Hi. In this video, we're going to take a look at a two-layer piece of B and a four layer PCB and what difference having four layers makes to EMC slash EMI or radiated emissions from the board? Now this is the Gigatron TTL computer which you've no doubt seen in previous videos and this is the original from the designers and it's a two layer PCB. There is no ground plane internally to this, so all the signal traces you can see they go horizontal effectively, horizontal on the bottom and more vertical on the top. And that's a traditional two layer PCB layout.

Yes, they've got some ground flood fill in here and stuff like that, but generally it's It's actually quite a nice neat layout. Get all the chips are in the same direction and it's quite a neat traditional 2 layer layout. But the problem with a two layer PCB is that well as we'll get into the finer details later, but it radiates like buggery. So I took this exact layout and I've done a video on this old link to it down below and at the end.

if you haven't seen it where I took exactly the same layout and actually produce a four layer PCB It's an absolutely identical layout. I've changed, none of the traces whatsoever. It's completely identical. The only difference is you can see in there that I've actually got a huge internal ground plane so ground and 5 volts internally for there all the bypass capacitors.

Everything else is absolutely identical. And what difference is this going to make to radiated emissions? Which is important if you're designing a product commercial product for sale. You have to meet various electromagnetic conformity requirements in there varies in different countries. We will won't go into that in this video.

so you can see our computers operational here. It's got a six megahertz main clock which is not particularly fast, but remember it's not the fundamental clock rate. it's in fact, you can have a one Hertz clock on this and it can still radiate like burglary because it's the edge rate of the signal. It's not just the fundamental frequency.

so we can actually use one of these H field magnetic probes to measure the near field emissions it's called and I might go into a bit of that later, but we can measure the emissions from this board and we can put it under here and we can see the difference between the two layer board and the four layer board. This is going to be neat. Now what we're going to concern ourselves with in this video is as I said, this is a near field magnetic probe. It's called a H field probe.

There's also a voltage probe which looks quite similar, but it actually operates quite differently and I might have to do a separate video on like the differences between these two, but this basically measures the radiated magnetic field from the PCB traces and the loop areas as we'll talk about. and I've got my jig of repeatability here, which just means that The probe will be in exactly the same position and then I can move the board over various points and I can look straight down through there to actually Center things and we can get a direct comparison and overlay various signals on both boards between the four layer and the two layer. how much difference do you think it's going to make? So we're dealing with their radiated emissions that actually come out of the board rather than conducted emissions which come out through our cables and including the braid and the shield of this, this will be conducted mode radiated radiated emissions. The USB over here I've got this powered by from an external battery pack by the way.

so to do these measurements will just remove any cables from there and we can get a direct comparison. So what I'm going to do is use my largest H field probe here at largest diameter because that's the most sensitive. You can't get smaller ones a bit. There's no point.

The small ones are really useful when you're checking out differences between like individual traces and you know stuff like that. And I've got that going into my preamp over here. That's a 20 DB gain preamp 3 Meg to 3 gig and we'll use our rogue old DSA 8, 1 5 spectrum analyzer. here.

We're using the EMI the electromagnetic interference filter type which gives us a industry standard 120 kilohertz resolution bandwidth filter. Here, you don't necessarily need that, but if you're going to sort of get ballpark pre-compliance measurements that you know try and match what you might get in the field, then you know 120 Kilohertz and our frequency span here will just go to 100 megahertz at the moment. I Can show you a wider span that just allows us to look at some nice detail in here and our amplitude. We're got units in DB micro volts that's just a bit easier than DB Millivolts doesn't really matter and we've just got a input attenuation of a fixed 10 DB here because depending on where we put it, put it right over chip we can actually see it saturate.

I Can probably show you that Now there we go out of range. it's a start. Saturates, Oops. And we've got a the input preamp on as well and I'm going to put it directly over the six megahertz crystal here and this is the response we get if you have a look at this first peak here.

Bingo! That's our six megahertz, our fundamental and if we skip through 12 18 and they're all the harmonics and that will extend all the way with LBJ it will actually keep going. If I actually change the span, let's go to 500 megahertz there. Boom! Look at that. it extends all the way right up.

