 for today is going to be about PCB design for RF circuits. Our next presenter is going to be Chris Gamal. Please welcome him with a big round of applause. All right. Thanks for coming today. So let's talk about some goals to start with. I'm just going to talk about how I got into the world of working with RF. I'm going to try not to bury you with math. I know that it's a hot summer day. The last thing you want to do is look at math equations. And then finally, the main thing is to get you some resources as well so you can do what I did and make the mistakes I've made, hopefully a couple fewer. So let's talk about you. Who here has electronics experience? OK, that's a good crowd. How about RF experience? Couple fewer, a couple maybes. Any ham radio operators out there? OK. And then the main thing is you might have some more experience than I do. That's the main thing I was getting at here is that I want to talk about the struggles that I've had as I've moved from DC to RF. That was the genesis for this talk. And show you what I've been doing and show you some of the mistakes I made. And hopefully you make fewer mistakes than I did. So first off, who am I? My name is Chris. I've been doing electronics design for 15 years or so. And I've worked in a bunch of different industries, semiconductor, and industrial, and stuff like that. And I really like the industrial side of things you get to really dig into problems. And you don't have the sixth month cycle that consumer does, although there's a lot of fun problems there too. I also host a show called The Amp Hour, electronics podcast. Been doing that for about nine years now. And I teach people electronics. Obviously teach people the DC side of electronics. I'm not teaching them RF because I'm learning RF. And that's called contextual electronics. So that's an online course where I teach people how to do some of this stuff. And my past, though, I started by working on this. Anyone know what this is? Raise a hand if you know what this is. It's an electronometer. So this is used by material scientists. Basically, you push 1,000 volts into a material. And then you see how much leakage current is coming through. And it can do down to femtoamps. So a really cool thing. But the thing is it's really slow. Like, it's really slow. 10 PLC power line cycles means you're getting about six readings per second. And inside, you're basically integrating everything. And by integrating, I mean you're integrating your measurements over and over again. And things are just going really slow. It's truly a DC measurement equipment. So not quite blazing. Of course, there are microcontrollers in there. There are some higher-speed signal things. But my focus, when I get started in electronics, this is my first real electronics job, was doing this kind of thing. It's so learning a lot of the DC measurement techniques. Over time, it's also morphed into working with sensors and other power supplies. Obviously, everything is a switching regulator these days. But we're not talking about blazing speeds here. We're definitely not talking about RF level frequencies. So what really changed here? And what changed is the world around us. And you've probably all noticed this, too, right? We're all wearing wristbands right now that have Bluetooth on it. It's everywhere. Every product basically throws in a Bluetooth chipset for effectively free, or it has some kind of connectivity method because that's a feature. And that means that every product is going to have RF in it, RFs and everything. Of course, RF is actually everywhere as well, right? We're always surrounded by signals, but every product has it as well, and it's super cheap these days. And so this is a chip I've been talking about lately. This is like a little 400 to 33 megahertz transmitter. And it's 15 cents if you buy 100 of them. I mean, these are levels of cheapness that are like, it's just you could throw it in anything. And so this was starting to motivate me. And I assume that that's why you're sitting here as well, or watching online. You're a little bit motivated as well by some of this stuff, and you want to know what to do with it. So you're interested in making things. You're interested in this cheap, wonderful hardware. And you're wondering how to actually integrate it. So let me tell you about my story about how I started with my first RF design. So the first thing I did was the most obvious thing, which was to copy the app note. And if you're talking to electronics designers, that's a very common thing. When I started my career, I was always kind of worried about it, like, oh, I'm just copying. I'm not doing anything original. But so much of electronics these days is just re-implementing what's there, because you're in a rush, or you're using what's working. And even if you put everything in exactly the same, there's still going to be problems. And I always say that electronics designers, they make their payday by making things work, even when they're supposed to be working. So I always guess about 80% or so of a circuit design might be, it's a reference design. You're copying a switching regulator. You're using a microcontroller, and you're copying the stuff there as well. And you're just trying to get it to work. And so that's the first thing I did. And there's a great talk as well by Mike Osman, who is here at this conference. He talks about this as well. This is a talk he gave back in 2015 with some rules about how to basically copy RF circuit and go through this. So I kind of think about my talk as an extension of that talk. Some of his rules were using four layers, using integrated components, trying to get everything as 50 ohms, following the manufacturer recommendations that's effectively copying the example circuits, and then routing RF first. I think these are all really good rules here. And I did a lot of