[SI-LIST] Re: [Expert system]
- From: Ralf Bruening <ralf.bruening@xxxxxxxx>
- To: "Mike Mayer" <mwmayer@xxxxxxx>
- Date: Mon, 4 Apr 2005 08:55:55 +0200
Hello all,
with a short delay (Easter vacation) I would like to add some
(caution, vendor biased somehow) comments to the "Expert System" discussion.
The motivation behind the UMR approach (which has chosen in the 90ies
the slightly misleading title "EMC Expert System" - misleading
because expert systems have been a real buzz-word in the 80ies/90ies for those
who remember) has been the failure of the EMI simulation kernel
builders to develop 3D methods which can do a full-wave analysis
of real (real means real in size/complexity) boards, so the goal has
been to allow an EMI analysis of complete (!) PCBs, not to match
measurements in a range of +/- 10 dB but to help identifying
potential problem causes for differential mode, common mode and
power-bus noise (and for me the discussion whether this is a DRC or
simulation or whatever is splitting hairs - or a matter of
abstraction, you can define everything which would be solved by a
computer as a simulation, couldn't you ?)
Therefore a match with measurement points to the wrong direction
(my feeling), of course I often hear this question either and at
UMR verification work has been done to prove the algorithms, but
even more its an issue whether the system can predict problematic
areas on PCBs and can be used to optimize the EMI, I know that
users of Mentor QUIET Expert and of Zukens Hot-Stage EMI (which
are the 2 commercial EDA products build our of the UMR algorithms)
have checked the tool results for various design versions
(problem detection and matching of the delta's in the field
results between the different PCB versions) - and thats where
the UMR approach can be used within the design process, not meant
to compete (or replace) neither measurements nor 3D full wave
simulation, but is a put parts of the black magic into the hands
of the design engineers - if the limitations (which are of course
there) are taken into account properly - and there are users of
the tools out there ;-)
Just to clarify what my (!) understanding is behind the 'EMC Expert
System', its rather different from what people in general see
behind expert systems, rules and artificial intelligence, the
common thing is that is does a diagnosis (for potential EMI
causes) on PCBs.
take my regards
Ralf
Thursday, March 31, 2005, 3:14:24 AM, you wrote:
MM> I was involved in selling one of the products Weston mentions, and here is
my
MM> experience:
MM> EMI folks would ask whether there were documented comparisons between the
UMR
MM> algorithm and actual measurements in a chamber for a real product. The
answer
MM> was no. I never knew of a seat sold based on the expert system prediction of
MM> radiated emissions. Everyone bought it for the DRCs.
MM>
=============================================================================
MM> Mike Mayer
MM> mwmayer@xxxxxxx
MM> -----Original Message-----
MM> From: si-list-bounce@xxxxxxxxxxxxx
MM> [mailto:si-list-bounce@xxxxxxxxxxxxx] On
MM> Behalf Of Beal, Weston
MM> Sent: Wednesday, March 30, 2005 1:09 PM
MM> To: si-list@xxxxxxxxxxxxx
MM> Subject: [SI-LIST] Re: [Expert system]
MM> Se-Ho You,
MM> Probably the best description of what the expert systems for EMI aim to
MM> do is from the UMR EMI consortium web page,
MM> http://www.emclab.umr.edu/consortium/Expert/pcbexpert.html
MM> "The goal of this project is to develop intelligent EMI modeling
MM> software that will work with automated printed circuit board layout
MM> tools to:=20
MM> "review and analyze printed circuit board designs;=20
MM> point out problems with the layout that could result in electromagnetic
MM> emissions or electromagnetic susceptibility;=20
MM> estimate levels of radiated EMI;=20
MM> anticipate ESD and radiated susceptibility problems; and=20
MM> provide circuit and board layout design advice.=20
MM> "Unlike numerical EM modeling software or EMC design rule checkers, the
MM> software being developed by the consortium examines boards in much the
MM> same manner as a professional EMC engineer. Potential common-mode and
MM> differential-mode EMI sources are identified and evaluated. The software
MM> looks for EMI antennas on or off the board and evaluates how hard they
MM> are being driven. It identifies any problems found with the board layout
MM> and estimates the impact of these problems on the radiated EMI from the
MM> system."
MM> I describe it like this; An EMI expert has experience and knowledge that
MM> enable them to find potential problems in a PCB design. These experts
MM> often manually scan a PCB design for configurations that they know can
MM> cause EMI. An expert system such as QUIET Expert automates the search
MM> that the expert would do so that the search is automated, consistent,
MM> and less prone to human error. An expert system cannot replace a field
MM> solver, engineering education, and good problem solving. What it does is
MM> relieve the EMI expert from the often mundane task of PCB layout
MM> scanning so that the human expert can spend more time on the "good
MM> stuff" like field solvers and lab measurements, etc. The design rules
MM> are only as good as the expertise of the human expert that specifies
MM> them.
MM> Another aspect of the expert system is that it points out a specific net
MM> or part and gives advice about how to fix the potential problem (design
MM> rule violation). This language should be understandable and applicable
MM> in terms that the layout design can use. Telling the layout designer
MM> that the PCB has radiation 8dB over the limit at 525MHz doesn't tell
MM> them what to change on the design, but "Net HISPEED_CLK is too close to
MM> the front panel connector. Move the trace at 100 mils from the
MM> connector." is something that the layout designer understands.
MM> The EMI DRC software that I know of are QUIET Expert (Mentor Graphics),
MM> EM Designer (Zuken), something from IBM, EMControl (Cadence; haven't
MM> heard anything on this lately. Is it still available?), and various
MM> internal tools. Does anyone know of more?
MM> Regards,
MM> Weston
MM> -----Original Message-----
MM> From: si-list-bounce@xxxxxxxxxxxxx
MM> [mailto:si-list-bounce@xxxxxxxxxxxxx]
MM> On Behalf Of yous@xxxxxxxxxxxx
MM> Sent: Tuesday, March 29, 2005 5:14 PM
MM> To: ray.anderson@xxxxxxxxxx
MM> Cc: si-list@xxxxxxxxxxxxx
MM> Subject: [SI-LIST] Re: [Expert system]
MM> To Dear Ray:
MM> I am not going to comment on the nomenclature either.
MM> I am looking at this expert system with research point of view. Question
MM> is what is expert system aiming at? I understood that it checks design
MM> rules.
MM> Does it mean expert system is quick-and-dirty checker? In the end, do we
MM> need to use full wave electromagnetics simulation tools? Does expert
MM> system "suggest" best design approach among many rules?
MM> Thank you very much.
MM> Sincerely,
MM> Se-Ho You
MM> Quoting Ray Anderson <ray.anderson@xxxxxxxxxx>:
>> The title of this post caught my attention. I'm not really going to=20
>> comment on where EMC research is going on, however the concept of=20
>> "Expert Systems" as related to EMC and SI has always struck me as a=20
>> misnomer.
>>
>> I've authored a number of SI related CAD tools and have used many fine
>> commercial tools whose end goal is to simplify a complex analysis or=20
>> design task enough that the average experienced engineer can use the=20
>> tools to their advantage. However calling any of these tools "Expert=20
>> Systems" or "Expert Tools" with the intent to imply that knowledge of=20
>> an expert in a field will be imparted to the user by merely utilizing=20
>> one of these tools is a pretty far stretch in my opinion.
>>
>> While I was at Sun our SI staff had numerous conversations on this=20
>> topic and concluded that these tools should be termed "Tools for=20
>> Experts" as opposed to "Expert Tools". You aren't going to hand a=20
>> complex tool like HFSS or Specctraquest for example, to a new grad or=20
>> 'Joe Garageshop' and expect to get the same results as someone with 20
>> years of experience in the field. (note: by citing a couple tools by=20
>> name in this paragraph I wasn't implying that those particular tools=20
>> are particularly difficult to use, but was just trying to give some=20
>> non-trivial examples)
>>
>> I'll accept that a tool for screening a design for EMC flaws and=20
>> red-flags may be a bit more amenable to the concept of an "Expert
MM> Tool"
>> than say a full-wave 3D FEM extractor, however, if the user is=20
>> depending totally on the tool rather than using it to supplement their
>> knowledge and experience, I can't really foresee a good outcome.
>>
>> anyway, that's my 2 cents for Friday...... :)
>>
>> -Ray
>>
>>
>> --
>> Raymond Anderson
>> Senior Signal Integrity Staff Engineer Product Technology Dept.
>> Package Engineering Group
>> Xilinx Inc.
>>
>>
>>
>>
>> yous@xxxxxxxxxxxx wrote:
>>
>> >To Dear All:
>> >
>> >Researchers in University of Rolla (UMR) and EMC related companies
MM> have
>> >developed "expert system" to check or detect EMI and EMC problem
MM> early.
>> >
>> >I wonder whether EMC engineers actually use. If they use, then I
MM> wonder to
>> >what applications they use.
>> >
>> >It looks like to me that there are lots of research going on in UMR,
MM> but
>> other
>> >than UMR, I cannot find any papers from any other research group.
>> >
>> >Thank you for your comments in advance.
>> >
>> >Thank you very much.
>> >Sincerely,
>> >Se-Ho You
>> >(cell) +1-303-818-4352
>> >
>> >
>> >
>>
>>
>>
>> ------------------------------------------------------------------
>> 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
>>
>>
MM> ------------------------------------------------------------------
MM> To unsubscribe from si-list:
MM> si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
MM> or to administer your membership from a web page, go to:
MM> //www.freelists.org/webpage/si-list
MM> For help:
MM> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
MM> List FAQ wiki page is located at:
MM> http://si-list.org/wiki/wiki.pl?Si-List_FAQ
MM> List technical documents are available at:
MM> http://www.si-list.org
MM> List archives are viewable at: =20
MM> //www.freelists.org/archives/si-list
MM> or at our remote archives:
MM> http://groups.yahoo.com/group/si-list/messages
MM> Old (prior to June 6, 2001) list archives are viewable at:
MM> http://www.qsl.net/wb6tpu
MM> =20
MM> ------------------------------------------------------------------
MM> To unsubscribe from si-list:
MM> si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
MM> or to administer your membership from a web page, go to:
MM> //www.freelists.org/webpage/si-list
MM> For help:
MM> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
MM> List FAQ wiki page is located at:
MM> http://si-list.org/wiki/wiki.pl?Si-List_FAQ
MM> List technical documents are available at:
MM> http://www.si-list.org
MM> List archives are viewable at:
MM> //www.freelists.org/archives/si-list
MM> or at our remote archives:
MM> http://groups.yahoo.com/group/si-list/messages
MM> Old (prior to June 6, 2001) list archives are viewable at:
MM> http://www.qsl.net/wb6tpu
MM> ------------------------------------------------------------------
MM> To unsubscribe from si-list:
MM> si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
MM> or to administer your membership from a web page, go to:
MM> //www.freelists.org/webpage/si-list
MM> For help:
MM> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
MM> List FAQ wiki page is located at:
MM> http://si-list.org/wiki/wiki.pl?Si-List_FAQ
MM> List technical documents are available at:
MM> http://www.si-list.org
MM> List archives are viewable at:
MM> //www.freelists.org/archives/si-list
MM> or at our remote archives:
MM> http://groups.yahoo.com/group/si-list/messages
MM> Old (prior to June 6, 2001) list archives are viewable at:
MM> http://www.qsl.net/wb6tpu
_____________________________________________
Ralf Bruening
Product Manager High Speed Design/Partner & Solution Business
Zuken EMC Technology Center
Vattmannstr.3
33100 Paderborn
Germany
Tel: ++49 (0) 5251 150 600 (direct -621)
Fax: ++49 (0) 5251 150 700
Cellular: +49 170 4228512 (alternative: +49 172 9070675)
Email: ralf.bruening@xxxxxxxx
Web: www.zuken.com
_____________________________________________
This message is for the designated recipient only and may contain
privileged, proprietary, or otherwise private information. If you have
received it in error, please notify the sender immediately and delete
the original. Any other use of the email by you is prohibited.
------------------------------------------------------------------
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:
