STM32 마이크로컨트롤러 RF 트랜시버 USB 최고 속도 인터페이스 PCB 설계

STM32 마이크로컨트롤러, RF 트랜시버, USB 인터페이스를 갖춘 소형 폼팩터 PCB를 설계하세요. 임피던스, 가드 링, 적절한 디커플링이 제어되는지 확인합니다. 레이아웃 및 라우팅에 대한 제조업체의 권장 사항을 따르세요. 제조 전에 철저한 점검을 수행합니다.

### STM32 마이크로컨트롤러 및 RF 트랜시버를 사용한 다목적 PCB 설계
* 🛠️  이 설계에는 STM32 마이크로컨트롤러, RF 트랜시버, USB 최고 속도 인터페이스가 통합되어 있어 PCB의 고속 신호로 인해 주의가 필요합니다.
* 🤖  이 보드는 처리 및 로깅을 위해 데이터를 PCB로 전송하는 RF 기능이 있는 로봇과 같은 다양한 프로젝트를 위한 것입니다.
* 💡  이 설계는 STM32L432 마이크로컨트롤러와 nRF24L01 저전력 RF 트랜시버와 같은 대중적이고 저렴한 부품을 사용합니다.
* 📏  레이아웃은 전원 및 커넥터 부분, 트랜시버 부분, 마이크로 컨트롤러 부분으로 구분된 회로도와 함께 디커플링 커패시터와 부품 배치에 주의를 기울여 세밀하게 세분화되어 있습니다.
* 🔌  USB 및 RF 섹션은 임피던스 정합 트레이스, 무중단 접지면, USB 통신을 위한 차동 임피던스 정합을 제공합니다.
* 📶  RF 라우팅에는 저전력 신호를 차폐하는 가드 링과 누화를 최소화하기 위한 세심한 배치가 포함되며, 레이아웃은 4계층 스택업에 맞게 미세 조정됩니다.
* 🎨  실크 스크린 라벨링, 구성 요소의 3D 모델, 제조 및 조립을 위한 적절한 파일 생성도 강조되며, 파일을 제작소로 보내기 전에 철저한 거버 뷰어 확인이 이루어집니다.

this is a design that I've been working on, which I think would be quite interesting to look at together. It incorporates an STM32 microcontroller, an RF transceiver, and antenna interface, as well as a USB full-speed interface. The design generally requires a bit more attention due to the high-speed signaling on the PCB, and lets us explore a few more advanced circuit design and PCB layout topics. The overall idea of this board is that it can be used for many different projects. For example, take a robot that has RF capabilities, and we need to log the data that this robot is picking up. One way of doing it is having an RF link, so the robot has an antenna and an RF transceiver, and sends data over-the-air, and we pick it up on this PCB here, process it using this STM32, and send it out to a host computer for logging using the USB interface. Of course, there are many more different applications we could use this board for, but as a general idea, that's what I had in mind.

Overall, where I'm using popular and inexpensive parts. For example, SC n 32 market controllers, which are fairly inexpensive, and this n RF 24 transceiver. So if I pull up the datasheet here, you've probably heard of it. It's an n RF 24 lo1, so low-power RF transceiver working in 2.4 gigahertz. And you can see it a lot in Arduino projects. But I thought you've got nice to use it because it's inexpensive. And then with an STM 32, the STM32 microcontroller I chose was this STM32 L for 32 KB. So this is part of their low power line. If you look at the ST website, it's an ultra-low-power microcontroller running at 80 megahertz. Because first of all, our power, we don't need much processing power. Essentially, the STM32 just provides a link between the RF and the USB host computer. We don't need much flash, we don't need much RAM, but we do need a USB full-speed.

interface. It's nice because this STM32 has the physical layer inside the chip already, so we don't have to do anything any more complex design. Okay, we've gone over the RF transceiver and STM32 microcontroller. I also thought instead of having a PCB antenna, it would be nice to have an SMA connector over here where we can attach various types of different passive 2.4 gigahertz antennas to this board, so we're not constricted to using just one antenna. Overall, I also wanted a relatively small form-factor. So this looks big here, but this is actually 15 millimeters in height and about 40 millimeters in length, so very, very small. I also chose to do this because there's quite a lot of interesting aspects of circuit and PCB design. For instance, USB differential signaling and just RF layout and routing in general. Okay, so Jeff, that's just gone over the overview of the board. Let's move on step by step how to ash your design is for what and what kind of design decisions I made. I usually start.

