Switch Mode Power Supply Measurements and Analysis

Hi, Johnnie Hancock here. Welcome to the Keysight Oscilloscope webcast series Today our webcast is going to be on making power supply measurements using an oscilloscope. That’s probably the instrument to use most often to characterize your switch mode power supplies We’re going to talk about a broad range of measurements beginning with AC power quality, current harmonics, and then probably the one that is most important to you is measuring losses such as losses in your switching device but power and energy as well as output ripple. Now we’re also going to talk a whole lot about proper probing techniques Your measurements are only as good as how you probe them, Which are really important especially for output ripple So you’re welcome to view the whole web cast or you can see time tags and subtitles down below. You can just click on them and just advance to the sections you’re most interested in Let’s go ahead and get into the agenda, so the agenda first of all we’ll give a quick overview of a switch mode power supply, talk about probe deskew, which is very important if you want accurate measurements and then we’ll get into three different categories: input measurements, switching measurements, and output measurements Now many people they’re not interested in and completely characterizing from A-Z from input to output Sometimes they are zeroed in on the input side, some folks are zeroed in on the output side but we’re going to talk about all the measurements We’re going to wrap it up with some probing recommendations So let’s go ahead and get started here first of all when you consider power supply basics The job description of a power supply is to efficiently produce well regulated and low-noise DC power from an input rail. Now there are two basic types of power sponsors. Linear power supplies and switching power supplies. Now when I first started my career 35 36 years ago almost all the power supplies were linear type and they have some advantages and there are still linear power supplies out there today In fact, many dc-to-dc converters are effectively linear power supplies. They have low noise low filtering requirements. Some of the trade-offs though: they’re always stepped down in nature, they’re inefficient, they’re somewhat costly and as far as a big linear power supply, they’re big and heavy and bulky and can generate a lot of heat. A switch mode power supply, some of the advantages are: they’re highly efficient, you can get more power out of a smaller package, there’s a variety of topology is just not just stepped down you can get step up but the big trade-off is they have moderate to high noise or output ripple. Now here we show a real basic simplified schematic of a switch mode power supply on the input side You typically have a full wave rectifier so it’s going to it’s going to convert that AC input which is typically line it could be a lower tie pace again but typically line here in U.S. 110 volt AC and Europe Asia South America often 220 volts It’s filtered and then it goes through the heart of the switch mode power supply which is that switching transistor there in the middle We’ll get into that in more depth a little bit later and ultimately it’s filtered again and rectified and you have a DC output. Now I’ve listed here some of the key measurements that we’re going to be talking about today some of the key measurements on the input analysis a AC input analysis now some of these could apply to AC output if you’re working with an inverter and some of the switching device analysis as well as the DC output analysis So some of the trends and power supply design today engineers: first of all they want more efficient power supplies – green energy, people want to improve power density, pack more power into a small package, increase for reliability control EMI, improve output power rail integrity That’s extremely important today for high-speed digital designs. Decreased thermals make it such as it doesn’t run so hot and reduce costs. Now all of this adds up to increase test time which is our focus today So before you make any power supply type measurements probably one of the most important things you need to do is deskew your voltage and current probe, so if you’re making power measurements it requires those two type of probes voltage x current = power The problem though is often your two different probes there’s a skew between them and you need to skew that out We offer a deskew fixture that you can use and you can press a button it

automatically deskews and see here if I click on this you see it automatically deskewed You can also create your own fixtures simply a resistor and a wire and connect your voltage probe across the resistor and your current probe, typically a clamp on right near the resistor on the wire and you can create your own fixture and manually deskew For all scopes you go into the program and you have the ability to manually deskew them. Line them up visually, I want to say all scopes, maybe I’m wrong maybe some of the lower end scopes don’t have that, but almost of your mid range and high performance scopes have that ability So let’s take a look at input line analysis AC analysis and it will step through and looking at switching analysis and then DC output There are three primary measurements that are performed to analyze the input power quality Number one is power quality things such as real power, reactor power, reactive power Current harmonics and inrush current let’s take a look at the first one power quality So what I’m showing here is a what we call a signals connection diagram on one of our InfiniiVision X-Series oscilloscopes It gives you some hints on how we’re to connect the probes some suggestions for you if you’re analyzing AC input in North America It’s 220 volts, so you need to probe can handle approximately 300 volts peak to peak, but if you’re working with European or Asian power to enter up to 240 240 volts RMS you’re going to need a differential probe that can handle up to approximately seven hundred volts peak to peak. Current probes typically for these types of American measurements it’s a clamp on type or Hall effect current probe. If you press the auto setup the scope will automatically set up the voltage waveform in the current waveform and turn on the power waveform Math function now to give you some hints here if you don’t have an