ON Semiconductor Corporation (ON) Earnings Call Transcript & Summary

Hello, everyone, and welcome to today's webinar on Ultra-High Density 300W Power Supplies with Totem Pole PFC and Integrated Driver GaN.. I'm Emile Wassel with on semi, and I'll be your moderator today. In today's session, we will review how the compact and efficient 300-watt power supply solution is enabled by onsemi's multi-mode Totem pole power factor correction controller and high-speed LLC controller, working together with the On-semi-driver GaN solution. [Operator Instructions] Today's presenter is Jonathan Harper. He's a member of the technical staff at onsemi and covers energy infrastructure and European marketing for the Advanced Solutions Group. Thank you so much, John. Floor is now yours.

Hi, everybody. Good afternoon. We're going to be talking about the ultra-high density GaN-based Totem Pole PFC and LLC power supply 300 watts. So it's a bit of a mouthful, so I'll just show you what it's all about. It's getting a power density of 36 watts per cubic inch using -- and this is the example. So people want to get power supplies small, efficient, and I'll be telling you how we do that. And it's quite a challenge to get this powered. Now right at the end, we'll be talking about a product or a solution, which we showed at APAC, and that's 240-watt power supply in these sites. It's in there -- it's being shipped over to us. I don't have something to show there, but just a size of 240 watts, using the same technology I'm showing for this other board here. So it's all about getting power supplies smaller. So you've seen this outside. You don't need to see look at a PowerPoint picture of that. You can see the actual -- I think that picture might be too small me say here, but you can see the real power supply. What it's all about getting high-power density is removing components you don't need like the bridge rectifier, and that's what the Totem Pole PFC does for you. And it's also to increase the switching frequency so you can make the transformer, smaller and the boost [indiscernible] small. So here's just -- here's the transformer for the LLC and here's the boost conductor here. Also makes the switching high frequency makes the filter components smaller, and that's what you could see here, this is one of the filter components here. And then there's another comes here to see the extra other components. So all the components are the smaller by having a far higher switching frequency. Having higher switching frequency normally means the losses, but you can reduce that by using GaN as we use it to an integrated drive a GaN to achieve this. And when we look at the overall efficiency curve, you will see that it's not just good at the peak power it's good at others. So let's talk a little bit about that. In the good old days, 15 years ago, you just worry about the efficiency of the maximum load. Good thing to worry about. It helps you define your overall size of your system, your overall cooling -- it's good to be efficient there. Also more important is to be -- have low standby power. It's also very important. And they're very important that the third point, of course, is having efficiency over a wide power range. Let's consider my PC at the mine time showing you a presentation. My PC is not doing much apart from sending a few messages down the Internet, it's not really overloading. It's doing a symmetric simulation of an LLC converter with a lot of detailed components in there, then it will be running quite hot, it'll be running at 100%. But most of the time is kind of running at these power level, showing PowerPoint slides, things like that, so relatively low. So what's actually very important is to make sure that the efficiency of line load conditions and medium load conditions and follow conditions is all good because if most of the time people are running their PCs 20% load, for example, we need to have good efficiency at 20% load, so you can have a good energy efficiency overall. So this is why we want to get good efficiency over a broad power range. So how do we achieve this? First thing is we use a total pulp, a total pulp PFC, we remove the inputs rectify. And the input rectify is a large item, and we replace it with a Total Pole PFC, I'll show you that as in the moment. So this circuit shows the common mode. Here we are the common mode choke the differentiated mode choke shows the right capacitors we're using here be protective. So it shows you the high-level major components of this. That's what I'd like to show you, and show you what you can trim and what are the main functions without going to a full circuit diagram, which may be going to a little bit too much detail. So this is the purpose of this diagram. As Emile will tell you will be able to download this presentation, join or after this event and get these diagrams and look at them more closer. So Totem Pole PFC. To explain Totem Pole PFC's benefits, you need to show what is replacing a Totem Pole PFC