by choosing my microprocessor and what pin out I need. So, what interfaces I need if I knew GPIO, SPI, I squared C, and I tend to do that in qmx or cube ID. So I will choose a microcontroller, which is an STM32L432. It will show me this view here of what pins I can use and this overall size of the chip. For example, I can choose SPI3 and I squared C, select a full duplex mass, and so forth, and then select all the pins. So that's the first step I do. In the end, I'll have some sort of pin out diagram like this where I selected all the SPI pins and the extra GPIOs which link up from the STM32 to the RF transceiver. I've selected the SWD debug interface, which will have pins out here. I selected USB, which will go to this USB connector here, as well as two LEDs, receive and transmit, just to indicate if we're receiving or transmitting packets. Okay, so once I've done that, I tend to

move on to schematic creation which we'll do in Chi cat. All right, so if we go, I've made a little project here called strf. Go to the schematic layout editor, and here's the schematic I've come up with so far. I tend to section my schematics, so I've segmented it into three parts. Here we have the power and connector parts up here, a USB connector. We have the transceiver part over here and the microcontroller part over here. I generally also put notes next to my schematic because then for future use on their what I what the hell I was actually doing. Okay, but let's start with the power section of the board, which is over here. So USB has a five-volt power connection turret, but we need 3.3 volts to drive for Percy sir and the transceiver. Now in the data sheet with STM32, it says that we actually need 3.3 volts pretty much exactly for the USB to work on the STM32. So that's one thing to take in mind because we're using USB to power this.

board we need to be a bit more specific would what decoupling caps we use. So the USB specification says the intricate aster c1 needs to be a maximum of 10 microfarads. If it's any larger, the inrush current when this board is plugged in first so the host computer will be too large and the host device might not like that. So a maximum of 10 microfarads at the input is given in the USB specification. Another thing to note is usually I would put in reverse polarity protection in terms of a diode or p-channel MOSFET, but since we're using a USB connector here, the chances of a she getting reverse polarity on the input voltage of very slim. So I thought I just emitted for this for the sake of board space, but in general it's quite a good idea just to put a reverse polarity protection on there. I've had a fuse of 100 Mille Anton and that should be probably more than enough for the power consumptions or current consumption of this board than that it's a fixed LDO.

regulator, which can drop 5 volts to 3.3 volts fairly efficiently, and that will give us a maximum - and milliamps are 3.3 volts. Okay, so that's kind of four very, very basic through through regulator. Another thing maybe to know is that I'm using a ferrite bead just to limit the high-frequency interference coming from a power supply. So essentially it's bait if almost effectively an inductor. So it really at high frequencies at a hundred megahertz, it looks like a hundred ohm impedance. Okay, so moving on, the next section I would like to look at is this STM32 section. And again, this is taking just information from the datasheet and application note provided by ST and making schematic out of it. So it's fairly simple. The pinout is transferred from QMX, so I showed the picture of that pinout. So I've just labeled all the pins using these global labels over here and put them on. Other than that, I use one decoupling capacitor of 100 nanofarads.

per VDD pin so we've got one two three here. That's why I have three 100 nano farad capacitors up here. I also tend to put one bulk decoupling capacitor fairly close to the general chip itself. So in this case it's a 1 micro farad capacitor. Alright, we also have this boot zero pin over here, which is PH3 in this case. And I pull this low. So the boot pin determines if the bootloader is started or not. In my case, I will be programming via SWD and not using UART or USB to program this ST. So I can pull the boot pin low. There are several application notes and the data to it'll tell you what boot pins you need to pull higher load to achieve which bootloader setting. But in general, if you want to use USB or you are to program it, pull this high or make it variable with a dip switch. But if you're just going to be using SWD, pull this pin low, the boot zero pin low with a 10k resistor. Okay, so next

I've got these two LEDs. I've chosen green, yellow, any color should be fine, and then of course current current limiting resistor. The way you can calculate the current limiting resistors: you have 3.3 volts at the GPIO pin where it is high. Red or yellow LED will drop about 1.8 volts. And then you do 3.3 minus 1.8 divided by your current requirements, maybe 2 milliamps, and I'll give you that the resistor values. So I think I calculated about 1 milliamp for each LED, and that's usually more than lovers fairly bright. Okay, also for USB, it's a bit hard to find in a datasheet, but once you know it, you won't forget is that the USB differential pair inputs, a USB D - and US pair D +, don't need any external termination or pull-up resistors. These normally there you'll see these 22 ohm resistors in series with the D minus and E Plus lines, but for this us for this

stm32 you won't need that they incorporate it into the chip which is really nice and if made a little note down here as well also to indicate that it's a four-speed device. You'll sometimes need a d-plus pull up in a pull-up resistor for about 1.5 kilo ohms but that's incorporated into the device as well, and that's given in the application note a and for 879 so it's a bit daunting at first to see what external resistors you need what termination resistors you need but once you have it once this is usually fairly the same for most ST microcontrollers. Okay, other than that I've this is that section pretty much done here I have the serial wire debug connector and if you just google arm 10 pin swd can see what the pinout is for a typical swd connector so if you just copy that over transfer to your schematic that's exactly what I've done and that'll fit with the st-link adapter okay this is something that isn't