We can go to like one gig or something like that, but it starts to and drastically drop off there. but you can see that it really extends all the way up because of that edge rate. Okay, so what I've done is that frozen that yellow reference there and let me take it away and let's plug in our four layer board and right under the crystal, it's exactly the same. Look at the difference here.

Yes, you still get all the peaks, but very significantly reduced broadband noise here. We're 5 r DB micro volts per division here. so we're talking about. you know, a good 5 10 15 DB difference.

In sort of like the bulk noise here and the peaks there simile like 5 or 10 DB down from that. So you know it's a huge reduction there and you wouldn't have thought so you thought. AHA it's all just coming from the crystal. It's exactly the same crystal in exactly the same location.

No, it's the radiated emissions in this case, the near-field radiated emissions coming from the traces and everything else on the board. But because we've got the ground plane in there, it's actually it lowers the loop area and does. We'll talk about this towards the end of the video, so stick around for that and it just lowers the radiated emissions. It makes a drastic difference.

15 DB is absolutely enormous. That could be the difference between passing and failing your compliance. and you know, costing you. You know, five grand or ten grand or something like that, you've got to Rhys pin your product.

You could easily spend tens of thousands of dollars because you've failed your compliance. So you know. And then you go Oh well. should have used a four layer board.

II Begin with, you'll see that probes actually quite a substantial distance above that board - but the heads are those radiated emissions. they're Aquila. Let's just try another random spot here, straight over the addressing mode decoder chip. Here this is the four layer board and that's our spectrum there.

you can see it was. It's much lower than we've got before. you can still see the fundamental and the harmonic spikes still in there. but it's going to be and pick up a whole bunch of broadband noise because this is a digital computer that's you know, refreshing the screen that's doing all sorts a process in the background.

everything's going all over the place and it's just you know, generating a whole bunch of wideband noise. Oh by the way, if you want to see what happens where we disconnect the power, it of course completely vanishes. Now let's try the two layer board and this is the two layer board. So four layer board in the yellow there and you can see that actually the four layer board actually has more prominent peaks in there because all the rest of its being kept more down in the noise by the ground plane.

whereas the two layer board has once again, you know a good 5 10 15 DB difference in the like, the average broadband noise level there. of course the peaks are once again, 1015 DB above the peaks on the four layer board as well. That's a huge difference. and let's try riding on top of the Rom here just for just for kicks.

So that's our two layer board. and there's our four layer board. Much much law just over the accumulator chip. There, that's our four layer board.

and there's our 2 layer board. So once again, you can see the broadband difference. It's It's quite remarkable. This is our accumulator chip again, over a 250 megahertz span.

This time, this is on the 2 layer board. You'll see what happens if we physically take the board away. I've got the board like a good like foot away from it now, but you can still see some of the radiated emissions there even though it's not in the correct plane. which we'll talk about in the minute.

There you go, there's our four layer board once again. I've changed the scale here to 10 DB per division now and once again, it's a 15 DB difference. It sort of gets a little bit closer up here. you know, around the like, a two hundred hundred and fifty to two hundred megahertz range, but still significantly under.

that could make or break your compliance for sure. And the thing with these H field magnetic probes and it's not like an issue with them, in fact, it's a feature is that they are dependent upon the orientation. They work in the plane. So if you've got your coil like this, it's picking up magnetic fields that are in that flat plane there.

So you'll notice that if we take this, there's a spectrum and that that's over 250 megahertz. And if we simply rotate that like that, it picks up different components. Look at that. so you can actually use that as a feature.

Using a smaller diameter wind, you can get down there and you can trace down. Oh, you're offending our components and traces better things like that. So I probably have to do a whole separate video on this. But yeah, it does make a difference.

The orientation We've seen quite a significant difference here between the four layer and the two layer board. makes a heck and a difference. like typically like broadband noise In this particular case about you know, 15 DB or so and that's a lot. But does that translate if you measure, say a 15 DB difference here? Does it actually with your near-field probes? Does that actually translate to a 15 DB difference on your EMC testing when you put it through the test house and you test it against the compliance standard? Well, the answer is unfortunately not these near-field probes, both the H field magnetic field and the electric field probes.

All this is as I said: the Near Field And whereas all of the compliance testing is done in the far field and I'll explain that in a minute because I have a Dave card. So what's the point of using these near field probes if they're not sort of like quantitatively equivalent to what they do in the test house? Well, the good thing about it is that at the design stage or maybe if you fail compliance or something or you need to or you're doing some pre compliance testing, you can go around your board and sniff all around your board and with the H field and the E field probe to see if there's any issues. see if they're You know anything's radiating wildly and stuff like that you can. You might be able to see a big spike or something at one particular frequency you might go or we need to knock that down even though you don't even know it might be compliant.

At the design stage, you might go well, you know, I'm not gonna take any chances and I'm going to knock that problem on the head now. before I send it across to the test house. So we'll briefly talk about near-field and far-field here and how it relates to the electromagnetic radiation. Now, A you might have heard the term electromagnetic radiation.

It's electro and magnetic, contains electric and magnetic components and you can look at it. This is like the standard visual representation of it. the electrical field might like would go up in the z-axis like this, and the H field is 90 degrees from that. So they actually propagate in different orientations.

And of course this is the wavelength. And here's a cute little animation just to show you how that works as it propagates down. Now what we actually have to look at though is what's called the wave impedance and this is where the difference between Nefi does Everything on this side and far-field is everything on this side. Now the wave impedance in ohms like this in there.

for this particular scale, please excuse the crew. Didn't have time to build up the scale. a lot of pain. it from 10 ohms to 10,000 here.

So this is where you have to define far field and near field. Well, the electric field and the magnetic or H field. There is a difference between H and B. By the way, B is flux density.

You might sometimes be here at chord B, but it's actually H magnetic field as opposed to induced magnetic field as opposed to magnetic flux density. Anyway, ain't going to the details. So the H or magnetic field actually has a very low impedance source in the near field, whereas the electric or a field has a very high impedance. It'll clarify that in a minute.

but basically it all comes down to the wavelength lambda here and this is normalized to 1 here and it's Lambda on 2 Pi, which is basically we're going to normalize to that value. And of course, let's take for example, 100 megahertz is a wavelength of 3 meters. so Pi on 2 that's about 1/2 meter. So when you get to 1/2 meter away from your product, this is where the electric fields and the magnetic field actually start to converge.

It's not really clean like this. There's a bit of you know overlap in here, and this is like the transition. There's going to be like a transition region in here where the two fields eventually combine and anything over roughly half a meter away. At a hundred megahertz, the electric and magnetic fields combine to give you a singular impedance, which actually happens to be 377 ohms in free air.

So anything over the wavelength on two Pi is deemed to be the far field and anything closer than physically closer than that like we just did with our probes. Here is the near field. Now this is why we have two different types of probes. One is the H field probe, the magnetic probe, the other is the E field or electric field probe, and the magnetic or H field is going to be generated by higher currents ie.

sources that have a very a lower in he danced. So for example, if you've got a lot of current flowing in a in a particular art race either due to an actual like heavy current switching or even very fast switching that's dumping a lot of energy into the bypass capacitors and the capacitance between the power planes and everything else, then that's generating typically be generating a magnetic field due to the low impedance and the high current, but very high impedance. Things that don't generate lots of current, then they generate electric fields and hence the bigger source impedance. So you'll generate electric fields from, say, just like a static power supply.

for example, your 5 volt power supply, whereas all you're switching stuff will dominate down in the H field here because there's lots of current being dumped into the trace or the load capacitance or the particular load itself when you switch in things. So that's why you need to use these two different probes and the magnetic field probes. They are sensitive to orientation like this and like that as well as I talked about on the plane. Whereas the electric field is not sensitive, you can just put that in any orientation and it's not going to make a difference.

So if I use my ear field probe like this and let's say I probe this power trace over here like this, you can see it's really not going to make any difference in the orientation that I put that in. It's just completely insensitive to that because there's no magnetic field coupling, it's electric field coupling and it's just purely the distance. But if you take a magnetic loop probe like this and I just change the orientation like that Wow That makes a big difference. It really brings out the peaks if I put it vertically like that.

if I put it horizontal, it gets more of the current flowing through the trace. And if we use our smallest H field probe, let's just have a look at let's say this like blank area over here. This is our four layer board like this. or maybe right over on the edge of the corner of the board over here like this.

and let's compare that with our two layer board here. Bingo. Look at that because we've actually got a power trace actually running right around this corner as well, which we actually physically reemerge and you can actually see that the power trace actually running all the way around there like that. So that's just going to radiate like buggery.