is replacing a standard bridge rectifier plus a boost circuit. Here's a bridge rectifier, which rectifies the AC voltage. And then you have the Standard Boost component, which is a inductor, a switch and a diode and, of course, the output capacity. That -- we'll talk about what that does in a moment. But we shall know about PFC. So we just take this for granted that this gives us better it gives you a benefit. Then the bypass diode that we're in rush in mixer diode. This is important because you need to make sure there's not too much in rush currently. Normally, one draw that in a boost circuit. But when you get to see the circuit diagram of the Totem-pole PFC, you'll see those diodes. And it's important to explain what they are in the Totem-pole PFC. So that's why we included in there. So what does it do is this power supply is hopefully discharged. That would be pretty bad from a thing as it was doing that. If I put that into the main, these capacitors are got to be charged up. So there'll be a large inrush current if you didn't have -- there's a large inrush current, of course, anyway. If you have a bypass diode and the inrush current goes to the bypass diode and are the induction this time. So it could, to some extent, protects these switches and has a little bit more controlled flow if you do that. So just through the bypass diode. Again, it's a minor detail, but you see -- it confuses people when they see the circuit of the Totem-pole PFC. So that's why I just want to explain. So let's see how you get away with not using a diode bridge. How do you do that? And this is the way I like to drive the Totem-pole PFC there's a number of ways of doing this. But by relating it to the standard boost circuit, it helps to understand much more easily why the different components of the Totem-pole PFC do what they do. So what I say is, I said, okay, we have a positive phase AC. That's all there is. So you have a positive phase AC, how would you generate PFC from it. So you'd have a again of boost circuit. The boost circuit obviously has now got a boost up form of very low voltage, but of course, you can do that. It's just change [TV] cycle and make sure you can cope to that. That's not the problem. Normally, with a rectifier the voltage should be a little bit higher. But okay, you've got to boost from a lower level, just programming to do that. it's no problem, right? You have your standard Boost circuit, which I'll mark S2 and S1 for reasons you see in the next slide. You have a bypass diode again. And then I say, well, I want to make sure really that this circuit only works when I'm in a positive phase. I'm going to put a diode here to make sure that if you have a negative phase, it blocks them. So this diode only conducts joint positive phase, okay? So I've got a PFC, which works fine for the positive phase and is totally switched off the [net] I say, okay, half the job done. I then want to add this -- okay, let's not add it to it. Let's just consider a negative phase. We'll have to the negative phase. It's a little bit more complicated. You have a negative boost circuit. So this is a negative boost circuit. So this works just as well as the non boost circuit. It looks a bit weird, but you will find out, it works exactly like a normal boost circuit. You have the bypass diode just like you did before, and you have this diode which conducts during the negative phase, so here and N is positive in the current flow this way through the negative phase because this is the negative voltage and this is the ground. So this is the higher voltage, this is a lower voltage, so flow in this direction. It's excellent one, obviously, conducts during the negative phase. So that's how you do that. So you combine these 2 circuits together. And what you can -- if you do that, you'll see on the next page, this is the standard Totem-Pole PFC actually using diode to the switches here. So we have the bypass diodes here and here. Now you can see what for the bypass diodes because people gets confuse, there is one way, why is it 2 diodes in there, which ones, false, which one is flow? These are just the bypass diodes, don't worry about simple discover a bit of surge can that's about it. Then you have the inductor. You have been the GaN positive phase, S1 acts as the switch I hope I'm not flip this way. I think I've flipped this round. But okay, this 1 axis is we're general positive phase. This is then the diode for the boost. And then during the negative phase, this one acts as a switch and then this one as the diode. The good thing about this is you're also using synchronous rectification, you're replacing what is normally a diode here with the channel MOSFET the reverse conduction channel the MOSFET. And we get also better efficiency at full low by doing that. So good efficiency all around. We still got these diodes in here. And every time we see a diode, we like to replace the MOSFET is what we do. We replaced the diode with MOSFET. The super junction MOSFET