Bill,
Google is your friend Lol!
https://en.m.wikipedia.org/wiki/All_models_are_wrong
About your other concern, I hadnât really noticed, but thanks anyway! :)
Bert
Sent from my iPhone
On Mar 15, 2019, at 3:58 PM, Bill Hargin <billh@xxxxxxxxxx> wrote:
Thanks, Bert and Yuriy. (Reminds me of a kids' show ...) Good discussion.
I wanted to make a public apology to Bert here. In my previous post, I
leveraged some information from some of his white papers and I used some
things that I adapted from his work without attribution. (Bad form on my
part, and I apologize.)
I was debating whether to respond to Yuriy based on my experience in the PCB
manufacturing part of the world. Bert's quote of EP Box (who is that?)
summarizes what I was going to say, relative to Rz vs. Ra.
Bill Hargin
Director of Everything
Z-zero ⪠Innovative PCB Stackup Design ⪠www.z-zero.com
billh@xxxxxxxxxx ⪠425-301-4425 ⪠Skype: bill.hargin
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx <si-list-bounce@xxxxxxxxxxxxx> On Behalf
Of Bert Simonovich (Redacted sender "bertsimonovich" for DMARC)
Sent: Friday, March 15, 2019 11:27 AM
To: hungdn.hcmut@xxxxxxxxx; shlepnev@xxxxxxxxxxxxx
Cc: si-list@xxxxxxxxxxxxx; 'Bert Simonovich'
<lsimonovich@xxxxxxxxxxxxxxxxxxxxx>
Subject: [SI-LIST] Re: Megtron6 vs. Megtron 7 vs. Isola MT40
To add to Yuriy's Comments, and try to explain why Cannonball model works.
Actually the roughness parameters Ra, Rq and Rz are derived from the
roughness profile, as measured with a mechanical prolifometer. A mechanical
prolifometer records the roughness peaks and valleys as it is dragged across
the surface of the sample being measured. It is essentially a 2-D
measurement. Think of a record player needle being dragged across a vinyl
record.
Ra is defined as the average roughness of the measured profile over the
sample length. Rq is the RMS value of the profile shape over the sample
length. And finally Rz is the 10-point mean value, which is the sum of the
average of the 5 highest peaks and 5 lowest valleys, of the profile over the
sample length.
As part of the electrodeposited process, the matte side of the foil is the
side facing the CU sulfate solution, as it gets electro-deposited onto the
large rotating drum. The drum side is the side in contact with the drum. The
drum side takes on the profile of the stainless steel drum and is fairly
smooth compared to the matte side. Afterwards the foil goes through a nodule
treatment process where it deposits tiny nodules, usually on the matte side
of the foil or the drum side. The standard nodule treatment is done on the
matte side but when applied to the drum side it is known as reverse treated
foil, or RTF. Almost all foil suppliers will give Rz numbers for their
treated side of the foil, and Ra for the untreated or drum side.
Yuriy is correct, in that there is no exact conversion between Rz and Rq.
But, in the words of E.P.Box, "All models are wrong, but some are useful", if
we assume Rz represents the average peak-valley profile over the sample
length, and if we model the peak-valley profile as a triangular waveform
profile, with the peak-peak height equal to Rz, then the RMS value of that
triangular profile over the sample length is simply Rz/2/sqrt(3). I use this
as a rough approximation for Rq from Rz and vice versa when needed.
My Cannonball model relies on a stack of 14 equal sized spheres stacked in a
close-packing of equal spheres arrangement over a square tile base with an
area of 36(r^2) - where r is the radius of the sphere. The unique thing about
the Cannonball model is that it simplifies the Huray equation to just 1
parameter equal to the radius of a sphere, for an N-equal size sphere model.
Because the area of the square tile base is determined directly from the
radius of the sphere, they cancel out in that part of equation, thereby
reducing the Hall-Huray surface ratio, or roughness factor to a constant =
1.56(pi) or ~ 4.6. This is the "SR" parameter for Ansysis and Cadence tools.
In Simbeor, RF1, as defined in the tool, will be ~ 8.33. By the very nature
of the Huray equation, these numbers are constant - regardless of the size of
the sphere.
So even if you want to use the material identification method, as Yuriy
described to try and come up with material properties, all you have to do is
use the appropriate constant for SR or RF1 (depending on the tool) and adjust
the radius parameter until you get a fit to measurements. Personally I have
found that when I adjust Dk due to roughness from engineering data sheets, as
described in my DesignCon 21017 paper, then tune the respective radius
parameter, you can get excellent results from essentially laminate supplier's
data sheets. Thet's it. Rz is not in the picture yet!
But, in my opinion, the material extraction process is really only good for
the samples from which they were extracted, the quality of the measurement
and for the software used for the fitting. They do not represent the
intrinsic values of the material, especially the roughness of the copper foil
used. All results really show is how well the software fits measurements.