But even if we go over just the ground plane there, you can see it's much much higher than we get with the four layer board. And this is why at the Emc test house, they'll test in the far field here because it binds the electric and magnetic fields together and basically the typical testing distances would be like one meter, three meters, five meters, 10 meters, for example. away. It depends on the type of product they're testing into, which are standard they're actually testing too.

but say, if you put it 10 meters away then you can have a larger rotating turntable so that your product rotates around like this on the turntable. and they can measure all the axes like this when they, while they have, they're super expensive. You know, buy a conical, super calibrated measurement antenna ten meters away measuring over say 30 megahertz to ten gigahertz far-field for example might be a typical measurement range. and then there'll be a standard like envelopes that you have to get under and also peaks and things like that and it gets.

You know the standard gets are quite complicated, but yeah, just the near-field testing that we do here. It doesn't really translate to the far field, but you can certainly I get an indication of whether or not you've got any nasties on your board. So why does this happen? Why does a four layer board make a huge difference compared to a two layer board when it's an identical layout? All the traces are exactly the same length, all the chips are in the same location. It's got just the same number of bypass capacitors, Everything's hunky-dory They should be identical, right? Well, it all comes down to loop area, which you've heard me talk about in many videos before.

And a huge general rule of thumb when you lay an hour Gords is not only to keep your traces as short as possible, but to keep that what's called the loop area. As small and tight as possible. So the tighter your layout and the tighter your loop area, the less problems you're going to have with EMI in generation and EMI and susceptibility to electromagnetic interference as well. So you have a source and you have a destination on your PCB a trace going from one side deep from the source to the load.

Now the loop area is actually the total area that includes the ground for that entire loop, but it also includes the power system with the bypass capacitors. and I've done a whole video on that actually showing the return path for currents and why you need bypass capacitors. I'll link that one in. It's really fascinating so that it has to do with the entire loop area and the bypassing.

and on two layer board like this one here the original one you just don't have the luxury of having or often don't have the luxury of having a very tight loop area for all of your signals. You might buy either accident or good design have them for certain traces, but when you've got you know what, two dozen chips spread over on this on a large board like this, you just can't possibly have every one of the traces having a short loop area so something's going to radiate somewhere. In fact, probably the majority of them I just have large loop areas and hence are generating a larger magnetic field. in this case, a bit electric field as well.

Electromagnetic field will say generating a larger electromagnetic field and hence why we see the huge increase in radiative emissions. So this is the original two layer board and what will go in here is will go and just inspect a single signal. Let's say the address 0 pin to the RAM chip. That's it because it's only got 2 connections on there.

Now we can actually go in in there and have a look at this. This is the RAM chip and this is the driving chip here for that address decoder. Now look, it's got a nice short trace there. Look at that.

That's really neat isn't it? And in this particular case, the under the driver chippers here and the ground of the RAM is over here. Now this is actually a reasonably short path for a look. It has the snake. It's got to go all the way around here.

That's a reasonably short path for a two-layer complex to lay a board like this, but if you turn the top on, there might even be a shorter path. Not yet like it might jump back over a veer here. Yes, it does. Look at that, that's handy.

They just happen to put in a veer here and a veer here, so the path is actually shorter. It goes through here like this and goes back over to here and that is kind of the shortest path. Otherwise it's got to go all the way over here on the bottom and all the way around on that green layer there. But still, that's relatively short.

so you might think that's not too much of a problem. But aha, what about the bypassing? When your signal transitions like this? the capacitive load, and the capacity of traces and everything else actually a capacitor? When you apply voltage to it, it first. when you transition up like that, it appears as a short-circuit and that generates a little gulp of current. and that current should come from the bypass capacitor.

So let's have a look at the bypass capacitor for these two devices. Look at this. Here's the VCC pin of the RAM chip. The bypass capacitor goes down to this ground here, and look.

The the two chips are actually sharing a very short power path, so that's almost ideal. So if you just look at that from a point of view of like the loop area, from a point of view of just the power pins, everything's reasonably hunky-dory for this particular trace. Remember, we've got hundreds of these traces on the board, each with their own loop area because they're all switching and they all intermix. That's why you get all that huge broadband noise measured across the spectrum now.