here, SiC MOSFETs, good for this application, too. And you can use a GaN would maybe be an overkilling. So in these devices here, these again are only switched on during the positive so -- so 1 of them wish on during the positive phase, 1 of them which is on the negative phase, and that's how that works. So we have the different devices here. One thing to point out is that one of the complexities of this circuit is around 0 crossing. Remember, I said you have to boost up from 0 box where you didn't have to do that on a normal circuit, right, as you've got the Indi Rectifier, the voltage doesn't drop that to 0. So the challenge is then 0 crossing what you do at 0 volts is actually very difficult to control the current. We have very, very short GT cycle to do that. So you get a bit of harmonic distortion around that point. When you get rid of that harmonic distortion -- so okay, harmonic distortion. And the other issue is EMI is if you try to move the harmonic distortion perfectly, you've got quite a lot of switching level, at that level, the low voltage level. So there are current levels and voltage levels, you're going to get a lot of the EMI around that point -- so it's important to balance out the EMI on one side, and we do that with a new large capacitor's to balance out the 0 across the EMI. But if you make those capacitor's too big, you'll get your total harmonic distortion, and that's a problem to do. So you've got to balance out EMI, we took out harmonic distortion with these capacitor's. That's why I've shown these capacitor's in there. This is an important part of the slide. So I mentioned what Totem-pole PFC is. I haven't said that about PFC, I like to put this various reasons I like to put this in this part of the presentation is what is PFC all about. PFC means that you make the AC current in phase and look exactly the same as the AC voltage. There's 2 things you need to watch out for. One is you've got to make sure that phase is the same as the voltage of the power factor correction. And then you've got to make sure that there's no -- the total harmonic distortion is the lower. So if you have a third or fifth order, seventh order harmonics going through your system, you get problems because those harmonics don't contribute to the power of the system. So if you have a third harmonic multiplied by any harmonic multiplied by fundamental end of a zero. So you'll get no power for those harmonics but those harmonics will cause current flow through our system. And there's another point is that the 3 end harmonics that the third, ninth, the fifteenth harmonics, they have to be odds there is no even harmonics in PFC side. They would actually circulate through the rephased system and one gets suppressed. That's a pretty big [indiscernible] to that. So the active front end in 3-phase systems is all about reducing total harmonic distortion. And this is the same on a single phase system that we're looking at here is total harmonic distortion is important which is. Power factor reduction in sorry, power factor. It's good to be close to 1, that's important to get the maximum power. But the overall total harmonic distortion is pretty important, too. So I talked about PFC at this stage because it's related to how you do the control. So how do you control boost PFC. The reason why I showed the two boost model of the Totem-pole PFC is that then you can easily understand from this how the Totem-pole PFC works because we have a normal boost operation during the positive phase, like I showed in the start, and then you have the similar thing happens in the negative phase. But let's have a look at the simple boost stage in this case, a positive phase. What I'd like to show here is what happens when you just think of Boost circuit just in 1 pulse and then just leave it because you often see different wave forms that people don't understand is going on. So I'll say just in 1 pulse and then leave it no difference on symmetric simulation. So I have my 1 pulse here from a boost circuit. This is the gate drive voltage. So when the gate is high, the output is low, the switches to normal. The current will rise linearly. Then you got a voltage here, which is caused by this net input voltage plus the IoT and the actual boost circuit, you get this voltage here and the current will drop to 0. And then when the current drops to 0, this will ring to the input voltage, the input voltage here is set about some 130 volts, just to show, then you can see that it rings down to 0. So if you leave just 1 pulse, that's what we get. Simulation is not drawn. It's that's what happens. So for larger power supplies, we use continuous conduction mode. And this one don't. It's a 300-watt power supply you can run in basically 2 types of discontinuous conduction. So we can switch off when it's coming to 0. That's what people used to do about 20 years ago. They realized that they switching a little bit later, they can reduce losses. The reason for that is that although the current