字幕表 動画を再生する 英語字幕をプリント Hello my name is Andy Reiter, I'm with Microchips Power Supply Applications Group. The topic we would cover today is our new CIP Hybrid Power Starter Kit. This starter kit is introducing the PIC16F176X and 7X families of freely programmable pwm microcontrollers. These microcontrollers have specific peripherals that are running independently from the core and are highly configurable so that it can configure them or interconnect them in specific ways to form some PWM controller architectures. So the advantage of this freely configurable peripherals are that you can tailor the PWM controller capabilities specifically to your very application. To allow designers to utilize these capabilities we also have introduced a new design tool chain, which is encapsulated in our development environment MPLAB X. So this is part of our Microchip Code Configurator tool called MCC, and is a set of sub libraries that allows you to simplify the configuration of multiple peripherals at the same time. We have set up this assembly in this demo here where we have a starter kit on the left and here on the right on the screen you see this MPLAB X environment with the specific switch mode power supply library subset for Microchips Code Configurator. So this tool now can be used to set up PWM controllers that support different control modes like peak current load control, voltage mode control, average motor control. You can combine multiple control loops, you can add in specific fault handlers and the device itself then supports you with setting specific thresholds like shutdown levels for maximum currents, voltage clamping, inputs under voltage over voltage lockout and further features by utilizing the internal DACs. So the variety of things you can do with these peripherals is really really wide. So the switching power supply library helps you to abstract the capabilities and tailor them to specific topologies. So therefore we also offer different sets of pre-configurations for targeting specific topologies. So once you have used that tool to set up a PWM controller for a specific application and topology, that where you're generating the code, the code is downloaded into the device, the codes that's generated with the device also includes functional aspects like soft start, so that you're having a nice behavior right from the beginning, so that when you fire up the power supply for the very first time that you can be sure that it won't burn. So once you have downloaded it, and this is what we are doing in this demo set, is then we have connected a network analyzer to the output, we are injecting an error signal and we're using our BODI 100 Network Analyzer to measure the total open-loop gain in the closed-loop system. So depending on the additional tools you might have used to design your power supply and to set up your compensation networks, so to achieve a certain stability of your application, you can then just measure directly from the kit the final results compared with your simulation and then you take it from there and in further some iteration steps you might have quickly achieved your goal. so before we go into the details about this kit and its software and design tool chain might be necessary to explain a little bit what hybrid power actually means the work hybrid is often used in different contexts like hybrid power supplies is something entirely different from hybrid power controllers so hybrid in our context means that it's a device which is somewhere between the analog control domain and the full digital control domain so analog controllers usually come as fixed Asics covering very specific topologies or specific applications so the conventional pol controllers for example or you have here switch switching regulator for flight controllers or resonant converters and for every single one of them so for every single application you might have a set of specific Asics to choose from while full digital control then digitizes even the feedback loop so everything is running and software the interface is an A to D converter so analog signals continuous time domain analog signals are converted into digital signals and then processed through software and then some PWM is adjusted so this the letter 1 full digital control gives you extreme flexibility in terms of what a controller can do while analog circuitry is might have a significantly higher performance but they are nailed to what they have been designed for so that section between both worlds that is what we're trying to cover with hybrid controls so it means we have a small device with relatively low power consumption but and in within the architecture we're combining digital logic with analog circuits so that we can leverage that in the advantages of both worlds in one single device so and when we are focusing on these hybrid devices then it's always a a device which is targeting the application would have a specific requirement for very specific features so the basic concept of these products is that in one chip you get a microcontroller core and a PWM controller peripheral that PWM controller peripheral always breaks down into three main major blocks consisting of the compensator so Erin play fire basically and the modulator so the modulator then determines which control loop control mode you're applying and there's always an additional fault level that helps you to make sure that the power supply stays within its safe operating range default is usually tied into the modulator or uses inputs to the modulator to shut down the passerby in case something is going wrong so when we're looking into the modulator so here on the right side we see a specific setup for a voltage or an average current mode and on the top we see an implementation for a classical picker remote controller so here on the upper writes