automated power option on your scopes First of all you need to scale your wave forms to near full screen to take advantage of all the bits that the scope has often get screenshots from our customers and you’re having problems with accuracy and resolution and I find out they stay scale the waveforms to one or two divisions well if you scale the waveform you got eight divisions peak to peak in the scale say one of your waveforms 2 divisions now you only using one quarter of your A-D or if you got an 8-bit scope now you’re only using you’ve got a 6-bit measurement So first of all scale the waveforms near full screen and then most scopes today ave a way for math which is simply if you want to power you do channel 1 x channel 2 So here the yellow waveform is voltage and I’ve labeled it Vi or instantaneous voltage. The green waveform is current. One thing to note about this with switching power supply it’s going to be pulsed like this, it’s not going to be a sine wave, so that the current comes out and it pulses. And then the power is just simply computed V x I, the instantaneous power, if you press the apply button on our scopes then it turns on a broad range of measurements to tell you about the input power quality Click here, ok the first one I don’t know if you can read the values they’re not the real power, typically the symbols P for power is simply the mean or the average power over in cycles. Now if you’re doing this manually on our scopes you’re able to select an average over N cycles or the average over full screen. Some scopes don’t have the N cycles option , in that case you need to get that measurement and make sure the measurement is over a discreet or integer number of cycles So this is the real power is really the only one that you can typically measure on any scope directly all of the other measurements are derived from other parameters such as apparent power is simply voltage RMS times I Current RMS, again over an integer number of cycles scopes typically have the RMS (IRMS), the reactive power is a computed by the square root of the apparent power squared minus the real power squared. Power factor is simply real power / apparent power. Phase angle is arccosine real power over apparent power and then CF stands for crust factor is just a ratio of peak over RMS for both voltage and current and so the the real advantage here of having

automated power app in your scope Whether it’s from a Keysight scope or one of our competitors is you get these things automatically so that you can get your answers much more quickly rather than automating with a computer or using a pencil and paper Another important measurement for line analysis is current harmonics. Now I mentioned earlier that if you looked at the current, it was pulsed, you had a positive pulse and you added negative pulse. It wasn’t a sine wave. Back in the old linear power supply days, the current was a sine wave. Well the problem with this pulse current that’s a characteristic of switching power supplies as it can inject harmonics back into the mains and that can disturb other equipment on your mains network power network, and so there are some standards that many different products must meet today It’s called the IEC 61000-3-2 specificiation three different subclasses: class a, b, c, and d. It really depends on the type of product that you’re developing and ultimately selling to a customer, it must meet these standards This was developed in Europe and what you can see on screen right here is what the scope does is it runs an FFT measurement on the current waveform and then what you can see in that in the top part of the screenshot there is a table The first column just list the harmonics, the second column lists the actual measurement of each harmonic that’s in voltage RMS, the third column lists the specification or the limits and then we show the percent margin and then whether or not it’s pass/fail. You can see in this example everything is passing. It measures up to the 40th harmonic, if it fails you’ll see a big red box and it’s a fail and in the fist marginal to be in yellow Now there’s two different ways you can view this information, what I’m showing right here is the tabular format, you can also select the bar chart format, and what I’ve evaluated on our competitors is that they pretty much show the same thing The last input line analysis measurement I’m going to show you show you here is inrush current. This is a single shot measurement you really don’t need a power up to do this measurement if you know how to run a scope you can set up a single shot measurement yourself with our power up – it sort of helps guide you along in setting it up and tells you when to turn off your power supply when to turn it on and so on And when you initially turn on a power supply or a product you turn on the power there’s going to be a surge of current because all of the input filtering on the input of the power supply, all the capacitance has been drained away and all these capacitors initially looked like a short circuit, so there’s going to be an initial surge of current. In this case I don’t know if you can read the number or not. It measures a peak current approximately 15 amps and the scope will automatically measure the maximum or minimum current whichever way it is – it could go either way Just what you can observe here the other waveform again is the input AC voltage you can see it begin to wrap up and then you can see the input current we got the spike We got a high level of oscillating current and then it begins to settle down to more the steady state Let’s now take a look at switching device analysis there for measurements were going to go through Probably the one people are most interested in is switching losses That’s where a lot of losses occur during power regulation Rds(on), slew rate, and then modulation analysis Let’s take a look at the theory of a switching power supplies and and what I’m showing here is the theoretical or ideal world you can see the yellow waveform is the voltage across the switching transistor we call that VDS or voltage drain to source and the green waveform is IDS for current drain to source. During the on state that’s where the voltage is zero across the transistor. It’s turned on in the ideal world the on-state losses are current times zero volts so you have 0 watts here in the off state That’s where current is zero and voltage is high, again voltage x 0 = 0 W, so the net losses in the perfect world is 0 watts, but let’s take a look at the real world Here we show during the on state VDS is not zero there’s some small saturation