strictly necessary is this essentially debouncing capacitor on the end reset pin. 100 nanofarads. And that just essentially helps to prevent parasitic resets. So if you, if someone touches the pin with the hand, it might reset the vice. That kind of protects it against that. Okay, so that was already the microcontroller side to the USB connector. So we remember here on the microcontroller, we have the USB differential pair - D - and D + and that goes to the USB connector here. And I've chosen a USB micro connector just for the size. Just before that though, I've chosen this USB LC 6 - SC 6. And that is to prevent against ESD. So we'll have a lot of human touch when plugging and unplugging the USB cable to the connector. And humans can carry quite a lot of static voltage on them, which might disturb discharge through the PS PCB. And that's what we definitely don't want. And this chip is an ESD protection chip. Fairly common that'll protect that.

so it has low the T vs. diodes inside in a nice little 6-pin package. But I'll show you more of that one we actually get to the PSP layout and PCB layout and routing. I'm good sneak peek it's this chip here. So we're routing from the STM32 through to that ship and then through through to the connector but more on than later. Okay, so back to the USB connector. The USB connector has strictly defined pin outs. So pin 1 is the 5 volts which then goes through the fuse through the ferrite bead to the regulator. We have our data lines data + and data - and we also have this are deep in which we don't need so I just use it no connect flag. Ground of course is connected to ground and the shield is an interesting aspect. So shield the opinions vary but according to the USB specification the question is do we need to ground this shield or not agree on the shield and the answer is the shield is grounded at the house side so if

you're connecting this to a PC, this will be grounded at the host side. Now, we don't want any current to flow through the shield because we don't only want any current or to flow from, for example, the host to the PCB and then to therefore induce noise into the PCB, and we also wanted to act as a shield. So we do not ground the shield on the PCB side. So again, add a no connect symbol here. For more information, there's plenty of information online. Just google shield USB ground or mock Randy. Okay, so the last thing we need to look at is the transceiver side, which is this whole section over here. A good thing here is to always follow the data sheet. So if we look at the data sheet again, which is over here, this will have a lot of information of just the general block diagram, the pinouts, but it will also free I can find it if you an occasion example. So we're using a single-ended matching network, so have a single-ended antenna. It'll tell you about the crystal you need to use.

and what frequency there needs to be at various decoupling capacitors, and pull-up, and pull-down resistors, and so forth. Essentially, all you need to do is transfer this schematic to Chi card, and that's pretty much exactly what I've done. I've seconded sectioned it a bit more, and we'll go through it, but essentially follow the datasheet, and if that application example matches what you need, in this case it does because I'm using just a single-ended antenna of 50 ohms, you can just transfer it over, and they'll give you the component values you need for the matching necklace, and so forth. But let's just go through it quickly. So here we have the NRF chip, we've got all the connections to and from the microcontroller, we have various pull-up resistors, and of course again the decoupling capacitors. So again, per VDD, we will use one decoupling capacitor and one bulk decoupling capacitor, which will be close to the

chip so the NIF transceiver we're using here also requires an external crystal, and the schematic for that is given in the data sheet over here. So I've pretty much exactly taken over and put it into click Add. Now the one thing you might have to change depending on the crystal you're using is the load capacitors on either side of it. So the crystal I'm using here, I calculated the load capacitors to be 12 picofarads on each side. The way I calculated is that on the datasheet for the crystal you'll use, you find a load capacitance. You take this load capacitance, you subtract the stray capacitance you'll expect on your PCB. So this will be anything from two to five picofarads, and you multiply that value by two. So I did that for this crystal here, and these are the values I got. For more detail, or just Google crystal load capacitance or PS crystal oscillator, and will tell you.

exactly why these capacitors are there. Okay, so the final thing we need to look at for this transceiver is the antenna impedance matching and the SMA connector. So if we go back to this 3D view, we've done the crystal here, we've done as STM32 microcontroller, and we've done this USB port over here. So the crisp, the RF matching part, antenna matching part is all this stuff over here. So we need to do that because this chip will have a certain output impedance. I don't know, it could be a kilo ohm or something, or 200 ohms, but our antennas generally have an impedance of 50 ohms, and we somehow need to match this 1 kilo ohm or 200 ohms whatever to 50 ohms, and this is what this matching network does. So it converts essentially an impedance here to an impedance here, and make sure both sides see a matched load, and this is very important because the signal voltages and signal powers we're seeing or receiving from the RF side incredibly

small and we want to minimize any reflections and power losses. Therefore, we can actually transfer the maximum signal strength. And this is what this network does. Additionally, and also, it converts impedances, but it also converts from single-ended antenna to a different differential input from this chip. So we have a differential connection here, but we have a single internet and 10 over here. And this network also converts single-ended to differential. But again, this is given in the datasheet. So essentially, it's this network I have transferred over here, which is quite nice because you don't have to do any of the complex calculations yourself. Okay, so this pretty much concludes the schematic section. And next we'll move on to component selection and then layout and routing. Okay, so now we're ready to select our components. I've clicked up here on the sign PCB footprints - schematic symbols. And this is open this window where we can actually assign footprints to every component in the view. So in general, because I'm I want to make a rather small PCB, I've gone for Oh 402.

components, or rather small SMD components, especially for the RF side of things, is actually quite important that they are oh for two components, and that is also given or mentioned in the datasheet for the NRF 24. So all these components you see here are oh for Oh - other than that that only plays where I've used Oh 603, so slightly larger surface mount components is for example the bulk classes of logic capacitors, so 10 microfarads are Auto 603 s, and also the resistors for the LEDs at the current limiting resistors. So anything which has to handle or higher currents or higher needs to dissipate more power, I generally tend to make them larger. Other than that I've also made the fuse Fei large, so that's a she 1206 in size. So over here that's 1206, just so in case it burns out and need to be replaced it's easy to replace. Other than that I've chosen a USB connector, which is fairly cheap, and an SMA connector, which is a 50 ohm.

connector rated at 2.4 gigahertz. So I can show you that looks something like this. This is just a ready run in through your model but the general view is here and it's an edge mount PCB connector so it'll slide onto the edge like this and we can solo down onto the pads. Okay, so that pretty much is a component connection. And so let's move on to layout. Okay, so now when the PCB editor and let's just talk about layout to start with. A good idea is to always do a rough sectioning of what components need to go where. So let's look at the 3D view. So if essentially three four different main sections, we have the USB and power section, we have the STM32 section, and we have the RF section. So three to four different sections. And first of all, you want to do the rough layout. So roughly take the components, put them in positions where you think they'll be later on, but don't be too finicky. So don't change the grid size very small.

and try to mess around and move these millimeters or less than millimeters. You want to kind of just do a rough placement to get a feel for how big the board is actually gonna turn out to be. So we want to turn the board size via the rough placement and try to fit all our components into that space. We also want to make sure that we're placing, if we can, all the components just on one side. As you see here, all my components on one side. This makes the assembly much cheaper, makes the troubleshooting much cheaper. You don't always have to flip the board up and down, and you can actually place this on a flat surface. Okay, we also want to place the critical components first. So this could be, for example, your centerpiece and your RF sections. You want to place the chips first, your RF layout first, the crystals first, and things like decoupling capacitors over here. So decoupling capacitors want to be ready or very close to the supply pins. With example here, we have the supply pins, 3.3 volts.

and not the decoupling capacitors. Are only fractions of a millimeter away from these pins. So place corporate critical components first. And then place less critical components. For example, less critical components could be this 3.3-volt regulator. Right now, there's no high-speed signals going through this or near this or this fuse. These are not entirely critical if you place them a bit off. So critical components first. And then the less critical ones. And try to fit that in the board space. We also want this ESD protection chip, which is this one here, close to the USB connector. So closer to the USB connector than to the STM32 chip. Regarding the RF layout, luckily if we look at the datasheet, there is actually a PCB layout example given. This will give you where you need to put all the RF matching the ICS, Barry's connectors, and crystals. It'll tell you where to put them. And this is exactly what I've done here. I've transferred the example they gave onto this PCB. And that should mean should hopefully work.

and be a good layout, okay, and generally manufacturers will give this kind of information in their data sheets, okay, so once you've done the rough layout, you've placed all the parts, you can determine the board outline. So I initially place all the parts fairly roughly and said, okay, this should fit in these kind of dimensions. So I ended up being 15 millimeters high and 39 millimeters wide. I also added, using the edge cuts layer, these rounded corners. I tend to think this is using this ad graphic arc tool. I tend to think it makes the PCBs look a bit nicer, okay, so now you've got the rough layout, you've done the edge cuts layer, you've got four dimensions sorted. Now you can actually go into the fine-tuning of the layout, so making sure decoupling capacitors, again, are close, but maybe they're not optimally close. So you can optimize things that actually go to go into a finer grid setting and move them around bit more finely. So fine-tune that for as long as you have

to because a good layout ensures that you'll have a much easier time actually routing this board. Okay, so I was fairly happy with my layout here, and it turned out that made my PCB routing tasks much easier, and that's what we're gonna move into now. Okay, so now let's get started with routing, but before we can start with routing, we need to look at the layer stack up. So in general, anything RF, you want at least four layers. So two layer boards will mean your controlled impedance traces will be very, very wide, and you'll generally have problems with decoupling and power planes and so forth. So what I tend to do is generally stick with a four layer board, and you'll be pretty safe. So if you go up here into the board setup, you can see the layer stack up. So the top, we have the front copper. We have an inner copper, secondary inner copper, and the bottom copper layer. Now I will be using the front copper just for signals. The inner copper layer will be ground. The second inner copper layer will be