in the current note section we find a subset of codes lay or labeled with PRG so that stands for programmable RAM generator so we are utilizing this ramp generator to generate a negative ramp which is modulated onto the reference input to our modulator that then is used usually in Picard mode for slope computation but it can also be used in voltage mode this is when you look below you also see a voltage range on the upper right where we use a positive ramp to generate an analog PWM signal the next therefore we find in this modulator are fast comparators so in the current mode section on the top we find that comparative between the RAM generator and our so-called Co G which is actually our PWM output logic block so that comparator is used to tie in the current feedback so that we can trip on peaks of the inductor current while in voltage and average card mode that comes very same comparator is now used to compare the reference from the Aaron profile to the artificially generated ramp and when the ramp exceeds the reference voltage level then it changes its output strips and and it's our latch and this then influences our PWM output so in both motivators we always use a PWM module and the so called CEO cheese or the complimentary output generator or in other devices it's called a complimentary waveform generator cwg but these blocks to is actually said they give to get to the capability to start out from a single PWM signal and split it up into complementary waveforms allowing you to adjust that x or you can split it up up to a full bridge drive so it also offers additional inputs for false signals so fault triggers that come from different peripherals or from external circuits can also be fed into the PWM logic and some additional glue logics and also would allow you to establish something like conditional fault handing so the third level is the compensator the compensator on these devices usually just a are just consisting of an errand to fire so this is actually a general-purpose operational amplifier then you get a deck to adjust reference and then the reference itself which is called the fixed voltage reference F we are so in addition to help you to condition signals for especially false signals of to define fault levels to trip on there is a an additional set of comparators and attack with a lower resolution so usually you get five bit tags that help you to set up thresholds that five bits might not be sufficient for in some cases so when you need a higher resolution then it is possible to use external resistors to create a window which is then still scalable with five bit and then you send it around a threshold range you would like to set up so once we have picked these three blocks we now can use these to set up an entire analog feedback loop here we see the first architecture so on the right we have a power supply so in this case it's just for the purposes of a sake of an example it's a boost converter that offers a voltage feedback coming from a voltage divider and we have a peak current signal which is taken from below the boost mains which using a shunt resistor so the voltage feedback now is fed into the outer loop portion of our architecture which is then our air amplifier so the amplifier compares that signal to our reference which is internally set by attack the output of that amplifier is fed into the programmable Ram generator and this part we are modulating a negative frame onto this reference voltage to apply a slope compensation for our inner p-card load control loop then we are entering then or we are taking this output of this program of a ramp generator feed it into a comparator that comparator compares that reference signal to the peak current signal that's coming from our shunt resistor and then that output of that comparator trips our PWM output logic and effectively true hates the duty cycle so to set up the switching frequency we are now using a digital PWM oniel which allows us to set up a fixed period with a maximum duty cycle so in case the comparator does not trip the pwm module in time it's the only time base of the pwm module which makes sure that the duty cycle is not hitting hundred percent and maybe with a bad outcome when your inductor saturates and then your power supplies effectively going out of control so one thing we see here in this particular example is because of these devices are highly configurable and can practically drive any kind of topology it is very hard to define our C Network values that might be necessary to set up an appropriate computation for every single different topology out there so therefore all of the compensation networks need to be external with these devices so we see here in the lower around the OPA so the operational amplifier that the input as well as it's the output is connected to a device pin so these device pins are usually sitting right next to each other so that it's very easy that it can place your RC components of the conversation that are very close to the device so after having established this basic feedback loop we are now adding the fault level so the fault level consists of a comparator with five attack in this case we are using it as an additional over current shut down and then we can feed that output of the comparator directly into the CG so that it asynchronously overrides anything that's coming from the software from the rest of the European architecture and we can truncate the duty cycle and turn off the PWM immediately as fast as possible so usually these comparators have a propagation delay of between 30 and 50 nanoseconds and this is the timeframe in which she can shut down can you respond to fault and shut down the power supply economy so what we can do with these PWM controllers now is a lot so as I said potentially you can support all kind of different configurations