voltage and it’s not easy to measure and that’s something one of my competitive colleagues talked about and we’re going to talk about that a little bit later about the difficulty of measuring that actual voltage level during the off state. I can click on it right here, you may not measure zero amps now it will be 0 and unless you have a problem with your switching transistor But you may not measure zero amps and that’s due to the current probe typically the current probe or the scope no offset error and then during the switching when it turns on and turns off the edge rates of these signals is not infinitely fast and the transistor will pass through its linear region and the net result is you will have, well let’s go back to this real quick you will have losses. Most of the losses during the turn on and turn off time you’ll have false losses during the off state and conduction losses during the on state Now let’s take a look at see what this looks like on an oscilloscope Now first of all let me explain why I scaled these waveforms wrong I’m not taking advantage of the full resolution my scope but I scaled them this way so that we could see them more clearly So the top waveform is VDS the yellow waveform, the green wave form is IDS, current through the transistor, and the purple waveform is the waveform math function which is volts times amps. So here we show the conduction phase of VDS. Here we show the non conduction phase when current should be 0 and here where you see the pulses on the power waveform is what we call the T off time are the time when it switches off and the T on there is no pulse they’re dependent upon a topology there could be a pulse or a power loss during this phase So let’s now actually make a power loss measurement using the scope The first thing I’d like to point out is this is the measurement that is probably most critical that your probes are deskewed because if you just get a slight skew between your voltage and current probe that’s going to change the power losses during the turn on and turn off time Another important important thing as we mentioned earlier that measuring that saturation voltage across the switching transistor is very difficult Our scopes have a precision offset calibration that can be performed, so this can really help improve the accuracy of making those measurements if we’re still going to talk about this more in a few minutes You press the auto setup and the scope automatically scales 2 switching cycles across the main time base window and one switching cycle across the zoomed in window and turns on the power loss measurements. Now at this point, I don’t know if you can read it or not, the power losses measuring in a range of about ten and a half watts, an energy loss of two hundred some odd I think it’s nanohenries but i can’t read it at this point you can zoom in and measure the losses of each of the phases- we’ll show that in a few minutes Now the last thing I’d like to point out, I don’t know if you can see it or not, notice how flat that power waveform is during the non conduction phase. If you use the default settings on the Keysight InfiniiVision oscilloscopes, we detect that’s the non-conducting phase and we simply set the power to 0 because if it’s not zero what you’re measuring is no error of the current probe it should be 0 now you can change that so it actually does measure the error. If you want to measure the error then we go ahead and just clamp that it 0 watts Another closely related measurement is measuring the effective resistance of your switching transistor if the FET type transistor this is called RDS or effective resistance drain to source when it is turned on with the scope does is it windows in on just the conduction phase and measures the average voltage and the average current during a window of time during that phase and then it computes RDS as the average over I average. Again it’s important that you have done a very accurate calibration of the offset error prior to this measurement. Now at this point when you’ve measured the RDS value you can then select use RDS to measure losses you click on that then the scope

switches back to the switching loss measurement which we showed earlier and we’re showing now and at this point it measures the losses based on RDS during the conduction phase as opposed to based on the actual voltage measurement Now if you look at the last bullet there are the one on the lower right it actually computes the power differently in each phase of switching and again I think all scopes from our competitors we do the same thing if you use the RDS factor during the induction phase is simply i square times RDS during the turn on or turn off phase It uses the conventional computation for power and that’s volts times amps and there during the non conduction phase again as I mentioned previously we clamp the current is 0 so it gives you 0 watts during the non conduction phase. Now at this point you can zoom in and here I am going to show how you can measure the losses during each phase. What I’ve done here is using the zoom time base I zoomed in a very accurately into the in this case the conduction phase and I’ve also zoomed in vertically just on the power waveform so that we could see the slope of it and there you can see the purple waveform is the instantaneous power equals the instantaneous current squared times the value of RDS on in this example I’m zoomed in on the tee off phase and then and and now it’s using just V times I to measure the losses during this particular phase Now let’s talk about some of the measurement limitations. I’ve been saying it’s very difficult to measure that saturation voltage during the conduction phase. There’s a myth out there that the biggest component of there is a silicon measurement resolution so many folks say if I only had a scope that higher resolution I could make this measurement much more accurately. Now higher resolution is not bad, but that is not the highest component of error The highest component of error is the scope or probe offset or position error A typical specification for offset position error on the scope and the probe it’s about a tenth of the division or you looking somewhere around one percent. Now if