of the voltage is 0 is you still got the charge on the capacitor -- and capacitors, that will be the copper capacitors of the MOSFET to any capacitors in parallel with that, you will have the -- even the boost convertor. It's in parallel high frequency thing you consider that as well. So you can -- you switch off at this much lower level and you save Power. Remember, we've removed the bridge rectifier. So what you find is that to get better efficiency, get better performance. You need to do little time changes on things. So each of these little contributions makes a difference. So we say, okay, we switch off on this. Say okay. That's good enough. We can run on that. We find that the frequency there will go up on the load light. So when the load is light. This pulse is then a little bit shorter because you want to have a lower current right here just over 8 Amps we don't want 8 Amps with time when you've got a sine wave coming, you want to have 1 Amps with low. And so what happens then is you can imagine you switch for shorter time, the frequency goes up. So we've reduced the voltage of the capacitor, but if the frequency of the switch goes up, you're half see the squared X losses will sooner or later they go up. So we say, okay, we're not going to switch on this one, we're going to switch, let's say, on the third one as we've shown it. Just click here where the [house works] nows its a bit funny. We click on the third minimum. We switch on 3 minimum instead. So that will mean the frequency will go down and you have a reasonable compromise between switching losses and switching frequency and switching voltage. You can work out an optimum, there's always going to be an optimum new workout. So you optimize that. So instead of doing this, which is pretty good. You do this, which is even better, and this is what our switches are all about. So Totem-pole is not just about just getting rid of those [redirectifier] diodes. It's about making sure you get the maximum efficiency at every point you can. So remember, we said for PFC, we have to have a variable voltage coming in sine voltage and make sure the current goes in phase without voltage. So just before we do that, we will see what happens when you have current at different levels. We have currency different levels. We have current at a high level here, which we don't have in our circuit, we go continuous conduction. If you want to have a Totem-pole PFC device, which goes a continuous conduction or there is one, just a very similar device NCP1681 but you would need to have 2 additional current sense resistors, current sense transformers sorry, current sense transformers, resistors will waste power. And that makes the system larger but it's needed for higher power. So for 500 watts, we recommend that, for example. So then going to -- is the currency a little bit lower. We will switch on the first minimum and the current stand really low. In this example, we switch on the third minimum. Obviously, there's a place in between where we switch on the second minimum. So what happens is you start off with the sine way. You start off with the pulse is looking like this, you switch on the third minimum and pulse is getting a little bit bigger in between here, you switch on the second minimum and then the pulse is getting much bigger and you switch on the first minimum. And this is what this multimode PFC does. So we have a standard multimode PFC NCP1632 which does this already. And this functionality is exactly the same, more or less exactly the same as in the total, more or less explained in a moment. So that's what we do. So here we are. This is then the -- who's showing the voltage pulses, the gate drive pulses, the PWM you see. You have a much higher frequency at the low current levels, current and the voltage of the same, right. It's going to be written on the cover, but the magnitude is the same. So you see that even when we're switching on the third minimum or whatever the minimum actually is, you would have a -- you're seeing a of frequency [awake]. That's why this frequency mode this multi-mode operation is very important to minimize the number of pulses and keep things going. But it also sees that 0 crossing, we have the problem, of course of boosting up from almost 0 voltage and you have to optimize the year my year. So you have to -- first of all, you have to basically stop switching during that time. That's the best option. Then you have the open loop pulses to drive the synchronous rectification. It's shown originally is a diode, that steering diode, which are shown at the start, which made our positive boost circuit only work in 1 direction. That's good place, of course, by the MOSFET. And then each other slow drive to make sure you get good EMI. So this is the difficult bit is a 0 crossing running into volt, that is same with are 3 phase PFC as well. That's the hard bit to do. We are NCP1680 controller manages that for you. So that's works really well. So that's when we have to