 and welcome to Desk of Lady Aida. Hi everyone, and welcome to another Desk of Lady Aida. It is Sunday, fall is nipping at our heels. It was a little rainy and cloudy here in New York, so we actually stayed in and did a bunch of engineering. Woohoo, so it's time for the Desk of Lady Aida. Yeah. So we can check what I'm up to. I did a whole bunch of stuff and I got a lot of samples in, but I thought maybe we'd start with the samples. Let's jump right in. Okay, so a couple weeks ago, oh, let's go to the overhead. We showed off the design for this infrared emitter that you can plug in a JSTPH2 pin cable in two and it'll turn on and be very bright and for an emitter and then we sourced the IR LEDs. Well, the final PCBs are now fabbed and thanks to Penguin. Let's see if I can get this to talk a little better. Okay, thanks to Penguin. It's got a beautiful silk screen with Futura Heavy. You can see the tune for it LEDs and then I added a spot for an optional five millimeter LED if you want to solder one in. Like if you were like, okay, I need it. I want a third LED because I want like full 360 degree infraredness and I also did the tester today. That's actually kind of experimenting. You know, it's like, I like to show when things don't work out because like there's a lot of times I do stuff and I'm like, hey, I had this brilliant idea and then it's like, that turns out not so brilliant, not so idea. But you know, we make these pogo pin testers and you know, we have to solder in these pogo pins that the P75s are kind of my favorite, the crown or spear tip pogo pins. The thing that's really annoying about these though is like you solder in each one and they like, you know, I like to give a little bit of give on the hole. Also like when you get holes drilled in your PCB for pads, like you know, you'd say 39 mils or whatever, but there was a little bit of variance. And also sometimes they expand them or contract them. And so even though the pogo pins fit fairly tightly, they do tilt a little bit. And so like we have to spend a bit of time getting them nicely centered, which is annoying. And I was like, oh, you know, maybe I could use these pogo pin strips that we got in the shop. They come in strips of nine. And I thought, okay, I could cut them down. Maybe I'll show, maybe we can go to the computer real fast and I'll just show it in the shop. So that's what I'm talking about. These are the normal pogo pins. And then I had these like kind of pogo pin headers. And we put these in recently and I was like, oh, you know, maybe like this could be a good alternative because, you know, oftentimes I have to measure or connect to like a strip that is on a feather header. And it's like, you know, 12 or 16 pins in a row. And so it's like, well, maybe these will be ready to go. Like they're, they come in a plastic strip already. So they're pre-aligned. They're at a point one inch boundary. But then I, what I didn't realize is I forgot to think, oh, the pogo pin ends are less than, they're, they're not as, they're not wider than the holes that we use. So if you go to go back to the computer, sorry, back to the overhead, you know, I designed this. So it's like, okay, you've got these holes here that kind of help you align. But these actually go straight through the holes in the PCB, which, you know, if you hold it down, it's fine. So, you know, we've got this test, this tester here, plug it in. It does work, you know, but it's not, I doesn't have that nice feeling when, if the pogo, it pushes up against the PCB other than going through the hole, it's a kind of a better feel. So it would be good for something that didn't have any, any through-hole holes to test, only had test pads, but like, you know, we tend to use a lot. Another thing is, you know, as you can see, it's, it's flickering because, you know, the connection isn't so good, you know, if you hold it down and you can see the IR emitting and detecting, it does work. It's okay, you know, what I might do is when I'm like quickly putting together a tester, maybe I'll do this for like the first one, but I think it's not a long-term solution. So I might try to find some pogo pins on strips that have like a crown head, although I'm not convinced I'll be able to find one. So this experiment, I tried it. So that's the IR emitter. Also put together some prototypes. So, you know, we've got these little boards that we use for making a little power supply. So this is a TLV62569. It's a 3.3 volt, one amp buck converter. But if some people remember a couple of weeks ago, we talked on the show about replacing the fan 5333 boost converter that I use for biasing OLEDs. Sometimes I just need like a high voltage to get 12 volts or something reference, not like high current, maybe like 10, 20 milliamps. So we've got this PCB with the TPS61040, which is one of the options we found. I don't think you can get any closer, but you know, it's two to six volts in. You know, for this