you’re really good typically a differential voltage probe is going to have a thumb wheel or a small screw adjustment that you can visually adjust and null out some of that error but what about the best you could possibly do is maybe half of this about 0.05 divisions So for example let’s assume that your error iss .1 division in this case we show a volts peak to peak of are switching of nearly 300 volts and the Scopes volts per division set up on that voltage waveform is about 40 volts per division and if the offset here is a tenth of the division then you’re going to have absolute voltage offset of about plus or minus 4 volts which is huge the saturation voltage for high power switching transistor. It can sometimes you know be well less than one Volt can be a hundreds of millivolts or even tens of millivolts and even if you if you’re really good and you can count out the 0.05 divisions then you still have plus and minus 2 volts so this is a big issue with trying to accurately measure the power losses during the conduction phase My recommendation is to actually get the specification out of the device data sheet of what is RDS on and here you can see one for a particular power transistor from Fairchild and you can see it’s a very low value, ranges from seven and a half milliohms to 9 milliohms, you can as earlier I showed measuring are two RDS on and then using that value you can also directly enter the value of RDS and this is how you’re going to get your most accurate measurements Let’s move on to another important measurement for switching devices and that slew rate that determines how fast a device is going to turn on and turn off. If you you can turn them on and turn off faster you’re going to have lower losses and we’ll talk about what the trade-offs are about now there’s a couple different ways you can measure it what we do with our automated power app is we actually zoom in during that switching time and we turn on a different map function this time not v x i but we turn on the differentiate or dV/dT if you’re

measuring the slew rate of the voltage waveform and you can see that purple wave form at the very peak is where the maximum slew rate is and I think we’re measuring somewhere in the range of about 19 million volts per second on this particular device. If people typically think of it is a 19 volts per microsecond our scope just shows you the millivolts so you can flip that around if you want another way to measure this You don’t need to power up some scopes do this is you can use the tracking cursors or markers on the scope so you can set those two markers near where you visually think the highest slew rate is and then and in our readout we also do it if it’s tracking is measuring volts delta volts and delta time we automatically show you what the slew rate is and this next slide here I just show you the slew rate of the current waveform it again it just uses a differentiate math function di/dT The last measurement we’re going to talk about person for switching devices modulation and what the scope uses for this is what is doing is characterizing the control signal that’s controlling the switching on switching off and that’s the big the voltage level on the gate of the switching transistor and we use a measurement called the measurement trend and what a measurement trend is it’s in this case we’re doing a trend of the duty cycle of VGS gate to source voltage across that transistor So the vertical axis on that purple waveform is duty cycle on the horizontal axis is time and so if you look at the yellow waveform is VGS there’s some modulation going on there I can’t see it but the trend waveform pulls it out and in this case I can see there’s some what appears to be a see modulation Now this various parameters you can select in this case I’m showing duty cycle and here’s an example of measuring the trend or the modulation of frequency of the gate signal in this case i’m showing an example of what happens when you initially turn on the power so you can see the voltage ramp up on the gate and then you can also see on the purple wave form the frequency ramps up and I think it ramps up in about four hundred microseconds until it gets to its steady-state switching frequency which in case this is from our demo board to switch is relatively slow about 70 kilohertz switching rate Now some of the trade-offs for power device analysis and bottom, by the way I I don’t know if questions are coming in yet i don’t have that window open, I am NOT a power supply designer I am an oscilloscope, some people would say, I’m an oscilloscope expert so when you’re throwing these questions at me I appreciate if you throw measurement questions at me and don’t ask me how to design a power spot because I’m not a power supply designer but these are some of the trade-offs they make switching power supplies they’re getting faster today Faster switching frequencies and what the benefits of that is you have it allows for smaller inductive devices which means you can get higher power density more power in a smaller package The smaller inductive devices have lower costs but the trade-off is is you have a faster switching frequency. Now the duty cycle of that turn on and turn off time when the losses occur when the when the transistor switching that increases now to resolve that you could have faster slew rate devices That’s going to lower the turn on turn off losses because now that delta time when it’s ramping up and ramping down as much smaller but the big trade-off is higher output ripple, higher EMI and of course higher costs for the faster switching device and that makes some of the output analysis what we are going to talk about more next more important For an output analysis we are going to look at real quickly five different measurements output, ripple, turn on turn off time, transient response efficiency, and the last one here power supply rejection ratio. So output ripple is actually a pretty simple measurement but there are some limitations and issues you need to be aware of so here I’m showing again this is our connections diagram you typically just connected passive probe most people do to the output DC signal here you can

