 Lunar New Year, welcome to Desk of Lady Aida. Hello, this happy dragon welcomes you to the year of the rabbit and the water rabbit you said earlier. Well, happy Lunar New Year to everybody. It's a wonderful time of year. A lot of people get to see family, eat some delicious dumplings, maybe get some nice presents in their red envelopes. Here at the Desk of Lady Aida, we are also making delicious dumplings, dumplings of electronics. Yeah, we got a lot of going on this week. We have some good top secret videos we've been posting up. We just filmed some of those, but this is our live Desk of Lady Aida show. So, yeah, ask questions during this, but you can also save them towards the end. We also do the great search brought to you by DigiKey, where we use Lady Aida's power of engineering to find the things that you want to find on digikey.com. Lady Aida, what is on your desk this week? All right. I'm glad you're at it. Of course, are there any news or updates? It shows as usual, right? Yeah, Liz is hosting her co-hosting show and tell, and then we're back doing Ask an Engineer while we were hanging out with some of the Raspberry Pi team on Wednesday. So, that is why we did not do a show. Okay. Got a lot of cool stuff ahead. Okay. Especially with Raspberry Pi related stuff. Sweet. Well, speaking of, let's go to the overhead. So, this week, one of the projects I worked on is I wanted to make some more feathers. I haven't made new feather boards in a while, and I thought, why revise the feather RP2040? But when I first designed the feathers, I think it's actually one of the first feathers, this one. The way I did it is I started with a dev board that has USB and charging and a reset button, and then I had a main chip, and then the idea was you'd have a different thing over here. So, for example, this is a Bluetooth module, and this is a, let's see, this is a 324, sorry, this is a SAMD21. So, this is a Cortex M0 SAMD21. Oh, that's right. It has this crystal. That's how I know. And it's got a Bluetooth module. Here's that same SAMD21. You can see all the parts on the left side are the same, and on the right, it has a SD longer instead. Or this version, also a SAMD21, you can see, again, left sides all the same, but this time it's got an RFM, Laura, or a radio module. And then, you know, originally this was a 324 based. 324 is good, but I wanted more memory in space. Cortex M0 is a nice update, 32K of RAM. That's still a lot of time for Circuit Python or other embedded, interpreted languages, or even if you just want to do slightly more complicated projects, especially IoT projects, if you're reading JSON data in or buffering data, 32 kilobytes does not last very long. The SAMD51 is a better bet, and we do have the SAMD51 here, but it turns out, well, first off, the SAMD51 is just a larger chip. There's more accessories on there. I could shove this over, but, you know, because I wanted to basically update the SAMD21 to make it better for Circuit Python use, because, you know, for example, the Laura radio, it kind of uses almost all the memory on the SAMD21. If you want to add other sensors and stuff onto this featherboard in Circuit Python, you run out of memory very quickly. So redoing it with the SAMD51 would be great, but I can't really get the SAMD51 right now. It's still, like, totally unobtainable. And so I thought, you know what, I should just move on and, you know, update it all to use the RP2040, and this is the original design of the RP2040 feather in pink. You see, it does take up a lot of space, but that's because, you know, you've got this large Neopixel and you've got this SWD header and a large SOIC flash memory. But maybe we can make this smaller, if we can make this as small as these SAMD21 feathers, then I could fit, you know, basically the same kinds of accessories, and it'd be great for use with Circuit Python, because that's kind of the only, you know, we don't, you can use the SAMD21 for Circuit Python, but again, it's, it's a bit of a cramped fit, especially these days, because we have so many things you can do it very quickly, you run out of memory. So to go to the computer. So when I, you know, we've got all these feathers, but when I did the Scorpio boards, the Scorpio doesn't have a lot of stuff at the end, but it does, you know, did have this thing where I kind of tried to squish the RP2040 over to the left to make room for this flash memory and I squared C stem and QT port, and then this level shifter, the Neopixel is smaller. So I took that RP2040 feather and sort of squished everything to the left to make enough room for this two by eight header and level shifter. So you could use this to drive, you know, eight channels of Neopixels, for example, but it's not this, this, this zone here is still quite small, I could probably fit one of those tiny micro SD sockets, but it's not large enough for like a bigger radio module. So my project this weekend was to really squish that stuff on the left and here is what I came up with. So to squish, there's a couple of things you can squish and some things you can't, right? Not everything can get smaller. Oh, you know, this rabbit is kind of in the way. Hi, rabbits in my circuit. This is a cute little, we're gonna give this as a gift to somebody with a kid that we know. It's a little happy rabbit. Happy new year. Okay, rabbit's out of the way. So the JST connector, we can't make smaller. The USB-C, we can't make smaller. The LED, there's not really much point in making smaller