(Sponsored) Inductor Hardware Design Basics (+Measurement & Modelling) - Phil's Lab #160

In this video, I'd like to go through the practical hardware considerations when designing with inductors. Typically, when we think of inductors, we might think of a basic inductance value. Let's say 1 micro Henry, 10 microhenry's, and so on. Maybe a package size and maybe some DC resistance. However, there are far more parameters and non idealities to inductors that I'd like to expose you to a bit in this video. We might be able to see some of these non idealities and parameters when going through a typical part search. For instance, here on Mousa, there are many, many different options to select from. We have inductance, tolerances, maximum DC currents, DC resistances, self-ressonant frequencies, package sizes, and more. This, of course, can seem very, very overwhelming, and what parameters you specify depends entirely on the circuit you're dealing with. If this is power, RF, audio filtering, and more. So, I'd like to give you more of an insight on how you can specifically choose and pay attention to these various parameters throughout. I'll also show you some useful manufacturer tools where you can get more information on these non ideal parameters and we'll perform some of our own measurements on real world inductors and then how we can model them in LT Spice and why that might be useful and I'll give you some practical considerations along the way. As usual, a huge thank you to JLC PCB for sponsoring this video for characterization and measurement of an inductor which you'll be seeing this video. I'm using some custom-designed calibration and test PCBs that I then had manufactured by JLCPCB and they did an absolutely fantastic job. JLC PCCB provides easy, affordable and reliable PCB and PCBA solutions which empowers electronics engineers to develop projects efficiently. JLCPCB is incredibly easy to use, affordable and reliable. The ordering process is incredibly simple. We simply have to upload our Gerber files, select some of the core properties we would like, and if you'd like assembly, it's incredibly easy to upload your blue materials and pick and place files and to select from JLC's and LCSC's very large component catalog. JLC PCB is incredibly affordable. We can see here six layer PCB 50x 50 mm is only $2 with included options of epoxy filled and capped VA. So you can do VN pad and even eight layers is currently at $2 for 50x 50 mm PCB which is simply incredible. They're very reliable in terms of quality and lead time because they have an all-in-house production and rapid turnaround. I'd strongly suggest checking out JLCPCB. Again, they're super easy to use, affordable, and reliable, and you can always count on JLCPCB. To help you with ordering from JLCPCB, there will be some links in the description box below where you can get various coupons for your PCB design projects, such as a $30 coupon for GLC PCB premium six layer PCBs. Let's start off by going through some inductor basics. We'll be focusing on SMD inductors in this video, but of course this applies to throughhole inductors as well. SMD inductors typically come in various sizes. These could be related to standard such as imperial or metric sizes such as 042, 0603 and so on. But of course, many inductors come in these somewhat custom sizes. I've shown some different inductors on the right hand side in these various images. Other than various sizes, of course, there are many different properties that make up an inductor. So there might be different SMD inductor types for various applications which then have differing properties for those applications needs. There might be specifically for radio frequency electronics RF for short for power which might be able to handle higher currents. We might have shielded or unshielded inductors different core materials and so on. The different types and properties of course have an effect of the performance in the specific circuit. When we're looking through Fson's mouse or dig key looking through the data sheets that these various inductor manufacturers provide, you will come across these main parameters which are the absolute basics that somewhat characterize these inductors. The first parameter will of course be your inductance and a tolerance. This might be 10 microhenre's plus minus 20%, 20% is pretty typical. So your inductor value can range anywhere from 8 microhenre's to 12 microhenre but it's normally at 10 microhenre. Then there's the rated current and the rated current is the maximum permissible current based on some sort of criteria. This might be saturation or it might be temperature-based. Saturation simply speaking means an inductance reduction with increase in current and temperature-based is how much the temperature increases typically maybe 40° Celsius that is the rated current based on that temperature rise. Other than inductance and rated current we will also typically get specified a nominal DC resistance which also has its own tolerance and that will be DCR for short. Depending on the size of inductor, the physical properties, the number of windings, the type of winding material used that DC resistance will vary quite considerably. There's also the type of core material that's used for an inductor. This could be a ferad based or powderbased core. And I'll leave a link to this Mag Inc. web page in the description box below, which goes over the essentials of inductor core materials and shape choices. So, this is a rather interesting read and summarizes the core material properties quite well. Other than these main parameters which to a certain extent characterize the inductor, there are many more parameters and effects that need to be

