Hi welcome to the Eev blog an Electronics engineering video blog of interest to anyone involved in electronics design. I'm your host Dave Jones Hi The other day on the Ow I mentioned these uh little prototype Pcbs I got from a company called Ited Studios Itad Studio.com and how I thought they were real. Great value for money and almost game changing and I was raving on about them. Hey, 32 m mm x 32 mm PCB double-sided uh solder mask both sides, silk screen both sides plated through holes uh 66 uh th design rules.

Nice boards and I got 10 of them for 16 Australian dollars delivered. You got to be kidding me I thought this was the greatest thing since sliced bread. Awesome. But and I and I tweeted that oh, you know they're they're pretty darn good value.

A few little minor alignment issues with them, but not a big deal. Man thought I'd struck gold with these things and I posted a photo and Well Turns out I was a bit off the mark because I had only given these a cursory glants before I started raving on about them. So I went and built one up today, sold it on the components, powered it up and it didn't work and well, you know Murphy's right happens every time. Circuit never works.

First go great. I get to do some troubleshooting so let's go for it and well, no, I went through all the problems and it turned out to be a problem with the PCB manufacturer. Let's take a look at it so let's go through the problem, shall we? I've got a little block diagram of the system that I was trying to get working. This is the PCB.

This is my new little prototype circuit. Doesn't matter what it does, totally irrelevant. Okay, but it's got a little AVR 0, Mega micro on it. Okay, and it's got a Uh RS 232 uh type uart type interface which goes to one of these little um external Ftdi.

Um, just you know. I've got one of these spark fun, um serial U interface boards, right? Really quite neat. Hooked up to the USB to the PC and it's supposed to talk through to the AVR and uh, I was able to program okay. I I was able to program my AVR okay using the onboard Um in circuit serial programming header.

Not a problem. Okay, so everything was looking quite sweet and I was quite happy with it, but it just wouldn't talk to the PC and ah, you know, I was scratching my head thinking oh, I've screwed something up maybe I've swapped the RX and the TX classic uh design mistake which I still make these days I still goof it up Murphy gets me every time. but I double check, triple check. It wasn't that and I started to look at.

You know the usual problems: a shorts on the board. you know, components in backwards. you know, incorrect components. whatever.

I went through all that. Not a problem at all. So I'd gone through pretty much everything I could without actually getting the scope out and actually probing around. So that's what I did I went and got my scope.

It was just a little, uh, standard bench scope down here and I hooked it up to a ground here. The board's got like another little pin interface on the side of it like that. so I hooked that. That was rather convenient.

So I hooked the uh, the uh oscilloscope probe the ground uh, point up to the ground there and I started probing around. but I didn't have to uh probe around very long because the thing just magically started to work and then I went D I know what that is instantly I knew it. I've seen it dozens of times before. It's a classic ground path problem because once you hook an oscilloscope well, there can be two things that happen when you, uh, hook up an oscilloscope and your circuit suddenly magically works.

Either a it's a ground problem because you hooked up the ground lead on there and you've got some ground system thing happening or it's uh, actually, uh, loading. either capacitive or resistive. uh, loading, but usually capacitive loading on the point that you're actually measuring. but I knew it wasn't the ladder.

it wasn't the loading of any point that I was actually measuring. it was the ground problem. And that's just a classic error that I've seen time and time again. and I knew instantly that the ground pin between the Ftdi uh uh, uart interface here and this one over here was broken.

Oh, and by the way, this board. this uh Ftdi board was actually supplying the + 5 Vols for this board under test because the board didn't. You know it doesn't consume much at all. So I was able to power it from the 5V interface directly on there.

and one of the things I did when I was troubleshooting this thing before I hooked up the oscilloscope. of course I measured that I was getting 5 Vols there and I was and I measured it directly on the pins and it was no problem at all. but I didn't think to actually probe any other point. So I knew this was one of those classic system ground problems because I didn't even have to uh probe the circuit.