because it's kind of in this corner here, but and then the crystal, I tried to see could I get a smaller 12-vanguards crystal? No. Two by 2.5 millimeters is as small as it gets. The tactile switches, you can't make them smaller. They're pretty much as small as you can get while still being functional. Although I did move the boot tactile button over to the left. It used to be, you know, it was over here on the Scorpio. I moved it over. The Neopixel, you know, you can make smaller. I made that the 1.5 by 1.5 millimeter and I kind of tucked it over here. Flash memory, you know, if you want to use four to eight megabytes, you have to use the larger WSAN. You can't use your XSAN. You can't use USAN. The USAN is really small. It's two by three millimeters, but they only go up to two megabytes. It's actually is like the physical amount of space in the package is required to put that much memory. I also thought about changing the regulators and, well, the chargeship of the regulator, you know, I could change, but I have so many of those in stock and they're not inexpensive. So, you know, even though I could have shrunk these down, I was like, well, you know, I think I'll keep them this size for now and I can always shrink them. I did go with smaller vias. These are 11 mil vias. Usually I like to do 15, but, you know, I know that it doesn't actually cost more to do 11. It's just I'm just kind of used to doing 15 mil vias. So I did 15 for the power and ground, but for signal, I went with, I think this is a 7-7 rule, which is still pretty generous. 7 mils traces, 7 mils distance and 11 mil vias. And then I did another trick. Again, I could have gone with a four-layer board, but four-layer boards actually do start to cost more money. I wanted to stick with two-layer because it's not a radio board. So it doesn't, you know, it's not a big deal. As long as I have a fairly good ground plane, I'm not worried about this chip. It doesn't actually draw that much power either. I have a regulator here that's 500 milliamps, but it's actually okay to I think go with a two-layer board as long as there's some ground plane on the top and bottom. Underneath the chip, I used this little gap area between the pads and the ground plane to put some vias in to kind of fan out the signals. I'll say if you do do this, just notice this yellow layer here. That's a white solder, sorry, silk screen. I do recommend putting silk screen because otherwise you risk the vias that they can't short against the ground pad if the ground pad extends out. So whenever I do this little trick, you know, where you kind of shrink the center pad down a little bit against mechanical, not even really thermal in this case, do put silk screen in that gap area also underneath modules. It's just a little bit of layer. It lifts, you know, separates the vias and the ground, but you don't have any issues with connectivity. Like you'll still have all the pads connect just fine. Okay, some questions. When will we put up the Eagle files for the bones? And then the JST itself, you can't shrink, but you could use a smaller JST or even a pair of pins to reduce the actual board for print, right? You could, but then none of our batteries would work, you know, and we've been talking the same batteries for like 15 years. So I don't really want to change. I think the feather standard, you know, you want to have the USB standard and you want to have the JST standard. I'd like to keep, yeah, I don't want to make something different. Okay. And what about the Scorpio bones files? Scorpio files are up. Okay. They're up on on GitHub. Okay. Next up, there's more. Were you able to sacrifice any of the capacitors recommended in the RP2040s reference design? No, I actually kept all the caps. But what I did do is I moved them down to 0402. So you'll see that this is an 0402 design. I kept the 10 microfarads as 0805 because the DC bias, you know, derating of a capacitor is affected by the size. The smaller you go, it's not like it's free to go smaller. When you go smaller, the derating increases. And so, you know, I would go to 0402 for one, maybe 0.1 microfarad, I don't mind or 0.01. One microfarad, I prefer to do 0603, but I couldn't fit it. But then the 10 microfarad, I really kept at 0805. I think that's important. You want to have your bulk capacitance is, you know, and that's the 10 microfarad. And I have like four of those. Have those as big as you can. Okay. And then I've never done a four layer board or RF board. Can you explain why four layers better for RF and which plane order would you recommend? Check your data sheet for the module or radio they're using because they'll actually tell you exactly what plane order. Usually it's, you know, signal at the top and then ground and then power and then, you know, bottom signal. But some of them have different, you know, opinions about where they want the ground plane. But usually they want their ground plane close by to the signal layer, which is on the top because that's where your traces are. RF, there's nothing inherent about four layers better for RF. It's just you're more likely to be able to get one solid ground plane close to the traces, which makes it a little bit easier to hit that, you know, if they want a 50 ohm trace to the antenna, or you have a bunch of 50 ohm traces, it's easier to match the impedance with the ground layer is nearby. There's a lot of online JavaScript calculators that'll tell you if your ground plane is this many mils away, how thick your