The reason I say this is laminate suppliers usually have 2 or more copper
foil suppliers from which they buy from and each of them will have slightly
different roughness based on their nodule treatment process. And furthermore,
there will be oxide alternative (OA) treatment variations, depending on where
the final design board will be fabricated. Each board shop has their own OA
process, and there is no guarantee the product board you carefully
characterized will be built from the same fab shop forever.
Over the last few years, I have also shown that you can indeed get excellent
results using Rz parameters from foil manufacturerâs data sheets in
conjunction with my Cannonball model to describe the roughness profile. In my
method, Rz is only used for the 10-point mean peak-valley height then my
algorithm converts that into a sphere radius to complete the Cannonball-Huray
simplified model, as described above. If the Rz is known, then the radius of
a sphere can be approximated as 0.06(Rz).
But you have to get the foil roughness numbers from laminate supplier, NOT
board shops. In the example of my DesignCon2019 paper, (
http://lamsimenterprises.com/Awards_and_Publications.html ;), I knew the
actual supplier of copper foil used on the core laminates of the test board.
When I used exact numbers from the foil mfr data sheet, and the effect of OA
on RTF foil, I got excellent results for both insertion loss and phase delay,
just by using data sheet values!
Mentor Hyperlynx and Polar Instruments field solvers have adopted the
Cannonball-Huray model within the tool so all that is needed is Rz. In this
case all you need to tune for material identification is Rz parameter in the
tool, if that is your end goal.
In the end, even though the Cannonball model may be âtechnically wrongâ
from reality, it is still just a model and useful, depending on what you are
trying to accomplish. The Hammerstad model, which has been used for many
years, by the way, is also "technically wrong" but was useful in its time. It
relied on RMS profile of the peak-valley triangular profile of the model, but
because of model limitations it loses accuracy after 3-15 GHz, depending on
the roughness of the foil.
The bottom line is my Cannonball-Hurray model gives you the simplicity of use
as the Hammerstad model, but with better accuracy of Huray model, regardless
what method you like to follow.
Bert Simonovich
Signal/Power Integrity Practitioner | Backplane Specialist | Founder
Email:Lsimonovich @lamsimenterprises.com LAMSIM Enterprises Inc.
Web Site: http://lamsimenterprises.com
Blog: http://blog.lamsimenterprises.com/
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On ;
Behalf Of Hung Dang
Sent: 15-Mar-19 12:31 AM
To: shlepnev@xxxxxxxxxxxxx
Cc: dmarc-noreply@xxxxxxxxxxxxx; si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: Megtron6 vs. Megtron 7 vs. Isola MT40
Thank you so much, Bert, Yuriy, Bill.
I will try to config on my simulation.
Và o Th 6, 15 thg 3, 2019 vaÃâ¬o luÃÂc 04:34 Yuriy Shlepnev <
shlepnev@xxxxxxxxxxxxx> Ãâã viết:
Hi Hung,
You need measurements for at least 2 line segments to identify 2
parameters in the Huray model with either GMS-parameters, SPP Light,
Gamma-T or eigenvalue techniques - just pick your method. There are no
other scientific ways to come up with the model parameters, even for
the worse-case scenario.
To start with, you can use the numbers from the publications I cited below.
The actual losses may be higher or lower - it will depend on the
copper foil and PCB manufacturer. Fair comparison of dielectric
performance can be done only with the same type of copper foil for the
same PCB manufacturing process.
There are two mechanical parameters that approximately define the
frequency of the skin effect onset on the rough surface - those are
peak-to-valley parameters Ra and Rq (or Sa, Sq if measured optically).
It follows from the physics of the skin-effect. When skin layer depth
is becoming comparable with the Ra or Rq, the losses due to the
roughness start growing (rough surface absorbs more energy). The
growth rate and maximal value are defined by the surface shape - there
are no mechanical parameters from manufacturers that can be used to
characterize that. One-ball or one-level Huray model has
2 parameters - ball radius defines the skin-effect onset and
Hall-Huray surface ratio or roughness factor defines the maximal
possible increase in losses due to absorption by the surface. There
are no ways to identify both parameter from the mechanical or optical
measurements (though there are some attempts). The reason is very
simple - the model describes absorption by a small sphere (or multiple
separate spheres) - it is a solution for a "spherical cow in vacuum"
(Paul Huray actually used this analogy at our presentation at
DesignCon 2015). Though, the formula captures the physics of the
skin-effect onset process. The stacks of balls in the original paper
was purely for illustrative purpose - no connections with the loss
increase formula. Though, unfortunately, some people took the
illustration too literally and it went as far as the model parameters
derivation from just Rz.
Considering availability and use of Rz - this parameter is simply not
relevant to the electrical roughness characterization. Ra or Rq cannot
be derived from Rz - as simple as that. See mathematical definitions
at https://en.wikipedia.org/wiki/Surface_roughness. It is easy to ;
proof or illustrate. Two surfaces with same Ra may have very
different Rz. Or, the other way around, two surfaces with identical Rz
may have very different Ra values - it follows from the definition.
Best regards,
Yuriy
Yuriy Shlepnev, Ph.D.
President, Simberian Inc.
2629 Townsgate Rd., Suite #235, Westlake Village, CA 91361, USA Office
+1-702-876-2882; Fax +1-702-482-7903 Cell +1-206-409-2368; Virtual
+1-408-627-7706
Skype: shlepnev
www.simberian.com
Simbeor - Accurate, Productive and Cost-Effective Electromagnetic
Signal Integrity Software to Design Predictable Interconnects!
2010 and 2011 DesignVision Award Winner, 2015 Best In Design&Test
Finalist
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx
[mailto:si-list-bounce@xxxxxxxxxxxxx]
On
Behalf Of Hung Dang
Sent: Wednesday, March 13, 2019 7:25 PM
To: shlepnev@xxxxxxxxxxxxx
Cc: dmarc-noreply@xxxxxxxxxxxxx; si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: Megtron6 vs. Megtron 7 vs. Isola MT40
Dear Yuriy,
As you said: "The foil manufacturers may provide some mechanical data
that are not relevant to the electrical behavior of copper"
I'm extracting 25Gbps differential traces by ANSYS HFSS, I'm confused
how to choose the suitable roughness model of traces.
What are the parameters or specification which we can base on that? I
need to cover the worst case of losses due to surface roughness.
Thanks.
VÃâ =o Th 5, 14 thg 3, 2019 vaû-o luÃȉââc 06:49 Yuriy Shlepnev
<
shlepnev@xxxxxxxxxxxxx> ô-Ãâ Ãâ viÃÂâââét:
Hi Dan,the
Meg6 was used in the analysis to measurement validation project
EvR-1 -
results were reported last year at DesignCon 2018 - see #2018_01 andThe
#2018_07 at http://www.simberian.com/AppNotes.php Complete solutions ;
for the material model identification and automation scripts for
that project are available on request.
Meg7 was extensively studied in another project with the results
reported at DesignCon 2019 - see #2019_01 at
http://www.simberian.com/AppNotes.php
goal in this project was to build statistical models for dielectricitself
and for the conductor roughness.larger
Note, that in both cases the identified Dk and Df were close to the
data provided by manufacturer. However, the actual traces exhibited
much
losses due to the conductor surface roughness. The boardwere
manufacturers in both studies did not have any numbers for the
electrical characterization of the copper roughness. The foil
manufacturers may provide some mechanical data that are not relevant
to the electrical behavior of copper. In addition, the copper is
also treated by the board manufacturer - there
no data on this process either. Any guess in such case would bethan
better
nothing, but if you want accurate characterization and comparison,have
you
to do the model identification with the measurements.Finalist
Best regards,
Yuriy
Yuriy Shlepnev, Ph.D.
President, Simberian Inc.
2629 Townsgate Rd., Suite #235, Westlake Village, CA 91361, USA
Office +1-702-876-2882; Fax +1-702-482-7903 Cell +1-206-409-2368;
Virtual +1-408-627-7706
Skype: shlepnev
www.simberian.com
Simbeor - Accurate, Productive and Cost-Effective Electromagnetic
Signal Integrity Software to Design Predictable Interconnects!
2010 and 2011 DesignVision Award Winner, 2015 Best In Design&Test
designs
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx
[mailto:si-list-bounce@xxxxxxxxxxxxx]
On
Behalf Of Dan Bostan (Redacted sender "dbostan" for DMARC)
Sent: Wednesday, March 13, 2019 11:27 AM
To: si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Megtron6 vs. Megtron 7 vs. Isola MT40
Does anyone have direct experience with all of those materials, for
above 32Gbaud?I used successfully Megtron6 in the past, but I am
looking for alternatives.Any pricing information would be
appreciated, as well.Cheers,Dan Bostan
------------------------------------------------------------------
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 forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
List archives are viewable at:
//www.freelists.org/archives/si-list
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 forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
--
----------------------------------------------------------------------
------
---
MS Hung-Dang Ngoc
Email: hungdn.hcmut@xxxxxxxxx <phandainghiamnc@xxxxxxxxx>
Phone: +84 1682 405 564
Skype/Facebook: hungdn.hcmut
------------------------------------------------------------------
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 forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
--
-------------------------------------------------------------------------------
MS Hung-Dang Ngoc
Email: hungdn.hcmut@xxxxxxxxx <phandainghiamnc@xxxxxxxxx>
Phone: +84 1682 405 564
Skype/Facebook: hungdn.hcmut
------------------------------------------------------------------
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 forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
---
This email has been checked for viruses by AVG.
https://www.avg.com
------------------------------------------------------------------
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 forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu