> >At  the 1999 PCB east conference I learned about the 20H rule.  This rule
> >simply put states that EMI can be substantially reduced by keeping the
> >power planes back from the edge of the ground planes by 20 time the
> >dielectric thickness between the two planes.   This would be a lot easier
> >to implement than vias .
<snip>

> This rule of thumb appears to be highly controversial among experts.

I don't like the term "expert", but they don't pay me because I make mistakes in areas like this.
My opinion is that this was a special case, and the special got lost from the case.. or to use a cliche,  this is a wives' tale.  "Rule of thumb" would be pretty loooooose use of that term.
Anyone who thinks this actually works is going to have to back it up with some data to convince me.

So, you might ask, "Why isn't this going to help in my general case?"
Here are a some major points:

What if your ground plane is as noisy as your power plane? 

Your ground plane, except in very few cases, can generally be counted on to be exactly as noisy as your power plane.  Period.

Consider the majority of cases:  A board with a ground and power plane on the inside with components and signal layers on the outside and sometimes additionally between the planes.  You attach little AC noise generators all over this board, and hook them up to power and ground.  Now, you hook this board up to a power supply and ground with two similar wiring methods, several many ÁH of wire to the power supply.  This is a little easier to visualize if you imagine a floating supply.  Now, what exactly is the difference between power and ground besides a DC offset?  Nothing.  This is even more true after you add a slew of decoupling capacitors, and have the planes capacitively couple to each other through the board material. 
I absolutely promise that both planes will radiate, for all practical purposes, the same.  Try it sometime.  Gripe at physics if this doesn't fit rub you the right way.  You'll notice more difference from individual traces on the board radiating due to individual characteristics.
(This is of course assuming that both planes are contiguous, slot radiation from splitting planes is a different case and will most definitely get you a difference in planes)
Yes there are other exceptions.  I'm quite sure that someone will come up with a few and state them here.  Nothing is ever absolute.  In some cases those power supply wires might even turn out to be antennas.

Back to my point, some will say, "my supply isn't floating, one side is grounded, so one plane is better..."  Nope.  Not after you connect it to ground with a bunch of ÁH of wire.  That RF isn't gonna go down that wire to ground.  It just won't.  Not any more than you are expecting it will cross your RF chokes :)

Now visualize this:  If you connect your ground plane to ground with a short 4" copper strap (Lower the Inductance Substantially), then yes, it will drain off, exactly to the degree that you provide a non-inductive/resistive path to ground.  And you will then have a differential between planes.  If you decoupled it right, not much of one though.  Depends on frequency and tons of other factors.  There are other ways of accomplishing this too, like screwing your board directly to the chassis.   Properly executed this can be helpful at getting your board to chassis potential RF-wise...

Mr. Baggett understands these concepts quite well:

You might say that the ground plane being as noisy as the power plane is
ridiculous.  Try an easy experiment:  use a working digital circuit that has
inherently noisy circuitry, such as a computer board.  Connect it to a power
supply that has an isolated negative terminal (typical bench power supply).
Hook up your scope probe and measure the noise on your power plane.  Now
connect your scope ground wire to the power plane and the probe tip to the
ground plane.  The ground looks just as noisy, doesn't it?  That's because
your ground reference is now the power plane.  The fact is, RF noise
radiation doesn't care what your ground reference is, because radiated
emissions are (by definition) radiated, NOT conducted.

Another question:  "What about phase differential between the planes?  Wouldn't that do something if they weren't the same size... make it into a patch antenna if they were wrong but possibly cancel if they were right?" 
Um, perhaps.  But wow, this is really stretching it.  What about Frequency?  Show me how, would be my answer.  Any antenna guys out there?  Antenna construction is also a Voodoo art :)

Mr. Wahab's original question was will the ground via thingy reduce EMI.  With knowledge of the above concepts, "No" would be the short answer, assuming lots of things.  Not generally, that's for sure.  Same answer for the 20H "rule".

-Frank



Frank Gilley
Dell-Star Technologies
(918) 838-1973 Phone
(918) 838-8814 Fax
[EMAIL PROTECTED]
http://www.dellstar.com * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * To post a message: mailto:[EMAIL PROTECTED] * * To leave this list visit: * http://www.techservinc.com/protelusers/leave.html * - or email - * mailto:[EMAIL PROTECTED]?body=leave%20proteledaforum * * Contact the list manager: * mailto:[EMAIL PROTECTED] * * Browse or Search previous postings: * http://www.mail-archive.com/proteledaforum@techservinc.com * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to