board, I just set it to be 12 volts outputs. I don't quite get 100 milliamps. So it's kind of my guess. I was like, well, you know, we'll see what I get. It turns out it's actually kind of limited of three volts. I can get 12 volts at like 40 milliamps, five volts in. I can get 12 volts at, you know, 80 milliamps out. So, you know, good for like an LED strand or maybe like a very small device biasing audio, you know, you know, if you need plus 12 volts for an op amp rail, that's not real to rail, you know, you want to get a high reference voltage with some current. So, you know, I tested on the load, electronic load that I've got, which I always recommend, you know, at electronic loads, you know, before I bought them, like it's kind of ridiculous to spend like a couple hundred bucks on this thing. That's just like a big resistor, but actually it's really, really nice to have a digital load when you're like, I really want to just test the power supply. You know, you've got your bench supply giving you your current in and the load, you can dial it in, you know, any milliamps and tweak the knob while you watch on your scope to see that the power supply stability. So I got this like BNK one, the BNK precision. It's the 150 watt. It looks like the 8540 power supply. So I do recommend it. And then I did another STEM board. I thought I showed this last week, but maybe I did. Maybe I showed it only on the ASCII engineer. So this just a little prototype. Also, you know, similar idea behind the IR LEDs, which, you know, I just dropped. You know, you have a JSTPHN so you can easily connect it to, you know, wearables or, you know, a micro bed or a circuit playground or you can plug it into a breadboard, but it's freestanding. And this is a little motor driver with a three or four amp end channel, FET, flyback, diode protection, LED indicators. I'll quickly just plug this in to test it out. Show it off. So you have three to five volts input. So let me see. Get this back up a little bit. This is five volts. And then this is ground. So this is just a common thing, you know, we're often like, somebody has a project, like I just want to like turn on a motor or a solenoid. I'm like, uh, you gotta like wire up your ULN 2803. And it's kind of a pain. So this is, you know, it just makes it a little easier. Uh, you know, you see the LED tells you, this red LED tells you that it's powered. And then it just, uh, it just turns on whatever it is, you know, up to three amps. So good for, um, you know, anytime you want to PWM or just control a motor solenoid is something that you just need that transistor and a flyback and maybe you want a couple of LEDs to indicate. So the MOSFET is a three volt or five volt output. You can do three, it's actually up to 12 volts to be honest. It's, um, the signal can be low. The power can actually be as high as I think 20. I think I use the, uh, DMG 3415. Which might be a, I think VGS can be 12 volts. I think maybe six, 12, six or 12 volts. And then, yeah, so the control signal can be that high. And then the power I think is 30 volts or something. It's up to three amps, very low RDS on. So, uh, you know, you're, you're pretty much free to do whatever. Okay. So now to the meat, no matter how much time. Okay. I'll go pretty fast. So, so again, I showed like this little tester, um, which, which I call, we call it drop and beep tester cause you drop the board on and it beeps. And it takes like, you know, it's less than a second and a lot of I squared C sensors are like that. You'd be surprised. You drop a BME 280. You can initialize it, do an initial reading, verify the voltages in, you know, two to three seconds. It's very, very fast. And so, uh, there's a little indicator LED and a beeper. And so people just test very quickly. Beep, beep, beep, beep, beep, beep. They can go through an entire panel almost instantly. And that's, that's great. When I'm doing more advanced boards, um, like let's say a, uh, let's see what I got here. This is a thing of feather. I can get a feather at four, which I don't know if I'll ever make again because I can't get these chips. But let's say I could. So to program these, um, historically I used a Raspberry Pi. So I'd have like, you know, Raspberry Pi computer here. And I would use a, um, open OCD or J-Link and I'd program it using Linux, uh, to run the commands to do the SWD programming of, of the chip. And then I would test the USB enumeration. And that was fine. Except the issue is that, you know, Raspberry Pi is like, they're just not, they're, they're fine as testers, but they're just not as, they're not super fast. Like there is a lot of operating system stuff that has to happen. I mean, with the Pi four, it probably would be as bad. When I was first doing this on the Pi one, it, there was some, it did take, you know, a minute or two to kind of like get open OCD running and you have to launch it and send the commands. And, you know, if I did bit banging it