see or measure of 15 volt DC supply output ripple is becoming more and more important today because any kind of ripple or noise and people just call it noise on the output DC signal if you’re if it’s trying to drive high speed digital devices that noise on the power rail is going to translate into jitter and timing uncertainty and it’s going to effectively closed your data valid window So output ripple is becoming a more and more important issue and is often called power rail integrity measurements now I mentioned herein the bullets here there are many probing limitations you want to measure millivolts of noise you’re typically going to have to use a one-to-one probe to to measure that low noise here I show pressing auto setup and what the scope is going to do is going to make it is going to AC couple that signal and then blow it up and simply measure the volts peak to peak as well as V RMS to give you a feel for what your output ripple is now let’s talk about what some of the probing limitations are now I mentioned we recommend using a one-to-one probe for low millivolts the problem with the 10 to 1 probe is going to limit the scope’s input sensitivity on most scope somewhere between 10 millivolts per division to twenty millivolts per division So if you’re trying to measure 2 millivolts you’re not going to be able to measure it very accurately another problem with the standard 10 to 1 probe is not only does it move your dynamic range up away from your most sensitive settings it also expands your scopes noise floor by a factor of 10 as well The problem with a one-to-one probe you get your maximum sensitivity so 1 millivolt perhaps two or maybe 5 millivolts per division on your scope but a one-to-one probe has limited bandwidth approximately thirty five megahertz also the one to one probe is going to limit your DC offset range and that’s why, as I showed earlier, we AC couple to measure that ripple Now we’ve talked to a lot of our customers about this and this is the big issue they tell us I want to measure the to DC component at the same time as the AC component or the ripple and I want to be able to also measure any slow drift well AC coupling is going to destroy the slow drift that you you may want to measure as well as the DC component. Now you can always measure the DC component and a separate measurement, but it’s that slow drift that’s going to be destroyed Now let’s talk a little bit about more probing techniques and then I’ll show you what we consider a very good solution for some of the issues I just presented here and and what I’m going to talk about right here is something that my competitive colleague really pointed out earlier so here we show a one of our passive probes and some of the accessories that ship with it and this is the accessory most engineers and technicians use to connect the ground to their circuit but that is also can be considered an inductor as well as an antenna Now let’s take a look at this comparisons of measurements while before we get to that this is the best ground new technique there are various accessories that you can get good foreground this is one that actually soldered into the board so it’s not the most friendly but if you can put in something like this on your board soldered in you’re going to get a very good ground short ground connection now I have four examples here. Here I am measuring the output of the USB power and in the first example here I’m using a standard 10 to 1 probe the probe that ships with the scope and I’m using that long and tender to make that measurement and in this case you can see that we’re measuring the output ripple of about 64 millivolts. In this example I’ve removed my antenna that ground lead and I’ve used one of the grounding accessories it ships with a scope it’s a little spring clip that’s got a very short ground and you can see that I reduced my output ripple in half so I didn’t really reduce my output ripple, I reduced what the scope is measuring the next example I’m using a one-to-one probe again with my long antennae and again I’ve reduced again now I’m down to about 21 millivolts and last year’s the best measurement I’ve performed I’ve got

it down to about eight millivolts again using a one-to-one probe and a ground clip now why is that noise so low Part of it is good probing techniques using that ground flip another issue may be that I went from a 700 megahertz 10 to 1 probe down to thirty five megahertz 1:1 probe this is what I just mentioned earlier is a new solution we just introduced a few weeks ago actually we call it the N7020A passive power rail probe it allows people to measure output DC signal integrity with high bandwidth This is an active probes that has two gigahertz band with it’s a one to one probe so you can take advantage of the maximum sensitivity of the scope and it also has plus and minus twenty-four 24 volts offset so you can measure the DC component the AC component and slow drift all at the same time and without losing anything now Here we show an example this is on one of our infinium windows-based oscilloscope the orange waveform is using that thirty five megahertz 1:1 probe you can see it’s noisier yellow waveform is using the new power rail probe You may not be able to see it but there are some very narrow spikes in there that have very high frequency that’s what the the one to one probe is going to miss, plus there is much lower noise on this there’s a webcast it was presented it was a month or so ago by one of my colleagues at Keysight it’s called measuring power rail signal integrity with oscilloscopes Now I’ve listed a you URL here that you can go to if you want to view that webcast it talked specifically about that this issue of the whole web cast is on power rail issues Another measurement for output analysis is turn on and turn off time again you don’t need a special power app on your scope to do this if you know how to run a scope is basically a single shot measurement So the top waveform there I’m showing turn on time so when you can you turn on your product or your power supply that DC is going to ramp up what’s nice about the power app is it automatically measures from the initial turn on time that it detects the edge crossing on the AC input until it reaches ninety percent of the expected value in this case it i believe it’s a 12 volt power supply and then the turnoff time measures automatically measures the time from when the AC goes away until the DC decays to the ten percent level if you don’t have the automatic power apt you probably have to just simply use your oscilloscope tiny cursors manually set them where you visually want want them Another