power or 3.3 volts, and the lower copper layer will be signal. PCB Cygnus is one point. It is one point six, and that all looks fine, okay. What we want to do is then route section by section, so we've laid it out nicely. Now we want to root, for example, all the USB. We want to route all the power in one section. We want to route all the STM in the one section, and we want to route or RF and into one section. In general, we want to keep the largest tracts we can and larger tracts for higher current traces, and the same goes for vias. Obviously, you want anything that carries any higher current to be as thick as you can possibly make it. For example, here, these power tracks over here are almost a millimeter wide, and any signal traces, for example, up here or anything coming out of these small paths, will be under 0.3 millimeters wide. We also want to keep high-speed signals from different systems away from each other. So, for example, here are the SPI

lines you want to keep that away from the USB or serial wire debug lines. So any high-speed signals, which aren't part of the same system, we want to keep separate. Also, every ground or power pad. For example, here ground should have its own via. So we have one ground via going here, one ground via go here, and so forth. And that's generally a good rule. Of course, sometimes we'll have situations where you won't have enough space, but generally one via per pad for power ground pad. The nice thing using a four-layer PCB is we have these copper pores in the middle, and we can just use a via to connect directly to this copper pot. And this gives a nice low inductance connection to ground or power. Okay, so on the topic of ground paws, I switched is view so you can actually see them better. You'll see at the front layer I have some ground pause, but they're mainly just use so Dan how to route everything manually, but in a first in a couple a as we said as a ground.

Paul has a huge uninterrupted grandpaw, except for this crystal here, but we'll get into that a bit later. A second inner layer, we have this large 3.3-volt grand pool, but nothing underneath the RF section, and we'll get into that later as well. And finally, the lower order back copper layer, again a large ground pool, but nothing underneath is RF section and nothing underneath the crystal. All right, but let's start with looking at the power section. So again, we want large tracts between this. This is the blow of the regulator over here, so we want the large tracts, and we also want, because this regulator is going to feed essentially the power into one of these ground planes, you want a lot of parallel vias connecting into that power plane and also into that ground plane. So a lot of parallel vias means we reduce the inductance. Okay, so that was a really the very simple power circuit over here. That's what you do. Pay attention to large tracts and parallel vias to reduce

the inductance into the feeding into the ground and power planes. So, over here, this is our STM32 chip and essentially our centerpiece of this board. And that's why I thought to require ice just to put this into the center board. So, as I said before, we wanted the decoupling capacitors first close to the supply pins. And then I routed the SPI connections to this NRF chip. And also, for example, the serial wire debugger to this connector. And lesser important connections, such as these LED connections. So, thus, you see the kind of author I did things first or due to coupling capacitors than any high-speed traces. For example, SPI and USB and serial wire debug. And then the less important ones will get into USB in just a second. But before that, let's have a look at the NRF chip again. They said before the layout is in the datasheet. All right, you can just copy that over again. Decoupling capacitors close to the supply pins and so forth. One important thing here is, of course, that we're using a crystal or rather of NRF requires

crystal for proper operation, what you need to pay attention to here is that nothing, no high-speed, or any signals in general, should be running near or below the tracks of the crystal. So you see how from no traces running underneath the crystal, no high-speed traces, everything's nicely separated. Another thing you need to pay attention to is that the ground pour is separated for the crystal. So if I switch to this view, you see that I have a huge ground plane for the rest of the board, but the crystal is separated. So I have drawn a little ground pool underneath it and connected it to the main ground pool with a very thin trace only at one point. And as several sources which recommend doing it this way, and it's worked well for me so far. But there's one thing you really should pay attention to. So I have no other tracks or any other ground pools underneath this, just the one on the inner play, and it's connected by this one trace onto this larger ground pull. So now we're coming to the more

interesting parts of the board, and we're going to look at the USB and RF sections. So one thing to mention right at the start is that impedance matching is very important. Now, in general for RF, we want a 50-ohm impedance traces. So, for example, here, all these traces are matched to 50 ohms. So let's look at that first. So you can see here the width is 0.2393 millimeters, and I'll tell you exactly how I got to that answer. So if you go to my preferred PCB manufacturers, GLC PCB, let's just click on quote now, choose a four-layer board, and then it will show up with this impedance calculator. And this is what I use because you'll need to know the stack-up from your manufacturer of the PCB. So I can see if this works. So every PCB manufacturer will give you this kind of information of how their PCBs are layered up. So if there's a copper layer, and then there can be a dielectric material or another copper

answer fourth the what the PCB manufacturer uses this and this is the stack-up for their board. And their stack-up will determine the width of the trace for a certain impedance. Luckily, they will usually also include an impedance calculator. So, I will choose I want impedance and trace space. I want a 50-ohm impedance trace. I'm routing it on an outer layer, and it's a single-ended on a four-layer board. So, I click essentially calculate. And for the second I'm using, which is this one here, the recommended trace width is eleven point five five mil. Particular line, and all you have to do is then go to board setup, go to present tracks and vias. And I typed in eleven point five five mil here. So, I just typed in eleven point five five mil, press ENTER. And it'll then convert to millimeters for you. And it turned out to be zero point two nine two three seven millimeters. And this will give you a 50-ohm Trek. And this is what I used to root the.

whole RF section here. So this is 50 ohms, this is 50 ohms, and so forth all the way into this chip here. Now what's also important is not just that your track widths are correctly set (and you can use an impedance matching calculator to do that) but also that you have an uninterrupted ground plane directly underneath the layer where you're routing. And this is what I have here. So I have the top layer, which is this red or this red copper here, but underneath I have an uninterrupted ground plane. So anyway, there are their signal, there's no, there's no disking, you discontinuities in the ground plane, and that needs to be the case for everything. Another thing you need to take care of is that you don't have a ground fill next to the traces for control Derpy's and traces because otherwise if you have a grand fill, let's say next to this here, you'll also get kind of impedance matching or microstrip effects between these planes, and that's not what you want. You want

only that effects to take place between the top layer and the next inner layer, which is a Grand Tour. The same goes for any controlled impedance or high-speed trace. So calculate the correct trace impedance and then also put a solid ground pour underneath. And of course, you can figure out the stack-up and what impedances and trace would you need from various different calculators. There's also one in ChiCAD up here, so PCP calculator. You can go to transmission line type in your PCB parameters, and it'll calculate what trace width you need. Okay, but keep those two things in mind: trace width and uninterrupted ground fill underneath and no ground fills next to it. Okay, so let's check if I've covered everything for that. Okay, and the same goes for the USB as well. So USB, you will see I have matched impedance traces, and I'll tell you about differential impedance traces in a second, but you see that the ground fill is uninterrupted underneath all the way to the connector. Okay, so here we looked at single-ended matching USB is a differential.

communication system or signalling system, and we need to do differential matching, but it's as easy as it was before. USB requires a 90 ohm differential impedance. We click on differential. We also need to click the trace space, so the separation between the tracks. So this separation here, and you're free to choose any of these values here. These are militant mils, and I chose 8 mils. And then you can click calculate, and it gives you ten point two eight mils trace width. So the trace paths was this 8 mils, and the trace width is always at about 10 mils. Here you can also enter that in the board setup. We type in the width, which was about 10 mils. I believe the gap was eight mils, and I just set the vibe after the same, and that'll this will then enable you to route differential pairs which are controlled impedance. Okay, so the way you can route those is click on route, and then differential pair, and then click on whatever you need to here, and it'll wrote it for you. Okay.

so one thing you might need to do with USB and with differential traces is to assume it match the SKU or the length of the tracks. Just clicking on route and then differential pair and routing the pair's does not guarantee that these the lengths of the individual traces will be the same, and that can result in different that the signals arrive at different times at the receiver, and that can be quite dangerous. The maximum SKU you can have or maximum difference in reception times for USB is stated in the specifications and is given as for NAT picoseconds, which is a fairly large margin for the speeds we're looking at here, but the way you can match the trace lengths is if you go to route tune differential pair skew, click on all these, and then just drag around, and you see it's trying to match the length of the trace, and that's what I've done here. So this little bump here is actually matching these two traces to be pretty much equal length, okay, and these apparently are equal lengths as well, so everything's SKU wise is looking fine.

as well. So just to recap, you want to calculate the crate trace width and trace separation to give you a proper impedance. You want to have an uninterrupted ground plane underneath, and you also want to make sure the SKU or the phase difference between the two differential lines is matched so you don't get any problems with the reception or transmission. Okay, so that's USB routing and RF impedance measured routing. Rather quickly, one thing you'll notice here in the 3D view is this ring over here. So you see it starts Senshi when this RF section starts and ends at the RF connector. And this is what is called a guard ring, and the guard ring typically surrounds critical RF circuitry. So low-power signals coming from the antenna onto the VCB. So these are incredibly low-power signals, and you want to shield them from any stray currents which may arise on the PCB. And you can use a guard ring to do so. So a guard ring will be essentially exposed copper around your