traces have to be. And basically, you don't have to be as thick. Okay, keep going. So yeah, basically, the thing I really sacrificed is first off, I went with 7-7, reduced the vias, put the vias underneath this pad, the big diode that is the past. So, you know, the places I could really move stuff. First off, you know, all resistors also know the resistors are carrying power. So, those immediately went to 0402s. I would have actually gone to maybe use resistor packs, but turns out like it was actually a little easier to route. I find it easier to route without resistor packs, unless it's just like they happen to be nearby each other. So, I stuck with just individual resistors, you know, I feel like if I'm going to pick something up just haven't be all the same. So, the place I actually saved a lot of space is here. Like again, there's this bulk capacitance. I didn't want to change that those. There's three or four, I think there's four 10 microfarad caps. There's one on the battery, one on the V-high line, one and two on the 3.3 volt line. Here's the other 10 microfarad cap. So, I don't want to reduce those. And the regulator, I really like the RT-9080. I think that it might be possible to get it in a 1 millimeter by 1 millimeter like XFN package. But I kind of wanted to just not do that. I guess have so many of these regulators and like I feel like you actually like I kind of like them as is. I was like I'll do that last like I kind of wanted to get everything else if I could adjust first. So, this diode usually this is an MBR040 and then I change this to a PMAG 2020. I might show off how I source this on the next great search not this week, but next week. Basically, the important thing about this is I want a fairly good forward voltage, you know, maybe, you know, less than 500 millivolts at about an amp. It should pass an amp because this is a power pass, right? The V-bust line, this is USB, goes into here, goes into V high. I don't want to lose more than about a half a volt here. So, this is a 4.5 volts. V-bust can be as low as 4.5. This can be 4.2 volts. And then I still have plenty of headroom into here. And also a very small reverse leakage current because this resistor, which pulls down this transistor and keeps the battery line from interfering. Sorry, it doesn't, sorry, it turns on the V-bat, P-fat when V-bust is gone, but when V-bust is here. This is powered and I just want to make sure that if V-bust is off and V-bat is on, this current doesn't back leak into here and put voltage on here. So, you know, some shocking data is going to have as much as like a milliamp or more of reverse leakage current. You want to keep this low leakage. So, I think this is like a 20, maybe 50 micro ampere leakage, which is very small. It means that this will be a very low voltage. The place I actually pretty much, you know, so I reduced this to, this was a SOD 123. It's now a, I think a SOD 320, yeah, 323F. So, a little bit smaller, but still can pass a good amount of current. And, you know, you don't want, you can't have the package to be too small because it does have to pass quite a bit of current. Have that low reverse leakage. Have the low forward voltage and dissipate any current. Because, you know, this could be, you know, an amp easily into here, like spiking. But what I did do is I reduced this. This was originally a SOD 23. If you go to my overhead, I'll show it, the two parts that I changed. So, these two here, maybe I'll do the zoom. Can I zoom? Oh, zoom. Look at, look at how zoomy this is. So, this is that NBR040. This was the large diode. And this was the transistor. So, this is a SOD 23. So, I basically shrank those down. Those were the, you know, again, all their zisters, these are 0603s and I changed these to 0402s. But those two parts of the power path, I also managed to shrink while still maintaining their, the capabilities that I needed. So, if there's no questions. Let's go to great search. Let's go to great search. This is a great search brought to you by Digikey. I need a fruit. This is when we use Lydia's powers of engineering and smarts and decades of using the internet and digikey.com to help you find what you're looking for. Lydia, what's on the great search this week? This week, I'm looking for a smaller package for a power p-channel MOSFET that I use for the power supply on my feather boards because I wanted to shrink my feather board down to make it a little smaller. In addition, I use a SOP 23 p-channel FET. But let's look at what I used instead. So, at the computer. So, this is the original feather design that I'm going to shrink down a little bit. This is a standard SOP 23 channel p-fet. I use the DMG 3415 or 2305 pre-standard. The important things for me are they can pass a lot of current. So, at least two amps. Even in this particular case, it doesn't pass that much current. It passes maybe like, you know, a one amp peak. But oftentimes, I use this same part in a couple different spots in design. So, I want to make sure that it still maintains at high spec. It also should have a very low RDS on. Some p-channel FETs have like, you know, two, three ohms. I want this to be 50 milliohms or less. I also want a pretty low VGS threshold. So, what's the voltage required to turn on? It should be about a volt or less because I'm dealing with three-volt logic. And often the VGS is like, you know, specced at like less than a milliamp of current. So, you know, I want to make sure that I'm well above that. And again, I want smaller than the SOP 23 I've got here. So, and, you know, the other thing like VDS, you know, in this case, is 20 volts. So, basically, it's a smaller version of this. SOP 23 is kind of the standard size for small power transistors. You know, you tend to go that or maybe SOP 223. And then you get until like the power DI package sizes. But, you know, SOP 23 is very standard. So, I want something smaller, but not BGA, because again, BGA is not great for my workflow. I want something that's like a, you know, DFN or QFN or whatever, but smaller in some respects to this. So, this is the part that I tend, you know, I've used historically. You can see why there's a lot in stock. You know, it's a very popular part for many people. 