highlight that. If you look at the internal functions of the devices, we have the line sensing. So we have -- you have to look at both lines to see the ways of polarity. It's got power supply for the device. And this is the feedback. This is a boost circuit where we have normally 400 volts as the Alpha voltage. So it measures that 400 volts for voltage divider, and if it's too high, it says, "I've got an overvoltage protection. There's things like soft overvoltage protection and a hard overvoltage protection. So it goes a little bit too high. It does a little. It takes a more measured approach at reducing our voltage. But if it sees it going up too highs as I come not messing around and turning off immediately. So this means you have a more reliable operation when you get sudden changes in the higher voltage. Also the voltage goes too low, it will also detect that. So this is what this feedback circuit does for you. We've got the skip mode. That means a very light loads. We want to make sure that the PFC is doing -- having a smart kind of operation. In this case, you -- it's -- operator still works, but it keeps the PFC going, but it's -- you have to work in a different way do [indiscernible] The 0 current detection of the current to measure what is that current 0 crossing detection. This is what we [TT] in does, it measures that current and slow. There's a fall out or pen -- it's something which says here the PFC is okay. Now it's the auxiliary. Remember, we looked at the first or third minimum. This is what detects that. Then you have to drive the circuit. There's no bootstrap driver in the circuit. So we use an external driver. So we have the fast leg drive signals in the fast leg drive drives the boost circuit drives the Boost MOSFET in a positive phase and the boost of diode in the positive phase. And also in light load mode, many you find now that boost diode maybe just leave it and done bother switching it, that saves power. Again, it's a minor difference in power, but that gives you an extra advantage. You need to do everything you can to improve the efficiency once you've taken out the big loss, especially in GaN itself has so little losses, you've got a conduction obsolete about it. So you have to look at other losses to get things more efficient. So here we have the slow leg drive signals. The slow leg is -- remember, that's the diode, which is switched on, only during a positive phase and we made that synchronous. So we drive that. And again, if we think that it's not worth turning that diode, that MOSFET on to be a synchronous MOSFET in parallel with our diode and just leave it leaves a diode. And then that's again. So there's a lot of smart control in these devices that do to make sure you get that excellent efficiency of light load. You have the driver. So we have -- we're switching GaN. We better switch it fast. So we have to have good delay matching, absolutely exactly where switch is imposed to have good matching and good propagation delay, which you can predict. And this is used for the fast [indiscernible]. The slower leg using a different device, but we could just as easily use this device in this [indiscernible] conduction as well. So again, good product, optimize the drive of the devices. So I talked about the PFC. Let's talk about the LLC converter here. So just show the LLC. So this we use a high-speed LLC because it makes the transformer smaller. So let's have a look at the LLC actually put the next slide, that's a better idea is here is the -- so we have the L, which is here. The L, and I'm pretty sure that this bank of capacitors here is the C. So you have a circuit with LLC. This is what it's called LLC. So whats this all about. You have a half bridge, which is switch with a kind of squarely. So as we're engineers, we say, square wave is approximately equal to a sine wave that's called the fundamental approximation. Then you apply a sine wave here because you say it's a sine wave. So it's first on that square wave. It's going to be a sine wave, and it's multiplied by 4 pie the top. So -- sorry, it's 4 over pie is the gain. That's why the LCC you see you've got a gain of slightly over 1 as a result of the pie. It's 1 of the reasons why you it's together. Anyway, so you put in the sine way and then you put it into some kind of network. So this is a network. So you have LLC, you have a transformer and you have a couple of diodes in the output stage, I'll show you that at the moment. And then you have a network I think this is also a good time to show the diodes here is the diodes you normally see is a diode coming out. So it's difficult to show that at this -- a this point, right? So the current starts here and ends up here. So the diodes goes in that direction. And -- but you want to have it reference to the ground, so you put in the other direction. I noticed the diodes are done so that they always work like this, and you can put this in and you basically reduce the losses of the diode. And then here, you have