for specific topologies it does not only cover fixed frequency you can also setup the peripherals to support user 80 controls like internal time constant of time quasi resonant converters resonant converters so the variety of possibilities is really really wide so for just to keep things simple let's just start with fixed frequency operation off a standard topology and this is what we are doing with this starter kit so this starter kit we have used a simple synchronous buck converter not only because it's safe to operate it easy to compensate but also because everyone is very familiar with this topology so this starter kit now comes with multiple options so that it can play an experiment and explore a little bit the capabilities of the part play with different configurations and see how you can solve different design challenges by using different modifications of a preset of provided examples so when we are looking into the hardware we find that this port is as a power supply on the lower side so this is where we find our synchronous buck converter so that buck converter has been designed for a power level of approximately 25 watts so the maximum current call from the outputs at 3.3 volt is roughly at around 8 m/s so you're above this is where we have find our controller so that controller is bigger than it would be necessary for that single topology so we have picked our superset device providing up to four independent pwm controller peripheral sets the reason why we have done that is to allow you to switch between different control modes so we have set up one PWM control of the voltage mode another one for pico remote control in the third one for average current mode control we use the peripherals of the fourth one to add more advanced functions one of them for example is VCR sensing so one of the current feedback options provided by the power plant is a RC network which is tied in parallel to the main inductor and which can be used to to use a lossless sensing technique which is fairly popular especially in low voltage here will applications to implement that function we need an additional append so as the op-amp is still available from the fourth PWM controller architecture we can now utilize it to create an additional option which then is tied into one of the other PDM controllers so that kit now is supported by the power supply library NCC so when we are now trying to utilize this extreme flexibility of the Percel we really rely on tools which try to keep the complexity as low as possible during the design process so for that particular purpose we have created this switching power supply library which is available in the microchip code configurator so the basic architecture so the ones of you who are familiar with MCC might know that it is a graphical tool that helps you to configure specific settings of one independent peripheral so usually when you set up an application you pick one person like a UART and then you use that school to set up moderate or very specific details of that very powerful and then you're generating hold and the generated code is then called during startup of the device and putting that function provided by the person in place so now as we have seen that a switchman power supply controller is more than just one single peripheral the configuration of multiple blocks can get very tricky so eventually you might end up with seven to nine different peripheral blocks that need to be connected in a very specific way to achieve a certain function so to simplify that design process we have now added additional layers to our MCC design tool so here on the bottom of that diagram we see that classical MCC layer this is what most of you are maybe familiar with and this is what contains all the different configuration sets of every single powerful so here in that list you see all the peripherals that are used to form that function of a PWM controller so here we find this program a program generator the complimentary output output generator the PWM module a timer comparators digital to analog converters fixed voltage references op amps and so forth so these blocks mell are merged into that functional blocks of a modulator here on the left for peak current mode control and compensator block which consists of an op-amp attack and the fixed voltage reference here in the right we another set and this is the model a box for voltage not control or average more control so once you have you know which control mode you would like to use then we can tie together compensator and model a the block to form a control mode block in this case we have two options the card mode control or voltage or average mode control now this block is still very generic and it's not really tailored to a specific topology so to make the right position on the switch configuration which kind of dead times you need we now need to nail down the topology you would like a pro apply this control block for so this brings us up to the next level which is the topology level until specifically for our starter kits we are using a synchronous buck converter block so that block now when we pick one of these blocks we know exactly what the PWM output configuration needs to be we know which feedback signals we need to consider and we can pick all the peripherals to form a PWM controller so that you as a designer and up with a system that is already pre-configured and the only remaining options that need to be picked to make this block work or where should the feedback signals go into the device where is the pillar of iam coming out of the device and what is your preference value for your feedback and what is the definition of your switch node so switching frequency and that times need to be specified so eventually instead of diving into the peripherals and spending quite some time to set