher value, which is OK.

But the challenge is that when you want to start this converter up, we will pre-charge that series capacitor to half the input voltage. And we do this to have a very smooth, well-controlled startup. Now, in this case, we have a 10 milliamp pre-charge current. And if that was to go into this one microfarad cap that was calculated, it will result in a approximately 625 microsecond delay to pre-charge that series capacitor before you can start switching and ramp up the output voltage. And you can see that in the scope image on the right, this is the delay due to pre-charge.

Now, a couple other things are mentioned in more detail in the paper and in some other documentation that we have as to what sort of types of capacitors to choose, and what are some other things to keep in mind. Like the DC bias effects, the ceramic capacitors, and temperature ratings, and things like that. And all that's included in some other documentation if you want to explore further.

Now, input and output capacitor selection is also very important. And we could spend a long time on that. But the main things that capacitors impact is not only your steady-state voltage ripple, but also your closed-loop bandwidth and your low transient performance.

And as you can see in the graphs on the right, these are Bode plot measurements for this converter with two different output capacitance values. One is 91 microfarads, and the other capacitance value is 138 microfarads.

Now, with a lower output capacitance, you will achieve a higher crossover frequency-- in this case, around 320 kilohertz. With a larger output capacitance, you will have a slightly lower crossover frequency. In both cases, you can see that the converter is quite stable, with over 50 degrees of phase margin. So either way, it's OK. It kind of depends on what you want to design for.

The paper has more equations that detail how to select for different low transient conditions. In this example, let's say we want to have, designing for a full load step, which means 0 to 10 amps with 1.2 volts out. And if all we want is a 36-millivolt excursion on the output voltage, sorry-- if this is for a load step down, then the minimum output capacitance would be around 127 microfarads.

Input capacitor selection is also similar. You usually can design it for your steady-state ripple, assuming that the bulk capacitance on your input bus is sufficient to handle the deviation on the input bus during a load step transient. And there are more details in the paper on how to select that.

Now, the feedback network is fairly simple, but also important. This is what you will use to connect from the output voltage to the f And there's a couple different ways you can do this.

You could start with a simple resistor divider, where you just have two resistors that connect the output voltage to the feedback node. And it's very simple. The downside is you don't get any phase boosts.

So often, what's recommended is that you put a capacitor-- you can see capacitor C1 under the feedback network number 2. It's in parallel with the high-side resistor R1. And this helps provide a phase boost.

And it's fairly simple. You have a little bit more flexibility in how you want to tune where that phase boost is if you put another resistor in series with that capacitor C1. And that's what's shown in feedback network 3. Of course, it comes at the price of more components.

Another trick that sometimes is played is adding a additional resistor labeled R4, as shown in feedback network number 4. That resistor can essentially help provide a very high-frequency, low-pass filter that helps improve noise immunity. Of course, the drawback of that feedback network is more components.

So take an example where we'll just compare the first two feedback network configurations. And the example parameters are shown on the bottom left. But what you can see, the benefit of having that phase boost capacitor is that it can really increase your phase margin and even your crossover frequency.

So if you just have a regular resistor divider, you might have 188 kilohertz crossover frequency with only around 37 degrees of phase margin. But with that phase boost capacitor in this configuration in this example, you'd have almost 50 degrees of phase margin and about 10 kilohertz higher crossover frequency.

OK. Let's move on to packaging and layout. Now with the TPS54A20 specifically, we are using a unique style of IC packaging called HotRod. It's basically a QFN package, but it's flip-chip.

So that means that the die, instead of being belly-up with wire bonding, it flips it straight down onto a lead frame. And this flip-chip design really helps reduce parasitic elements, especially parasitic inductance. You can see a picture of the converter in the middle with the pin assignments on the right.

Now, thermal vias are, in this case, placed in the power ground strip that goes right through the middle of the IC. This really helps with heat removal. And it's important to have PCB ground planes that this is connected to. Essentially, they're acting as the heat sink. These ground plane connections also help with the ground return currents.

Now, here's an example layout of the converter. As you can see, all of the components can fit on the top side. The picture on the left shows the IC, the input capacitors on the left, the series capacitor to the right-hand side of the IC with the two output inductors. And then the output capacitors are on the top.

Now, the red indicates the top layer traces and pores. In blue, you can see that the feedback trace, in this case, is going on the bottom layer. And there's different R's and C's for setting different values of different components.

A key thing, really, is you want to have a very compact, tight layout. So that means place your input cap and your series cap right next to the IC. That's how you can reduce the high-frequency loops area, which helps to reduce switching losses.

You also want to place your gate drive and boot strap caps close to the IC for a very similar reason-- you want to reduce the parasitic inductance. And it's also very important to have good connections to power ground through thermal vias. It really helps to improve your thermal dissipation and provides good ground return paths.

Now, this is a board layout example of essentially the same layout that we showed on the previous slide. What's highlighted here is the switching loops. Switching loop A is the switching loop for phase A, and switching loop B is the switching loop for phase B. And you can see that here we have very small areas in those loops, and that really helps to reduce the switching losses and parasitic inductance.

You also may want to have small switch nodes. This helps to lower EMI. Again, you want to make sure that your power stage and bootstrap caps and all this is compactly laid out together. You don't want to have some of these components on far-flung edges of your board and ensure a good ground return path using ground planes.

Now, this here shows the thermal dissipation of the same layout that I showed you before. In this case, this is a condition of 12 volts in, 1.2 volts out, a full load current of 10 amps, and operating at 2 megahertz per phase. Now, there's no airflow in this case, and it's just at room temperature.

You can see here that the IC is, as is usual, the hottest component in the converter. The inductors, on the other hand, are relatively low temperature. And so they are still not a thermal bottleneck.

Often, people are concerned that OK, well, if you reduce size so much, then you're going to cause a big thermal problem. Well, it turns out that the power loss challenge, or the heat dissipation challenge, is still basically the same as it was before. Because you have basically the same heat that you have to dissipate through, basically the same size with the IC, but you don't have any bottleneck, even though you're using much smaller inductor sizes.

OK, let's sum this up and head to the summary now. The total solution for the series capacitor buck converter that you get when you put all these things together is incredibly small. As an example, let's compare just the inductor on a 10-amp buck converter EVM on a competitor's evaluation board.

You can see here that the inductor size is almost 500 millimeters cubed. What you can do with a series capacitor buck converter for the same 10-amp application, as shown on the right, is you can fit that whole converter in a smaller size. So in this case, it's around 300 millimeters cubed.

That includes the PCB thickness that's shown in this picture. So the total solution volume is smaller than just the inductor that is required on a conventional 10-amp buck converter that's used in a competitor's EVM. So the size benefit is astounding.

So just to summarize this presentation, we've presented a high-frequency converter that enables incredible size reduction and performance improvements. Conventional buck converters have fundamental limitations that are challenges to high-frequency operation. The series capacitor buck converter has very unique properties that enable high-frequency operation without compromising on efficiency.

We've presented a few design guidelines for this high-frequency series capacitor converter, and just demonstrating how easy it is to design and implement. And if you're interested in further details, I encourage you to look at the paper or other collateral that's on the TI's website. Thank you very much.