Welcome back to The Factory, where we explore all the little details about PCB assembly and electronics manufacturing. Today we're looking at a few prototypes that I have in work, and we'll also talk about a few design constraints and a few ways to work around them.

Alright, here's a render of prototype number one. This is a nice simple one. We have a Micro:bit card edge connector up the top here. I don't have a render for that. And we have two PiicoDev connectors down the bottom. And we also bring out the clip connectors of the Micro:bit to replicate those, so that when you plug the Micro:bit in, you can still access those rings. The connector that you can't see is this guy right here. So you can see there are two locking tabs and two alignment pins. And then there's just 40 surface mount pins.

From an assembly point of view, this is a really, really easy board. There's just these two connectors, which are the same, and these two pull-up resistors, which are the same. So there's two unique parts for the machine to place. However, this connector, this Micro:bit card connector, is a little too long for the NeoDen K1830 to recognize. We've found that there's a maximum 40 millimetres that the vision system will pick up, which means that this part needs to be hand-placed.

So I think an appropriate workflow for something like this is that we can send panels of this board through to get the two connectors, the two resistors, and then we can have the belt between the pick-and-place machine and the oven stop and allow an operator to hand-place these connectors very quickly. And these alignment pins help out a lot with that process.

So to keep the manual operations to a minimum, we're just going to hand-place this part andThen send it through the oven. We still need to be able to stencil solder paste for all these pins and for these two locking tabs.

If I come into PCB New here, this is the footprint that we're working with, and I'll zoom in on one of the mechanical pads. And so you can see that there's the hole routed for that locking tab, and the pad is actually much wider on the back here than it is on the front.

And the point here is that when we stencil solder paste onto this pad, we're creating this large deposit at the back here, which is actually underneath the connector. And so as this goes through the reflow oven and that solder paste melts, it'll be drawn up, and the idea is to fillet this locking tab. It may not fillet the whole way around, but it will at least get one of the locking tabs and create a good mechanical join.

Flipping over to the back layer, you can see the asymmetry here. So on the back layer, because we're not doing any soldering here, this pad is just the appropriate size for that locking tab. And the side where we had solder paste has this much larger, kind of like a reservoir pad.

So we haven't tried this technique of assembly before. I'm excited to see how it goes. I do know that some other manufacturers use this technique for these kind of exotic parts. So there might be a little bit of tuning with, say, the size of that pad or the size of the slot. Obviously, the closer fit that the slot is to these tabs, the more likely you are to get a good fillet and probably with less solder as well. So there might be a little bit of tuning, but that's why we prototype.

So this first revision is basically ready to be sent off. I am noticing a little bit of an opportunity here, though. This SDA trace here, thisIs the data line of the I2C bus. I see a little bit of an opportunity to perhaps rotate this part and take one of those junctions out. I can bring this trace up, change its angle, and then go through the resistor like that. And clean up that leftover. I can bring this signal out, go through the via that was there before, and come back to this bridge.

Okay, so that's one thing taken care of. And I could also do a similar thing for the clock line as well. I could bring this down a little bit and bring this trace over and into the pad. And route it through the pad like that. I'm going to leave this T-junction here, but as far as I'm aware, T-junctions are not really a problem. I mean, if you really want to go to town, you can just add additional trace material to ensure that there's no 90 degree or smaller angles. So here, that T-junction has been broken into two angles of 45-degrees on each side.

Just taking a wander around the board, looking for any low-hanging fruit. You might think that this is strange that instead of just going straight up to this via, the ground trace meanders around. This is the ground label silkscreen here, and I don't want the via to be in the silkscreen. However, there is possibly an opportunity to just drag that via just in between the lettering, and the drill hole now probably won't disrupt the print. I think that might be okay. In fact, it might be better to bring that via even over here to keep that length just a little bit shorter.

And lastly, it's not a big deal on this board because I've just routed around the problem spots anyway, but I've noticed that the hole footprints that come with KiCad, they only have the clearance around the mechanical hole, which means that I could bring thisGround trace to about here, and this would be okay. This would pass the DRC. However, in general, that's undesirable because when you put the fastener through that hole and tighten it up, the head of the fastener can rub the solder mask away from that spot. It could wear it away or damage that trace if the fastener is over-tightened. So it's better to just keep traces right away from fasteners if at all possible.

So bring some clearance to this hole. This is a 2.7-millimeter hole. I'll select the pad, which is the hole, and hit E for edit. Then go to the Load and Clearance settings, and I want to give the pad that 2.7-millimeter clearance. But actually, that's the diameter. I want to apply that clearance to the radius. So I can just enter 2.7 divided by 2. We can see that thin yellow line has now come out to the edge of where the fastener head will be. If I bring my ground trace inside that area or attempt to, you can see that the drag function won't even allow me to drag that trace in now.