up all these different blocks functional blocks you now practically just configure the entire system on a very high level by just using physical values coming from your area application so we have seen the basic concept of the MCC smps library I would say just let's give it a try and configure or switch mode power supply so for that purpose I take my starter kit and I plug it it with my using the micro USB cable and what happens now when I do this when we look into MPLAB X as soon as the tool connects we see the product page coming up so MPLAB X is detecting the board and the board comes with a specific feature that it mounts a thumb drive on which all the information is stored that is now displayed in that window so now from that window on you can directly click on go to the kids home page and on that development kids home page you find all the documents and collaterals we offer with the device so as of today what you find is a user guide for the kids the configuration user guide for the switch composed by libraries which concludes the installation process the library itself so did not be already included in your envelop X installation you can always download the latest version from that website and then you find further documentation like schematics and the detailed data sheet of the chip we are using which is in this case the pic16f877a so let's get started so to set up a power supply controller on on that chip what I do first is I create a new project so I create a standalone project continue I picked the device I'm using but just pick 16 F one seven seven nine okay next now I have to check when my tool is and here I think to find the CIP hyper tower starter kit so I select this one click next choose my compiler and then I need to give it project name so I call this CIP high HP test choose a directory and click finish so now with in this new generators project I'm now opening MCC and as well as the mcc modules have floated I can save the configuration I'm going to build in a specific file which is usually stored in the project directory so the very first thing and this might be a little bit out of the scope of normal power supply designs is I have to determine which frequency I would like to have running on that device so to make sure that all the peripherals including pwms interrupts input capture you are and everything else you might use in your application works at decent performance we pick the eight megahertz oscillator and we enable the PLL which then gives us an effective CPU frequency of thirty-two megahertz so based on that timing this is the base for all the other timings we are going to set up in on that chip so later on in the modules we will find specifications for switching frequencies for dead times and so forth and all these timings always refer to the specific frequency we have set in this window so from that point on we are ready to go and dive into the configuration of the power supply peripherals so to make our life easy we are now referring to the switching power supply library so when we look into the device resources we find here this SMPS power controllers so when we open those then we find either soft blocks or the highest level would then be power supply topologies so when you recall that diagram we have just seen at the bottom there were all the peripherals then above they were already pre-configured sets of peripherals forming modulators and compensators above that we found encapsulated blocks for control modes and above that there was our topology so we going to start at the highest level let's say this highest abstraction layer and as we see there are a couple of different topologies already with can pick from in this case we pick a generic synchronous buck converter double click it and now it's loaded into our project resources when we load that project item then first we end up in a short page with some short description about what this block is all about the second one is then our configuration window so if you consider the number of peripherals who would have to put together for one PWM controller so all of that is now already pre-configured so we know exactly which peripherals we need to pick how they need to be interconnected because we know what you are going to design so the number of parameters then need to be set now comes down to a very very short list so this now is starts with the switching frequency so for this particular starter kit we pick 500 kilohertz the maximum duty cycle this determines some certain clamping value so for that starter kit it's 90% but it's really highly the application dependent and it really depends on the topology and your safety levels you're trying to achieve so we leave it at the 90% for that for our demo setup here the reference voltage relies on your feedback signal so the voltage divider we used on the starter kit gives us feedback signals of exactly 2.5 volts when the converter is producing 3.3 volt at the output so to achieve that 3.3 volts stable output control we need to set the reference at 2.5 volt then we have leading-edge blanking so and that depends on the current feedback we are going to use so the starter kit offers three different current sensing options the first one is a current Stinson's fauna which is located above the high side switch so when we switch over into the third top then we see a simplified schematic that gives us some indication where we find our current sensing option and how feedbacks and interconnects are done so here we see that complementary PWM that's connected to the highside and low-side switch of our half bridge then the current sense gives us the inductor current so the first option would be this is the one we're going use would be the currents in some format so that guy is located here just before the hiset switch that location that is indicated