20 volt, 4 amps, SOP 23. Again, you know, about 50 milli ohms, RDS on. And a little bigger. Hold on, sorry. 50 milli ohm RDS on, 20 volt during the source, 4 amps continuous. It doesn't need to be 4 amp continuous, but you know, I tend to use this for a couple different things. VGS on is about a volt or less. I don't care as much about gate charge or input capacitance, because I'm not turning this on fast. I'm not using this as like a switcher. I'm usually like turning it on and it's keeping it on or turning it off and keeping it off. So, I don't care. The capacitance effects is how quickly you can turn on or off, because you have to pull current on off the the FET capacitor. But in this case, it's not that important. Okay. So, let's go for a surface mount. So, I'm going to find similar stuff. A surface mount active P channel MOSFET. There are some specs I'm going to sort of do manually. So, let's see what we've got. My internet's a little slow, but hopefully we'll, we'll be able to look it up. If not, I will just go to the, oh, here you go. Okay. All right. So, let's look at what the specs that are important to me are. So, first, I want it to be in stock. Thankfully, the chip shortage has calmed down a little bit. You can actually get some parts. I don't want marketplace components. We'll have some not worried about. Let's see. So, VGS, that's, that was something that's important in it. There's, you can see that it quickly goes up to two, three, five volts, especially if you want, you know, the bigger transistors, but I want something again, you know, roughly about a volt or less. So, I'm going to select that. For RDS on, you can see, you know, they can get pretty high. I don't want like, you know, you can see that there's some that are like, you know, five ohm or more. I'm going to try to stick to, you know, 50, you know, the one I looked at before that I'm comparing this to was 52 mill ohms. So, I'm going to keep it nice and small. And I'm going to wait for the, this thing to update. All right. So, now we're down to like 153 options. The drain to source voltage, you know, that's how much voltage you can switch. For this particular board, the feather, everything is three volts, you know, let me try again. For my feather design, it's three volts, but, you know, I use this in my metros that are switching 12 volts. So, I'm going to, I'm still going to stick with like 12 or more. I think eight's a little low, because often I have a nine, you know, I have nine volts somewhere in my setup. 12 volts is pretty common. And then let me try again to apply. And you can see that there's a lot of SOP 23s. This is very common. Okay. So, okay, got those filters going. So, next thing is I want to filter out all the SOP 23s, because, you know, the only thing is I've got about 150 options, 150 results here. And first up is that DMG 2305. Again, that part that I would get the most of, I mean, there's tons in stock and it's low cost. But I want to not have the SOP 23 size, because it's too big. So, let's go to device supplier package. There's also package case. I think they'll do package case. So, I don't want any TO, SOP 233s or TO size chips, these are too big. I'll kind of pick everything above here, because I don't want the BGA. I don't want the SOP 23 variant. I don't want the powered DFNs, because those are kind of big. And I don't want the SOICs or T-SELFs. But I think the others are good. Let me see if those come up. Okay, great. So, now you've got some smaller chips. So, the thing is actually each one is kind of, they can't come in like all different sizes and shapes. There's a few common sizes. If you go to the right, you can see the sizes. And some of them, at least they are pretty clear. Like, for example, this one says two by two. So, it's going to be two by two millimeters. This is 20 by 15, which is smaller. That's two by 1.5 millimeters. There's micro-fat. Basically, let's look at, because now we still have like 66 results, which is less, but it's still quite a few. Let's look at pricing at 5,000 pieces. So, it looks like Toshiba has a couple really good options. Basically, you're going to pay 10 cents for this transistor, which makes sense. It's kind of a nice transistor. There are, you know, a couple different sizes, but it looks like two by two millimeter actually is not that much smaller than sub-25. So, what I did is I ended up deciding, you know, I'm going to stick, I'm going to deselect. So, I'm going to select, I'm going to select all these, because these are the sub-package types. I'm going to deselect two by two. So, anything that was 20, because I was like, well, if I'm going to go 20, I might as well go with like a, you know, it's not that much smaller than that. So, 23. So, deselect all those, and then this is kind