the same approach. It's just a negative phase. So this is the -- this shows synchronous rectification. Let's get back though to this network, right? So what this network does is it causes a certain gain characteristic. So I mentioned from our fundamental approximation, we've got this 4 divided by pie term, and that is going to be over plus over 1 is 3.1, so 4 of the 3 is going to going to give you a gains curve. We also got a bit of residents because of the circuit. So because of that, you can get a gain just above 1. What you also see, unfortunately, is that a gain has no change of frequency doesn't change much. If you could give another example of what was going to show you a -- flyback converter. And I showed you the transfer function of the flyback converter, how the gain changes as you change the duty cycle. So you have the duty cycle starting at 0, you have 0 gain and the duty cycle is 1, right? They'd be linear, right? You can change -- you can a gain of 0.1, 0.2 or 0.7, all right? And then the terms ratio at different gains. But you can't have a wide range of -- so you have a wide range of gain of flyback converter, but you don't have that wide range of gain on this LLC. So you see it also is relatively insensitive to low the load and you do the transfer function acts like a damping the Q factor in the gain characteristics. So you -- where you have the second resonance point or which is this one here. And if you operate around here, you can get a very low change in frequency. So you can operate in a number of places but often around here is good. But the disadvantage again, you can't have a wide output voltage range, okay? So if you have a fixed input voltage ranges, the output of the PFC is 400 volts. You can't really change it very easy. You have a fixed 48-watt output, that's perfect. If you have to have a USB Type C where they tell you to go from 3.3 volts to I think 21 volts. That's difficult. You're going to need to have intermediate, but converted to help you out there. So this is the gain curve. So why do we do all this stuff, right, is making it so if we change the frequency, we can change the gain. The reason we do this is we do 0 voltage switching on this device. So if you take this device, we see that we're switching the square wave. This is all square wave. And here, we just forget that we said it's a sine wave, we using a square wave now. So what happens is, at this point, this device has switched on and current goes from here, here round here. We switched both of them off on both on at the same time, and we have the current going to the diode up through here, through here too. So the current is going negative direction here. So the voltage here is at minus 1 voltage. And then you -- so at this point, just before you switch on, you've got the full current going through it, but the voltage here is then minus 1 volt. When you turn it on at minus 1 volt, and switching losses up very well. We also find out because the effect of capacitor switch off losses are quite low as well. So the main contributor is the [output capacitor] of the device. So that's exactly the reason why we use this difficult approach. It gives you the advantage of soft switching, which makes it more efficient, which means you can run at a high switching frequency, Which means you can make transformers more [indiscernible] than you can get some pretty small white base on the LLC. Let me have a current mode control. Current mode control is something which is standard on PWM controllers but it's not so standard on LLC controllers. So what does current mode do. Let's look at what voltage mode does first. It means better to explain that. So let's say we have a change in the load here, right? So if you're a change in the load, the -- it will say, "Hey, I want to have more power." You're getting more gain. So you get more gains in Standard PWM circuit, you have a high-duty cycle, we're getting more gain by having a lower switching frequency we saw in that curve. So we reduced the switch frequency. So reduce switching frequency and then this system settles, and it takes a long time to do it. So -- and the transfer function of this is quite complicated when you got voltage. If you do a current mode, and that's where we be little bit beyond the scope is, where transfer function is simpler easier to compensate for and better. So that's hard to show here, but you get a better done -- better [non transient] response. It is easy to show even though it's less of a usual event is a [line transient] response. So if you have a current here and you use this current in this way for and you say, "okay, I'm going to stop on that current hits a certain level." We will go in a little bit explain briefly where that current comes from later. We said that the average current level, if that's suddenly higher, let's say it's some 10% higher. We would have 10% of the load gain, which we stick to free a little bit more, and you do that immediately. Because if you had a voltage