here in is just idolized indicating that we are always measuring the inductor current in truth the second option would be the shunt which is located between the inductor and the output capacitor so at this place and then we have a third option which is called TC our sensing so we are using a parallel RC network which is sitting in parallel to the inductor and that would actually give us that indicated current position although this is more an an emulated current feedback signal so precise but is lossless because it's not sitting really in the power path so it doesn't create losses that is the advantage of DC are the penalty is it's not that accurate so the three different current sensing options have three very different characteristics so the advantage of a current sentence former is that it's very fast and it gives you a very nice resolution so it's a large signal so the signal noise ratio is very good the downside is because it's located above the high sites which is that we only see half of the current waveform so we only see current feedbacks when the current when the switch is closed and current is flowing through that current transformer so we don't see the discharge current towards the node and the output capacitor the second sensing option the shunt has the advantage that it shows us the entire current waveform however the shunt needs to be very small to create as little losses as possible so we need to amplify that signal to get it up to a size that we can actually work with it so this is done with a high set current sense amplifier these amplifiers have limited bandwidth which usually results in a tiny phase shift and usually some damping on the amplitude so instead of a razor-sharp triangular current waveform you might get I mean something that's a little bit more sinusoidal and face shifted the third option the DCR sensing is equally sensitive so the signal won't be very large so we also need to amplify it and it is a signal which need also needs to be taken from the high side so we need a high cycle or higher voltage differential amplifier so the advantage in this low voltage buck providing 3.3 volt is that the output voltage is still below the supply voltage of our device so what we can do nervous we can utilize an on-board amplifier as differential amplifier for the DCR sensing so these are the three options we have in this first configuration we will start out with the current sensors form so the Continentals former has a very particular signal waveform so every time the switch closes it creates a tiny spike so we see a turn-on spike and that peak of that spike might be higher than the peak or peak current mode control is triggering on so we have to make sure that this peak is not accidentally tripping our comparator we're achieving this by setting up a so-called leading edge blanking window so this time frame we are setting will disable the comparator during that time between the start of the cycle until the end of our leading edge blanking period so when we are going to set this up okay on this particular design we know already it needs to be 200 15 nanoseconds so the 250 nanoseconds are counted from the point our internal pwm module is generating the leading edge so there's the first delay we have to consider is the delay between our signal generator and the propagation delay through the complementary output generator logic to the through the pin driver towards the most driver the propagation delay of the most driver until the MOSFET eventually turns on so that delay is already 55 milliseconds and then we see that spike of that current sentence former and the width of the spike is roughly 120 nanoseconds and to be a little bit on the safe side to account for additional switching noise we add some safety margin this altogether that delay chain that brings us up to 250 nanoseconds so the third setting we have to make is dead times so we have a half bridge converter that means when we are closing both switches at the same time simultaneously we're practically shortening out the input voltage and that will potentially or very probably blow our switches so we have to make sure that this never ever happens this dead time setting is based on a high resolution delay line which is part of the complimentary own output generator so we cannot just dead times with a resolution of five nanoseconds per tick so when we set up a value of three then we get 15 nanoseconds dead time so here in this mask all the values were entering our physical values so using physical units so in this case we're just entering 15 nanoseconds if you try to enter 12 it might sit there with 12 but eventually it will just be 15 anyways so your choices are 5 10 15 20 and so forth so the rising edge with 50 nanoseconds sounds very short and that the reason why the dead time needs to be support is the MOSFETs we're using on this starter kit so these are high speed high speed silicon MOSFETs which turned on in roughly eight nanoseconds so between five and eight nanoseconds is typically rise and fall time of these MOSFETs so they're extremely fast and the edges don't change so any Miller plateau or any additional delays so we can position them very close to each other to ensure a most efficient operation of our power stage so the rising edge is 15 in this case the falling edge that time is named with 60 nanoseconds so the mismatch between both dead times also mainly come from that delay between our signal generator edge through the entire output logic MOSFET or our master travel delays and then we have to account for all these additional and late delays in addition to our appropriate dead time setting so the best way to figure out how what the appropriate dead time is is always starting out with a very safe setting which is usually derived from the data sheets of the components here