would be slower, but if I use a J-Link it was like 600 bucks. So what, um, we did, um, is a couple of years ago, Tak and I, um, who will, to USB, we ported the SWD, I think maybe K-Town did this too. We ported the SWD programming protocol to Arduino so we could use a board running Arduino to SWD program a chip. And, uh, we got the Sanbi-21 going, the Sanbi-51, NR-52, NR-51, we did actually very recently, um, and the RP-2040 we can do with, with mass storage, but for SWD programming basics of making these Trinket M-Zeros, like you, you can program an AVR chip with SPI very, very easily and you don't need very complicated hardware. It's going to be a little bit weirder, but as I started to get to Sanbi-21s, I was like, I can't have a Raspberry Pi computer for every time I program because it's very also very easy. Somebody knocks the Raspberry Pi and the USB cable comes a little loose and then I have to reboot or it gets corrupted. I wanted a very solid state solution. So we came up with this idea of using, um, you know, one board running Arduino very quickly to BitBang SWD to program these boards and then connect over USB and then just make sure that the USB enumerates because if it does, it means basically you programmed everything correctly. And so that is how we came up with the, the teensy based tester brains idea. This is actually a fresh one, but basically you've got a teensy 3.6 and the reason I like the teensy 3.6 is because it has USB host, which is important because again, I have to USB enumerate. And as an SD card slot, which is really nice because then the firmware I'm burning is on the SD card slot. So people can change out the SD card very easily. It's like, oh, you know, update the firmware, update the program or the test, just change the SD card. They don't have to, you know, the people running the testers don't need to know anything else. So that's all great and wonderful. And I've, you know, we've made like hundreds of these, but there's a problem which is, you know, there's no teensy 3.6s anymore right now. And it's not clear when we'll get more. And I kind of had a stash, a stash that like about a year and a half ago when this started happening, I was like, okay, like I'm gonna, you know, keep a hundred for us so we can continue to make boards. The problem is that we're out and the teensy shortages isn't quite over. And I was thinking, you know, I could port the code to the teensy 4.1. It would probably be okay. I did make a couple of tweaks to how USB works to optimize for the way you do testing. But I also thought this was a really good opportunity to maybe diverge because even though maybe teensy 4.1s are kind of available right now, I just sort of learned my lesson, like not to depend on a dev board that I can't control because it's like, if I can't get teensies for a project, what's one thing? But if I can't run the AFruit factory because I can't get a dev board, that's kind of a bad scene. So yeah, the teensy 3.6, you've got the USB host. You've got a, you know, this one doesn't have the power transistors. There's a couple of power transistors here to control some voltage outputs, reset button and 16 by two LCD. And this is actually all through a hole put together. We just hand solder these whenever we need them. And so we've got like a big collection of these teensy brands. But as I mentioned, I wanted to design something different. So let's go to the computer. And this is the original. And like again, this was worked wonderfully for years. And I would still be using it if I could get these boards, but I can't. To think that's important is, you know, I have a bunch of standard test pins, I squared C, VBAT power out, V measure in, ground, target reset, SWDIO, target, SWD clock, target. Number 28, which is kind of like an analog input slash extra GPIO whenever needed. You are in a three volts power line. And so over the weekend, I kind of designed a totally new board. And one of the things I asked TAC to do is there was a Japanese person or some person who spoke Japanese who wrote PIO code for the RP-24 to do USB host. And it's not amazing USB host, but it's good enough to do like CDC and HID, it seems, and enumeration, right? Which is again, if the board enumerates, I consider the USB working. I don't need to like actually transfer data on the channels. If it shows up and I get the right VID PID, I'm like, great, the USB connector works, the USB pass the resistors, the crystal, everything. You know, you have to get a lot of things working right in order to enumerate USB and identify the VID PID correctly. So the RP-24 doesn't have as good of a USB host stack as the TC-36, but you know, if I can move 50 to 75% of my boards over to this, then I can scrounge the TC-36s for when I need them. So this is a cowbell. And another thing is as I was doing this board, I was like, you know what? I'm a little tired of us hand soldering these. What if I make it surface mount and then