20H Rule

[imnavajas]

I am reading about the 20H rule for a multilayer design and I found a lot of controversia about it...

As reading sources, I have taken 'trusted' ones based on papers, and summing up, most of them state that 20H actually does not improve EMI in the design, even more they could mess things up.... as an example I attach one of the sources from IEEE (of course, as far as you read you can find something saying the opposite!).

I know Robert recommends 20H rule, but is it based on expertise? in such a case how is achieved if is it worth to use it or not? Was the design EMI tested?

I wonder if somebody else has had issues with this and could give some recomendations.

Thanks,

Nacho.

[robertferanec]

This is a very interesting topic. I am very curious to see the discussion.

I had a look at some of the articles, and found Minimizing EMI Caused by Radially Propagating Waves Inside High Speed Digital Logic PCBs by Franz Gisin, Zorica Pantic-Tanner. Very interesting - especially the picture 18 (see the attached picture). It's a shame, that a simulation with two GND planes and one middle PWR plane pulled back is missing - because that could be the picture we are looking for.

It looks like, the rule should not be used for 2 layer PCBs, but I would say, it should be fine if PWR plane is between two GND planes. It looks like, 20H can have also other effects on your PCB e.g. according to the Franz Gisin, Zorica Pantic-Tanner article "... it reduces PCB resonance effects ...".

In our designs, we pull back PWR planes in most of our PCBs, but we normally pull it back only 1 or 2 mm under GND plane and we place PWR planes between GND planes. We have not seen any EMC issues because of that.

I like the following sentence from 20-H Rule Modeling and Measurements article: "Like most design guidelines, the 20-H rule can be helpful to those who understand its origin and purpose. However, it can create more problems than it solves when misapplied".

What do you think?

[imnavajas]

Looking at the radiation pattern, what 20H rule does, it looks to me, it is change the radiation direction. I miss a color scale in order to see the dBuV/m range and if they are appreciable. Because if the lightter the most radiated EM, 20H rule gives you more EMI in the upper direction of the picture.

What you think?

[robertferanec]

I am thinking, there is another picture microstrip / stripline and I would say, maybe the simulation with PWR plane placed inside between two GND planes could look like that .... see this ... in this case, the 20H could really help:

[imnavajas]

This looks like pretty good, but what about at the edge radiation?

[robertferanec]

We need a simulator expert - to make all the pictures we would like to see

[imnavajas]

And what about the plating edge solution? That has an EMI negligible result besides a higher internal EMI confining, but if you look at these images with 20H there is internal EMI too...

Well, I assume al high speed signals should be done in internal layers to avoid EMI...at least one via -> antenna on the Top/Bottom layer by high speed signal on the board, but this is always there not matter what you use!


What do you think about PCB edge plating and ground planes shortcircuiting?

[robertferanec]

As you noted, I always try to route the high speed signals on internal layers. I also always try make the layer under TOP and the layer above BOTTOM to be GND plane - so everything is between two solid GND planes.

I have not done edge plating, so I am not really sure about that.