these. And so let's talk about my first design. So my first design was a cellular module I was working on for a previous employer and had a little microcontroller, had a SIM card on it, and it worked. So not much of a story there, it just kind of worked. But why did it work? That's kind of the main thing we're trying to get at here. Why did this just work out of the box and with simplicity that I was applying to it? I applied those five rules that Mike was talking about, but the main things to know are the module on board, it was pre-matched with the antenna as well as plug-in. Effectively, the RF section that I was using was connecting the output of the cellular module, modem, to a UFL antenna connector. And then I was using an antenna that was already precaracterized as well. So I had a very, very short distance where I was connecting the output to a port where I was going to plug an antenna into. I basically minimized my interaction, I kept my RF trace short, and I didn't really worry about it past that because this was just a test board, but it worked enough for me to be excited about it. Kind of keep trying and you need those dopamine squirts to kind of keep going. So can we do this all the time, right? Can we always just keep keeping traces really short? Can we just copy the datasheet? Can we follow those five rules? And the answer is, yeah, maybe. I mean, you might be able to. You might be able to just integrate components like that, but it really depends on your design goals. It depends on what you're trying to get out of it. It depends on what you're trying to do with your end product. When you're starting to move into a more uncharacterized zone, say you're making a PCB antenna, you want to make your own PCB antenna where the antenna is built into the board, that's going to be a little bit harder. You're going to need to do a little bit more characterization and testing. If you're going from a differential to a single-ended, you might have to do some more stuff there and do a bunch more math and use more components, but it's really the complexity of your circuit that's going to have the most impact on your success in that realm if you're just trying to do that stuff. Let's talk about, though, why I wanted to learn more about this in the first place. The first one is I'm a consultant. I do designs for people. And the thing that kept me up at night is I was going to just copy the datasheet, do this thing, but then I'm going to get it to the bench and it doesn't work. And there's nothing scarier than the thought of being there and just trying to hold your finger here and you hope that it works. And you open the window and hope that changes something. And you don't know what you're doing. You're basically cargo cult kind of trying to impact your environment without actually having any control over it. And that was a big fear for me. I didn't want to be in that situation. That's doubly true when I'm sitting at a compliance test lab and they're charging $10,000 a day. And I still don't know what I'm doing. So I wanted to basically be able to understand the underlying because that could really impact the ability to get a product out into the world. And then finally is kind of a feel good one as well. I want to make sure, obviously, FCC and CE compliance, these are things that they're testing the same thing that you're not stomping on someone else's bandwidth or you're not operating out of range. But that's also something I don't want to mess up someone else's day because I've got a poorly designed circuit. So a lot of people said that they had RF experience in the crowd, which is great, but I'm going to assume not everybody does. So we're going to go through some of the RF concepts that are kind of tricky and stuff that still kind of trips me up a little bit as well. First off is the frequency domain. So this is a lot of the things we talk about when we're talking about RF and higher frequency things. We start moving from a time domain to a frequency domain. And this is a topic that I think there's a lot of math behind it, but I promise you not to get too much into it. So some of it you're going to have to take on faith. I'm going to show you this is a scope trace. So the top one is a square wave. You might have this as a clock or a similar on-off signal, say you're blinking an LED on and off. What you have there though is you have different frequency components that actually exist within that. And so the bottom part is the actual frequency component of that square wave. Now that doesn't really mean much, and it's really tough when you look at them side by side. They look like they're on the same axis, but they're actually not. So take that a little bit on faith. But I also found a great animation, and we'll hope that this plays. It looks like it's going to. I love this because it's basically showing you the components. Is it playing? Yeah, it's playing. So it's showing you the components that actually exist and make up. They basically sum up and make a square wave. So this is a simplified version. They make up the square wave, and they exist in these different amounts. So the largest sine wave is the main frequency that exists within it, but then there's all these smaller frequencies as well. And you see how that kind of translated to be between the time domain and the frequency domain. We could basically, if we have this information, we can then go and perfectly replicate that same square wave by knowing the frequency information. So this is kind of a tough concept to get your head around, but it is one that will be very, very useful. And I'll talk a little bit more about an instrument that kind of helps you to visualize this as well. I think that went through a full cycle. Another tricky topic is just kind of components. We're trying to squeeze a lot of electronics into a 45-minute talk