All all I had to do was hook up the ground on there and all of a sudden it started working. It talked, no problems at all. I was talking to my AVR from the PC Excellent, Not a problem. Take the ground, the oscilloscope ground clip off this pin here, take it off and it stops working.

It just dies. So I knew I didn't have connectivity on the board between this ground. Point Coming in here and this one over here because an oscilloscope is actually grounded like that. A bench oscilloscope.

okay is unless you've got one of those, uh, fancy isolated ones. uh, you know, tectronics and many other ones. or you're using an isolated probe. the BNC connector on your the BNC input connector on your oscilloscope is actually connected through to Main's um Earth that's that's a ground symbol.

I should probably do the uh Earth symbol like that and because I had it hooked up to a PC as in a desktop PC They also are Main's Earth and the USB connector and the other connectors on there. The shell of that and the ground and the circuit ground inside the PC is connected to Main's uh Earth as well. So instantly you've got that connection going back through your PowerPoint through to your powerboard and coming back. and if you got them on different circuits, it might actually go all the way.

You know, if you got a house here, draw a little chin. It's not a very good house, but anyway, it can go all the way through your house, but it will eventually joint back down to your uh, common Earth Point down here. but they will actually connect up like that. and that provided the uh ground path needed to complete the circuit and that's why it instantly started to work.

The signal Integrity is going to be complet and out of crap because it's going to go all the way through your powerboard and back out. It's going to be awful. But because this isn't you know U interface isn't particularly high speed. it was able to work no problems at all.

So I was asking myself how could this be because I know I did a DRC a design real check on this board before I sent it out and there were no breaks in the ground at all So I knew I hadn't goofed up my PCB loud I knew I didn't even have to go back and check that because you know DLC it, you know you r on it and it works. Okay so I knew they were connected. so I thought there must be a PCB manufacturing issue and I thought okay one of the Vas is you know is I got a problem and it didn't connect through. Okay, it happens.

Now there's one interesting thing to note here is that I was using the desktop PC which is uh, they're all Main's Earth reference but if I was using my not notebook uh PC then it wouldn't have shown up instantly when I connected the ground probe over here cuz notebooks aren't Main's Earth reference. they're powered from isolated plug packs and if you're powering them from batteries uh as well, then it's going to be absolutely completely isolated no doubt about it. so it wouldn't have shown up like that Eureka moment as soon as I connect up the uh ground point I wouldn't have seen I would have had to start probing around more and you know it just would have led to some weird effects that uh weren't instantly recognizable when. I hooked when I actually completed the ground here with this circular uh system ground.

Loop So just something to watch out for. It's always a track with young players. Don't ever forget to check your grounds! I Decided to put the board under the microscope and here's what I found. So here's a microscope photo of the actual it board that I got delivered down the bottom here and on the top is exactly the same part of the board in the original Gerber file that I actually sent uh to them to manufacture the board.

Now normally they would, uh, take that Gerber file and they wouldn't alter it at all assuming unless there's something wrong with it they wouldn't touch it. They would just use that to manufacture the board now. um just some Basics about uh Gerbers like this, you'll notice that there's no holes up here in inside the actual pad like that. Now that's uh, that's common.

You will not find holes in the Gerber files themselves because they're specified in a separate NC drill file. That's why this uh via over here like this. um is not. You know you can't actually see it up there.

Uh, it's not that it wasn't in the Gerber file, it's just that it's specified uh in the NC drill file instead. Now, Uh, Now the real problem here. Okay, as you can see, there is, uh, copper in in my Gerber file. there's copper going between these pads.

This is what's called a flood fill I've I it's it's actually connected to ground and I've flood filled all of my board with this uh, grounded copper and it snakes its way through all these pads and up through the V's and all over the board. Okay, and that helps. Um, actually get the ground connectivity on the board because um now this ground Point down here. Okay, this one down here is actually the one that I hooked my hooked my oscilloscope probe up to and that I mentioned.

Okay, and as you can see, it does go through, it is plated through and goes through to the other side. same with this one next to it. But uh, because this boards particularly uh, dense on the other side is a very densely routed, there wasn't room to actually connect that ground through to somewhere else. So I've actually got to rely on the fact that um I'm using the flood fi it's It's got to go around here like this.