we can just do a run of 50 whenever we need a batch of boards because each product that we use as this tester brains, I need like three testers, like three, you know, cause we often have three people testing at once or we have backups. So I thought, you know, instead of hand soldering these, which also sometimes there's a loose solder joints, let's put it through the pick and place machine. So this is a Raspberry Pi cowbell connector. So, you know, you put your Raspberry Pi Pico RP 20 here and you know, it's just very low cost available and there's like, you know, people have made versions. So even if the original Pico isn't available, there's others, if I have to do some wifi test, of course, you can use the Pico W. It is not the SD card built in. So I attached a micro SD card here. There's a lot of pins. So this is not a big deal. And then I connected a USB type A jack to two pins and you checked it out and said, yeah, as long as the D plus and D minus just have to be next to each other, you can use any pins. And then down here, I've got the LCDs. This is the large LCD on the original, and I use a surface mount connector for that. On the original tester, we have so many pins and again, the SD cards built in that I didn't have to like try to do some pin saving tricks. But on this board, the RGB backlight actually didn't have enough pins for it. And so I actually put down a WS-211 and Neopixel chip, which we stock just the SOIC version so I can Neopixel control the RGB backlight. And then I actually also was like, well, sometimes I might want a TFT, I need more resolution. So the 1.9 TFTs that we use, kind of bar style, it's a color TFT, I put down a connector so you can use either. You can use either the TFT or you can use the LCD. Then the LCD is more legible, it's simple, but sometimes it's like maybe we need to show more data than two lines of text. So that's an option, reset button. This is the power control for turning on and off the USB five volts and then the VBAT pin. This is my little like, I always mess up the arcs and TX, these lines here going into the microcontroller board that I'm testing, I always mix them up. So I just put a switch here and I just, it's like if I get it wrong, instead of having to redo the tester, I'll just flip the switch the other way, it'll swap the two arcs and TX pins. So it'll always be right one way or the other. One two hole piezo piezos can't be surface mounted, they are done by hand. And then tack was like, oh, put a semi-QT connector on there, on the I squared C lines. You never know, maybe some other sensor or device would be handy to have controlled here. So I had a little spot and I added it. So this was just me like kind of messing around. I also put in a 18 pin FPC cable that has all these test lines. You know, I think for now, I do like to have big chunky wires going between the tester brains and the tester board, the device under test. But you know, there could be some future where we use a small flex cable instead, just to simple, so it's less likely people twist wires or anything or they come loose. We ever consider making the RP2040 tester a stocked item? Love ATFT USB A breakout and extras. I don't actually think so. I don't mind publishing the files but I don't think I would fabricate and sell this because it's- Who would test the testers? It's first of all, it's hard to test the testers. Because we don't test the testers. Then we'd have to test the tester before we could test the tester. And second, it's a very specialized piece of hardware that's just for me. I don't want to have to deal with if I revise it, I have to tell people and then there's some guarantee of support. So you know, I'll probably publish this online. I think if I was going to make a board for like USB host, it would be different. I'd make like a, you know, a Pie Cowbell or something rather than this kind of monstrosity which also might have mistakes in it. So anyways, this is the tester. So I just, you know, I kind of routed it over the weekend. I think I have some really good improvements. You know, I use every single pin. I love the RX-TX Fliprk, it's so embarrassing. I love that I added a STEMI QT SD card and the TFT option instead of the LCD for some situations I think is going to come in very handy. And then we have to port, you know, the Arduino code, make sure that the library for the SWD programming still works on the RP2040. I think it will. But you know, the RP2040 is very fast. The reason I went with a TNC36, of course, it's incredibly fast and bit-banging. You know, SWD, you can bit-bang really fast to program a chip. And it actually makes a difference. Like if you are bit-banging at like 10 megahertz versus 100 kilohertz, it's going to make a big difference when you're transferring, you know, you do the chip erase cycle, sure, but then when you actually have to blast that firmware over. So if you have to do a read back, right? Because you not only have