using so look into your into the data sheets of the most drivers try to figure out what the corrugation delay is there look into the most read data sheets look for the typical rise and fall times when driven by a driver at a certain voltage and then you get some good ballpark number where that data might be at a little bit of safety margin and fire up your power supply for the first time look at the scope measure it and optimize it so this is the more typical a little bit pragmatic but not a useful way in setting up these values good so once we have gone through that list so now the last setting here would be this computation so slope computation is not available yet because we first have to have a peripheral which we need to program so at that point there's nothing else we can do so now we have all the settings we need to at least set up the architecture we're doing this by selecting one out of four pwm controllers so the peak current mode control controller is located here in a specific corner of the device so the pin out of the device is divided in four sections so I'm here in the upper left this is where we find the in and outputs of the first PWM controller then here is number two number three and number four so that peanut has been arranged in a way that when you're designing multiple power supplies that it can route the signals on to the device in the most ideal way without crossing other lines especially trying to keep analog separated from digital and so forth to prevent any conflicts with signal integrity so on the starter kits the peak current load control is wired up to PWM controller number three so I pick that one from the list and now I can upload all the modules that are required so I'm uploading the configuration for all the people peripherals so this takes a moment and here we are now in the project resources we see decide that one power supply topology we have added first there is now a list of peripherals so now we see all the sub blocks so which is then here for example the comparative digital to analog converter and op-amp for the area fire our fixed voltage reference at the complementary output generator the ramp generator and the basic period is switching on PWM so and as you soon see every time I click on one of these modules I see all the configurations which are available for that particular powerful so if you go into the register map you see this there are endless options of configurations that could potentially be done but as we have done all this already all these settings have been done for you ready so I can just skip that peripheral level go back into my sync rectifier and now as attack is loaded and I know what the resolution of the deck is now we can set a new revenue rate of our slope compensation ramp generator so for that particular design we need a slope of 0.35 volt per microsecond so that ramp that is modulated onto the reference of the air amplifier now decreases over time with a ratio of 350 millivolt per microsecond so now we are all set so I save my configuration and now I generate the code and in which the help would we know until the process has completed and I see this is stunned first when the progress bras have disappeared and when I find a line that says save configuration' in final my path so when I now switch over into the project and I open the header then I see that the MCC generated files folder in which all the search and possible configurations have been generated - and then there's another folder called SPS and in this folder I find software functions that already have been generated in addition to the basic peripheral configurations so what we providing as a standard generated code block is a soft starter team especially in half bridge configurations at high input voltages if you would just snap on the PWM with a certain duty cycle the chances are very high that it will just blow up or not to be switches so usually half which drives especially with high speed MOSFETs like those needs needed is vitally need to be soft starter so and that code is now part of your project already so this is a function you can call the right key and you don't have to be concerned about your some sort behavior anymore so when we go into the source files we find also a main which contains the main program and years our MCC generated block with all the peripheral configurations and here also this which will power supply software function blocks so why did I look into my main the first line within Maine is system initialize so inside that system initialize I see the configuration of my pin out oscillator settings what's your timer settings and then all the peripherals for most which one power supply PWM controller including the to sink park and PCM C Block software blocks which include the soft start within Maine there's nothing else to find so there's just a while loop so the only modification we have to do manually is to decide when that power supply should be turned on so where to place my soft start function call so the first thing is we have to figure out how that function is called so we have to go into the PCMC header to learn how that function Hall looks like and how its named so it's called PCMC underscore soft start easy to remember and we would like to soft start that over to start the power supply immediately it just before we enter the while loop so I place that function call here and this is it now we are ready to go so now I hit compile compile the project just to make sure that the generated code doesn't produce any errors and I get a built succeed and now can program my device okay so now we see programming verify complete so the program has been successfully programmed okay so now here we are with a programmed board now we have to test if it's really working now I'm connecting this port to the power supplies