[SI-LIST] Re: Stack up for EMI reduction, plane resonance and u-s trip radiation etc etc

  • From: steve weir <weirsp@xxxxxxxxxx>
  • To: "Michael E. Vrbanac" <vrbanacm@xxxxxxxxxx>,Chris.Cheng@xxxxxxxxxxxx, si-list@xxxxxxxxxxxxx,lnima@xxxxxxxxxxxxxxxxxx
  • Date: Wed, 11 Feb 2004 00:08:01 -0800

Michael, I haven't seen Nima's posts in this thread which was not about a 
specific board problem.  At least not to the best of my 
knowledge.  Additional comments embedded.
At 01:01 AM 2/11/2004 -0600, Michael E. Vrbanac wrote:
>Chris, Steve, Nima,
>
>Jumping into the middle of the discussion...  some thoughts on a number of 
>comments/
>subjects brought up... including the original one...
>
>First, Nima....
>The information you've given is not really sufficient to give any of us 
>sufficient grounds
>to base many of the presuppositions that have been offered.  A lot of the 
>ideas offered
>are certainly possible but giving us a stackup and a few frequencies is 
>not enough
>to draw a certain conclusion.  I think someone early on said something to 
>that effect.
>I've arrived at a test site to look at a client's product issue and it 
>usually starts with the
>information that you've supplied.  There's a lot more to know in order to 
>put a problem like
>that to rest.
>
>I have several questions that I hope will help you:
>
>1.  Are we talking a single board or multiple boards?  Anything else 
>notable about the system
>that might contribute to the problem?
>2.  Have you proven that it is related to the 700 MHz signals?  Are there 
>any other
>operations on the board that might generate these?
>3.  How does the radiation problem manifest itself?
>4.  What has been done that hasn't worked?
>5.  Are we talking about broadband noise or all harmonically related signals?
>A comment you made suggested that there might be other signals present also.
>
>Just a suggestion:  Don't just change a stackup just because there is 
>nothing else you
>know to do.  You'll need to do a lot more detective work to solve this and 
>if you still
>can't get it, let us know what you found out and maybe we can help you more.
>
>Various comments/subjects found in the thread...
>
>re: "Any HF noise that resonates in the planes can find its way out, by 
>going up
>vias to some structure above the board."
>Yes.  When this propagates outside the system, this is known as "common mode"
>radiation... a very nasty and difficult problem to solve on the whole.  It 
>is such because
>everything tied to that common point, a plane usually "ground", will be 
>energized as
>"one" making it difficult to track down the real culprit causing the 
>problem.  Common
>mode radiation is not the same as "plane resonance", though.  A resonance 
>implies
>that a "standing wave" is present in some form.

Considerable energy from what would be only a standing wave can and will 
find its way out under many conditions.
Agreed that CM is a separate issue of plane potential versus the chassis.

>re: excessive power/ground/signal pins... (understood to be plane 
>perforations, etc.)
>That's almost always causes a real nasty problem that will cause common mode
>radiation.  Hence, my next point...
>
>re: EMI reduction (in relation to components on PCB and the PCB itself)
>It is the package that radiates and it does so against the reference which
>is the PCB to which its attached.  If the local reference inside the chip is
>at the same (or very nearly the same) potential as the PCB underneath it, 
>there
>will be little or no radiation to speak of unless there are other issues 
>left unresolved.
>No stackup can fix this by itself unless the potential between the 
>elements of the
>problem is brought to very low levels or effectively "zero".  The key here 
>is to reduce the
>PCB to package impedance at the spectrum of interest.  But this is not 
>"news", this is
>EMC 101. Examine the ways to reduce that impedance and you'll have the 
>solution to
>fixing the problem...and you'll be on your way to fixing the decoupling 
>issue we all talked
>about a few weeks ago as well.