here. But capacitors, basically, when you're at DC, they're blocking everything. There's no signal that's going to be going through. As you go up in frequency, the signal is going to basically be able to pass through the capacitor. Again, it's not physically passing through, but it's effectively passing through. And this is a really tough concept to get. The main thing to know is that the transition between blocking and not blocking happens because of the value of the capacitor itself, the capacitance of that component. In the inverse way, the inductor allows frequencies to pass through at the low frequency, or DC, and it blocks it at a higher frequency as you move up into the RF world. And so these two things kind of play off on another, and they interact. Again, this is a lot of electronics theory trying to squeeze into a small talk, but hopefully you'll take some of it on faith. Another thing is that in RF, a lot of times we're working in log scales. So you'll see a lot of decibels, or dB, and dBm, which is a more, dB is ratio metric, and dBm is more an actual value. But the idea is here that we're operating in orders of magnitude. And that's really important. You'll see an example a little bit later here. So you'll see a lot of things referred to in dB. And that can, again, be kind of hard to wrap your head around, but it's an important concept because the scale that we're working with of signals, when you transmit a signal and when you receive a signal, it's very, very different in terms of the magnitudes that you have there. As an added bonus, though, you basically, when you're going through a system, you have different gains going through. If you have a 10x gain in something, you have a 20 dB addition to it. So if you're starting at a zero power signal and you go through a 10x stage, you have now 20 dB. You put it through another 10x stage. Now it's 20 dB plus 20 dB. Now you have a 40 dB signal at the output. And so that's kind of nice because it does simplify some of the math once you get past the confusion thing. Wrong way. Where'd it go? And that's the last thing about this is the power is how we're often going to be talking about RF circuits. It's not just current. It's not just voltages. Those obviously make up power together. But we're going to be usually talking about power. And so that's an important thing to know as well. When we're looking at RF subsystems, what we're usually talking about is not degrading the signal. So if we have gain and that's something that we want to maintain over time, we want to make sure that the signal is not being interfered with by the things that we introduced to it. So if we have a connector that has minus 3 dB of gain, that means that we're making the signal worse on the other side of that connector. And that's bad. We want to make sure we're not doing that. Similarly, if we have an output and an input that are not well matched together, and we'll get to that here in a second, that also is going to degrade some of the signal quality as well. So let's talk about what that signal path looks like here. So on the left side here, we have a TX. We have a radio that's transmitting, that's TX. Going through a cable, going through an antenna. And yes, there is a little bit of gain in the antenna as well. But you see in the middle here, there's that red sloping line. That's basically as it's floating through the air and going from a CCC camp cell tower to your crazy camp phone. There's actually some loss in that. And then, again, it hits the antenna on your phone. There's some gain there. And then there's a cable, and then it's at the receiver. All of these things basically say that you're going to be either adding or removing signal by going through these different stages. And the main thing is we want to make sure that we're not adding any degradation or noise as we're going through our systems. Just back to the logarithmic side of things. So again, to give you the scale difference here. So if you have a link budget that's 100 dB, and that means you have a signal. So what you do is you divide that by 20. There's a five there. You take that 10 to the fifth. I know that's math again. But basically, we have a signal that's output it. If we have one watt signal output at a transmitter, that means we would be receiving 1,100,000th of that at the phone in this example that I gave. And so that's kind of why we want to not introduce any other degradation in that, because it's going to make it less likely that we receive that signal at our phone. How do we get through to this? So again, how do we make sure we don't degrade unnecessarily? Well, what we do is we make sure we're matching the impedance of these inputs and outputs so that we don't introduce signal degradation. Now, this is a new concept when I was moving from DC to RF. Now, obviously, it exists at lower frequencies as well. But we want to basically, it's much more critical, I think, at RF, because you are playing with much different impedances as well. Now, one of the baseline things is that you need to have perfectly matched impedances input and output, right? Output might be an amplifier. Your input might be an antenna. You want to make sure those things are as perfectly matched as you can so you're not introducing unnecessary degradation. So let's talk about a counter example. This would be a bad situation that you'd have here. So you have an ESP32. It's a little tiny Wi-Fi module. And then you have a PCB antenna on board. And those two things are not well matched together. When they're not well matched together, what's going to happen? Well, first off, it might not work at all. Or you may perceive it not to work at all. Because you have a Wi-Fi router over here. You have your ESP32 over here. And it's not actually talking to your Wi-Fi router because there's no signal getting from one to the other. So