to write the firmware, but you have to read the firmware back. That's where I think the speed of the RP2040 and maybe we could PIOFI the SWD because it is, again, this is weird bit-bang, you know. It's not a classic interface. It's a clock with one IO pin, it's bi-directional. Excellent. Okay, the request. And an additional one is PyCalvel USB host. Okay, fine. Yes, maybe we'll do that. Okay, so that's what I've been working on. So that leads us into the great search. Here we go. The great search brought to you by Digikey. And for this time of the week, every single week, one lady to use their power of engineering to show you how to find things on digikey.com. Lady, what is a great search of the week this week? This week I'm looking for a surface mount USB connector for this tester that I've designed. So let's go to the computer and I'll show the part in question. So the original board that I'm revising had a through hole, had all through hole parts, which was, you know, great and everything, but these had to be assembled by hand. And I kind of, I'm a little, I don't want to do that. I want to, you know, make 50 of these. Once I get the test design done, because we use these in so many boards, I want to just do a run of 50 and not have to hand solder them anymore. So this is like the classic through hole USB A connector and it's great and everything. But for this new version, again, I really want it to be as SMT-ified as possible. And I did have like one, you know, you can see this as an SMT-USBA. And I'm certainly going to go with this connector. This is just the connector that was in my library because I'd used it before. So I put it down as a placeholder, but let's go and find a possible, some possible alternatives for the through hole USB A in surface mount on Digikey. So let's go. Well, not translate, but Digikey, not Digikey. Okay, so let's go to USB type A. That's the thin, long one. You're like, why don't you use USB C? I don't actually know how to deal with the host side of USB C. I just want to do USB A because it's like, I know it. I just want ground, D plus, D minus and five volts. Okay, a lot of options, connectors. Here you go. So I want USB connectors. It even has the photo of the USB A connector. So I want connector assemblies. It was about 3,000. So let's take a look. Yeah, these look right. I mean, this is actually starting to look like USB A's and all sorts of shapes and sizes. This is kind of a cool vertical USB C, classic USB B. All right, so let's start with, we only want the active products. Okay, and then we only want one port. So that's a really easy. Second filter. Next up connector type, right? Like all the connectors that are sort of similar, DVI and HDMI and ethernet, whatever, they're all kind of in the same area. And there's, you know, there's micro B, micro A. I've actually used like micro A, but I really do want to have just that standard, chunky, classic type A connector. I don't need to save space. And moreover, I want to be very user-friendly because people are gonna be using it, might be like, what is this weird, like micro A, B is like a very weird connector, to be honest. Okay, so I selected the A, there's a couple versions here, but I selected all three of them. I'll also pick up the dash. Okay, and then we're down to 600 components. Number of contacts. So actually, no, we'll get that later. Okay, plug a receptacle. We want receptacle because we're gonna plug in a USB cable into this. Cool, see how we're doing. Okay, yeah, we still have some good options here. These are all looking about right. After I do a couple of searches, I like to go down and just make sure, like, am I on track? Okay, so let's now do the mounting style because that's what you want, surface mount. So there's board edge, cut out surface mount. Select those. We don't want free hanging or panel mount and then surface mount, through hole. Now, a note, if you see through hole after surface mount, it's still mostly surface mount, but it might have a through hole board guide or tab or something. So it's okay as long as the surface mount comes at the beginning. So let's apply. Okay, now we're really, we've really cut it down to 85 bits. Okay, so the next thing is, some of these are USB two or three, sorry, USB three or four or whatever. And so they'll have more contacts than the classic four contacts. I really want just four contacts. I don't want anything special. I don't want USB three. So let's do that. And then, those are like a lot of them are in stocks. That's kind of good. So let's look at normally stocking. And it's excluding marketplace. So you get down to 45 options. Okay, cool. So next step, these actually look pretty good. So there's a couple of things here that I noticed when I looked. So you see how this has like a little, like a hooky thing. So even though this is surface mount, I feel like the pick in place is gonna have a lot of trouble getting those springy bits to push