But the problem is the ground for this. Here's where we might come. A gutter. He goes up here.

oops, Where else does it? Go look at this. It's got to go all the way over here. all the way over here, or all the way over here. If it's just on this layer then it's just snakes its way back through there and it comes back through the trace.

That's a huge Lou period. Now of course if we turn on the top layer we might be lucky and we might get some vir stitching in here. Let's have a look. actually this one's not too shabby.

The ground from here goes on the bottom layer, the green layer over to here then jumps up onto the top layer. The red goes through this via and there's another one up here as well. goes down here and then can drop down to the bottom layer like this and then go to the pin like that, but you know it's it's a bit higgledy-piggledy It's not ideal. there's only one.

they might only be one lousy vir in there which is going to be higher impedance at higher frequencies. It's got more inductance right? So that is the problem. The inductance of your veers and all the and your inductance of your ground paths and your loops and all that sort of stuff comes into play. So you know we're still kind of within this area.

but it's going to be much higher inductance and it's just running all over the shop. And if you were the person to be layout person, you've forgotten. If you didn't put these and like via stitching in here like this then Wow It'd be like traveling all over the board and the loop area would be larger and larger and generating a greater electromagnetic field for a given particular switching current. If we go over to the four layer board, we have a look at exactly the same signal like this.

you'll notice that well, there's no more flood-fill ground planes on the top and bottom because the whole board is one ground plane. So now the trace is exactly the same, but this ground pin can go all the way over to here. This ground pin not only it's the same length path, but it's lower inductance because it's a big solid ground plane going right over and then that's solid. Ground plane is also connected over that entire area right up to this bypass capacitor here.

And likewise for the power because we're going to big power plane on there. so the loop area is a little bit smaller in here, but it's much lower inductance so it's going to be much more effective. And also you've got the shielding provided by the ground and the all planes which makes a difference, but magnetic electric fields. We won't go into the details, but yeah, it makes a big difference.

and that loop area if you don't keep that small, Faraday's law is going to screw you over and the greater the area. Just like the reason why, you have two different sized magnetic H field probes that they've both got one loop in there. There's just one loop. That's it.

but this one is a larger diameter so it's more sensitive. It works in the reverse. If you've got a larger loop on your PCB, it generates more and it receives more if it's bigger. And that's just a singular example of one particular trace.

Remember, you've got hundreds of different traces, each with their own loop area. and let's have a quick look at the power system on. Once again, we're back on the two layer board. So this is actually the VCC or power pin.

You'll notice how it's just running loop. It's got a run right around here like this. So imagine if you had a source up here and a destination down here and it's got to go all the way back through that power and the local decoupling. You know it wasn't right.

And thus, ground planes was split. And that's the problem. Split ground planes are always a killer because they're instantly going to generate more loop area. and then if you have tracers running across splits, that's really bad for AMC And there's all sorts of stuff like that so you know.

Look, it's just. it's just a problem. and it's nothing wrong with this layout. This is a good to good layout to layer board.

they put in via stitching in there to try and sort of. You know, shorten the grounds everywhere. when you're flood fill a two layer border like this. Yeah, you should just pepper everything with this.

Sometimes you can screw it up and it might cause a loop where you didn't want it to go or something like that. but in in general, yeah, the more you stitch those planes together the better you know, the luckier you're gonna get and you're going to avoid Murphy And you're going to, hopefully you know, not be too bad, but there's absolutely no competition compared to a solid ground and solid power plane across a board. Now if you had a three-layer board you would do a three-layer board because they manufacture them into in even numbers. but let's say you did and you only had the ground.

then you'd still have the power like this and you might still have at lower frequencies. for example, because the bypass capacitors work at different frequencies and they have different responses done videos on that so the lower frequency current has to flow in much bigger loops then the higher frequency stuff does. whereas if you have that nice solid ground and power plane then the loop areas are going to be much much smaller. Check this out.

I Found one that's worse. This one has a source in two destinations. It's the y0 pin. I don't know what it does, it's a TTL computer so it connects this chip here, this one and this one here and the ground is actually pretty good if you look at this.

see between this chip here, goes over to this one here, right? So it's up and there, so right. So that's a relatively short ground path. But let's look at the bypass capacitors, which as I said, is where all that high frequency stuff comes from the high frequency bypass. and that's why you have the bypass capacitors.