of getting intense. I've got like 15 filters going on here. Another thing I like to do is I like to filter by quantity, and I go down. All right. So, now filtered, and now I, wait, did I actually do the filter or did I click away? Hold on. Oops, I did not actually filter. One second. Maybe we do that filtering. So, deselect the two by two sizes. Okay. Oh, I forgot to click apply all. That's why. All right. Let me let it chug. Okay. And then, so let's apply it. So, I had a couple. There are two good options that I saw, basically. So, one, I liked, this one was kind of nice up here. This is the UFM, which is actually kind of like a sought 23 size, I think. And the, hold on, let me check. There's still quite a few. Oh, I like the SSM series. That was the one that was kind of interesting. So, it's quite a few of these SSM series. They come in a few different sizes. There's also DFN 1616, which was nice and small. UFM, I think, is actually about the same size as the sought 23. But what I liked is when I started like filtering by quantity available, this actually popped up, which is the DMG 3415 UFY. And the reason I thought this was nice is that I actually already use the DMG 3415 in sought 23 in the same like situation, like it's one of my alternative parts that I've used. So, this is just the DFN 2.0 by 1.5 version, which means it's, it was basically like swapping compatible. It's just like shrunken down and has little pads. One thing to note though is compared to the DMG 3415, it's slightly less specification, like it's not identical specifications. But a lot of the important stuff is like, for example, the RDS on is nice and low, 40 milliohms. The threshold is one volt. The VGS is eight volts. The only thing to watch out for because I've used this before is that means if you're using it as an ideal input diode for polarity protection, the VGS has to be the same as the VDS. So just be aware, you can't use a transistor with eight volt VGS max with like a nine to 12 volt power supply. But again, for this feather, it's three volts. So I'm good. So I ended up really liking this one. And I spec'd it out and this is what it looks like. So, you know, compared to, this is what the SOP 23, it looks like is how big it is. And then that's the DFN. So you can see, it's like, you know, about a third of the size. The only downside is you can't route a trace through the middle here because there is, well, not without putting down like some soap screen, you know, to end no stop. But this is the thermal pad. So usually you want to keep it open. Also, this gap is pretty narrow here. And I actually made it a little bit wider. So you do have to have like a fairly, you can't use it with a rough PCB process. It has to be like seven, seven mil or six, six is better because the gap is quite small. So I'm using this, I think it's going to work. I like that it's the same, like core part number, and it has almost the same specs. So I think this should let me make this design a lot smaller. I've moved that SOP 23 and the NBR054 up here and, you know, to the right here to clear up a lot of space. And that let me move this crystal to the left. Okay, it looks like the tiny, looks like it has tiny side pads out. Possible to hand solder or is it way too small? I think you could solder this by hand, but you'd want to use a little bit of paste and hot air. Check out the last great search for where we cover how to get hot air in a syringe, sorry, paste in a syringe for doing this kind of hot air or hot plate work. Okay, that's a great search. That's right. All right. And that's the desk of Ladiator for tonight. A couple odds and ends before we go. Someone wants to know in one of the other chats, is it possible to program microcontrollers with C-sharp? Yes, I think there is C-sharp. Like the .NET microcontrollers, a programmable C-sharp. Okay. And does our Bones Scorpio board hint towards an Adafruit Feather RP2040 with onboard Wi-Fi? Yeah. One of the modules here is the ESP32 Mini, which you can see on my screen. And that would fit into that spot here as an airlift board. You can see it's actually quite nice. It has like, you know, a little bit of space here for some traces and capacitors. The only thing is, you know, you were like, oh, why can't you just use the Pico Cypress chip? I can't get that chip for the pricing that Raspberry Pi can't. I can only get the ESP32. Okay. And then last up before we go, I just wanted to say three million thanks to everyone. We are a 100% woman-owned, no VC funding, no loans, no venture capital, open source software and hardware company. And we just hit three million orders. Every single order was important and special. So thank you so much, everyone in the community, our entire team. These orders were shipped up by human beings with great pay and benefits and more. And that's one of the things that's been really hard to do. And we're one of the few folks left that do all this. So I just wanted to say thanks so much for being part of it. I reached out to the person who got the three millionth order, gave them a nice gift certificate, thanked them so much, and of course they ordered a Raspberry Pi because we make it possible for people to order Raspberry Pi. It's a real person. It's a real person. So anyways, I just wanted to say thank you so much, especially for the folks who have just given us words of encouragement and more, watched the shows and all that. Very much appreciated. And that is our show for tonight. Bye, everybody. Bye-bye.