through because there's, you know, they push through and then they spring and hold up against the PCB. So this is surface mount, but it's, I think it's, I feel like it's hand assembled. I don't know, we might not have to be where something like this is definitely surface mount only. Like there's no through hole sections. There's also a few that have something like this where if you see the plastic, this is literally going through the board, which I don't want. This is what was categorized as board edge cutout. You know, the board itself has a rectangular slot. This means that you can have like the, you know, the connector in line with the PCB, which is kind of sometimes for enclosure design, it's actually quite nice. Because you have a clamshell with two halves cutout and then, you know, the USB goes in the middle and half is on each side. Honestly, I don't, you know, for my, for my uses I don't want that. So I'm gonna get rid of the board edge cutout options. And the next thing is I do want to pick and place this. So I'm going to go only with cut tape, did you reel and tape and reel. A bunch of these come on tray and I'm not gonna, it's like, no, I need to just, I don't want this to go through the pick and place machine. So let's apply. Okay. So finally what I did is I actually, you know, some of these were good. So this is actually kind of what I have already now, but I actually looked at quantity available because this gave me a good idea of what was a good, well-supported one. And so again, this is close to what I have on the board itself. The issue is, is that, you know, there's a lot of leverage there. It's very easy. Somebody pulls on the PCB and they pull on that, the PCB or the, on the connector and those USB type A connectors on cables are like amazing levers. They're like long and they're like extremely strong and they're molded. I feel like somebody just yank and they would just yank the, you know, the pads right off. This one's a little bit better. It has two sets of pads. So I feel like this would, you know, if you really wanted to do service not only this would probably survive better. Again, these are good, but I just don't know if the pick and place would be able to get this through to the PCB. So what I really liked was these emfinal connectors, which were also very popular. And the reason I like these is they do have a through-hole tab, but the tab doesn't have like a massive kink in it. It's like, I feel like it could, it's, it'll, it'll push through and, you know, you can even make the slots a little wider. So maybe there isn't anything in the way, but then you have a paste on the pads and it kind of solders it in place, sort of like how surface mount, USB type C connectors work, right? You have, it's pick and placed, and then you have these slots. So the slots, you know, you have paste on top of them and they just give a little bit of mechanical through connection to the PCB and gives it some strength. So I really liked this connector quite a bit. And what's interesting is if you look at the datasheet, they actually have two versions. They say, look, if you have a low-cost machining process, you can drill a hole. Look, if you can have a hole that's plated, but it's only good for like 1500 cycles, but they're like, look, if you can get a slot, an actual slot, you'll have less leeway. This is good for up to 9,000 cycles or if you use, I guess, this other part, 12,000 cycles. So I thought that was actually kind of interesting. They actually gave you two options for how to have footprint for this USB connector with different reliability. Either way, I don't, you know, I'm happy with 1500 cycles or I'll do the slots, you know, just for the heck of it. But I rather like this parts. This was my pick. It's actually expensive either. It's, you know, 75 cents, less than a dollar in quantity and comes on a tape and rail. That's a great search. Okey-dokey, here's a question. Is shield always counted as identical to the ground pin for the contact count? No, that shielding is usually separate and you don't necessarily want to connect it to ground. There have been a couple of good app notes which I can't recall off the top of my head, but with USB in particular, I do not ground the shield. And sometimes you would put a ferrite bead on the shield, perhaps, but I usually do not ground it. And that's, I think, pretty standard behavior. But I Google for like USB shield grounding to find out why. I think it gets grounded on the host side not the device side. All right, so you'd also want to search for five pins there. Well, the USB-A connectors only have four pins. That could, again, it's the contacts or the inside part of the USB, not the shielding. Okay, all right, everybody, that's our show. We'll see you all throughout the week. We have a bunch of cool stuff planned. And Morris, you online, thanks for spending your Sunday night with us. Thanks, everybody. That's Morris Desk of Lady Aida. Bye.