and the load which is part of that development kit so for that purpose I'm stealing the input voltage from that input terminals connected to my freshly program port and I take the load module we have here on this side and connect it to the output and then here on the display you see an operative of 3.3 volt at already close to four amps so regulation seems to work to make sure that it's really working we're just changing the load turning it off and it's the regulating to 3.3 volts so seems that we have an active controller that controls the voltage to a constant level overload how well it really works our controller what did eventual performance is and our bandwidth the transient response that obviously requires a little bit of more thorough analysis and for that purpose we have a network analyzer which is connected to that kit so I will just remove these blocks again give it a look the load module back in and the software which is currently running on that kit and that kit is identical so this is exactly the same basic p-card mode control we have just generated so when we are now letting this board run then we have on this side our network lies analyzers running and it's continuously measuring the loop so the way we do this is that we have a tiny resistor on top of our output voltage divider so voltage dividers at the output I have usually resistors in the kilo ohm range and we put a tiny 10 ohm resistor on top of this high resistance and then we use a injection transformer so this is a one-to-one current transformer and that is connected to the signal generator within our OD analyzer so this frequency now is applied to the primary winding is injecting a current that current is then showing up in on the secondary side and is injecting an error signal in that Tang ohm resistor on top of our voltage divider so this artificial error signal is now traveling through the voltage divider entering our feedback loop that's their amplifier the area fire starts to respond to that error signal and is injecting this error into our modulator and so our PWM starts to be to modulate against that error that is showing up at the feedback input and so our output voltage starts to change so the network analyzer now looks at the signal that has been injected on one side and then looks at the output of the power supply on the other side of the Tarin ohm resistor and just measures the sinusoidal response of the system and measures the difference in amplitude which gives us the game and the phase shift which gives us the face and plotted as a bode plot and the port button allows us to do stability analysis where we can measure it across our frequency the slope of the gain at the crossover frequency the face at the point where in the game hit zero and the value of the gain when the face hits the minus hundred eighty degree sole point of instability so the so-called gain margin so then with phase margin gain margin in Kosova frequency we now have all the parameters that determine the total bandwidth and performance basic performance parameter instability of our power supply so before we have a closer look at the measurement let's briefly revisit what we are actually going to measure so this is a block diagram of our system consisting of the power plant which is the transfer function G of s in the S domain and then we have the second transfer function which is our compensator H H of s in the S plane so our feedback in our system is closed with a voltage divider and this feedback signal is then fed into the compensator and the compensator and is and comparing just into a reference and the difference is then applied by our modulator to the plant and that changes when we change the duty cycle for example the optical burst goes up or down and any transients that might hit the system from outside or no transients or from the input will then be compensated by our control system so to make sure that this system is stable we have to have a look at the total transfer function of this entire block from input to output and this is the gain the closed-loop gain so g closed-loop equals the plant gain over 1 plus the plant gain times the compensator gain so when we look at this term and one thing becomes obvious that thing is always stable as long as G of s times H of s is different than minus one because imagine when this tends to go to minus 1 then we have 1 plus minus 1 this goes to what 0 so the gain of the system will go to infinity which means that the system is coding getting out of control and the gain becomes infinite and the game the system is gone and will most probably blow up in our case so this is why we call this G of s times H of s equals minus 1 is the sole point of instability so in any other condition the system is de facto stable and only in that point it becomes uncontrollable so when we are making a stability analysis the whole stability analysis is now about how far are we away from the sole point of instability so and this is the so-called open loop gain within the closed loop system so while the system is operating a closed loop we are only having a look at the open loop gain of the system so now when we have a closer look at the bode plot so what you are seeing here is a continuous measurement of the open loop gain and closed loop system so this term of open loop in a closed loop system comes from the open loop component that looks for how far away are we from the sole point of instability which is the point when the system hits a phase lag of - degrees so in this region the system should not be able to respond to any transient in any reasonable form so we need to have a large negative gain in that region the bode measurement is as always when we look at the face there's always one kind of misleading thing about the scale of the face and this is the minus 180 degree line