The following transcript and video were originally presented at EMC LIVE 2016 by Lee Hill. For access and information on all EMC LIVE events, both past and present, please visit https://emc.live/.
Kenneth Wyatt:
Hello, and welcome to EMC Live 2016. This presentation is about visualizing, localizing and troubleshooting EMI signals with real time spectral analysis. I’m Kenneth Wyatt, senior technical editor for Interference Technology, broadcasting from our studio in Colorado. And we are the creators of EMC Live. EMC Live is a free three-day event focused on practical EMC pre-compliance test design and developing standards side of EMI engineering. And today’s presentations will include compliance testing with pre-scan analysis tools, EMC testing to make your job easier, fundamentals of EMI, pre-compliance testing, and we’ll be updating you on mill standard four 61 gene.
So join us for all these topics today. Just register at the EMC Live 2016 website, emclive2016.com. This webinar is presented by Lee Hill. Lee is a founding partner with Silent Solutions, LLC, and is an industry expert in electric magnetic compatibility and provides EMC troubleshooting services, design reviews, and training to a wide variety of industries nationally and around the world. He also teaches graduate level classes in EMC at Worcester Polytechnic Institute, University of Oxford, and for the [inaudible 00:01:44] EMC society’s annual Global University and EMC fundamentals program. He earned his MSEE in electromagnetics from the Missouri University of Science and Technology, EMC laboratory.
Attendees, before we begin a few notes for the thousands of you who have registered for this event from over 70 countries from around the globe, and may not have attended an EMC Live event before. First, since you already registered, you will automatically be sent an email the next week, when a copy of Lee’s recording and presentation slides are all to download. So no need to email us. Second, the presentation is interactive. So please do send in your questions during the course of Lee’s presentation and we’ll answer as many of them as possible at the end.
To send in a question, just use the chat button on the right side of your screen. And if you wish to put away the control panel during the presentation, simply click the orange arrow button at the top right area. Click it again to see the chat channel once again. Lee’s presentation lasts about 35 minutes and it will followed by a 10 minute question and answer session. And now Lee Hill, of Silent Solutions will begin the presentation. Over to you, Lee.
Lee Hill:
Great. Thank you. Well, welcome everybody. And we’re here to talk about the real-time spectral analysis. I’m going to just give you a very quick introduction here and we’ll be going to launch right into it. My job at work, I’ve spent about half of my time with hands-on troubleshooting of EMI problems in hardware. Some of our customers care about, what I call passing the stupid EMI tests. They care about regulatory noise problems, and some of our customers don’t care about regulatory noise problems. They want their systems to work okay. They want their sensors to work good when motors and power supplies are turned on. So we work on a wide variety of noise problems from regulatory noise problems, functional noise problems, about half my time is spent on troubleshooting those things.
About half my time is spent doing design reviews of schematics circuit boards, cables, connectors, and enclosures. And the other, the third half of my time is spent teaching, for-profit classes, in-house classes and open enrollment classes. And as Ken mentioned at Worcester Polytechnic Institute. All of our courses include live demonstrations. I think it’s just not possible to teach or learn EMC without them. So in this presentation, I’m going to include a number of videos of actual hardware demonstrations. I’m going to try to bring the fun and enthusiasm of live presentations that I do to this online forum.
That’s going to be a little challenging and we can’t actually give you the sound of the demonstrations through the video files. That’s not quite possible yet with the technology. So for extra fun, I’m going to be making some sounds for you as we watch the video. So be sure to be listening for those sounds, that’ll help you understand what’s going on in the video. So just a little extra note about what’s to come. All right. So we’re going to talk about a few things here this morning or in the evening where you might be, or early in the morning where you might be. We’re going to review some old measurement technology. We’re going to do a quick review of EMI signals. I’m going to talk about why I am so excited about real-time spectral analysis and particularly persistence mode.
I’ve been in the EMC engineering for about 25 years. And while we’ve seen improvements in technology, in graphical user interfaces of our measurement machines and it’s easier to get files in and out, and we can remotely control our spectrum analyzers now. There have been really few, revolutionary changes, very few game-changers in the field, and this is one of them. And, a couple of years ago, I got a spectrum analyzer with real-time capability, but I was afraid to push the button. Nobody taught me how to push the button. And so once I pushed the button, I got trained how to use it. I found out that there was a world of looking at EMI signals. There was a new world there that I just wasn’t aware of before, and it really has changed the way I approach EMI troubleshooting.
And that’s from a guy who’s getting older, who’s been troubleshooting prompts for a long time to actually say that there’s something new and different. That’s remarkable given the amount of time we’ve been working on, say regulatory noise problems in our field. And then of course, as I said, the demonstration videos, all the presentations, I do always include hardware demos. You see there a tabletop of me, all the stuff I’ve got for teaching a class. So I’m going to try to bring some fun into this presentation by playing some videos of the actual hardware demos. I the old days, we used to use swept-tuned spectrum analyzers or stepped tune receivers.
If you have an old spectrum analyzer with a green screen, you know who you are. Your machine definitely is a swept-tuned spectrum analyzer. And up until a few years ago, this was the dominant technology in EMC measurements for regulatory concerns. We call them swept tune spectrum analyzers, or step tune receivers. That’s because if you look down at the bottom of the screen here, there was a local oscillator in that machine that was swept from a low frequency to a high frequency. So here comes the first sound effect, from low to high. And wait, and wait, it takes time for that local oscillator to sweep, and it takes time for it to start at the low frequency and go back up again.
So there’s a significant amount of time when the machine can’t measure a particular frequency or when it can’t measure any frequencies, we call that blind time. While the machine is sweeping a wide frequency range through a narrow little window, that’s our resolution bandwidth, machine can only look at a very small number of frequencies at once. And then other times it’s completely blind. If the signal is present, that we’re looking for all the time, if it’s always there, then we’ll always measure it. It doesn’t matter when the machine is blind or when it’s swept past that occurring signal.
We know that a signal will be there when we get back to it, but there is a significant blind time when sweeping or stepping, it takes a long time for the machine to do this. So just to take a quick look a little bit deeper at this, there’s really three parts to measuring an EMI signal. There’s the acquisition. We pipe a signal into an attenuator, a pre-select or a preamp and a mixer for the old swept tune spectrum analyzers, or step tune receivers. But the process is the same, regardless of the type of machine. We’re going to go acquire the signal, and then we’re going to measure or make our machine respond to that signal. We use some resolution bandwidth filters at RBW setting. Now when swept tune and step tune machines, those are physical resistor capacitor filters, and they take time to respond.
We have to wait for them, they’re like a mechanical system, they’re like a spring. We have to wait for that mechanical response time. That takes a lot of time. It takes time for that local oscillator up here to sweep, and then to get back to low and zoop, and wait zoop, and wait. It takes time for those resolution bandwidth filters to respond. They have to stretch out like a spring and come back. They have to respond. And then we have to do our display magic. We have to show the information to the user. So this acquire, measure, display or acquire, respond and display, that all takes time. And there’s significant portions of time where the machine just can’t see what’s going on.
So in the old days, let’s take a look at this. What does this look like? With a very simple animation down the bottom of the screen, you see a curve that represents a resolution bandwidth filter, and the horizontal line represents a frequency axis. What does it mean to be sweeping across frequency with a narrow window? Well, it’s slow and let’s go try to measure a signal that’s blinky, that turns on and off through this narrow resolution bandwidth filter. What happens? There’s a signal turned on and off, blink, blink, blink. And then do we actually integrate the power that signal? Do we actually capture it? Is the machine at the frequency it needs to be to measure that signal?
Well, sometimes it is, sometimes it isn’t, and it takes a long time and that machine is blind when it’s going back and starting that sweep again. So this is what we used to use, slow and not so good at measuring these intermittent blinky kinds of signals. Now, what does it mean to be slow? How slow is slow? Well, let’s take a look at swept tunes spectrum analyzer versus new technology, fast Fourier transforms spectrum analyzer. When would this really matter? Well, suppose that we’re working in the automotive industry, and we’re trying to measure the impact of a piece of onboard electronics to our onboard radio. So we’re worried about noise reaching an antenna on the vehicle, an intended FM broadcast receiving antenna, or an intended television broadcast receiving antenna on our vehicle.
Well, we’re going to try to measure a weak little signal cause our receivers are very sensitive. So to do that, we’re going to use a narrower resolution bandwidth to drop our noise floor down, so we integrate less power. But we want to see lots of frequencies. We want to understand what signals are working at the antennas. So we’re going to look at a wide span and then make changes to the vehicle to understand what’s going on. We need to do that over a wide frequency range to understand the noise problems so we can solve it.
Well, if we reduce our resolution bandwidth, if its bandwidth gets small, that means its response time gets slow. We’ve got to wait a long time for it to respond. We have to slow down our sweep time and that slow sweet time just makes it a real difficult task to find noise prompts, to observe cause and effect changes. That slow sweep time just makes it really tough to identify signals. Even over a narrow span, never mind the wide span that we would like, how slow is slow? Here’s a recent vintage Rohde & Schwarz ESCI. This is a machine that’s still it’s only been out of use maybe five years ago and, but it’s a sweat tuned machine. I’ve got it set to a span of 500 megahertz. I’m using a narrow resolution bandwidth that might be required by the standard that might be required to find the smaller signals.
And my sweet time is required to be five seconds. That is forever. I’ll never see the response across that whole spectrum to a turning on and off of a system or the movement of a cable very quickly. I have to wait forever to understand what happens. That’ that’s a big handicap when I’m doing hardware troubleshooting. If I just simply go to an FFT based machine, a newer machine, and you see under the mode picture here, you see auto FFT and you see the sweet time is 27 milliseconds over essentially 500 megahertz of bandwidth. So FFT lets me sweep faster. So just a survey of old technology. So let’s take a look now with the same sort of animation and FFT based machines. So, this is going to be a little different. It’s going to be faster.
We’re going to measure lots of signal frequency simultaneously, in parallel. It’s like having lots of peak detectors and lots of quasi peak detectors and lots of average detectors all in parallel across this chunk of frequency, that chunk this rectangle here that could be tens of megahertz in width. Now the part of the magic is instead of waiting for a physical resistor capacitor filter to respond, we’re going to calculate the equivalent filter response. We’re going to use compute time. I’m going to substitute compute time for a mechanical or an analog response time. So we’re going to be able to make those measurements much faster. That respond or measurement phase of the EMI measuring process is going to be faster. There’s still going to be some blind time though, we can’t do this instantaneously.
So we can’t guarantee that we’re going to find a signal within this chunk of frequency. We still can’t guarantee. It’s going to take time to go off and do those measurements and display. And we’re only going to measure each frequency within that chunk of spectrum once. If the signal is sometimes they’re in sometimes not, we might miss it. And we’re also going to miss the opportunity to apply statistics to understand the nature of each different signal within that a chunk of frequency. So here’s my animation. The rectangle means, the machine at a particular frequency measuring and the red line is the actual signal blinking on and off. You see sometimes the machine is present and capturing that signal and sometimes it’s not. So that means even with a modern FFT machine, you go out and buy a brand new one today, one of the best ones on the market. It’s still possible that a little miss a signal.
Now you might be thinking right now, “Well, it’s only a special case.”Almost 99.99% of the time, I’m going to measure the signal. I’m not going to miss anything.” Well, that’s what the demonstrations are for. We’re going to take a look at some very simple signals here in a little bit. The last technology that we’re here to talk about is real-time fast rate transform. It’s the fastest type of EMI measurement machine out there. There’s dedicated additional hardware to do that FFT. It’s going to look at a wide chunk of spectrum simultaneously. How wide? Well, it depends on the make and model of the machine, the manufacturer and the model number, but let’s say it’s either 40 or 80 megahertz wide.
Now be careful. I was at an EMC fair, an EMC convention in Germany, in February. And there were many, many vendors showing different types of real-time FFT measurement, spectrum analyzers. And they all had different real time bandwidth. If you were allowed to set the span, that is the start and stop frequencies of your spectrum analyzer wider than the real-time bandwidth. Well, what does that mean? That means that there’s 40 or 80 megahertz chunks that are being put side by side by side by side. And it takes time to do that side-by-side, that stitching. So I asked a person at that show. I said, “What does that mean? You have a span of 500 megahertz, but you only have 60 megahertz of real-time bandwidth. What is that? What is that? Part-time real time?” And the guy said, “Well, yeah, I think it is.” And he laughed and he smiled said, “Yeah, it’s part-time real-time.” Well, there’s no such thing as part-time real-time either you can see the signal, either your guarantee the signal intercept or you don’t.
So, on the machines that I use, in real time mode, I can’t set the span wider than the 40 or 80 megahertz of promised or guaranteed real-time bandwidth. Now what’s going to make this machine fast, it’s like the regular FFT is that we’re going to calculate the response to signals. We’re going to calculate the resolution bandwidth filter response instead of waiting for it like we did with the old resistor and capacitor filters. This machine though, this real-time means if there’s a signal there, we’re going to measure it. We’re going to capture it. If a signal occurs within the 40 or 80 megahertz real time bandwidth, we’re going to see it. The other things that the dedicated hardware gives us, is it gives us the ability to calculate statistics. We’re going to measure those frequencies many times within that chunk of spectrum. Now here’s the animation that I just turned on.
You see that signal is sometimes on and sometimes it’s off, the machine is always able to measure that signal. We’re going to find that signal if it’s present within that 40 or 80 megahertz chunk. So we’re going to make sure we capture it. And then, as we’ve got this dedicated hardware, we’re going to make those measurements within that chunk of spectrum many times, which is going to allow us to do something really fun. It’s going to allow us to do statistics. So a very quick review on signals very quick. And then we’re going to move into the demonstrations, video demonstrations.
EMI signals that we measure, they have some funny names. The familiar digital clock signals with a finite edge time. We know what they look like in the frequency domain. We call those continuous signals or continuous wave signals. We say they’re periodic, they’re boring, they always do the same thing over and over again. We call them power signals. They have a finite power. Those signals are there all the time. They’re easy to measure. And then you hear people at EMI test labs, yelling, “That’s a broadband signal.” “No, it’s a narrowband signal.” Or you see regulations talking about broadband and narrowband. What the heck does that mean? Well, we’re really doing is comparing the line spacing. Some people call it pulse repetition rate, or I would just say the fundamental frequency of a periodic signal. That’s that line spacing.
If we choose a resolution bandwidth wider than the line spacing, then the measurement machine, the spectrum analyzer is going to integrate the power of both of those lines. And so if we make the resolution bandwidth still wider, we should expect to see a bigger measurement value. But if we have more than one line inside of the filter response, that means we can’t resolve individual lines. So we can’t figure out the spacing. So we see a smear, a bulge across the frequency spectrum and that’s people yell out and they say, “Oh, that’s broadband.” When you hear people talk about narrow band signals. What they’re saying is, the resolution bandwidth setting is narrower than the line spacing.
That means if we make the resolution bandwidth even narrower, we still only measure the power from one line, the measurement value doesn’t change. If we make it wider and we keep it narrower than the line spacing though, but we make it a little bit wider, a little bit wider, we don’t capture any more power. We’re still measuring the same power from that one line. So the measurement value doesn’t change. And if we make the resolution of bandwidth setting narrower than the line spacing, it allows us to resolve individual lines. So that’s what you hear people talking about, broadband, narrowband signals, these continuous signals.
One more signal that’s really important because I’m going to demonstrate that an example of a single real pulse, we’re going to talk about a theoretically and then a real pulse. An example of that would be a single electrostatic discharge. A theoretical perfect impulse, a mathematical function. It takes zero time to go from zero to full value. And then it takes zero time to go from full value back down to zero. That’s what it looks like in time on the frequency axis that needs an infinite smear from left to right of energy. So it’s constant value from negative light to positive light or from DC to light if we have a one-sided spectrum. Now, I can’t demonstrate a theoretical perfect impulse. It’s only theoretical. What can I demonstrate? Well, what I can demonstrate is a single real pulse.
I’ve clicked a couple of bullets ahead here just to jump to the bottom picture here. A single real pulse, and maybe it actually doesn’t have a zero edge time, like I’ve shown. Maybe it takes a little bit of time to get from a zero value up to a max value and a little bit of time to get from max value back down to zero value, but it takes forever to repeat. It never repeats it. Essentially, the next pulse comes forever ahead in time. So what does that mean? It means its frequency is really, really, really small, which means its line spacing is really, really small. So that means the line spacing of a pulse train. If we make the period infinitely long, the lines start to merge together and we get a continuous spectrum instead of aligned spectrum. That’s what a four year transform is.
And so we see this continuous spectrum over the right here. This is the energy spectrum of my single real pulse. Now what we’re going to do is, we’re going to take a look at a spark, a source of a sparky noise. It’s going to be a single pulse. I’m going to make a little tick, tick, tick sound. And that’s going to be a little spark coming out of my spark plug. And we’re going to look at a chunk of that spectrum in real time. That chunk is the real time FFT. That’s the real-time bandwidth guaranteed by the spectrum analyzer I’m going to use. So that’s what we’re going to take a picture of it and look at that next. And so here it is. I’ve got a black box with a red button.
This is something I really shouldn’t carry with me onboard an airplane when I travel because they don’t like to see this when I go through the security line, the black box with red button. I have to go pack this away and hide it. But when I use this for demonstrations, what do I do? I use this for demonstration to make a little spark noise. It’s something that would go, let’s say, inside of your gas grill or the stove in your kitchen to ignite the gas. When I push this red button, you’re going to see right here, a little spark jumping across the center electrode and this wire. And I’m going to let you know when the spark happens, cause I’m going to make a sound and you’re going to watch right here. Here we go. Ready. Tick, tick, tick, tick, tick. There it is. Tick, tick, tick, tick.
And you saw that spark. I hope right there. That’s the noise source I’m going to use, in the next hardware demonstration. I’m going to go and listen to that spark first with an FFT based spectrum analyzer measurement mode. And then we’re going to push the real-time button and we’re going to visualize that noise signal two different ways. So that’s where we’re going. You’re not going to actually see me push that button in the next slide. We’re going to try to watch this pulse on the spectrum analyzer. You’re going to be able to see that spark, just like I showed you here. Can spectrum analyzers see it? Let’s find out. We’re going to look at it, as I said, two different ways with the standard FFT and then real time.
So this is a video that I’m going to play. You can see the spectrum analyzer is set to spectrum mode. That means standard FFT mode. There’s an antenna plugged into the front of the spectrum analyzer. I’m looking with a center frequency of 100 megahertz, with a 40 megahertz span and I’m using 100 kilohertz resolution bandwidth filter. So very similar bandwidth to what you might use for measuring, radiated emissions. In the foreground here is the spark plug. I’m showing it to you. I can’t focus on both at the same time, but I want you to know that it’s happening right during this video. There’s nothing been changed, this is the actual video.
So I’m going to go press the button. And I’m going to make the tick, tick, tick sound. So you know that the event is happening in this video. I want you to watch the max hold line. Let’s see if we can see this pulse or not. Here we go. Tick, tick, tick, tick, tick, tick, tick, tick, tick. Didn’t see anything on the max whole night. There’s the antenna. This spark is just a few centimeters away from the antenna. Let’s make sure we didn’t miss anything. Let’s play it again. Let’s watch the yellow max whole line. Tick, tick, tick, tick, tick, tick. Didn’t see anything, a brand new FFT based spectrum analyzer with a spark right next to the antenna. It can’t hear it. That’s not at your try that at your place. You will find out the same thing. It completely missed that signal, even though it had a very regular period.
Tick, tick, tick, tick, tick, tick. I could see it. I could hear it. The spectrum analyzer couldn’t. Have you ever worked on hardware that has a noise source like that? Have you ever worked on, let’s say a medical laser, have you worked with machines that have a brush DC motors that are making little sparks all the time? This ought to be an easier signal to measure than something from a brushed motor. Those sparks are not maybe very regular. Occasionally they happen. Sometimes they don’t. There aren’t a lot of systems where you can actually see and hear the noise with your eyes and ears. And then also see it on the spectrum analyzer. This ought to be easy, but it’s not. It’s not at all easy for this machine to see, or any FFT machine to see in just standard FFT mode.
So now let’s look at this with real time FFT. The only thing that I’ve done is, I’ve pushed the real-time button, everything else is the same. Center frequency of about 100 megahertz, 40 megahertz span, 100 kilohertz resolution bandwidth, same antenna. All I did was move my finger and push the time button. And you’re going to see in the foreground again, not in good focus is going to be my little spark plug generator, and I’m going to make the sound so you know what’s happening. Now, what did we say was the spectrum of a single pulse? The spectrum of single pulse should be a continuous energy distribution. We should see a line across the screen. It’s not going to be flat. We’ve got a weird antenna with a weird frequency response, and we’ve got a pulse generator. That’s got some wires attached to it. That’s going to modify its amplitude versus frequency profile. But we’re going to look for a line across the screen when I make my tick tick sound. Here we go.
Tick, tick, tick, tick, tick, tick, tick, tick. Did you see it every time? Yeah, the same machine just in real time mode. All of a sudden we can see a signal that we couldn’t see before. It was invisible and we knew it was happening. And we were in max hold and we couldn’t see it. Now in real-time mode, let’s play it one more time. Tick, tick, tick, tick, tick, tick. Every time the spark happened, we saw the spectrum from that signal. That is the big change in measurement technology that we didn’t have before. We can now see signals that we couldn’t see before, but it gets better than that. But I just want you to remember, you actually saw the signal from that spark. We couldn’t see it in FFT mode. Then we went to real time and it was easy to see.
The next exciting part of real-time FFT spectral analysis is this thing called persistence mode. Persistence mode is going to add some color to the display. It’s a totally different way to view and measure and identify EMI sources. And when I say totally different, I mean that is one of the only big changes in EMC engineering. I’ve seen in 25 years. Let’s take a look at graphical description of what we’re going to see. And then to the right is a picture of what we’re going to see of the real signals that I’m going to produce for you here in just a minute. So what we’re saying is, with dedicated real-time measurement hardware, we can not only measure a signal once within the real time bandwidth, we can measure it many, many times. So if we can measure it many, many times, that means we can calculate some statistics about its amplitude, and about how often it reaches a particular amplitude.
So if you take a look here at the bottom left, you see that at each bucket or each location on the frequency axis, we’ve measured a single 100 times. So what will that look like on our display? If the signal has reached a moderate level here where my mouse pointer is right now, if it’s reached this level 70 out of 100 times, then it’s a pretty bright orange. It was frequently that value, but another signal right next to it, just a little bit lower in frequency reach that value very infrequently. Only 10 out of a total of 100 samples. So it’s the same height, but when it reaches that height, it’s a blue color like you see over on the right-hand picture. If you take a look at the very middle of the screen, you see 100 measurements and you see that it reached a very high amplitude, and it reached that amplitude 80 out of 100 times.
So now we’ve got additional information about the nature of a signal based on how often it occurs at a particular frequency. Just to summarize here, real time FFT and persistence mode, that persistence is the color and how long that that color stays on the screen. We’re going to make many measurements at each frequency. The trace color shows how often a signal occurs at a specific frequency and amplitude. There are different words that people use in the industry. So you should be aware of that. They say the color is the hit rate. That one set of words, people use how often a signal occurs at a particular frequency. The height is the usual information you’re used to before. How strong the amplitude of the signal. Other people use the words, histogram, the signal histogram versus frequency.
Other people will use the words percent occupancy. How often is the signal at a particular frequency and how often does reach a particular value? A totally new way to look at signals. Now, I want to show you again, the sources that I’m going to use. I showed you a picture of my little black box with a red button and the spark. Now I’m going to show you my other sources here. I have three noise sources here. I’ve got a USB mouse that has not perfect appealing on its cable so we can measure some noise current on his cable, with his current probe. So its cable runs through the current probe. I have a brush DC motor that’s being controlled by a pulse with modulation motor controller. That’s off the screen. So I have a pair of power wires running through the current probe.
And then last, I’ve got a little demonstration board with a 20 megahertz oscillator. So I’m just going to show you that, what it looks like, when this actually works. Click and you see the led come on. Now, I’m going to turn on the motor driver and you’re going to see this. Well, you’re not going to see the sound, I hear it. And you see that flag moving around, indicating the motor is moving. All right. And then, off-screen I also plug in this USB mouse so I can turn each one of these sources on independently. So we’re going to take a look now at multiple signals, multiple sources at a single frequency. We’re going to be able to resolve these individual sources at one frequency. We just didn’t have the technology to do that in the past. So we’re going to demonstrate real-time FFT persistence mode.
Some people use the words signal under signal, one signal underneath another. We’re going to be able to see a weak signal underneath a strong signal. So just to be clear about what we’re going to look at, multiple signals occupying the same frequency, but they’ll have different amplitudes. And what’s going to allow us to visually distinguish one from another is that they’re going to occupy that frequency a different percentage of time. That’s the 10 out of 100 samples or the 90 out of 100 samples that I showed you in the previous slide. The three sources, again, that I’m using a DC motor with the PWM controller, we’re going to call that broadband noise. The USB mouse is actually going to produce some broadband and narrow band signals. And then I’ve got a 20 megahertz digital clock. That’s going to be a narrow band signal.
So those are the sources we’re going to look at next. I’m showing you the front of the spectrum analyzer. You’re going to see me manipulate things rather than showing you just a video file, the screen. I wanted to show you the actual physical demonstration. You can get a feeling for what’s going on. So we’re displaying a 40 megahertz of bandwidth. The center frequency is about 37.5 megahertz. The 40 megahertz is chosen because that’s the real time measurement bandwidth of this machine. So I’m starting off. What I’m going to do is start off measuring in FFT mode. We’re going to see what we can see when we turn on all the noise sources, and then we’re going to look at it in real time mode and see if there’s a difference, to see if this might be helpful for you.
In the screen, we’re going to start off by looking at one. I’m going to turn on one signal at a time when I play this video. I’m going to turn on the mouse and then the motor, and then the oscillator. Let’s do that. I’m turning it on. I’ve just plugged in the mouse and you can see one or lines pop up. Next, I’m going to turn on a motor and the motor is making a [inaudible 00:32:45], and I just turned on the oscillator signal just turned on right there. So we saw a little bit of elevation of the noise floor when the motor turned on. We definitely saw my 20 megahertz source turned on. We have a 20 megahertz line and a 40 megahertz line. Well, it’s not really clear where these things are and how I might localize them. Maybe with a near-field probe. It’s still not a very good picture to try to judge stronger, stronger, or weaker weaker if I’m moving cables or moving a near-field probe around.
All right. If I was faced with this in a real situation, I didn’t have real time. What would I do? I’d maybe turn on max hold. Let’s turn on max hold. Let’s see if it’s easier to visualize with max hold. That’s not super easy to try to track down if these things are blinking on and off. Let’s go to real-time and see what things look like here. Oh, that’s what it looks like. What looked like a jumble of blinky lines before is now really, really clear. Here’s the motor noise from the brush motor and the PWM signaling. This group of four or five lines here that are wide, this is what I would call a broadband noise from the USB mouse.
Do you see that they’re weaker than the broadband noise from the motor? They’re underneath the broadband noise from the motor, but we’re able to simultaneously visualize a strong broadband signal and a weak broadband signal at the same time. And we had no idea of this amplitude versus frequency, distribution, in the FFT, the plane FFT view. And look at this, that mouse signals jump out, the broadband singles jump out, the narrowband harmonics from the mouse jump out underneath the motor. So just think if you’re troubleshooting a complicated system and you were moving one cable or turning on and off one circuit board, you could instantly know which source was being affected by turning a part of the system on and off, or by moving a cable, or by adding a little bit of your finger capacitance to a particular part of a PC board. This stuff just leaps right out.
And the additional 20 megahertz lines just leap right out. So the next thing I’m going to do is to turn these sources off one at a time. So I’m going to turn them off one at a time. Now I’m going to turn off the oscillator and I just turned off the motor. And now I just turned off the mouse and everything went away. Let’s turn it back on one at a time. The motors turned on now, you see the broadband noise. Now I’m going to turn on a mouse clear as day underneath that broadband. And now I just turned on the oscillator. That visualization is completely different than what we’re used to seeing. I can see a broadband source, a weaker broadband source, and a narrowband source underneath, the larger broadband sources. This makes troubleshooting so much easier.
In fact, we have a new way to recognize signals that we didn’t have before. We couldn’t see this signature before in standard FFT mode until we went to real time, and then they just leap off the screen. Let’s just play the first part one more time here. This is in a standard FFT mode. We saw the mouse get plugged in. We saw the motor get plugged in and now the 20 megahertz oscillator is going to turn on. There is a 20 megawatts oscillator, and the second line is turned on. We’ll put it in a max hole because that’s what we’re used to doing to find challenging sources. We try to see how big they are and try to get a feeling for what they look like. And then as soon as we go to real-time, they just completely leave out at us.