considered when you're designing a real world hardware. For instance, the inductance might have a tolerance, but the inductance might and probably will also change depending on how much current is pushed through the inductor. The inductance might change depending on what frequency of signal that's passed through an inductor. With changing temperature, the inductance will change and more. Other than that, we have something known as a self-ressonant frequency, which we also can see with capacitors. And this self-ressonant frequency is due to a parallel winding capacitance between the windings of the inductor. And we can see the windings in the top picture quite clearly. And the self-ressonant frequency is a point which we'll see later on after which the inductor doesn't look like an inductor anymore. So you might not want to use the inductor close to or above itself frequency. Furthermore, the DC resistance again has a nominal value and some tolerance. But this will also change with temperature for instance. Then as we briefly talked about before, there is this saturation behavior which depends on the core material, core material type. Again, simply speaking, saturation is an inductance reduction with increase in current. Now depending on the core type, this could be a very rapid inductance reduction. For instance, with fried cores or with powder cores, this could be a more smooth, somewhat more linear reduction in inductance with increase in current. But saturation is something we typically want to avoid in most applications. Saturation you can think of is that as we get to saturation, the inductor looks more and more like a short. I'll leave some links in the description box below which go over saturation a bit more detail. As you can see, there are many more parameters to inductors than just the inductance and maybe current rating on its own. And these parameters need to be considered and chosen carefully depending on the situation. For instance, if we're looking at a buck converter, choosing the magnetics, choosing the inductor for that switching supply will be a rather different affair than choosing a radio frequency inductor for a matching network. Say that's running at 1 GHz. That again is different to choosing an inductor for power supply input EMI filter. And then again, that's different to choosing an inductor for an audio filter. So depending on the application, you do need to consider these parameters quite carefully. One aspect we briefly touched on and I'd like to go into a tiny bit more detail now is that of drating. When we're talking about drating, we are talking about operating points and those operating points and conditions changing the inductor's behavior and characteristics. An operating point might be how much current is passing through the inductor. It might be what the temperature is of the inductor or what frequency of signal we're passing through the inductor. This is often not given directly in data sheets, but rather you'll have to do a bit more search or maybe have to measure the inductor yourself to get this information. I strongly suggest checking the manufacturer's detailed specifications. And I'll show you a website called Marata Simurfing, which is of course specific to Marata itself, but this gives you all of these details. And these graphs, for instance, are what I've pulled from the Marata Simurfing website. These graphs are for a particular inductor which I'm showing in the bottom right hand side. It's a Marata 0603 SMD inductor with quite a high inductance for its package type. So 47 microhenry's normally 20% inductance tolerance, 20 milliamp saturation current and 180 milliamp temperature-based current limit and with a nominal DC resistance of 2 1/2 ohms. The first graph on the left hand side shows how the inductance changes with current through the inductor. So with no current through the inductor at the far left hand side of the graph 0 milliamps you are getting the nominal inductance value of about 47 microhenre's but this value drops rapidly as we increase in current. So as we reach this 20 milliamps of saturation current inductance has dropped by approximately half and as we go high in current up to the maximum temperature-based current limit 180 milliamps our inductance has now dropped to about 10% of its original value. The middle graph shows inductance versus temperature where we also have a variation maybe not as drastic as inductance versus current but this will also of course change on the specific part you chose. On the bottom left hand side we have the temperature rise versus current. So as we increase the current in the inductor due to its resistance and power losses we will get increased temperature rises. Similarly for the DC resistance we also have a graph. So as the temperature changes the DC resistance changes away from its nominal value. So nominal value might be at room temperature around 20° 2 and 1/2 ohms. As we go above and below that DC resistance changes as well. And finally we also have frequency dependence on the top right graph. We have an inductance versus frequency graph. And our nominal inductance might only be valid at 1 MHz where it's typically measured at or 2 MHz or below. We'll go into more detail regarding impedance or inductance versus frequency graph and actually do some measurements later on. But overall, it's important to keep in mind that your current, your amber and temperature, and many other parameters play into the inductor's behavior and can introduce these nonlinear effects that will affect how you design your hardware and how your hardware performs. Some specific examples that highlight some issues with durating. For instance, you might have a