without actually having measurement equipment, you won't actually know what's happening. But to your perception, it's not going to work. Another thing, though, is it may work, but it may not work as far as you want it to, right? So it only works if you're within a foot or within a meter. Sorry. And you expect it to work at 10 meters instead of 1 meter. And then finally, you're going to have a less efficient system as well, right? You're going to have energy that gets either radiated locally on your board and not radiated towards your Wi-Fi router. Or it's just going to be dissipated as heat. So what is impedance, though, right? So I know I'm jumping around topics a lot here. But what we're trying to do is we're trying to get an output like an amplifier and an input like an antenna. These two things to be kind of talking to one another. And we want them to do that by matching their impedance. What is impedance, though? Impedance is basically how it lets, how passes or blocks electromagnetic energy at various frequencies, right? So if we have a 2.4 gigahertz Wi-Fi thing like the ESP, we want to know that our antenna is accepting that energy at that frequency the same as our amplifier is outputting that energy at 2.4 gigahertz. You may be thinking, well, why do we have to do this at all, though? Shouldn't an antenna just work? Shouldn't all this stuff just work? And the answer is, yeah, it should just work, but that's not how the world works. The world's imperfect, right? You may remember the iPhone. iPhone 4, it was a well-designed phone. I think it was the 4. But it was a well-designed phone. And when you held it in your hand, though, your hand basically interfered with the antenna. It was changing the impedance of that antenna. And so the matching was all messed up. And you didn't get the same kind of distance to the tower as you thought you would. This is a similar kind of problem, obviously on a much larger scale. So it's not just a hand. It could also be, is it a more humid day? Is there other stuff in your environment? Is there a tree blocking it? Is there other things that are you in a city? Are you in the woods? All of these things can basically impact your signal path and also, on a much closer scale, how your antenna actually works. So even though you may have an antenna that comes out of the box that says, yes, this is definitely 50 ohms, you are good to go, there's manufacturing variability as well. And you basically want to be able to not only measure what's happening, but also then go and modify it so that they're well matched up with one another. So as any simple example, a pine network is one way to normalize your output and your input. In this case, a 1,000 ohm source, which is on the left here. And then you have your little networks of inductors and capacitors. And then your load, in this case, would be an antenna at 100 ohms. You want to normalize these things at 50 megahertz and a bandwidth of 6 megahertz. Basically, there's a bunch of calculations you could do. Again, we're skipping math. But a bunch of calculations you could do, and you can find values that allow you to match up your signal source, in this case, an amplifier, and your antenna. And they work better together. It's going to have the maximum output that you can have here. Let's talk about some of the measurement tools that would get to this point and understand this in the first place. I had mentioned this a little bit, but a spectrum analyzer. This was, again, that animation showing going from a time domain to a frequency domain. So a spectrum analyzer just basically measures the frequency domain. So you put in a signal into your spectrum analyzer. You're going to see a chart of the frequencies. And this is really, really useful as you start to look at something like Bluetooth, where it's hopping over all these different frequencies and stuff like that. But really, anything that's outputting its own signal, a phone is going to generate its own signal again, a Wi-Fi chip is going to generate its own signal, you then could take that signal and then look at what it's actually outputting in the spectrum and look at it on a spectrum analyzer. You could also do that what we were looking at earlier. It was a slightly different thing. That was an oscilloscope. So that's showing the time domain and then it has an FFT, which is a measure of the frequency. And it's basically a very tiny software version of what we're looking at here. Another thing we can get is a VNA. So Vector Network Analyzer is something now where you're not just sort of a phone might be outputting a signal on its own, right? It's got a battery, it's pumping out RF energy. But if you don't have a thing that's actually actively outputting energy and you want to measure it, you want to measure the characterization of an antenna by itself, you might want something like a VNA. And this is where these things get really, really useful. So I've pulled in some of these low-cost VNAs here and low-cost is very relative. On the top left there is the mini VNA tiny, it's 400 euro. Top right is pocket VNA tiny, 400 euro. And these are some of the bottom of the line, not the lowest, but pretty low. Then the D-PACE is the bottom left. That's a 2K, HP 8753A, starts at 2K and kind of goes up depending on what you have in there. And you're like, wow, that's really expensive. But so I've been shopping around for all these things. I've been on eBay and I've been looking at all these different things. And I was really shocked to see a VNA that was about 550 euros. I saw it and I was so excited about it. And I was like, wow, this thing looks beautiful. Definitely got a bit on it. And then I realized that I actually was looking at the wrong thing. This was $550,000. So if I could borrow some money from people in the audience, we could definitely go in together on this thing. But