mode and you have that, it will say, "okay, my voltage, I'm getting less power certainly and going through all the dynamics as we move, right?" it takes a long time to respond. We had a voltage mode and a line transient response. And if you -- but if the current goes up. The one trick here is that the current shape on LLC is a little more complex. Remember, I said that it goes to the diode. That means the current style so is kind of negative as shown here it also is negative, right? Which current you measure? Do you measure here or here or here. What you actually do in our circle is we measure the [indiscernible] charge. It's actually charging. So basically, putting an integrator added up and if the integrator current is the thing which determines the switch. So we get excellent line transient response. And the reason you get a good load transient response gets much simplified control dynamics of both load to output in line output transient functions. So current mode control really helps you a lot. And one good -- other good advantage for the current mode control is you don't get these weird mode of operation where you have the -- basically what happens when you go below resonance, so you can go below resonance on normal control, but with -- and you go into a hard switching mode, which is there on the devices. You avoid that hard switching mode which happens we go with a resonance when it uses current mode because current mode control effectively improve -- as impacted and benefited this. So it's a high-level block diagram. We see the current sense, which measures the current, the feedback, which measures the output voltage. So those 2 things are combined for the control as we just mentioned, a voltage temperature protection, you have skip mode, and that's quite complicated to explain, but it's very important to have a low power consumption mode. We go first into a mode where we do PWM control. It's current mode control, if you do that, will explain. And then you have a special skipping mode and that's quite -- that really is quite detailed to explain beyond this discussion. The high-voltage line sensing. And in this case, we have a bootstrap circuit driving our sort of upper and lower devices with half bridge high side. So the devices are driven directly. It's a motor control logic for helping -- saying with -- turning the PFC whether to go in low power because PFC itself doesn't know where it's going to go low power. So as our application engineering. What was the main problem when people do this design. So the main -- that's pretty robust design, there's not much you can do wrong. What people often do wrong is to don't follow the their recommendations. They have star connections everywhere and things like that is you've got to be star alight, connections correctly and make sure we got the right loops. So follow the recommend -- please follow the recommendation for the layout. So all the recommendation. Synchronous rectification. We -- at the last point, we need to have synchronous rectification. So I'm not sure how well you can see this on the screen. It's probably pretty small, but we're using no inductors MOSFETs here. And of course, the NCP4306. So again, I showed you earlier, we have the diodes in normally have discrete diodes and then you replace the MOSFET and again, you can get -- improve the efficiency. I said sometimes it's better not switching them because the switching losses in the gate driver are higher than the benefit you gained from switching to device on the first place, and this is what it's light load threshold. So you can set the light load threshold with the resistor. So below that, they said, okay, you're running a light load, I'm just going to not bother to and that's going to give you again better light load efficiency. So there's a lot of details in those small things but gives you again better efficiency. There's 2 resistors we saw that, which set the minimum 1 and then in [indiscernible]. Because of inductances, you can get spike. There's a little spike here, which we've shown from the inductive ringing. This would cause a false triggering if you didn't have this. And so you have a minimum on-time minimum of time to run very well. So what happens is you have the here at this point, the diode conducts, right? And then you turn the MOSFET on and then you get a lower voltage. That's why you turn the MOSFET on, first place. And these MOSFET's have got very low audio, very, very low voltage, that is very, very sensitive to inductance. So you get wrong triggering or incorrect triggering if you don't have a low inductance layout and low inductants devices. And why is this a kind of resin. You can see where we've got a resin current depending on operating merging might see different ones. But this is a good example of resin currents. As we somewhat close to resins, you would see this way for. And then this is the current through there. And yes, this is the current wave from, of course, this is also the threshold. And