important measurement is transient response analysis What that is what I’m showing in the upper screenshot there is one of our help menus if you press and hold down a key is going to pop up the help screen and explain the measurement for you when you when you initially change a load on the output of your device under test There’s going to be a transient that will occur on the DC that’s driving it so if you have a sudden increase in load that means the current is going to go up there will be a sudden decrease in the output voltage until it can really regulate it and get it back up to where it needs to be so what were showing here on the lower wave form the green is the current you can see it going from a low-level stepping up to a high level the yellow waveform is the voltage you can see as I just mentioned it will take immediate dip and then come back up and settle to its output level so we zoomed way in there on that DC voltage and the scope automatically measures from the initial time it goes out a percentage setting that you can set until the last time it comes back to within that tolerance setting that you have told the scope to measure to now the last measurement people often think about making and then perhaps the most important is what is the efficiency of my power supply there’s multiple ways you can do this you can take the measurements the very first measure and we talked about where we measured input power quality it will measure real power for you then you could take a a standard multimeter

and measure the DC output both voltage and current and then just compute efficiency you can also automatically do it with a scope however is going to require to current probes one on the input one on the output our particular app it allows you to measure the efficiency of AC to DC which is most typical for switching power supply as well as AC to AC, DC to AC and DC to DC so it’s pretty simple measurement input real power divided by or back that up output real power / input real power and then its measured in percent now the last measurement we’re going to talk about here today is power supply rejection ratio often term PSRR now this is a measurement you don’t typically think they could can be measured on an oscilloscope so power supply rejection ratio some people call it power supply ripple rejection what is it it provides a measure of how well a device such as a dc-to-dc converter a linear voltage regulator LDO how well does it reject various frequency components at the DC input and so if you have a rather than you know noise input noise is broadband some of that is going to sneak from the input to the output this allows you to injecting known frequency and measure what is getting from the input and the output and the way power supply rejection ratio is computed its 20 log V n over P out now some people argue no it should Vout over Vin that’s actually 20 log Vout over Vin is gained this is rejection so it’s Din over Vout now what it requires it requires a disturbance signal or a source are InfiniiVision x-series oscilloscopes have a built-in function and arbitrary waveform generator you can use that as the disturbance source but it also has to be summed with the DC input you can’t just can directly connect the function generator to the DC source that has to go through a summing network and one here that are recommend is from a company called pico test this is the peacoat SJ 2120 line injector other alternatives you could a see you get couple with inductors and capacitors each of those sources the DC source in the AC source but you have a very limited frequency band you can also divide your design your own unity gain amplifier to stick in there the easiest thing is just to get one of these boxes from pico test and stick it in there and is so it’s got an AC input and a DC input and it’s got a single output so let’s take a look at the measurement on the scope the PSR our measurement so you first of all you connect waveform generator output to this line injector you’re summing network and then also you connect your probes up to measure being and Vout now we recommend that for measuring g out again you use a one-to-one passive probe because if you’ve got a lot of rejection on your converter that you’re running the tests on the output voltage level is going to be extremely low and that’s one of the limitations of using a scope to this measurement typically use a network analyzer then you can enter your frequency range you want to test from were able to test up to 20 megahertz you’re in a vertical plot scaling and then enter the amplitude of the test signal that you want to use for your disturbance what is the AC value peak to peak of the AC disturbance signal before I folks show the final measurement here this particular screen shot i’m showing the measurement at one particular frequency i believe this was 10 kilohertz now you probably can’t real read some of the scale factors here but the channel one that’s the yellow waveform net to be in this is disturbing signal the scope is set up at a hundred millivolts provisions so this is about a half half of volt peak-to-peak input channel two is that 1 millivolt per division and that’s the green wave form that’s V out and so in this case you can see we’re measuring about a quarter of a millivolt or 250 microvolts that is not easy to do on an oscilloscope and what the scope does in order to dig that waveform out of that the noise

floor the scope itself we use averaging 22 eliminate all non-correlated signals such as random noise from the scope and we also bandwidth limit – 20 megahertz and then the scope steps in multiple frequencies from the low frequency setting to the high frequency setting that it measures the rms of input and output and then it computes the PSS are 20 log be an overview out and then plus the results so here you can see the the measure their can complete measurement it just does it one time it takes approximately one minute to do this sweep the common thought I’ve heard about what’s the best of scope could do they say about 50 DP and that may be true if you’re not using averaging or some of the techniques we use here I’ve been able to measure in this case here i’m showing about 74 DP of rejection i think around 20 kilohertz in this example but i would say the maximum dynamic range is about 70 TV so the bottom line this is what I’ve been calling a poor man’s doctor network analyzer