filter on the top right hand side which is a power supply filter or you could see this as a basic LC filter that might filter an audio signal where we have an inductance series and a capacitor in shunt parallel configuration afterwards which forms a lowass filter. Now if the inductance changes due to derating the current through it the ambient temperature of course the filter's cutff frequency which is given by 1 / LC in radians/s will change because of this change in inductance. Even small variations in current through this or signal applied across the inductor might give inductance changes which might lead to nonlinearities which you might not want in your signal because you get distortion. Similarly for power supplies for instance this buck converter shown here L2 is the magnetic element our inductor which is chosen based on certain operating conditions is chosen so that this buck converter is stable with in the combination with the load and the output capacitance and the feedback network. But as our load demands change, if we have low or high current, the inductance might change as well, which in turn changes the stability of the system, the response, for instance, trans response. We might get increased ripple as we approach the saturation current and much more. Just to highlight, it's important to consider these effects because these have realworld effects on your hardware. Other than derating, I'd like to talk a bit about inductor modeling because you can use inductor modeling to simulate your circuits and ensure they'll perform as intended when you actually come to designing real world hardware. Typically, our inductor symbol is shown on the left hand side. It's simply a coil with a nominal inductance L. This is what we might learn in school or we might learn at university that this inductor is basically a frequency dependent resistor and that's about it. However, other than the derating we saw previously and these various parameter changes, saturation and more, we also have frequency dependent effects that aren't characterized by this simple inductor model with just a simple element. For instance, if we're going to the RF domain, we are working at much higher frequencies. The frequency domain performance or frequency response of an inductor becomes more and more critical to be able to model to make sure the circuits work as intended. It turns out that any inductor is not just an inductor. It's not just this ideal circuit element. And you might want to represent your inductor by a different model. And I've shown one of these types of models as an equivalent circuit on the right hand side. Keep in mind there are different models and different model fidelities required depending on the application. This particular model I'm showing you here does not capture the saturation effects on its own as well as the derating effects. And it also doesn't capture the variation and tolerance just by this static model on its own. But it does characterize some of the frequency dependent behavior of an inductor more closely. Let's go into detail in this model. Around the inductor, we have now added three other basic circuit elements. At the bottom, next to our inductor. In series with inductor, we have RS, which is a series resistance, which is a model for a coil wire resistance, effectively our DC series resistance or DCR for short. Above that, in parallel with these series elements, the inductance, our nominal inductance L, our series resistance R is, we have this parallel capacitor CP. CP is used to model the interwinding capacitance which is typically very small between the individual winding of the wires around the core. And finally right at the top we have this parallel resistance which is there to model the magnetic loss. And this might typically be a value in the order of 10 kiloohms but of course depends on your inductor. So we've gone from just having a single inductive element to adding some DC series resistance, parallel capacitance, and a parallel resistance to create a higher fidelity frequency domain model that might more accurately correspond to a real inductor. If we move over and plot this impedance of this new modeled inductor versus frequency, we might get a graph such as this. At low frequencies, the DC resistance RS dominates. Of course, here the graph is showing 1 MHz, but of course, we're talking about close to DC, which this graph isn't showing. But the basic idea is that the DC resistance dominates at low frequencies. Then we move into our inductive region where our L, our inductor dominates until we hit a self-ressonant frequency, which is our maximum of our impedance versus frequency curve. And this is where we get this self-ressonance between the winding capacitance and our inductance of the inductor. Once these impedances cancel out, we're left with a self-ressonant peak. After which the inductance no longer dominates, but rather we go into this capacitive region where actually the winding capacitance dominates and our impedance now goes down and down again because the winding capacitance at higher frequencies acts like a short circuit across our inductance. Already from this you can see that we have an effective frequency range where we get inductive behavior. Once we're at the self-reson frequency and above, this nominal inductor behaves more like a capacitor. Again, due to the fact that we have this parallel winding capacitance. So at RF, let's say at 1 ghahz as a signal frequency, our inductor behaves far more like a capacitor, which might mess up our circuit behavior. Other than acknowledging this new frequency response model, of course, we want to

figure out how do we even calculate these various component values. How do we calculate the parallel resistance? winding capacitance? And how do we calculate the series resistance? Keep in mind so far the data sheet probably has only told us what the nominal inductance is plus some sort of tolerance and maybe the DC resistance. Finding these model parameters actually turns out to be fairly straightforward once you've acquired the impedance versus frequency curve which can be a bit more involved and does sometimes require more specialized equipment. But I'll show you how to do that in just a few moments. We'd like to find these model parameters because we want to run some simulations to figure out if an inductor is suitable for our specific circuit and application. We need to find the inductance and that's of course fairly straightforward. That is given typically on the manufacturer website or in data sheets. Keep in mind if you'd like to incorporate derating effects, you'll have to include the effect of DC current bias or saturation as well as temperature derating because these effects will decrease the nominal inductance L and shift of course your self-ress resonant frequency as well. The series resistance our DC resistance is easy as well. That's pretty much equivalent to the impedance at DC. So low frequencies, let's say 1 Hz, 10 Hz. Reading off the value of impedance at those low frequencies gives us essentially the DC resistance RS. The parallel capacitance and parallel resistances are tiny bit more involved but still fairly straightforward. We have our self-ressonant peak and at our self resonance we can approximate this to be a parallel resonance circuit. So approximately that frequency is 1 / 2 pi the square roo of LC that's in hertz. We know our nominal inductance we can rearrange for CP or we know that at high frequencies our CP dominates and we know the reactance of capacitor. We know the frequency. So we can plug in those values. The parallel resistance RP we can get from the impedance peak magnitude. We're going to ignore the DC resistance and basically the peak impedance magnitude is approximately equal to our parallel resistance which models our magnetic loss. Now I'd like to show you how we can use a BOD 100 vector network analyzer in combination with one of my impedance test boards that I've made to measure the impedance versus frequency curve for an inductor. On the left hand side very simple setup I'm using a two port shunt through measurement setup. And if you'd like full details on this measurement setup as well as calibration and how this works, please go and watch video number 151 on my channel. We can use this two port shunt impedance measurement setup with our test board to acquire impedance versus frequency of our inductor. For this, we do need to perform some calibrations. So an open, short, and load calibration. And then finally, the last segment of this PCB is the device under test where I will solder my 0603 47 micro Hendy normal inductance inductor to and perform those measurements. So, let's move over to the Bode measurement software and I'll show you what this looks like. Now, we're in the BOD analyzer suite. I've hooked up my Bodi 100 vector network analyzer to my two port shunt test board. I've performed all of the calibrations and now we're ready to perform an impedance versus frequency measurement of our Marata 47 microhenry 0603 inductor. I'm doing a sweep from 100 Hz to 50 MHz. I've set up my attenuator, receiver bandwidth, and so on. So now let's click on a single measurement and we can watch the impedance magnitude on the top graph and the phase versus frequency in the bottom graph. So let's see what this now looks like at low frequencies. So close to 100 hertz, close to DC, we are getting a flat impedance before inductance kicks in and we move into this inductive realm. And this is as expected as we move higher and higher in frequency. If I move my cursor at the peak here, this is our self-ressonant peak. This is where the interwinding capacitance and the inductance cancel out and we are at our self-resent peak magnitude. After which the capacitance starts to dominate and we're moving into our capacitive region which is now outside of the measurement range of the instrument which is above this 50 MHz. But at least now we are capturing this low frequency DCish region inductive self-ressonant peak and then capacitive region. Similarly on the phase of course we're at 0° phase and as we move into this inductive region we're getting closer to plus 90° phase at the self resonant frequency we traverse through 0° of phase and then we move towards - 90° because we're now in the capacitive region. We can use the Bodhi 100 analyzer suite to help us out with figuring out these various component parameters. For instance, on the right hand side, if I'm looking at trace 2, I can change the format from phase to ls, which is inductance. And moving more into the inductive region, we can see if we're looking at the trace curves on the top hand side, we are around our 47 microhenry nominal inductance, which is exactly within the tolerance of our part, our 47 microhenry part. So, our nominal inductance is about that. And we can see as the frequency increases, this inductance actually changes as well. and also decreases. So again, something important to keep in mind. We can also change our curve from inductance to resistance and move to the far left side of the curve. And let's say at 100

hertz, we're getting a resistance of 2. 5 ohms, which is exactly within spec for this particular inductor. Normally, our DC resistance is given at 2. 55 ohms. Keep in mind, this is without any DC derating effects. Of course, whatever source level we feed through our inductor, whatever we're using for testing will and might may have an effect on the inductance we measure. So, keep in mind not to have a too low or too high source level. And of course, we're not doing any sort of DC current derating for this. But now we have our inductance value and we have our DC resistance value. So, 2. 5 ohms and about 47 microhenre of inductance. What we're interested in now is figuring out our capacitance, our parallel winding capacitance and our parallel resistance due to magnetic loss. And we can get that from our peak. So remember from our slides at the self resonance, we have approximately the formula F= 1 over 2 pi LC. We can rearrange for C at that particular frequency with our inductance or we can look at the high frequency region where the capacitance dominates. The way we can get the peak is go on the impedance magnitude, right click, cursor one, find maximum. Or we could also go on the phase and see where we get the zero crossing. So right click, cursor one, find zero. And we can see our peak frequency, our self-ressonant frequency is around 24. 6 MHz. Typing this into a calculator. So 1 / 2 pi f 2 * l. converting that to poparads, we are getting approximately 0. 9 poparads of interwinding capacitance. Lastly, then remember we talked about impedance peak being approximately equal to our parallel resistance and we can simply read that off magnitude cursor. So we're getting about 11 kiloohms at that impedance peak and that's going to be equal to our parallel resistance. To see how this is now useful in simulation and for circuit design, I've moved over to LT Spice and imported our non ideal and pretty much ideal inductor properties into a simulation. The bottom left circuit, I simply have a current source feeding a DC resistance plus my nominal inductance value of 47 microhenry. So here we have a pretty much ideal circuit just modeling DC resistance plus our inductance. And I'm feeding in a current magnitude 1. And once we probe the node to the left here, we can get the equivalent impedance looking into the inductor. On the right hand side, I have the same setup, but I have a capacitor at the output and a voltage source instead of a current source because this now forms an LC lowass filter. At the top, I pretty much have the same circuitry. So one for impedance measurement on the left and one for a filter configuration on the right hand side. But I've included our not ideal parameters such as the winding capacitance and our magnetic loss resistance in addition to of course our DC resistance as before. In this way, I'd like to show you what the differences practically then are between a more ideal and a less ideal inductor model. I'm running an AC simulation from 10 herz to 1 ghahz. So I can click run now and let's plot the impedance of our ideal inductor with DC resistance and this is exactly what we expect. I'll change the graph magnitude the left hand side at DC we get our DC resistance we get this a somewhat flat band up to a corner frequency and then our inducted inductive region at infinitum. If we plot our more non ideal impedance in addition we can see we get the same low frequency performance. inductive performance up to or close to our self-resin frequency after which these models differ quite considerably. Our non ideal model has our self-resin frequency as we calculated at around 24 MHz with a magnitude of about 11 kiloohms due to this parallel resistance after which this winding capacitance dominates and we get a reduction in impedance magnitude. Now, if we apply that and check out the differences between these two lowass filters, let's say the more ideal one, we have a 0 to dB pass band with a cut off frequency at about 700 htz. Comparing that to our more non ideal filter, these again are very, very similar up to and close to the self-res frequency after which our filter that incorporates these parallel elements now flattens off and tapers off and gives us a minimum possible achievable attenuation rather than a decrease in amplitude. as we go up and up in frequency. So if you're looking at certain attenuation above the self-ress resonant frequency, you need to be careful because of this parallel winding capacitance and the reduction in inductive behavior above that point. Below or away from this self-ressonant frequency towards DC, these filter models are quite similar. So this is a model you might want to use if you're running RF simulations. You might want to include even more information, more detailed capacitances, inductances to model the situation more appropriately. Whereas for a buck converter simulation where you're running at a switching frequency of let's say 500 kHz. The DC resistance plus inductance model might be completely appropriate. I'd like to add a couple more words to spy simulation. However, if you're including these non idealities, it is far better rather than having all these external circuitry and components to right click