Ah yes, those devilish little details again.  As in frequencies present in 
the package now go way beyond what we can convey through the package 
interconnect to the PWB.  At frequencies substantially above 100MHz I know 
of no way to tightly couple any IC to the plane beneath.  The best I know 
how to do is shield the overall structure.  If you have a secret for 
overcoming package impedance at 100's of MHz, and beyond, I am all ears.


>re: plane resonance
>Fortunately, all the things necessary to make these problematic are not 
>present in
>most instances.  In fact, it is my opinion (because I haven't proven it 
>but it would seem
>to be true), that they cannot be entirely eliminated.  They can only be 
>mitigated.  The best
>we can usually do is make sure that the driving impedance into the 
>resonant circuit is
>much lower than the driver can "drive" at the circuit's resonant 
>frequency(ies).  This means
>that close power/ground plane spacing is appropriate to help prevent these 
>as it drives up the
>capacitance which drops the "differential mode" power structure 
>impedance.  It is also a
>reason why extensive PCB grounding with close PCB spacing to the chassis 
>does the same
>in the "common mode" sense.

I have always seen this as a matter of reducing inductance for the first 
choice.  If we hold L constant and increase C by altering Er, then the SRF 
falls by 1/sqrt(Er_new/Er_old), and for a fixed resistance, Q also falls by 
a similar amount.  It is actually a little worse than this, as the reduced 
SRF drops R due to skin effect.  However, for the same constant R Q falls 
directly with reduction in L.

>re: ustrip radiation
>All things equal, a microstripline will radiate more effectively that a 
>stripline. A study of
>the field lines surrounding these structures should show why.  That is not 
>to say that
>the microstrip will always be a problem, though.   Well designed 
>microstriplines should
>perform well but at times may require more "peripheral field containment" 
>than a well-designed
>stripline.  It is a matter of the skill used in the design of each and 
>what sort of signals are
>presented upon them that determine whether they will be a problem or 
>not.  In summary,
>microstriplines are not inherently bad to use but they might at times take 
>a bit more care
>in using them under certain circumstances.


Agreed, we can design well tuned or intentionally poorly tuned antennae in 
microstrips.  The point that had been circulating concerned the containment 
properties of prepreg.


>re: "Ans : Beaten to death, package is the choke point. EMI noise radiates 
>from package not PCB"
>
>Almost but not quite true.  The package does seem to be a problem but to 
>eliminate the PCB from
>culpability is not totally accurate.  Structures very similar to the 
>package problem can exist in
>PCB designs and the effect is the exactly same but usually at lower 
>frequencies.  Folks on the whole
>are getting better at this avoiding this and it doesn't show up as much 
>these days but it can and does
>happen.  Just think of the PCB just being a "macro" version of the package.

The point of discussion was the amount of energy transferring from the 
package to the PWB at high frequency.  The LPF characteristic of the PDN 
looking from the package out effectively isolates the core from the PWB 
with a cut-off somewhere in the 10's MHz to about 100MHz, with rare 
exceptions proving the rule.  Please see comments above about decoupling 
the package to the planes.

>re: " Ans : Managing the return path and reference plane not the 
>decoupling caps"
>
>Again, almost but not quite true.  While managing the return path and 
>reference planes are vital in
>these matters, so is managing your power that supports these 
>signals.  What is not generally
>realized is that when a chip is "starved", its waveform changes and the 
>driver may experience
>instability and generate signals at much higher frequencies than 
>expected.  That sad truth is that
>you have to manage the signal, the signal path and return, and all 
>energies used to produce and
>propagate the signal.

The working assumption in the conversation was that the IC is powered 
adequately at least up to the package cut-off frequency.  The issue of 
debate was given a well fed PDN into the IC, whether the decoupling network 
is or can be effective at multi-100MHz frequencies.

>Best Regards to all,
>
>Michael E. Vrbanac
>

snip


------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List technical documents are available at:
                http://www.si-list.org

List archives are viewable at:     
                //www.freelists.org/archives/si-list
or at our remote archives:
                http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu

Other related posts:

  1. Home
  2. si-list
  3. Archive
  4. 02-2004