Look, this one goes up and goes up to here and then it's connected to this bit of ground up here and that bit of ground is like split on this half. It's split by this power rail running right through the guts of this. so unfortunately that's like it's going to have a hard time getting backs that a loop area is going to be much larger for this actually trying to get back between these particular chips and then the bypass capacitor for this chip here is over on this particular ground, which is not via stitched over to this one which has to go all the way. It doesn't even go all the way across there.

Maybe there's some vir stitching in there, but it's It's like connected to a totally different split ground plane and it might make its way back here. Gordy Piggledy somewhere. But yeah, that's no wonder this thing has a lot more A.m. radiation than the four layer board because the four layer board would just that top and bottom under the chip with very low inductance, very low impedance, ground and power loop paths for the bypass capacitors and the signal because the return path wants to take the lowest impedance and I'll have to I think I did do a video on This takes a lower impedance.

so if we have the ground and power under here, then the return current will actually follow and float. The high frequency return current will flow and follow this particular trace or it'll try to. But if you don't let it because you've split your ground plane right through, yeah, you've got a huge loop area. It's just spewing out the radiation.

So there you go. It's all about loop area when you're laying out boards, and when you're the difference between two layout, four layer, everything else, but not only just loop area, it's particular frequencies of different types of bypass capacitors, different currents, different capacitive loads, and things like that. Different whether or not it's generating electric field or a magnetic field or a kind electromagnetic combined field, and things like that. Depending on the source, impedance and the load impedance, and all sorts of things, it gets a really complex.

So does this mean that two layer boards are just inherently horrible and you should avoid them at all costs? Well, no, you can actually do quite decent layouts and approaching the performance of a four layer board on a two layer board if you're careful and you're lucky with the layout. Unfortunately, with a design of this sort of complexity, this many number of chips spread over this, you know, convoluted arrangement. Yeah, you're just going to get issues a four layer boards just going to be. You know it's just going to bury the performance of for this particular design here.

but there are some good practices you can do on two layer boards like try and keep your ground and power on top of each other wherever humanly possible. Don't try and split the grounds, do extra vir stitching and flood fields and stuff like that to try and keep it as tight as possible. In fact, a well laid out two layer board might actually have less radiated emissions I even have better performance then a poorly laid out four layer board. So it, you know.

But for heads, just go into four layers. Having the ground and power planes just makes it much easier and you're less likely to screw up. Just remember, split ground planes really bad. So if you've got to chop up your ground and power planes on your four layer board, that can cause problems too.

but like there's lots of when you go down the rabbit hole on designing PCB layouts for EMI EMC performance. there's just almost an infinite number of things to consider. I Might do some more videos on that if you want me to let me know in the comments down below, but there have been and there are like entire books devoted to just doing this sort of stuff. So yeah, it can get quite complex and there might actually be a follow-up video to this if I can organise that.

I've actually comparing these two layer and four layer boards in an actual EMC test house or maybe on an outdoor test area called an Oates and an outdoor area test site. So oh, but that obviously requires a lot of planning and facilities to do that. I Don't have that here so you can do that in the far field. you can do like a far field test in your lab you can get.

you can buy an expensive Emc antenna, you can sort of roll your own, but they're a bit how you're doing and but anyway, you can give you a decent indication you can put in a meter two meters away you go. Companies even make their own RF Anechoic chambers and and stuff like that, but you know you can do that sort of stuff. but just using your H field and your F your probes going around your board, you can like and just check in for any sort of nasties hidden in there that can really save your bacon when you go for testing later. So while there may not be a direct correlation between the near-field test we did here and the far field ones you'd get in a MC compliance testing, it's You know it's not a bad sort of correlation.

so the two layer one would absolutely definitely perform worse than the four layer board in a true EMC compliance test over the full spectrum, if there's no doubt about it. So anyway, hope you liked that video. Hope you learned something if you did. Please give it a big thumbs up and there will be more videos hopefully coming on this as soon I Want to show you how you can construct your own near-field probes and also I'd like to do something getting a heat map of H field radiation on a large board like this.

So maybe we'll have a do-it-yourself project for that as well so we'll see what happens. Catch you next time.