Well, I'm pretty happy with how this looks for now, so I'll apply that change to this other hole, and then we'll move on to the next one.

All right, here's a really simple one. This is a breadboard adapter. Simple, just a breakout. We have a four-way header that goes to our four-pin JST connector for PicoDev, and just some pin labels. So you can plug this into your breadboard and break out the signals.

The things to note here, you can see that this header has a bit of a left-right, left-right pin configuration. I think this is called tiger claw, and when you purchase these, you have to decide whether pin one starts on the left or on the right. And depending on supplier availabilities, maybe sometime in the future, suppliers haveOne but not the other. So just a good kind of future-proofing, I think, for this little project, is to create the footprint with symmetric pads. So yeah, it means that you have pads that get pasted and reflow soldered that don't actually have a component sitting on them. But that can really de-risk us for in the future, where maybe due to supply chain, we can only get one instead of the other. This footprint will just accept both.

And I was a little surprised to find that when you order these on tape and reel, this is the size of the reel. Look at this thing. It's so big. This is a 56-millimeter tape. A 56-millimeter wide tape for a full... Let me see if I can find one for you. Look at that little guy in there. You can see that little plastic cap. That is the plastic cap that goes over the pins so that this can be picked up by a pick-and-place machine. And it's just in this enormous tape.

I mean, obviously, this makes sense for the manufacturer. They're designing for manufacture, so they're just going to put all their headers into a wide tape, and this can accommodate, I don't know, maybe like a 20-pin header. Then they have to make one reel of tape, and it can be used for all their parts. Makes a lot of sense for them. For us, not so much. This is a very wide tape, and that means it needs very wide feeders.

So, I mean, not only is this quite wasteful, like look at all the packaging for just these little pin headers. Those feeders are very expensive because they're not in very high demand. So here's our answer for that. This is a prototype tray that we've made for a larger header. This is a 20-way, but the idea is the same. If we can, like, laser cut or 3D print or some other manufacturing method, if we can makeA tray for this little 4-way header, then we can get the headers in, say, bulk or shipping tray packaging, and we can just load them manually into our precise pick-and-place tray. That means we don't need to buy a 56-millimeter wide feeder. I think those are like, oh, like $1,000 or $2,000.

So for the last board, we were talking about routing and angles, and you might see a bit of a red flag here. I'll go to the back layer. Look at this. So I need to do something about this right here. This is the scenario that I was talking about where if you have an acute angle, that can build up deposits of etchant, and that can slowly over time degrade the track or even while etching make the track much smaller. All I'm going to do for this is drag that out like this. I'm going to keep it as this straight line. Oh, no, there's still a sharp angle there, though. There we go. I think that's appropriate. Yeah. I'll do the same over here. That looks okay. And these remaining two look okay. I'm just going to leave these as just straight routes. Keep it simple.

And the last one for today, this is a shield for the Arduino Uno R3. So again, using our same pick-and-place idea, we're going to do a single-sided load of parts on the bottom side of the board, and that'll allow us to place these four headers. We have a 6, an 8, an 8, and a 10-pin header all loaded onto the underside of the board to allow the shield to go onto the Arduino Uno.

So to keep things as a single-sided load, that means we need to put all the other parts on this side as well. So that's where the PiicoDev connectors will go and some logic-level shifting circuitry because PiicoDev operates on a 3.3-volt I2C bus and the Arduino Uno is a 5-volt device. So we just have aLittle bi-directional logic-level shifting on board as well. And returning to the top side, you can see there are positions for stacking headers, but these won't be populated because these are through-hole parts. But it still leaves you the option to stack on top of this shield.

And so what do you do with the rest of a whole shield? I mean, we only need this small area for our PiicoDev breakout. We're going to put in a large prototyping area. So this has a bit of a breadboard-inspired bussing. So these groups of three are all connected in groups of three. There's a bit of a power bus in the middle, the 3-volt bus over here on the left, and the rest is just prototyping area. Also a couple of 2.5-millimeter holes for mounting a PiicoDev module on top of the shield as well.

Because all these headers are smaller than the maximum size that our machine can pick, these can all come off similar trays as I already showed you before. So we can make a parametric design with this grid and then just change the parameter for the length of the tray and make a tray for each size.

Thanks for joining me on this little fireside chat about little considerations like the maximum length of component that your machine can recognize. Working with usable-size packages is all little lessons that we're having to learn along the way.

So there you go. Let me know what you think of this format of factory video, where we do a little bit more of the behind-the-scenes work and look at how these designs come together. If you have any advice on how some routing could be improved or neat tips and tricks that I'm not aware of that we can start using, I'd love to hear from you. Just open a thread in the Core Electronics forums. But with that, until next time.Thank you for watching.