on your inductor and enter your series resistance, parallel resistance, and parallel capacitance into the inductor model itself here rather than having external circuitry because LT Spice can handle these properties that are internal to the inductor far faster and you get far faster simulation results with that. The way I've drawn it here is just for demonstration to make it a bit more clear. Keep in mind, however, you will have also seen in the inductor properties dialogue that a series resistance defaults to 1 milliohm if no series resistance is given. That's important to note. Also, if we go to the hammer on the top left and go to hacks, there's also this check box which is checked by default which says supply a minimum inductor damping if no R parallel is given. So, you might want to uncheck that and use your own parallel damping values. Finally, I'd like to leave you with some very generic part selection guidelines for inductors and what you need to watch out for. Of course, this whole thing can be very overwhelming, especially if you've just started out with hardware design. But here are some basic tips. The first thing is, what type of application are you dealing with? Is it RF? Do you have to take care of self-ressonant frequencies? Is it to deal with switching power supplies where maybe current and saturation is more of an issue? or audio filters where you have to avoid nonlinearities and you're not too worried about current handling or getting close to these self-resing frequencies. That will determine the type of inductor, type of core type of properties, you will need to choose. In general, if possible, it's mostly advisable to go with shielded inductors rather than unshielded because they have typically better EMI performance. Package sizes. As you go larger in package sizes, typically this will give you lower DCR. depends of course on your inductance values and more. But also larger packages might allow you to have larger current handing capabilities. They might give better temperature performances and more. I'd also suggest choosing lower profile packages where possible also for EMI reasons. Core materials and please refer to the document in the links in description box. You might want to go either with a ferite or powder or other type of core depending what properties you need. Also with respect to saturation characteristics, ferits typically have a far more aggressive saturation characteristic than powder coils. In terms of tolerance, you will mostly see 20% tolerance or maybe even worse tolerance inductors, which is usually fine and standard for power applications, but do calculate the allowed worst case for your specific application. For instance, an RF filter with 20% tolerance might not be terribly great because that influences the cutff frequencies or pass band frequencies quite considerably. The DC resistance unless needed for damping or other reasons I suggest going with the lowest possible DC resistance possible. For example, for power conversion that will reduce your DC power loss and reduce your temperature rises as well. Operating point again is very critical. The current through the inductor will change it inductance and increase its temperature which again has an effect on inductance and temperature itself. Calculate your worst case conditions and adjust your nominal inductance value that you're not outside of the spec for your circuit you're designing. Stay away from saturation current ratings and always take the ratings we talked about into account. I'd strongly suggest using vendor supplied tools. For instance, for Marata, they have SIM surfing, which we've previously seen in some videos on multi-layer ceramic capacitors, but they also have SIM surfing for power inductors as well as RF inductors. And these tools are super helpful for weeding out inductors, which might be suitable for your application. And here you can choose an inductor, plot the derating curves, plot frequency characteristic curves and much more. So rather than just going by nominal values, please ensure that you go into a bit more detail depending on your application. Thank you very much for watching this video. I hope it was useful and hope it gave you a bit more insight into the more non ideal parameters of inductors, why this is important, how we might be able to measure inductors and their impedance versus frequency response, and how we can then use that in simulation. If you like the video, please leave a like, a comment if you have any questions, and don't forget to subscribe to stay uptod date with any latest PCB design, hardware design, and embedded systems videos. Thanks again for watching, and I hope to see you in the next one. Bye-bye.