It's got to go up. It's got to go through here and up through this little narrow part here around here and through this, uh, via here onto the other side and then to the rest of my circuit. now. Um, so I'm pretty much Reliant upon this part actually going through here and this part going through here now.

Um I've this the amount of copper going between these pads is actually quite large. It's actually um, eight th which is quite quite a wide amount of copper. Now the um, uh, because that's what 8 th what was left over after I Used the It studio Uh design rules which were 66 which means that um I can do a six th trace and a six th space. Now this Trace down here is actually six th wide as you can see and they've produced it.

They haven't touched it at all. and the clearance around the pad here is also 6 th. So I met their design rules so they shouldn't have to touch this. uh Copper at all.

They shouldn't have to alter the Gerba files in any way. they should just actually produce the mask and and produce the board and that's it. But somebody at it has decided, no. We're going to take um, all of the ground plane and we're going to expand it.

We're going to expand the uh, You know this. The spacing around between the Um pad or the Vaa and the ground plane isn't enough. We're going to expand it even further and look down the bottom here. Look at this.

You can see this and there's actually a break in the copper there and it's left. What's called, uh, dead copper. Soort of this is here is an example of dead copper. Okay, it goes through like this and it's just floating now.

Um, it's not connected to any other part of the circuit. That's why they call it dead copper. Now you know that's not usually a big deal unless you're talking really critical. um.

circuit design. Uh, but in this case I was riant upon my ground going from my ground pin all the way through here, up through here, up in into this. Uh VI up here. But look right there.

Here's the break. there. it is right there. and that is what.

Uh, stopped my circuit from working because there was no electrical connection between the input. This is my uart. um, input here. the the ground and the 5.

Vols and and the transmitt receive come through here and in the ground. The the 5 Vols and the transmitt and receive went through to the rest of the circuit. But the but the ground didn't because there's a huge break down here. Now why they've done that? Why they've decided to expand that um, copper around the pads and the Vs I Have no idea.

It's crazy. Anyway, they decided to do it and it screwed up my ground plane now. um, my ground uh, connection. Now if these pads down the bottom here, if they actually connected through the ground on the other side and was routed off somebody else somewhere else, I might not have noticed.

Um, that you know a problem at all. it would have just worked. and unless I went through and actually visually inspected it I wouldn't have known that there was this break here and there was this, uh, large um, clearance around the pads. It's just crazy.

It's something to watch out for. and um, yeah, I didn't do a thorough inspection of these boards before I assembled them, just gave them a quick glance and they look pretty good. I think I was uh Carried Away by the super low cost of these at $160 each including tooling which is incredible I got carried away and well I didn't look at this and there enough. but it's a major major problem which can cause real headaches and screw up your design completely.

Even if these uh pads down here were connected on the other side, there could been a reason that I wanted the current going through this way as well. and then you can get all sorts of, um, subtle ground problems. If only you know if the current's not going where you expected on the border, it's being routed somewhere else. So definitely something to watch out for a when you're laying out boards and B when you're getting them manufactured like this.

So there you have it. The problem was with these It studio boards. Someone there decided to mod ify my Gerber files and actually expand the ground plane around those pads to the point where they went from an eight thoul uh Trace between there which I was relying on to nothing a break. It's crazy I You know I can understand maybe a TH or something like that but to totally eat away an E th Trace going through there is just insane God Unbelievable.

So if you're going to get these boards from it Studios it Dirt Cheap Great value for money if you don't get problems like this and they shouldn't I'm going to have to ask them why on Earth they're actually expanding their ground plane I already had the six THL clearance they required. The spec is six th Trace width and six th clearance between well, copper and one any copper and any other copper. So the ground plane and a pad or a V or something like that. so God it was already in place I didn't have to expand the Uh Power plane on there I don't know why they've done it and it screwed up my board and it was a pain in the ass.

Catch you next time.