[SI-LIST] Re: Routing Signals Between PWB Layers - Part 2
- From: Doug Smith <doug@xxxxxxxxxx>
- To: steve weir <weirsi@xxxxxxxxxx>
- Date: Sun, 07 May 2006 10:21:44 -0700
Hi Steve, Lee, and the group,
I am traveling (in Warwick, RI at the moment). My board was not meant
to be a good design but illustrate a specific effect if attention is
not paid to good board design.
The board has no bypass caps because the planed are DIRECTLY shorted
by the two SMA connectors and at the two load resistors. Obviously, a
real board should have many more than four bypass capacitors.
Steve, I have a data point on the cavity resonance thought. If the two
plains are shorted all around the edge with copper tape, the ESD
response of the path changing layers gets much better, about the same
as the path that stays on one side of the board. I don't have the data
handy. I suspect the emissions would also decrease.
Test boards to illustrate a specific effect such as I have done are
the basis of many papers over the years at IEEE EMCS and are valid. By
keeping the experiment very simple good information can be gleaned
about a specific effect. I remember a simple experiment by Henry Ott
on image plane effects (using a chassis member) to reduce emissions
that was very good. In a real system there are a lot of complicating
factors that need to be considered, but his simple experiment was very
useful.
The effect in my article is very important and suggests that good PWB
design principles such as appear on this list often are really
necessary. My simple experiment just illustrates what can happen if
care is not taken. I think your objections to my article may be one of
reading more into the article than I intended.
Lee, listen to my podcasts on the subject. There is much more
information there than I can type into an article meant to be read in
a few minutes.
Doug
steve weir wrote:
> Doug, Lee,
>
> I have a keen interest in this area. It plays to all: cost,
> performance and robustness of designs. I think the contrast between
> Doug's experiment and previous but very different tests Lee has
> published point out very important effects of coefficients when
> examining what happens when signals change layers. Both experiments
> provide insights. What the results can mean varies on:
>
> 1. The spectra of the excitation. ESD is very different than signals.
> 2. Bypass strategy, stack-up, live devices, stitching of ground layers
> in a > 6 layer board.
>
> I think Istvan, Lee and I all agree that the stack-up and bypass
> strategy can make a big difference. For uniform planes, matched edge
> termination theoretically ELIMINATES all cavity resonance. Cavity
> resonance is the clear source of the noise AFTER the incident edge
> measured in Doug's experiment. I think Doug's experiment does a great
> job of showing that:
>
> 1. On a board with thick dielectric, we can easily excite a simple
> cavity with ESD.
> 2. We have a strong motivation to suppress modal resonances.
>
> All of the measures that Lee speaks of tend to shunt and dissipate
> energy and so mitigate the situation. So, if we are trying to develop
> rules, we need to quantify these effects and the countermeasures.
>
> ESD energy still couples even with perfect edge termination eliminating
> all modal resonance. Big fast nasty things like ESD discharges make for
> very different results than one or a few signal lines. For low cost
> boards that typically tend to be line cards and such connected to the
> messy outside world, this is an important consideration. The cavity is
> thick and relatively high impedance. Oscar may touch on it Tuesday when
> he presents on designing to withstand 100V/m fields at the SCVEMC meeting.
>
> The situation gets more complicated but much better when stack up and
> stitch is done properly in high layer count boards. In those cases, the
> worst case would be referencing one of the inner Vddx with transition to
> reference of a Vddy. The net effective cavity height would approximate
> the sum of the Vddx / Gnd, and Vddy / Gnd cavities, typically 6-8mils in
> current construction, but could be 2-4mils with BC1000 or BC2000
> respectively. Applying thin layers is how some people have used dollars
> to fix boards that either had compliance issues or in severe cases
> functional issues. Whether cheaper methods could have been used had the
> design been planned better is arguable on a case by case basis. One of
> Chris' favorite mantras is that if we maintain a constant reference from
> die out then the cavity(s) won't give us any trouble even with very fast
> signals.
>
> On high layer count boards with well-stitched gnd joining the outer
> plane layers, we have a nice Faraday cage with a high resonant
> frequency, where leakage is limited to the edges and the antipads. With
> good edge termination the edges don't leak much. What doesn't leak
> isn't susceptible. Cross talk within the structure is a matter of
> coefficients: signal edge rate, number of SSOs, stitch density, and so
> forth. The only rule that makes sense to me is: If you are not going
> to do the homework, then risk is reduced by not transitioning through a
> cavity. That can be an expensive rule. It makes more sense to do the
> homework.