the reason I'm putting this up on here is to show kind of the scale difference, right? You can go from a 400 euro dollar thing all the way up to a $560,000 thing. And the thing that you're paying for between the two is the one that the 400 euro thing is gonna work more in the 200 megahertz to one gigahertz range. This is showing at a 67 gigahertz range. It's super, super complicated. All of the things that happen in between, all of the losses that I've already talked about, basically everything has to be perfect in this machine. And I think it's a good example of showing kind of the scale of what has to happen. And what you're really paying for when you buy a VNA is the frequency range, right? It'll scale pretty linearly. As you move up in frequency range, your price is gonna move up as well. So personally, I'm targeting things a little bit lower. I wanna do things that are commercially available, 900 megahertz, 2.4 gigahertz, maybe five gigahertz, but those are the ranges that I wanna play with and maybe you wanna play with as well. Because I wanna play with all the radios that are out there that are cheap and abundant and kind of self-contained. The last thing you might need is Calcut. And what this does is basically it gives you an open, a short, and a known load. And basically you can calibrate the equipment that you have that I showed, right? The VNA, you can go and calibrate that thing. And that's important not only because it gets warm, you're in a warm tent or you're out in the sun or something like that and you're using your VNA, you wanna make sure it's well calibrated at that point. But you also might have cables in between. You might have a test setup where you have a cable from point A to point B and you wanna make sure that that cable is in part of your measurement. And that's something that a Calcut can help with as well. So let's talk about some measurements real quick. Again, I know we're on a summer day, we're talking about measurements, how fun. But I think these are really important. Smith charts are basically, it's basically a collection of measurements in a visualization tool. And so this says things like impedance and reluctance on it. And I'm gonna be honest here, I'm not using Smith charts all that much, but I didn't wanna show it because the VNAs kind of allow you to have these outputs, right? You can go and match things a little bit easier. A lot of the, when I showed the C's and the inductors and capacitors that had different values, you can go and find those by using something like a Smith chart and that's really a powerful thing. VNAs also show you things like S parameters though. So return loss, insertion loss, power transfer and reflected power. These are all things that basically you can go and characterize your systems with. So your antenna, your output, maybe you have a gain stage. Basically you can use a one or two port network analyzer to go and measure these different parameters and characterize your system. And that's really useful for when things don't work. And again, that was one of my big things. Return loss is a very common one. So this is also known as a reflection coefficient. And basically this is something you might do when you might plug in an antenna here, right? So this is showing an antenna, right? And it's showing that it's reflecting power at most frequencies and it's a little unintuitive from my perspective. When it's flat at the top, it's saying that's where it's reflecting most frequencies. But at 2.4 gigahertz, it's actually, there's a bunch of power that's not being reflected back. And it's basically showing that it's tuned for this frequency. This is a way to show that this is a well-tuned antenna and how narrow or wide that dip is, as I was going to show how selective the frequencies are. This also plays into VSWR, which is the voltage standing wave ratio. This is what a lot of ham radio people use as well when they're characterizing antennas and you can use a little bit more broadly. But here's an antenna, or sorry, here's a equation here and you see the gamma that's in that equation, that's that funky looking R thing in the equation at the bottom. That is basically the last thing we were looking at here. Sorry, sorry. That is the return loss and that then plays into VSWR and it's basically helping you figure out how well matched it is. So basically as you get a lower and lower VSWR, you have a better antenna. Okay, so that was kind of the measurement stuff. We talked about some of the basics of RF and now I wanted to kind of transition into like the title of the talk. We needed that build up, all that kind of background information here. But the reason I started thinking about all this in this first place is why does DC end up kind of change as you move towards RF? What are some of these assumptions that I've always had as I was doing specifically layouts and new designs and thinking about how I'm gonna put my different components together? How does it all play together? The first one is that a wire is just a wire. So if I hook up a component over here at A and I put a piece of 24 AWG wire over to B, that the signal's just gonna get there. And that's obviously not true at DC either. These are all very simplified versions, but it's definitely not true as you move up in the RF region, right? Because every wire is actually an inductor, especially as you move up into higher frequencies, right? Inductance comes from the magnetic field that are happening as you pass current through a wire, you start to form these magnetic fields here. And they resist, it's basically the resistance of the magnetic field as you're moving current through it. So as you try and turn off the current, the magnetic field is now gonna push back and start and try and it's gonna spike the voltage effectively. And you may recognize the thing on the right here, if you've ever wrapped a piece of wire around a screwdriver as a kid and you attach it to a battery, you're gonna make an electromagnet. That's the same kind of idea here. If you look at a lot of inductors, what they look like is coiled wires around some kind of substrate as well. And so the main thing to know is that, yeah, a wire is not just a wire. At DC, it doesn't matter because usually the inductance is low enough that it's not a huge deal. And there's a pretty practical example here that many people, especially getting started in electronics, will understand. As you're using a breadboard, breadboards usually operate fine. Even up to 10 megahertz, they're operating fine. If you're blinking something at 10 megahertz, you're probably not seeing it. But say you're operating a clock at a 18 mega, 328. That works fine at eight megahertz. But if you're trying to pump it many, many more megahertz than that, you're gonna start to have degradation issues. And that's because of the inductance as you're starting to wire from one point to another. Basically, not just the wires that they show here, but then the strips of metal that are underneath the breadboard as well. And that brings us to our next thing, which is in DC, of course I never actually thought this when I was designing stuff in DC, but that PCB construction, the stuff that makes a printed circuit board, the fiberglass and the copper and stuff in there, aren't as important as the components that I put on there. So the resistance of a trace, or the inductance of a trace, isn't as important as the resistor that I put on board there. And that is definitely not true as we move up in frequency. So this is a four layer stack up here. This is showing how you, and it's kind of a side view, so it's kind of hard to see here, but you basically have four layers at the top. You're gonna run all your traces and you have your components actually solder down on top. If you haven't been to the hardware hacking village, you can go and see some of the surface mount stuff there. I recommend that. But you have traces on the top side. Inside, you might have a, you should have a ground layer, which is usually the second layer. Then you have signal and power on the third layer. And then you can have more traces on the bottom and also some more ground. So these are kind of the normal stack ups here, but the stack up, the actual dimensions of that stack up, how far one thing is from another, that's when it really starts to play together much more at high frequencies than it does at low frequencies. Again, these characteristics all exist at low frequencies, but you don't have to worry about them as much. Or you care about them in different ways, I suppose. So as you start moving up in frequencies, you start to have some more complexities that kind of get introduced here. So this is pretty much the only math that we have to do here, but this is, and it's in a calculator too, so from online, and the links are all here, but say you have a trace that's four mils wide or 0.1 millimeter. The track is about 50 millimeters long or about two inches, and your height above ground, right? So what we're saying is from the top, top layer, layer one to layer two, I was looking up just a standard stack of a set that's about 0.2 millimeters. And so basically you put all these things in there. That means that track has an impedance, an inductance rather of 63 nanohenrys. And if we take that at 2.4 gigahertz, that's effectively a one kilo ohm resistor, effectively. Again, this is operating at these different frequencies, but the idea here is that that actually is a more significant amount of impedance that you would have normally thought from just having a wire. If a wire was just a wire, that wouldn't be a problem, right? You assume that it's zero ohm in the ideal case. It is definitely not the ideal case though. So first off is very important to note that yes, your ground plane should always be the lower, the right below where you're running your RF signals. That's very, very important. You need to define that as well, and you need to pay attention when you're buying a PCV as well. So this is a stack up from JLC. And this is showing that 0.2 millimeters between the top layer, the top layer copper, and the next layer down. And that's important to know that. Not only that, you also need to know the, I think the calculator I was using had actually assumed the dielectric, but basically you're setting up these two physical things in the real world, and you need to understand how those things play with one another. Again, you needed to care about that at DC, but you really need to care about it as you're moving up in frequency for RF circuits, and not even just RF circuits, as you're doing high speed digital, you're also gonna have to worry about it there with controlled impedance and things like that. Okay, another assumption, not, again, a very simplistic assumption for DC is that a capacitor is there to store charge. And yes, that is true, but the idea here is that a capacitor isn't just a capacitor, right? And so these are smorgasbord of different capacitors up on the top left. The top left is through-hole capacitors, and the top right is surface mount capacitors. The top left, when you have through-hole capacitors, they actually have wires literally attached to them. That is what those leads are. And again, we've already talked about this. Those are inductors, effectively. The same thing happens though at low-frequency. You can never really get rid of that inductor, and what we have below there is the equivalent series inductance, the equivalent series resistance, parallel resistance, and then the actual capacitance itself. This is kind of a bulked model of what a capacitor is. And what happens is, as you start to move up in frequency, you start to see effects that you wouldn't have seen at lower frequency. So at lower frequencies, it just acts like a capacitor, it charges up slowly as you move towards DC. At true DC, a capacitor is just a block in the road, right? In more practical terms, below 10 megahertz, it's just gonna charge up at a different rate, depending on its capacitance. Well, as you move up in higher and higher frequencies, that inductance starts to really matter. And what happens here is that the impedance, right, which is a complex mix of inductance and capacitance, that starts to invert as you move past this corner frequency. And so your capacitor, which you thought was a capacitor, that blocking thing is now turning into an inductor, and that starts to really mess up some of your circuits. Your filter might change, your signal path might change, and you really just need to keep that in mind. Another, this is kind of what inspired a lot of thinking for this talk as well. I was looking at layout for how I would do layout for a really sensitive amplifier. Maybe we want to make sure that it's nice and you have ground planes around it, that's fine, right? That's a lot of RF as well. But we would actually cut out around the different circuit elements, so you wouldn't have currents kind of creeping in. You wouldn't have these different things creeping into the circuit. And I say that with kind of like scare tags, because I mean, yes, that is how it works at DC, but that's not really how it works as you move up in frequency again. So some of these are noise reduction techniques that you might use in analog. You might try and isolate some of your parts of your circuit using physical means, but RF don't care. That's kind of the summary there. So here's an app note from Hank Zumblin, one of my favorite applications engineers from analog devices. And this is what he's showing here, basically you cut out, you see that L there, you're cutting around the precision analog section. And that is less effective when you have high frequencies there. So you need to basically at low frequencies, the cutout means that the signals and the noise that you might be generating with a switching supply won't get around the cut, but it basically, as you start to move up in frequency, that same thing has other effects. And so that's like, at high speed, it will treat this as a capacitor and quote unquote travel through to the other side, but what it's really doing is it's radiating. And what I'm showing here is an antenna that's actually made by cutting a slot in a plane, which is the same thing I was doing. So you could start to, as you cut slots on planes to isolate things, you could also be creating radiation points, you might be failing EMC, you might be pushing some energy into the RF spectrum that you didn't think you were, and not necessarily the best idea. Another assumption here, current follows the path of least resistance. This is kind of a well-known thing, I suppose, but it really is, is it cares about impedance, it doesn't care about resistance. And so what this is showing that, if you have a meandering path here, it's going on this squirrely trace and it's going through a resistor, it might, it would go, the current would flow back to the source directly, but what actually happens, so at DC what happens is that this is true, this is the path of least resistance, it might be flowing through a ground plane, it might be flowing through a VCC, something similar like that, but as you start to move up in speeds, it starts to travel underneath the signal, underneath the trace itself, and that starts to have other impacts as well. So if you started to go through a region of your board that you weren't supposed to be going through, that could have some impacts. And the main thing to know is that at higher speeds, it takes on complex terms, so the inductance really starts to matter in this case. And so as the inductance matters more, where the signal is going to travel is going to change as well. And this happens because as, so anyone in the audience remember the right hand rule? Anyone? Physics? Yeah, okay, a couple people. So basically this is the right hand rule in action, so as you have current going through a wire, the magnetic field's going to curl around that wire, but the place where the inductance is going to be lowest is where that cancels out. So if the current can flow anywhere in the universe, it's going to want to flow directly underneath that signal because flowing back underneath that signal, it's going to cancel out, I don't have two right hands, unfortunately, it's going to cancel out the magnetic field as it was flowing that direction, and that's going to be the best place. So it's going to flow actually underneath, right underneath that line. Okay, so that was kind of the DC stuff. I had also promised in this talk, how does this all apply to Bluetooth and cellular and Wi-Fi and lore and all the new features that kind of come out here. So I wanted to disambiguate all these things. Basically they're all the same thing. So all of these lessons that I've been learning and hopefully you'll be learning or you've already learned, they all apply because the RF energy, basically as you add antennas and as you add amplifiers and everything else, they're going to be operating at different frequencies, but they're going to have the same fundamental rules applied to them. The things that might change are the length of the antennas or the shape of the antennas or the amount of power that you have to put into the system and things like that, but a lot of them are really just brand names and even more than that, a lot of them are actually software stacks that are on top of radios. That was something that took me a long time as well, right? So Bluetooth and Wi-Fi are both 2.4 gigahertz radios. If you didn't know this, a lot of times your phone is actually using the same radio to do Bluetooth and Wi-Fi. That kind of blew my mind when I learned that. And if not the same radio, then possibly the same antenna. Anyways, I wanted to show just some of the frequency ranges we have here. So again, Bluetooth is running 2.4 gigahertz, Wi-Fi is 2.4 gigahertz and five, Lore and SIG Fox, these are kind of low power, wide area networks, 433, 915, 868, depends on where you are in the world and what you're using. GSM and cellular have these multiple frequencies here, but the idea is I wanted to show there's, you know, there's a bunch of set frequencies that you might be dealing with. A lot of them are, you know, either as GSM goes or cellular goes. A lot of them are spectrum that's been allocated by your local governing body, FCC in the US, and I'm sorry, I don't know the European name here. But the idea is that those are kind of pre-allotted to these different technologies. But then there's the more open ones like 2.4 gigahertz and 915 in the States and ISM band and things like that. And the idea is that you're gonna be operating with all of these different frequencies, but the underlying things are the same. So let's talk about some resources that you might wanna look at in order to do what I did, I guess. Some of these are some books that I enjoyed, Practical Guide to RF and Mixed Signal. This is interesting because it's from the perspective of the people who were doing FPGAs for a long time and now FPGAs have radios going on boards as well, right? And they're moving from an eight layer board to a 16 layer board because they now need to have the interplay of these different layers. And so I'm coming at it usually from the low level DC. They're coming at it from the high speed digital and seeing the difference in RF characteristics and what they have to do to their boards is pretty interesting. Another one is just a generic RF circuit design. I enjoy this one. And then the final is a planar microwave engineering. Thomas Lee, Thomas is just like a famous guy in the industry, he's a microwave professor out at Stanford and it's just a really approachable book. Even though there's a ton of math, there's lots of very funny footnotes that helps as you're crunching through math as well. Oh, sorry, another one is YouTube. I love watching YouTube videos. I think that's kind of the highest bandwidth transfer of information. Alan Wolke is W2AEW. He's got a great channel. That's how I learned to, so I started learning Smithchart things and he's a hand radio operator. Tons of other great information there. And then Shariar who does the signal path. That's how I found out about one of those VNAs. He has a great review of those and then he also looks at the 67 gigahertz, $500,000 once. He of course has to give them back at the end. And then finally, a really accessible way to get started for a lot of this stuff. I think things like Mike Osman's, Grace Cocky Adjut's, HackRF, that's a SDR. And so you just kind of get a feel for what's out in the world. Another one is GNU radio plus RTLS SDR. That's like a $20 thing. And they're actually here somewhere in the camp. I think they're giving a workshop today too. And then there's analog devices, Pluto. These are all existing things. You're not gonna be building radios, but you're gonna be interacting with radios on the software side and that'll probably get you more into the hardware side. So a quick thank you to Jeff Kaiser or Mightyome. He helped me create these slides. And Derek Kozell who's actually the GNU radio person did this stuff and thanks for coming. So thanks to Chris for a very interesting talk. We have some time for questions. So if you have a question, there's a microphone right over here. You can step up and ask your question. But we also have a question from the internet apparently. No, no question from the internet. Okay, question here of the microphone. I have a question. Is it better to run RF through inner layers or outer layers? The question was inner layers or outer layers. I think the main thing is as long as it's referenced to ground is the main thing. Inner layers versus outer layers you're gonna have different characteristics because it's basically an outer layer is dealing with the dielectric of the board and then dielectric of air. So it's gonna have different like strip line characteristics. But usually what I've done so far at least, it's been real simple. It's been top layer, like Mike says in his presentation too. You do that first so nothing's in the way. Usually I think of inner layers, you're doing that because a lot of things are in the way. But I wouldn't wanna go through any vias or anything like that. I wanna just get it as close as possible to the output port and getting it there. So your mileage may vary. Yes, another question. Yeah, thanks for the talk. Are you aware of any FPGA like devices that use RF circuits? So the one that I was looking at was the, I think the Zinc has an RF front end now. They're not gonna be cheap though. That's the main thing. So I would recommend that. I mean, I'm a big fan of like the open tool chain stuff that's happened with Ice 40s. I think there's some of that stuff. Go pop one of those 433 megahertz radios on there. Otherwise though, not really sure about anything. At a certain point though the, so that was like a transmitter, right? You're basically pumping data to that. The fact of like an FPGA working with a transmitter is, you know, it's just gonna be pushing data to it. You could do the same with a micro or, you know, you could key it in your one if it could go slow enough, of course. But the main thing is that the transmitter itself is gonna be usually separate from the FPGA. I don't know of any integrated ones, especially not at the lower end where I'm usually looking. Any more questions? Final question? Okay, if not then please give Chris a big round of applause. Thank you, thanks for coming.