So you know that whenever I do EMI troubleshooting, when I’m measuring emissions, I always push the real-time button now. So just to summarize what you’ve seen, the presentation on the old technology and the new technology, previous generation spectrum analyzers and test receivers, they use swept tuned or stepped tuned architecture. Now it’s true today that if you go off and buy a new spectrum analyzer it pretty much guaranteed that you’re using an FFT based machine, but it won’t be real-time unless it says so. And then you have to try to figure out what its real time bandwidth is either by looking at the maximum of span that’s set or by asking the manufacturer.
They’re okay for measuring continuous, not intermittent periodic signals and narrow spans. So the signal is always there, they’re not particularly challenging to measure. They’re okay to use when you always know when in time and where the signal is, but I actually showed you a case where we knew when the signal was present. We turned a spark generator on. Tick, tick, tick, tick, tick, and we still couldn’t visualize it with even an FFT machine, but they can’t reliably display pulse spectrums.
They can’t visually distinguish between two or more broadband signals in a single piece of spectrum. And they can’t display narrowband DMI signals under a larger broadband signal. In the past, we’d have to throw away the broadband signal by reducing resolution bandwidth to have a chance of finding a weaker narrowband signal. And that would assume that we knew the narrowband signal was there. Because it was covered up by a broadband signal. We don’t have to worry about that anymore with a real-time. And they’re blind during large parts of the sweep or step into well time. They’re challenging to use for EMI bug except for trivial signals like narrowband clocks that have 100% occupancy. Summarizing real time, 100% signal intercept within that real-time bandwidth. If it’s there, we can see it. FFT machines without realtime are going to miss signals. I demonstrated that to you.
Persistence mode gives us a new, powerful way in real time to visualize signals that were previously tough or just impossible to display. This is the big change in technology, that’s silent. We’re really having a lot of fun with. Non periodic blinky broadband signals. I’ve given you some examples, some real examples of real sources to show you how incredible this new visualization is. And of course, we get that ability to clearly display signal under signal situations, which required us in the past to what we say, stand on our heads to manipulate the settings of the machine, to have a chance of finding them even if we could find them. Even if we knew that the signal was there, we be challenged to find them.
The visualization, it’s really tough to describe in words, even with static pictures, I think the pictures that I’m showing up at the top of the screen, it’s not exactly obvious what that means until you see the real measurement in real time to actually see that sources turn on and off and see that, persistence display live. So that is the end of my presentation. I want to thank you very much for attending. I think we’re going to try to bring in some questions now.
Kenneth Wyatt:
Thanks so much, Lee. We’re running a little short on time, but I think we have time for one question. And, from one listener is asking, is the use of real-time FFT EMI receivers mandated in the regulatory standards because many EMC test labs still use a swept tuned EMI receiver for formal compliance. Can you share your views on that?
Lee Hill:
First off is I’m not a regulatory expert, but I would say that’s maybe not the right question. Time is not mandated. Principally it’s use right now is for troubleshooting. It’s not a technology right now that is used for final qualification measurements.
Kenneth Wyatt:
Very good. So, attendees, this is your last opportunity to send in your questions. Please enter them now in the chat field, on the right side of your screen. We probably will not have time to answer any more. However, if your questions were not answered, don’t worry, we’ll capture all those questions. Send them over to Lee and post all the questions and answers on the EMC Live 2016 a website within the next week or two. You’ll need to sign up for the interference technology newsletter, and that’s how you’ll be notified when the questions and answers will be ready. If you did enjoy this presentation, you might also wish to download our new realtime analyzer mini-guide available free on the interference technology homepage. We’re also looking for your vote for product of the year. And you have until this Thursday, 12:00 noon Eastern time to join over 800 others to cast your vote.
See if you can pick this year’s winner. And tune in to the Ken and Jerry show this Thursday at 4:00 PM Eastern time where the winners will be announced. I’d like to thank Lee for his time and expertise today. I’d also like to thank our platinum, gold and silver sponsors Rohde & Schwarz, Amplifier Research and Keysight Technologies respectively for their support in making this event possible. This was a very useful and very informative presentation. Thank you all for your attendance and thank you all our attendees around the globe, as well as our technical support staff in Pennsylvania. I’m Kenneth Wyatt from Interference Technology. Join us just 15 minutes from now, if you’d like to get a good understanding of pre-scan analysis tools, which is coming up next. This promises to be very interesting. You still got time to register for this next event in the next 15 minutes. We’ll see you again very soon. Thank you