if this doesn’t replace a network analyzer and network analyzer would do a far better job as much more dynamic range better nocella scope but if if this is good enough for you it’s a pretty simple and inexpensive solution so let’s review what we’ve learned today power supply design optimization requires lots of testing and characterization . i’ve been trying to make here today scallops so scope license options for customized power measurements it can decrease your time to test we believe the power options available on on our oscilloscope provide the most comprehensive portfolio of power related measurements are power measurements pride provide setup diagram step by step instructions help screens and so on to make some of these measurements that can be pretty difficult a lot easier also probing and grounding techniques are extremely important if you’re trying to get important trying to get accurate measurements and key set offers a wide range of voltage and current approach to to meet your needs these are some of the recommended probes for power supply characterization I certainly didn’t list all of our probes the ones i have highlighted there and blue or some of the more popular probes for power supply characterization for a high voltage differential probux see the one listed there in 2790 a up to fourteen hundred volts there’s an error on this it says 51 or 101 that’s actually a 50 to 1 or 501 to get that fourteen hundred volt range and down at the bottom you can see this new power rail probe listed there and and again this is not a general-purpose probe for other applications is specifically a power rail probe for measuring ripple on DC key site has a broad range of Silla scopes ranging from handheld scopes you 60 1600 all the way up to very high frequency all type high frequency band what’s up to sixty three gigahertz the Scopes primarily focused on today or the ones you can see highlighted there in green and innovation series we have four different main products in that series of 2000 3000 4000 and 6000 x-series various man whats as well as sample rates the 2000 series does have does not have the automated power up that all of the others do I also showed one measurement on the infinium this is our new infinium s-series it has a very low noise front end and 10 bid a 2d this this is a extremely good solution along with that power rail pro to measure power rail integrity and I put that power rail probe is also compatible with infinite vision x series as well we have a lot of technical resources available you can go to key sites web site and find some of this information i think the one I would recommend first might be that switch mode power supply measurements application know that goes through a lot of measurements i talked about today it’s typical of you can see I’ve got a picture that happened out there it hasn’t been updated for keysight yet still as that agilent’s park on there so at this time I think I’d like to turn it back over to Bill and open it up for Q&A the first one are these probes for the AC current all compatible with other scope manufacturers

some of them are some of our probes have what we call the Auto probe interface and that’s a key side auto program interface and in those cases those probes can’t be is but also some of our probes have a standard BNC connection and some have battery operation or an external power supply those particular probes can be used on anything under scope ok here’s one more specific than 20 scope probe compatible with the tektronix cops unfortunately not ok what are the low and high frequency step points for the PSS are our sweep I believe the low point is 10 Hertz the the highest frequency is 20 megahertz that’s the maximum man with the built-in generator on that on these scopes ok I you know what scopes offer wage on all of the infinite vision next series so the 2000 x 3000 x 4000 X and 6,000 x 24 thousand and six thousand x offer to output channels and also double the output drive capability although for PSS our measurement you don’t need high output drive your typically disturbing with a very low output signal ok how do you effectively measure the current in milliamps using a current probe I know you have to calibrate it to measure in the range of 10 to 100 milliamps even after reducing the off set and calibrating the probe I can still see some noise on that measurement it is a challenge kinda mejor lo milliamps or even microamps first of all you talk about calibrating the probe this is something I didn’t mention one of my competitive colleague colleagues really hit on this a few months ago there’s a giggles button so probes will build up in that magnetic element on them and they will drift in and build up the DC offset and you have to continually degauss them and then recalibrate the DC offset now if you’re trying to measure low milliamps and sub milliamps a typical clamp on current probe is not the best solution there are other techniques using since resistors a sense resistor technique which is basically what a multimeter does to measure current you can use your own you can put in a small resistor in the circuit making sure that the value of the resistance does not substantially affect the operation of the circuit so it could be something like a a hundred million ohms and then measure the voltage across that and and convert that the current we also have a probe we introduced about a year ago that is is automatically made for this it has a sense resistor technology and also you can embed these connectors on your board and then you can plug in the probe and it’s got to make before break technologies such that you don’t have to break the line if the protein not plugged in the line is is is made once you plug the probe in it switches in that sense resistor before it breaks the circuit and so you can continue operation another unique thing about this particular current probe at this point I can’t off the top of my head we recall the model number of it it also has dual outputs and it’s an ideal probe for measuring low drain current three such as mobile devices so you do you take a example such as your cell phone you got it sitting there turned off I’ve got a sitting in front of me it’s always draining a little bit of current and then it wakes up every once in a while ago and and and pops up pops his head up and says hey is anybody out there do I have any calls coming in and then it goes back to sleep well that the difference in the dynamic range can be huge and so when it wakes up this measure a relatively high current and it goes back to sleep it’s still activity their problem is you can’t measure both at the same time this particular probe has two outputs one that is high sensitivity the clamps so you can measure down in the the tens and hundreds of micrograms range and then another output side that has a higher current setting ok I do the input line analysis tools work on the DC input line input line analysis