RF circuitry and it will type be tired to a low impedance ground. That's why I have all these vias here to tie it to ground. So an example where you might see that, you'll see that in loads of RF systems. But here a great example is the hack RF. And you see here you have an RF input and all the IR section is shielded by this guard ring around it. And this is exactly what I'm doing here as well. So we have a copper essentially Paul connected with loads of ground vias to the ground plane underneath, okay? And that in a nutshell is over RF routing. Yeah, again, I'm probably sick of hearing it now, but uninterrupted ground plane matched RF traces impedance matched traces and make sure the skew for differential pairs is matched. Yeah, and you should be pretty safe. And again, follow the manufacturer's recommendations. I also have a video on controlled impedance traces, which goes into a bit more detail of how these calculations come into play for the calculating the correct trace widths.

yes just to reiterate a point that it's good to keep clearance between everything and an especially high-speed traces. So you see, I've left quite a bit of gap between any high-speed traces. For example, VAR f and any other high-speed traces, for example, the USB or the SPI on there on the board. Does anyone try to - I know this is a very small board space but - right to maximize the clearance between high-speed traces. So he minimize capacitive crosstalk and cross coupling. Okay, so we've that pretty much concludes the routing part. So all the other important aspects have been highlighted. Maybe one thing just to mention here, you'll see I have some vias and pads. And if usually was shunned upon that you shouldn't be doing vias and parents because they were difficult to manufacturing and soldering. But generally, manufacturers these days can handle that perfectly. And it just enables, again, multiple views in this ground pad enable a thermal connection for one. So you'll have better thermal dissipation because these views will

then connect into this large ground plane for heat dissipation. In this scenario, it's not a problem because we're not going to be dissipating very much heat or power, but it also enables these multiple wires. Enable a low impedance connection to ground as well. Okay, so once you've done the majority of the routing, the next thing, of course, is to make it look a bit prettier. And one way of doing that is adding silk screen. So, for example, I've added just labeling to make sure everyone knows what these LEDs are for. So, it's a reception and transmission LED. I've marked that this isn't a serial I/O debug connector. I've put a little sense you logo or a little silkscreen image on to label what this product actually is. Put my website on the back, of course. Mark that this is pin 1, yeah, just to make sure people know what's going on with this board. Another thing for assembly, which is quite useful, is to mark the locations of pin 1. So, you see these little dots here mark the

location of the pin one of each IC. So pin one here, and one here, and so forth. I've also marked the polarity of the diodes. Sort of an anode is marked with positive here, and that just helps for assembly or for any troubleshooting. Okay, so that's the silkscreen, making it look a bit prettier than before. Another thing you can do here for this 3D model, not all parts will have a 3D model associated with them. So for the 3D model for this, as a makin actor, I asked yet to Google to try and find a step model so I can import this into Karkat. And the way you can usually do that is, usually manufacturers will have 3D part models in step formats or various different three formats. You can just download that from the manufacturer's website, or of course just Google the part name and 3D model, or use sites like GrabCAD and stuff like that. Okay, so once you're happy and you've checked the design multiple times, your

tricks at this chromatic the layout and so forth made sure everything looks nice. The ground paws look good. You've done that. The DRC check and made sure there's no errors. It's time to export it for manufacturing. And the way you do that is file plot generate your files. First of all, generate your file and then you want to generate all the layer files to the front copper inner coppers there copper silver and so forth. Just click plot rates all the Gerber's. And then you can send that off to your PCB fab house for manufacturing. But if you want assembly, you will also have to make other output. So if it'll whole fabrication output you want the footprint position files generate protein president fell here. And this will generate a file a Commerce separated value file of all the ICS and where their Center points are. And this will help the pick-and-place machines at the fab house manufacturing house to know where to put all the components. Another thing you to do back in the schematic over here is generate a bill

of materials. So, what proponents, yes, you want to be used for various capacitors, resistors, and so forth. I have a more detailed video going into more detail on how to actually do that and export these files properly. I wanna link that in the description as well, but once you're done with that, you're happy and exported all your files. One last thing I always do is do a Gerber viewer check just to make sure Kai card has exported the files properly. So, I'll just load all those in, will import them, and I'll just go through every layer to see if all the copper fields are correct, if all the vias are in the correct place, the board outline seems okay, and just go through those individually and so forth. Okay, so once you're happy with that, just send them off to your fab house, and then it's time to test the device when it arrives with you. So, I hope this has been helpful as an introduction to RF and kind of high-speed signaling layout with USB and STM and NRF, and if you have any

questions please do leave them in the comments. thanks a lot.

KiCad STM32 + RF + USB Hardware Design - Phil's Lab #5

Analog and digital electronics design, PCB design, control systems, digital signal processing, and more!

Website - https://www.phils-lab.net
Patreon - https://www.patreon.com/phils94

Hardware Design Courses - https://phils-lab-shop.fedevel.education


Phil’s Lab

RF 트랜시버가 포함된 STM32 마이크로 컨트롤러 기판 설계

이 설계는 STM32 마이크로컨트롤러, RF 트랜시버, 고속 신호를 위한 USB 인터페이스를 PCB에 통합합니다.

이 보드는 로봇 및 데이터 로깅을 위한 RF 기능 활성화와 같은 다양한 프로젝트를 위한 것입니다.

설계 결정에는 대중적이고 저렴한 부품의 사용과 다양한 안테나 옵션을 위한 SMA 커넥터의 통합이 포함됩니다.

보드의 전원 섹션에는 전압 요구 사항, 디커플링 캡 및 역극성 보호에 대한 고려 사항이 포함되어 있습니다.

STM32 회로도 설계하기: 핵심 사항

3V로 전압을 효율적으로 강하하는 레귤레이터, 고주파 간섭 제한을 위한 페라이트 비드 및 STM32용 특정 커패시터 구성을 사용합니다.

-  SWD 프로그래밍을 위해 부팅 제로 핀을 낮게 설정하는 것과 LED의 전류 제한 저항을 계산하는 것의 중요성.

-  STM32 칩에 종단 저항기 통합, USB ESD 보호 칩 사용, USB 커넥터의 엄격하게 정의된 핀아웃 등 USB 설계 고려 사항.

PCB 설계를 위한 적절한 접지 및 트랜시버 설정

호스트 측에서 실드를 접지하면 전류 흐름을 방지하고 PCB로의 노이즈 유도를 줄이는 데 도움이 됩니다.

트랜시버 데이터시트를 준수하는 것은 부품과 값을 올바르게 선택하는 데 매우 중요합니다.

매칭 네트워크는 임피던스 변환과 RF 신호 전송에서 전력 손실을 최소화하는 데 필수적입니다.

PCB 설계를 위한 부품 선택 및 배치

저항기와 커패시터를 포함한 SMD 부품의 선택은 전류와 전력 소모를 처리하는 데 매우 중요합니다.

레이아웃 프로세스에는 중요한 부품을 먼저 배치한 다음 덜 중요한 부품을 배치하는 데 중점을 두고 여러 섹션에 부품을 대략적으로 배치하는 작업이 포함됩니다.

PCB 레이아웃은 RF 부품에 대한 데이터시트의 예시에 따라 안내되며, 대략적인 레이아웃이 완료된 후 부품 배치를 미세 조정하기 전에 보드 윤곽을 결정합니다.

PCB 라우팅: 레이어 스택 업 및 트랙 설계

임피던스 트레이스를 더 잘 제어하고 디커플링을 개선하려면 4레이어 보드를 사용하는 것이 좋습니다.

일반적으로 신호용 전면 구리, 접지용 내부 구리, 전력용 보조 내부 구리, 신호용 하단 구리로 구성된 레이어를 쌓아 올립니다.

라우팅할 때는 서로 다른 시스템의 고속 신호를 분리하고, 더 높은 전류 트레이스에는 두꺼운 트렉을 사용하고, 각 접지 또는 전원 패드에 고유한 비아가 있는지 확인하는 것이 중요합니다.

전원 섹션에 병렬 비아와 큰 트랙을 사용하면 인덕턴스를 줄이는 데 도움이 될 수 있습니다.

라우팅 프로세스에서는 디커플링 커패시터와 SPI 및 USB와 같은 고속 트레이스의 우선순위를 정하는 것이 중요합니다.

PCB 설계에서 수정 및 RF 섹션의 적절한 작동

간섭을 방지하기 위해 고속 신호를 크리스털에서 멀리하고 크리스털을 위한 별도의 접지 공간을 확보하는 것이 중요합니다.

임피던스 계산기를 사용하여 트레이스 폭을 결정하기 위해 모든 트레이스가 50옴에 맞춰지는 RF의 임피던스 매칭은 매우 중요합니다.

라우팅이 수행되는 레이어 아래에 중단 없는 접지면이 있어야 하며, 임피던스 및 고속 트레이스 제어를 위해 트레이스 옆에 접지가 채워지지 않도록 하는 것이 중요합니다.

USB 차동 쌍에서 적절한 임피던스를 달성하려면 트레이스 폭과 분리를 계산하는 것이 중요합니다.

동일한 수신 시간을 보장하고 신호 지연을 방지하려면 차동 트레이스의 스큐를 일치시키는 것이 중요합니다.

저전력 RF 신호를 차폐하고 표류 전류를 최소화하려면 가드 링과 무중단 접지면을 사용하는 것이 필수적입니다.

PCB 레이아웃을 설계하려면 열 방출을 위해 접지면에 연결하고 낮은 임피던스 연결을 위해 여러 개의 와이어를 사용할 수 있도록 해야 합니다.

구성 요소와 핀 위치에 실크 스크린 라벨을 추가하면 조립 및 문제 해결에 도움이 될 수 있습니다.

제조를 위해 파일을 보내기 전에 디자인을 꼼꼼히 확인하고 거버 뷰어를 사용하는 것이 중요합니다.