then you turn it off and get good efficiency level. Just I'd like to always look at the data here and saw a pretty good example. Unfortunately, it doesn't say what type of SMP device we use, but if you use T0-220 on the top side, you see that the premature turnoff of the switch because of the inductance. Again, you say why is inductance so simple. Well you're measuring in millivolts, right? So that does make a big difference. If you use a good inductance device, you gonna get much better [frequency]. And again, if you notice here the difficult to see here, but you can see 1 side, you don't have a non pins sticking out you kind of soldered it together in one point. You see a lot more common [indiscernible] package. Here Q1 instead of having 1, 2 -- okay. It's 1, 2, 3, 4, right? It's -- they're all bank together on the package, why? low inductance. We've had a separate to be better inductance. Inductance is really important. And that means don't mess lay out. This is the recommended layout, which you should use when you use these devises. There's an awful lot of features in the NCP4306 when unusual operating amount of LLC, which shouldn't actually happen with our current mode control of a voltage mode devices. It supports that really well. And there's a lot of detail on the very simple device, the NCP4306. So last least I want to go on to the ultra-high density power supplies. I said we can get a power supply like this -- so we tried 2 approaches. One is to use the LLC because I think the LLC is really good, but for USBPD approach. So analysis is great when you have a fixed output voltage. When you have a variable output voltage like USBPD, USBPD is where the low says, "today want 3.3 volts, but tomorrow, 21 volt or whatever." So it's you can request different voltages from that without changing anything. It's very powerful approach. But that's going to be difficult to deal with for LLC controllers. So we need a follow-on backstage and this is our solution now, and it's 30 watts per square inch with the USBPD. So we now -- you can look at the circuit later on -- we have then the this size, which is -- this is in centimeters, whereas at [indiscernible]. We are 89 millimeters by 51 millimeters by 21 millimeters. So it's just short of an inch stake like be as thick as this, of course, high to the components in this size, the 240 watts with USB is a simualted USB but it's pretty cool, very, very high power debts, 40 watts per square inch -- cubic inch, I'm sorry. And that's very impressive. We're using then the -- here, what we're doing is we're using a quasi resonant blowback and [indiscernible] might -- so -- but this is, again, it's something I'd just like to talk about or mention at the end of the presentation. So we're developing new products, new solutions all the time, getting that power density down, getting the efficiency up and coming up with solutions which are programmable, make a lot of sense. So with that, I'd like to hand back to Emile.

Sorry about that the way -- thank you so much, John, for that great presentation. Before we end today's webinar, we do have a question, Jon, if you'd like to answer that.

Yes, sure. There's a question from -- can you apply this approach to 3-phase power. The approach on the PFC is fundamentally different on 3 phase power. 3 phases, you could use 3 channels to do that, but I would maybe recommend you instead of using the 3 phases of that, which it is actually possible to do that. But the rectifier approach is on common use of approach or a bidirectional approach using 6 switches would be beneficial. Often to 3-phase people are now looking for putting energy back into the grid. So being a rectifier using MOSFET so the diodes. So it's kind of TMPC kind of backwards. There is one approach use for an active 6 switch for 6 switch converter an active [indiscernible] is used. So you could use it. If you want to have something which works with 3-phase and single phase, then I've seen that news is used for internal charge, something like that. So that can be used, but I would just like to say that's -- yes, it's possible, has its applications and lower -- so we can actually do that. But there are also alternatives to consider like 6 swich are more front end, active front end and the PN rectifier. So thanks.

Yes. Thank you. And before we leave, I would like to ask everyone to take our survey, which should be centered on all of your screens. Your feedback is very, very helpful to us. Also, like John had mentioned earlier, if you would like to download today's slides, you are able to do that in the related content box on your left, which I am highlighting on all your screens down as well. Thank you so much for joining us today. On behalf of myself, John and onsemi. We really appreciate you taking time out of your day to join us with this webinar. We have a lot coming up in the coming months. So please register for that, and we hope to see you soon. Thank you.

Thanks very much, everybody. Bye.

Bye.