ON Semiconductor Corporation (ON) Earnings Call Transcript & Summary

Hello, everyone, and welcome to WE meet @ Digital Days. My name is [indiscernible], and I will moderate this presentation. We are very pleased that you took the time to participate in our virtual conference. The topic of this presentation is: Meeting Challenging Efficiency Standards with Bridgeless Totem Pole Power Factor Correction. The speaker is Alessandro Maggioni from our partner from onsemi. But before we start, I would like to point out one thing. You have been muted during this presentation. This means that you can't ask questions via microphone during the presentation. But nevertheless, you have the opportunity to ask questions during the presentations at any time via the chat function, and you can find the chat function in the webinar control panel. This presentation will be about 30 minutes long. The chat questions will be answered in a Q&A session following the webinar. There are about 5 to 10 minutes in addition scheduled for this, and if we are unable to answer all of your questions within this time, we will answer them via e-mail afterwards. And if you still have any other questions left, just mail us at [email protected], and we will, of course, try to answer all of your questions promptly. At the end of the presentation, you will be asked to participate in a feedback survey. We would be pleased if you take the time to fill out the survey and help us to improve our events. You will also receive the link to the presentation in the next few days, and the recording of this session will be available at our website shortly. But now it's time to hand over to our speaker, Alessandro, and I wish you an exciting presentation.

Good afternoon. Can you see my screen? So thanks, first of all, good afternoon. So thanks to Wurth Elektronik teams for these opportunities and welcome to the session of Digital Days. My name is Alessandro Maggioni, so -- and I'm responsible as Regional Marketing Manager for the EMEA region for the Advanced Solutions Group in [ San Jose ]. In this presentation, we will try to highlight how to achieve the Bridgeless Totem Pole Power Factor Corrector, the most challenging efficiency standard required in -- by the market. So this is the agenda of today. So we will start with a...

We have one remark, we actually cannot see your screen right now. Maybe you can try it again. Now, perfect.