is actually a serum at 0 degree phase in that ball plot the reason for that is that the small signal model defines that monitored minus 180 degrees at the reference of the system however the measurement now is done at the output of the power supply so feed other side of the island and then we have that phase lag of 180 degrees throughout the system doesn't really matter it's just if you're wondering why we're always talking about minus 180 degrees as so point of instability and here in that scale we look for that point at 0 degrees that's actually the same thing so what that Buddha plot here in an active measurement is showing you is actually more for more tuned towards stability analysis then a theoretical description of the system so that is the only thing that's maybe interesting to keep in mind so let's perform a measurement so as you see it's running here at the low frequency range we see a lot of noise and that is the small signal model so you're achieved best and most accurate results of the transit function when the signal you're injecting is as small as possible so that it you don't drive some components like especially the air amplifier into nonlinear regions at its margin of its total input reading range so this is why we always try to keep that injected error signal as small as possible and at low frequencies this is getting tricky for the measurement equipment to see the variation of the sinusoidal curve that is so small that the noise floor starts to have a large influence on the measurement result so this is why we see in that low frequency range it's always very noisy so luckily that measurement tool we're using here the Buddhahood home that has a feature that allows us to shape the gain of our input stage so that we can increase the gain to reduce the noise but as you see it is may be good but still not good enough so we would have to perform many many measurements of exactly the same operation point and activate an averaging over that period to get rid of that noise to also achieve a better more accurate result in that region but for the moment we are not really interested in this local frequency range we are more interested in the total performance of our feedback loop which is more about stability so the stability criteria we are now looking for is start with the crossover frequency of the game so when it does the creating of the game which is the red line cross zero dB so this is pretty much exactly at 30 kilo Hertz we can make a more precise measurement if we go here into the scale and say okay please give me the number at zero DB and we see exactly it's at thirty point one point six kilohertz the other point that's important for its instability at area is the gain margin which we find when the face which is the blue line now hits minus one at eighty degrees or in our plot C or the green line so we do the same I go into that table put zero and the cursor gives me thanks the number of minus 14 DB which is a sufficient gain margin for a power bond so when we are let me stop the measurements to keep it really quiet so knowing we analyze it one of them the third stability criteria would look fine four would then be the slope of the game at the crossover point so ideally it should be minus 20 DB per decade so we're looking into that decade so here we have a point which was very close to 10 DB so that is easy to take and then it should be a linear minus 45 degrees over minus 20 DB per decade slope and when we draw a line when we imagine the line that has a slope of minus 20 DB per decade then it would be at minus 10 DB a ton of kilohertz and the crossover point would right be in the middle of that slope so we exactly hit that slope of minus 20 DB per decade so it's a perfectly schoolbook stable system so the last and foremost designers most important value would then be the phase margin so that is determined to be at the crossover point so what's the phase lag at the crossover point so we take that server cursor and we've framed when we reset this to zero again so we find a phase margin of 60 degrees gain margin of 14 DB add a crossover frequency of 30 kilo Hertz and we are meeting so we have really good stability at the crossover frequencies of phase margin a margin and the slope at the crossover frequency all three major stability factors are in a really good region for a real stable power supply so the bandwidth we have set at 30 kilo Hertz is a little bit is at many of fifteenth of the switching frequency which is also a good ratio and which allows us to have enough safety margin to account for variations in inductor component values capacitor component values and so forth so what we have covered today is we saw a short introduction of the CFP hyperpower starter kit our demo setup the design procedure using our switch from power supply libraries in the microchip code configurator and finally how we generate the project we built our pwm controller downloaded into the device bring it into the system and then measure and analyze it so I hope also that the measurement results have given you some idea about the performance level you can expect from the analog peripherals on board of these chips it is really a straightforward pwm controller as you can expect it to find on any other ASIC in terms of performance nevertheless have a look at the data sheets and every further information you can find on the descriptions below just click on the links follow them and you will find all the documentation about the board the devices that are imposed by libraries and further information on generic topics of simulation of switching power supplies and design practices thank you very much for watching hope to see you soon
B1 中級 CIPハイブリッドパワースターターキットによるSMPS設計 (SMPS Design with the CIP Hybrid Power Starter Kit) 59 1 Red に公開 2021 年 01 月 14 日 シェア シェア 保存 報告 動画の中の単語