tools I don’t think so so if you wanted to measure you could use the efficiency measurement will automatically measure DC line and put it to measure the real power but you can’t can’t use the AC line analysis on a DC input current harmonics er I suppose you could do the current harmonics although it should be pretty low for the next question it would be nice to know which probes are compatible with other scope makers not sure if there a particular collection that would be the best way to do that is is go to our website take a look at the probes we have and and I we get you can look at the picture or I think we typically say does it have the teesside auto probe interface and and you can just look at the probe itself and it’s got a big fat input that plugs into the BNC or if the probe looks like a BNC connection and that is compatible with other vendors scopes ok discussion we have for measuring AC noise on a rail should you limit the Scopes BW man with a band with that’s what it is they want to see a higher noise content on the rail or they don’t have gotten over well it depends on what you’re looking for you can if you’re only interested in in noise up to a certain bandwidth yes you can limit the bandwidth the the the infinium s-series scopes has several bandwidth limits that you can set all the way down in hardware i believe all the way down to 20 megahertz so there’s a 20 megahertz selection a hundred megahertz election and I off the top of my head I can’t remember where all the other selections are and if one of the hardware selections doesn’t work for you we have a way for math function for low pass filter and there you can select any bandwidth you want to limit your bandwidth now if you’re trying to measure the highest frequency possible coupling or noise or perhaps switching noise it’s in there you may want to open up all the way so you can see your highest frequency now something I didn’t mention is you know we talked about a switching frequency in the case of our demo board that we use for some of our hands-on workshops it’s only switching at 70 kilohertz but if you run an fft on that you’re going to see you have multiple harmonics of that 70 kilohertz that go up to much higher frequency so there’s there can be frequency content due to the switching that’s going on far beyond the actual switching rate ok here’s another one how do you measure the ground difference on the scope for example you have gate driver on digital ground and the driver MOSFETs are on power ground how would you measure it effectively i used a differential probes to measure ground difference I think you could do that yeah so if you have isolated sides of the power supply so they’re isolated you could put a differential pro between them to measure what the difference in ground is and just realize that the differential probe does have an input impedance so it’s not call Donna CLE isolated so it does connect a big resistor could be 10 mega ohms between one side any other so i think that would be one technique but that’s that’s something i haven’t done not that familiar with it so ok does keysight sell the pico test accessory that you showed earlier for injecting the AC distribution signal for the PSR our test no we don’t and in fact our vet Dr network analyzers also recommend some of the pico test accessories the best way to get some of those if you’re interested in is just go to the website and it’s simply pico test.com ok are there any alternatives to measuring current other than the traditional clamp on current probe well i think i covered that a little bit earlier basically using sense resistor technology either you come up with your own own sense resistors is a or as I mentioned we we also have that new probe I think we introduced about it a year ago that has sent resistor technology ok I think he’s sort of mentioned that so this can uh your new power rail probe be used with the PSS PSS our measurement not today we just introduced this power rail probe today I recommend using a a

one-to-one probe which limited bandwidth but it’s written it’s a relatively low bandwidth measurement anyway the power rail pro the advantage is if we were to use it for the power supply rejection ratio the advantages is that it would have you could use it it it it drives into 50 ohms which 50 homes on the scope typically has a lower noise floor and so it’s going to give you more accuracy and then also a couple of our scopes like the 6000 series as even a lower noise for into the 50 ohm so it might be making those low level output measurements better but today it is not compatible perhaps in the future we may enhance the product to include this new power rail broke for that particular measurement ok we’re running a little over here so this will be the last question what band with scope is required for switching power supply measure my advice here well first of all switching rates are getting faster so the slew rate sus of many or power devices are getting faster and faster which requires higher bandwidth and it’s a it’s a challenge for all scope vendors to keep up with those requirements about the the highest bandwidth available today and a high voltage differential probe that will cover in the range of a thousand volts is about 200 megahertz and so right there you have limited the measurement bandwidth of your system to the bandwidth of the probe so it doesn’t make a whole lot of sense getting a multi gigahertz scope to measure something with a couple hundred megahertz probe now that’s that’s talking about specifically doing the switching analysis now is that we talked earlier about the power rail measurements that’s a 2 gigahertz probe so you could get two gigahertz are higher scope to make those types of measurements you’re trying to look at extreme power integrity and high-speed digital system all right well that concludes today’s presentation yeah yeah