Okay. So sorry for that. So here is the agenda of today, so we will start first with a small introduction of the AC/DC power supply market, where our use, how big it is and the trend. So then we will talk about PFC or power factor correction, so why it's needed, so the solution can be implemented and the control techniques, especially in the CrM or critical conduction mode. High efficiency. Why there is a growing demand of energy efficiency solution? So a comparison between bridge and bridgeless totem pole and which perform better. We will have a look on our solution for Bridgeless Totem Pole PFC Controller, so our own NCP1680 based on a CrM control mode. We will also spend some time and some slide introducing our evaluation board, based on NCP1680, able to achieve up to 300 watts with very high efficiency. And then at the end, of course, some time is available for any questions and rapid answers, so let's start. AC/DC power supply landscape. AC/DC power supply are extensively used in several vertical markets, including industrial, automotive, medical, smart cities, which is quite wide, [indiscernible] and cloud. More recently, the growing adoption of the wide bandgap devices such as gallium nitride, GaN, and silicon carbide and has directly impacted expansion of AC/DC power supply market adapter. Over the past few years, only the advancement in this technology, so GaN and SiC technologies have increasing become more and more feasible, also for commercial application. AC/DC power supply are still the major power supply product type in the market, so accounting for more than 85% of the market revenue, so this was back in 2014. We have the DC/DC converters accounting for the rest. This is [ raising the drive for ] the manufacturer, so as the demand of all of these products in the above market segment is increasing. So the production of AC/DC power supply is also rising. The back of these factors, so along with the focus on the revamping the electronics sectors, the global AC/DC power supply adapter market is expected to obtain a market value of around $12.9 billion, so this by the end of 2027, and is expected to grow at a compound annual growth rate of around 5%. In Europe, so the so-called Europe 5, so this means Germany, U.K., France, Italy and Spain, all of the biggest share of the Europe power supply market. This is owing to the strong growth of industrial electronics, also industrial automotive sectors in these countries. PFC, so now let's talk about why the solution and different control techniques. PFC, why? So that's the problem, okay? So AC power has 2 components, so the real power of the active power, so also sometimes called average power, which is expressed in watt. And then we have reactive power, so usually expect -- expressed in reactive of volt, amperes. As power is transferred along the transmission line, it does not consist purely of a real power that can do the work once plastered to the load, but rather consist of a combination of real and reactive power, so which is called upper-end power. The power factor of an AC power system is the same as the ratio of the real power assembled by the load to the upper-end power flowing to the circuit. A power factor of less than 1 indicates the voltage and the current are not in phase, so this means reducing the average product of the 2. And usually the current waveform does not follow the voltage [ width ]. Here, we have an example of waveform on SMPS without [ NPFC ], so here, you can see on top the voltage waveform, and here, you can see the input current wave. Here, voltage and current are more or less in phase, so this is great. So this means that I will have a good power factor, unfortunately, wrong. So measures here, the power factor is actually 0.6. So this resulted not only in power losses but may also cause harmonics that travels down to the neutral line and disrupt other devices connected to the line. Here, you can see -- in the same condition, but adding PFC stage, that current and voltage waveforms are very close each other, so with an improvement in the THD reducing all of the [ others ]. How to move from left, so this is the -- without PFC, from the right, which is the solution. So you need to shape, remember, you need to shape the input current in order to match the input voltage wave. So how? You set a switch mode that would convert the stage between rectifiers and the bulk storage cap. So now, we have a simple example of why you want to have a very good power factor in your AC power line. So here, you have 2 glasses of beer, so both are objectively full. And if you go to a pub or in a bar, so you will pay, in both cases, as a full glasses. But on the left side, you have a very poor power factor, so around 0.6, like here. While on the right side, you have a very good power factor, so well above 0.9. So I think everyone prefer to have the right one, or am I wrong? Control techniques. This is one of the control technique. There are several control techniques in the power factor correction. So CrM, or critical conduction mode, is one of the example, and here is we will multiply. So remember the goal. So the goal is to shape the input current to match the input voltage waveform. So critical conduction mode means that inductor current always goes to 0. Here on the left side, you can see a typical block diagram of a boost converter. So with internal reference multiplier. So the current shaping control varies the power switch, as we described here, so in the step 1 to 5. [ Digital ]. The internal reference multiplier, here generated the reference voltage, so called VREF, which is a scaled version of the rectifier AC input shape, coming from the AC input beam from via resistor divide. When the power switches on, so the current to the inductor ramps up until the voltage across the shunt resistor, so which is this here, is equal to the reference voltage. When the power switch is off, the current to the inductor ramps down until the current reaches 0. This is the result that average current is then equal to the peak current in the inductor divided by 2, it's the triangular waveform. So long story short. So basically, you are banging from the inductor [indiscernible] current, [ sends through ] resistant shunt, to the inductor 0 current. This is one of many examples, so, unfortunately, we don't have enough time to cover everything in this presentation. High efficiency. Which is better, bridgeless versus bridge? So why there is this kind of demand? There is a high demand of energy efficiency solution. Over the past decade, environment concern across an array of industrial domain have forced the players in the global AC to DC power supplier adapter market to lean toward the development of energy efficiency [ component ]. Most of the world's electrical energy is supplied through the AC/DC power supply, so as we saw in the first slide, so more than 85% of the total power supplies. And they are found in almost all [ means ] our equipments and devices. So this, meaning, that [ very ] efficiency has a huge impact on the operating cost as well as contributed significantly to the emission. Here, you can see the typical PFC circuit with a conventional boost. So the input bridge rectifier found in almost all the DC/DC power suppliers contribute losses that present a challenge in achieving the highest possible efficiency figures. However, removing the traditional diode bridge, the PFC [ fab ] and the boost diode in favor of bridges totem pole PFC around -- then announced the efficiency for the use of active switches. Most commonly, PFC use a boost converter to derive a DC level higher than the main voltage peak from rectifier [indiscernible], so this DC level is typically 395 volt for power supply design for a universal input, so 90 to 260 volt AC. It is then regulated using an insulated DC conversion stage to produce the required DC output voltage for the PSUs, for the power supply unit. A valuable byproduct of this is that the line requirement. The current follow the line voltage waveform so giving in theory at least, [ AC ] power factor. The approach [indiscernible] is effective, and PFC can be designed to operate in several modes, so the continuous, discontinuous or critical conduction mode, so the acronym are CCM, DCM and CrM, which are largely defined by whatever energy in the boosting adaptor fully exhausted during each cycle. Normally, there is a 2% loss within the CDC stage and 1% in the line rectification and PFC stage. Up close, this is much closer to 2% during the operation and below line. With close to 4% losses at low line, there is a significant challenge in meeting, for example, this 80-plus Titanium standard level that requires to have up to 96% efficiency at [ 130 ] volt AC input at 50% volt. So this is a specification which is commonly used, for example, in [indiscernible]. Efficiency. Bridgeless versus bridge, how to make a boost PFC more efficient? You always have to remember the golden rules, so better efficiency with fewer devices in the conduction path. On the right side, you have the typical PFC with bridge diodes. So you can see here in orange, you have the positive half cycles, so when -- which of the devices are conducting. And in the dark gray, you have a negative half cycles. So here you have the input bridge, you have the boost set and the boost diode. You notice both in the half cycle, positive and the negative cycles, you always have free devices as conduction. On the left side, here, we have the bridgeless totem pole, so as you can see its output. So the topology consists of 2 half bridge configurations. One half bridge commonly refer as a fast leg here on the right side, which switches at the PW frequency. On the other, commonly refer as the slow leg switches at the [indiscernible] frequency. The fast leg switches perform the role of the switch in the diode in the classical boost PFC, so that is the switch's function to regulate the output voltage and shaping to the input current to provide a high-power factor and new harmonic construction. The slow leg switches perform the role of the diode bridge in a classical boost PFC, so active switches with low [ or no ] resistance are utilized instead of diode, resulting of an improved efficiency. The totem pole PFC operates with only one slow leg and one fast leg device in the conduction path, whereas the conventional boost PFC, as we discussed before, operates with 2 bridge diodes and 1 active switch, [ almost ] diode in the conduction path. So this means that fewer devices in the conduction path and active switches replacing bridgeless diode allow totem pole PFC topology to achieve the higher system efficiency and the power density than a classical boost PFC. Also, totem pole PFC is a derivative of a standard boost PFC convertors where the bridge diodes are removed and therefore, the booster circuitry, that interferes with the AC input. So the key challenges with this approach are detection of the line polarity, current cells and current limit implementation. Further, totem pole PFC is a simplest boost converter, so as opposed to the standard booster PFC implementation. So therefore, a zero current detection scheme needs to be implemented. As you can see, an [ arc ] rectifier is [ live ]. All the board mentioned requirements are, of course, implemented in NCP1680. It is also important to note the similarities to a standard boost PFC. For example, the criteria from the selecting the boosting that of the output capacitor and the boost tech are exactly the same between the totem pole PFC and the standard booster PFC. Now, let's have a look a little bit more in details about our NCP1680 Bridgeless Totem Pole PFC Controller. Here, we have a table, which is explaining the main topologies for power factor correction. Here, we can go from a standard interleaved booster, semi-bridgeless booster, interleave critical conduction mode totem pole, and the new bridgeless totem pole solution. So which are the advantages? First of all, the transistor. So for a CrM, you can choose whatever kind of FET you want: standard silicone, super junction FET, wide bandgap so GaN, or silicon carbide. While for the CCM mode, so this is for high power, we are strongly suggest to use only wide bandgap GaN or silicon carbide. The switching frequency, so which is much higher. In CrM, we can achieve up to 500 kilohertz. Theoretically, up to 1 megahertz is possible as well. The advantage are, of course, on highest efficiency due to also to the low component count. In case of using a GaN, you also have 0 Qrr, so this means are pumping up again a little bit higher in frequency. And then, we discussed before, also you can have the highest power density cost. The cost is more or less aligned with the other solution. Of course, considering that this cost can be also, let's say, various depending on which kind of power FET that you are using. And then you can see the big jump in terms of efficiency, which is very close or even higher than 99%. Which of the application you can use for -- or you can have for a bridgeless totem pole PFC? Basically, everywhere. So everywhere, will you need a PFC. Can be data centers, [indiscernible] power supply, high power LED street lighting, industrial power supply, chargers, PSU, UPS, high power adapter, 5G telecom power supply, external power adapter. So basically, here, you have to think that you can use it to match all the applications where, of course, a PFC is needed; and then efficiency and compactness, a real, a critical parameter. So let's have a look a little bit more in detail of the device. So the NCP1680 is a critical conduction mode power factor correction controller. I see, which is designed to drive the Bridgeless Totem Pole PFC Controller. Remember, so the bridgeless totem pole PFC consists of 2 totem pole legs: a fast switching leg driven by the PWM switching frequency and the second leg that operated at the AC line frequency. The topology eliminates the diode bridge present and the input of conventional PFC frequency, so allows significant improvement in efficiency and power [ debt ]. The NCP1680 is capable of constant on-time CrM and valley synchronized frequency foldback. First of all, [indiscernible] we have a proprietary current sensing architecture and proven control algorithms. The NCP1680 allows for a cost-effective solution without jeopardizing the performance. Which are the main feature? First of all, we have industry-first mixed-signal controller which is based on a state machine core, so which is dedicated to totem pole PFC topology. Easy to implement compared to existing totem pole solution. Cost-effective, fast go-to-market, and no, you don't need to expensive software updates. It's based on a CrM architecture. So during the CrM, if the load is high, the switching frequency is low. Now when the load reduces, the frequency increase, so this is because the inductor current is not going to the load, so this allows to optimize the performance across the whole load range. Frequency foldback is the first line of defense, so we reduced the switching frequency at light mode, but when you still have the valley. So the second line of defense is the valley switching, so we actively detect valleys to check the safety. It's based on a proprietary current sensing, so we have proprietary sensing of the zero-cost detection resistor, so we can also detect the peak current [ day ]. We also use it for current limiting [ ecotech ], which is a more elegant solution. Traditional totem pole use a digital controller like MCU or a DSP, so which are necessitated of firmware and are more complex. So they are also controlling the switch timing using [ all 5 asset ] sensor, so our solution replaces this [indiscernible]. So [indiscernible] that over the control loop is internally compensated and this means that also the external component can be -- some external component can be [ removed ]. Of course, it also comes with the main safety feature required like burn-out detection, power factor authenticators, sort of power [ good ] for the cascade application, soft and faster voltage protection, bulk under voltage protection, so internal shutdown and a cycle-by-cycle current. Which are the market and application? Basically, it can goes in all the applications where a PFC with high density and high efficiency, up to 350 watt maximum is needed. But here's the limit of the CrM topology, not of the device. And then it is housed in an industry-friendly SOIC-16 lead. Which are the totem pole scenario today and which is advantage from us? So again, as we discussed, so existing solutions on the market are based on complex MCU-based solutions. So customer needs to write software code, costly current sensing method and to implement this topology. 1680 is the industry's first mixed-signal controller state machine core, so which is dedicated -- fully dedicated to totem pole PFC topologies. So now you have to choose your path wisely. So today, if you want to look at totem pole PFC solution, you have only 2 choice. Or you can go with mixed signal, and this may move to onsemi NCP1680, which is based on a simple sensors reinstalled the current sensor, or you can move to the rest of the market. So MDS, you have to work on the full ecosystem. So this complex solutions, [indiscernible], software development kit, firmware and so on. So on the other side, NCP1680 is a simple, integrated, coding-free and excellent light load efficiency and standby power. Moreover, we just received, a couple of weeks ago, the Electronic Design 2021 PowerBest award. Now, let's have a look, a few slides about the demo board based on NCP1680, so 300-watt demo. The demo, as we discussed, is up to 300 watts, and it is then utilizing 50 million GaN FET. So the evaluation board -- motherboard and the daughter card are shown here, so you can see the daughter cards put in vertical and here is the motherboard. So here you can see them and inspect, so input voltage range which is universal from 90 to 265 VAC, the line frequency range is 47 to 63 hertz. So the output voltage is the standard PFC voltage, so 395 volt, and the output power is up to 300 watts. And then also, you can see here all the components provided by our friends at Wurth Elektronik like inductor, differential, the PFC inductor, the common mode choke and the [indiscernible]. As shown in the picture here, the slow leg switches are high-voltage silicone-based FET, so also known as the super junction FET. And the fast leg switches on the daughter board, here, [ unannounced ] mode gallium nitride, so GaN devices. Since the NCP1680 employs a CrM control architecture where the inductor current reset back to 0 before the next switching cycle, low reverse recovery charge Qrr of the super junction FET can also be utilized for fast leg, albeit with slight interior technical performances, but better cost structure. As a controller, the NCP1680 is completely agnostic to the fast leg switches technology, but wide bandgap devices such silicon carbide or GaN are strongly recommended for optimal performance. SiC is a good choice for lower frequency application while GaN is an excellent choice for both low frequency and high frequency application. The NCP1680 version board is designed such that engineer is interesting, this novel topology can easily probe value signal and learn the intricate of totem pole PFC. The motherboard includes multiple IO connector and touch-point to simplify instrumentation and away from capture during the evaluation process. The fast leg, half bridge, is implemented on adapter [ code ] here, where the fast leg switches are driven using our unseen NCP51820, an high-voltage GaN half bridge driver, up to 650 volt, while the slow leg switches are driven using the NCP51520, so high-voltage silicone FET half bridge driver. The NCP1680 employs another current scheme where a simple resistor placed in the return path by ground and AC ground, is utilized for current [indiscernible]. So the 0 current detection resistor is further utilized to drive the control of the synchro switch in the fast leg. Additionally, the NCP1680 requires only a single auxiliary [ winding ] to sense a switch load balance in the positive half line cycle and switching of peaks in the negative half line cycle. These other schemes result in the main boost switch being turned on with a minimum voltage across the switching is also improving efficiency and reducing the amount. Let's have a look now on some data, so here is the efficiency data. So you can see the efficiency plot measure on the bill. So here, you can see that across the world of load range, especially with the high line, we are quite well all above 98.5% which peaked close to 99% with a full load. While for the low line, of course, it's a little bit lower, but it's always well above 97%. Here, I would like to highlight the value at 50% load. So as you remember, in the first slide, so 80-plus Titanium standard levels require up to 96% of efficiency at 250 volts input at 50% lower. So here, we are quite well above -- more than 2.5% higher than that specification, and this is more important. It's very important for efficiency and power consumption. You can see also the plot diagram for the THD. The device has a feature, another feature of the 1680 that is derived from the 0 across the voltage is THD enhancers. How it works? Simplifying, basically, it produce a small on-time extension proportional to the time duration of the 0 cost detection voltage and is less than the medium threshold. Typically, it's lighter load and the near to the AZ -- to the AC 0 crossing, so the amplitude of the inductor current reduces and the 0 cost detection voltage will be spent longer duration below this medium threshold, so this is resulting in a larger on-time extension. The THD enhancer work also in conjunction with an internal [indiscernible], so this allowed the NCP1680 to achieve a THD which is below 10% at the medium to a de-load across the universal input rate. Here, you can see also the power factors. So for the low line, we are very, very close to 1. I think he has not really [ write ] it, but we are up 99 point something percent. While for the low line -- for the high line, sorry, we are well above 9% -- 90% from the half of the load. So very -- quite very good power factor correction. One important things when we're talking about [ booster ] density is also the thermal -- the thermal images, so thermal scan. The NCP1680 version board and daughter cards were also evaluated for thermal performance. So in the worst condition while operating at 90-volt AC and 300-volt. This actually is the worst shape -- worst case scenario. The thermal image of the fast leg, GaN FET, so the boost inductor, and this slow leg sequence FET are shown here. So these images were captured in a 25 degrees ambient environment, so with no [indiscernible] high flow. The high efficiency performance of the totem pole PFC is evident here in the device temperature. So we have the fast and the slow leg switches both measure below 60 degrees, a modest -- so this means a modest 35 degrees rise above the temperature. The daughter board PFC -- PCB card is also designed in a manner that eliminates the need for additional heat sink to be mounted to the board. If you remember from the picture of the board, there was no heat sink at all on the board. The PCB internal copper plate function as a heat sink and the temperature rise of the fast leg switches is well-controlled by this copper plate. Of course, if you want to look more deeply on our NCP1680 web page, you will find the user manual, the data sheet, of course, of the board, and also some materials about tip and tricks and how to get the best performance from this device. I think I'm perfectly on time, and then we can have some times for any questions and answers.

Yes. Perfect. Thank you so much, Alessandro. I really liked it. I hope the audience, too. So as you already mentioned, you have now time to enter your questions in the chat or question window, so feel free to enter them. Let's check the questions. Maybe the first one. Why is wide bandgap semiconductor needed for totem pole PFC?

That's a good question. So in which case, we talk about, especially CCM, critical -- continuous conduction mode totem pole PFC being a half switch topology and GaN has no inner and body diode, so there is no reverse recovery charge, so that would contribute to switching losses. Hence, enabling ultimate PFC efficiency to be achieved. The body diode of a silicon carbide, MOSFET, still has a non-0 reverse recovery charge but is much lower, for example, compared to a standard silicone MOSFET. So these expect that efficiency of the totem pole PFC to be a little higher with GaN compared to the silicon carbide, and of course, much higher compared to the standard silicone.

Thanks a lot. Another question, I think it is related to the evaluation board. Why are you limiting the application range to 350 watts?

This is not a limit of the demo board or the device. This is a limit of the topology, so this CrM. And it's a limited of -- because if you remember how it works, so the current inductor moved from 0 to the peak, so this is just a limit of the CrM topology. It's not a limit of the device itself. So if you want to move to higher power, then you need to use a different [ control ] mix, for example, multi-mode or a CCM.

Okay. Thanks. And another question. Why CrM is a balanced approach?

CrM is a balanced approach because -- well, okay. If neither CCM or DCM present an ideal solution. So many designs use critical conduction mode which is suitable for up to a few hundred watts, as we discussed just the previous question. So CrM manage the switch frequency to operate on the border between DCM and CCM, so in response of a change in line voltage and low current. This benefit from low thermal losses and, as a peak current is meant to be double of the average current, so the core and the conduction losses remain acceptable.

Great. Thanks a lot. Another question. Can this topology work for 3 phase PFC?

No. PFC is not working as -- for example, I interleave PFC. It's only for single phase.

Okay. Another question to the frequency. Which is the switching frequency or the range of frequency?

So the device, it comes with different options, so it's a trimming option. It depends on the part numbers you choose. So in this case, we have released 2 versions. One is clamped up 230 kilohertz, if I remember well. The other one is above the 270. But of course, we have the possibility to go much higher, even close to 400, but not yet released.

Okay. Let me check. There is a question. How about surge voltage EMI robustness for this totem pole PFC topology?

Sorry, can you repeat the question?

Yes, sure. Of course. How about surge voltage EMI robustness for this totem pole PFC?

It's quite well compared also to the other PFC topology. So we already performed several tests on this -- even on this board, and it passed quite well on the test. But this is something we can discuss even offline [indiscernible] for more information of test.

Okay, great. I would say we will answer all the other questions afterwards via e-mail, so don't worry, all of your questions will be answered, of course. But for now, thank you so much, Alessandro, for your interesting presentation, also for the question-and-answer round. And thanks to all of you for your attention. Yes, you will get the slide of this presentation afterwards via e-mail, and also the recording will be listenable on our website shortly. A short hint from my side. At 2 p.m., there will be another Digital Days' session, which is called Filter Components, Mode of Operation and Selection Criteria from our Field Application Engineer, Mario Moller. So thanks to all of you. Thanks, Alessandro, and have a nice day. Bye.

Thanks a lot. You too. Bye-bye.