>
> Scott published a paper a few years ago that is available on the
> Teraspeed web site characterizing S11 and S21 for transitions through a
> thick cavity with varying stitch spacing.
>
> Regards,
>
>
> Steve.
>
> At 09:27 AM 5/6/2006, Lee Ritchey wrote:
>
>> Doug,
>>
>> I concede that it is possible to create a condition where EMI is worse
>> when
>> signals change layers as has been done with this demo. However, the
>> conditions in this demo are not representative of what a good 4 layer
>> designs look like. As some have pointed out, your demo does not include
>> any decoupling capacitors, which real designs have. It does not contain
>> any active circuits, which real designs have. It does not contain any
>> plane capacitance, which real designs have.
>>
>> My fear is that some will get the impression that your demo applies in
>> these cases and lead them to invoke some terribly restricitve design
>> rules
>> when they are not necessary.
>>
>> Your tests would be more applicable if you did your tests on a real PCB
>> with real bypassing and such on it. If you get the same result, then
>> it is
>> valid to claim that layer changing is a detectable source of increased
>> EMI
>> in a 4 layer PCB. It does not follow that this is true in PCBs with more
>> than 4 layers, by the way.
>>
>> Even then, the fact that there is a detectable increase is not evidence
>> that this is a problem that should be worried about.
>>
>> There is a big difference between detectable and important and we should
>> make sure we don't cause worry just because something is detectable.
>> Next,
>> you need to show that this is important if this experiment is to go
>> beyond
>> the academic.
>>
>> After all, vias are detectable on a transmission line, but only in rare
>> cases are they a problem. Another thing that worries engineers when it
>> shouldn't in almost all cases.
>>
>>
>> > [Original Message]
>> > From: Doug Smith <doug@xxxxxxxxxx>
>> > To: SI-List <si-list@xxxxxxxxxxxxx>
>> > Date: 5/3/2006 1:44:20 PM
>> > Subject: [SI-LIST] Routing Signals Between PWB Layers - Part 2
>> >
>> > Hi All,
>> >
>> > I have been writing and recording again and have posted part 2 of my
>> > article on signals that change layers in a PWB.
>> >
>> > Abstract: Printed wiring board, PWB, signal paths must often change
>> > layers in a board stackup. Under some conditions this can cause
>> > problems. An emissions example is used to illustrate the conditions
>> > where changing layers can cause problems.
>> >
>> > The link to the article is the picture of the emissions plot at the
>> > bottom of the home page at http://emcesd.com .
>> >
>> > There is also an audio discussion of this article on my podcast site:
>> > http://emcesd-podcast.com where the direct link to the article is:
>> >
>> > http://emcesd-podcast.com/2006/may/2006-0503.mp3
>> >
>> > Can't download mp3 files? Download the following instead:
>> >
>> > http://emcesd-podcast.com/2006/may/2006-0503.dcs
>> >
>> > After download, change the extension from .dcs to .mp3 and the file
>> > will then be able to play on most computers. Since the last HFNews
>> > there have been two new podcasts posted.
>> >
>> > Doug
>> > --
>> > -------------------------------------------------------
>> > ___ _ Doug Smith
>> > \ / ) P.O. Box 1457
>> > ========= Los Gatos, CA 95031-1457
>> > _ / \ / \ _ TEL/FAX: 408-356-4186/358-3799
>> > / /\ \ ] / /\ \ Mobile: 408-858-4528
>> > | q-----( ) | o | Email: doug@xxxxxxxxxx
>> > \ _ / ] \ _ / Website: http://www.dsmith.org
>> > -------------------------------------------------------
>> >
>> > ------------------------------------------------------------------
>> > 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 FAQ wiki page is located at:
>> > http://si-list.org/wiki/wiki.pl?Si-List_FAQ
>> >
>> > 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
>> >
>>
>>
>> ------------------------------------------------------------------
>> 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 FAQ wiki page is located at:
>> http://si-list.org/wiki/wiki.pl?Si-List_FAQ
>>
>> 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
>>
>
>
>
--
------------------------------------------------------------
___ _ Doug Smith
\ / ) P.O. Box 1457
========= Los Gatos, CA 95031-1457
_ / \ / \ _ TEL/FAX: 408-356-4186/358-3799
/ /\ \ ] / /\ \ Mobile: 408-858-4528
| q-----( ) | o | Email: doug@xxxxxxxxxx
\ _ / ] \ _ / Web: http://www.dsmith.org
------------------------------------------------------------
------------------------------------------------------------------
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 FAQ wiki page is located at:
http://si-list.org/wiki/wiki.pl?Si-List_FAQ
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:
