Hi Yuriy,
Thanks for the very enlightening paper. Yes, I also suspect that the RLC
models given by some tools are based on methods other than field-solving,
such as tline model, PEEC, and other models (eg, the one proposed in your
paper).
Anyway, I would say partial inductance is a very "individual - dependent"
concept. Different people can get different results. This is due to the
differences on the loop areas of their setups. Pls correct me if I am
wrong.
Best Regards,
Tramp
On Fri, May 22, 2020 at 6:48 PM Yuriy Shlepnev <shlepnev@xxxxxxxxxxxxx>
wrote:
Hi Tramp,
Let's start with a transmission line segment made of PEC conductors (1D
model) - there is no voltage drop in the conductors. However port voltages
at the opposite ends of the segment may be different (depends on the length
and wavelength). This demonstrate physics of the distributed models. Now
make just one of the conductor PEC, the other is real conductor. Again,
transmission line model will predict local voltages at the opposite ends.
But can we use it to measure voltage between the opposite signal terminals
knowing that there is no voltage drop in PEC? The answer is yes, but only
in
case if the segment length is much smaller than the wavelength. Check it
with a numerical experiment.
PDN simulation tools are based either on 2D Maxwell's equations or use 2D
transmission plane models that are similar to the transmission line models.
They predict local voltages between two metal planes. If one of the planes
is PEC and another is real, voltages at two ports can be used to compute
voltage drop between two points on the real plane, but only in case if they
are electrically close (distance is much smaller than wavelength). Again,
verify it with numerical experiment.
As far as I know, "hybrid full wave" or 2.5D (in this context) tools are
based on transmission line + transmission plane models.
Now, can additional PEC plane be used in such "hybrid" tool to measure the
voltage drop in one of the planes? The answer is yes, but with two
conditions. The first condition is the distance between two points - it
should be much smaller than wavelength. The second condition is parallel
transmission planes must be coupled. It means that voltage drop between two
real planes (first plane pair) creates voltage drop between one the real
planes and PEC plane (second plane pair). Coupled transmission plane model
can be used for that -
https://www.researchgate.net/publication/261263227_Coupled_2D_Telegrapher's_
equations_for_PDN_analysis
A simple numerical experiment can reveal if something like that is
implemented in your tool. Create stackup with 3 solid rectangular planes -
all real or one of them is PEC. Add ports between both pairs and see if
voltage applied between two pairs create voltage drop in another pair.
Planes should be solid for that experiment - no vias or holes. Now create a
hole and repeat the experiment.
In 3D solvers voltage drop can be measured between any two points in space
under the field conservativeness condition - the distance between the
points
should be much smaller than wavelength. Voltage drop in a plane can be
measured with ports referenced to any suitable point in space. It can be
point on existing metal object or on added PEC object. If we care only
about
difference of the voltages at two points, ports may be reference to a point
in empty space (a virtual reference point). Transition to very low
frequencies may enable very weird things, if you have microwave engineering
background, but only if the model permits :-)
Best regards,
Yuriy
Yuriy Shlepnev, Ph.D.
President, Simberian Inc.
www.simberian.com
Simbeor THz - Accurate, Productive and Cost-Effective Electromagnetic
Signal
Integrity Software to Design Predictable Interconnects!
Simbeor SDK - The Industry-First Signal Integrity Analysis Automation Kit!
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
On
Behalf Of Tramp
Sent: Thursday, May 21, 2020 11:34 PM
To: Chase, Yun
Cc: wang1.1@xxxxxxx; si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: how to use s-parameter to simulate power-aware SI
issues.
Actually in lab the concept of adding a PEC layer can also be done, I
think. Just keep in mind that the VNA measurement is localized and the
distance of localization increases as freq goes low.
Back to the RLC extraction question: Do those 2.5D fullwave tools
(HyperLynx fast 3D, SIWave, etc) actually extract the sparam of the ground
net by setting ports wrt. PEC then convert the sparam into RLC?
Best Regards,
Tramp
On Thu, May 21, 2020 at 5:40 AM Chase, Yun <Yun_Chase@xxxxxxxxxx> wrote:
Just catching up on this particular thread... few questions and pointsis
were made that caught my interest...and some thoughts came to my mind...
that doesn't happen a lot but self-isolation seems to have that effect
lately...
The original poster asked "What should I do if I want to probe the
ground's voltage...?"
Well... my question in response to that is... probe Ground Voltage with
respect to what???
Probe Ground Voltage with respect to:
1) the Grounding strap on the lab bench?
2) the Ground socket in the wall outlet?
3) the 10 foot (3 m) flag pole stuck near the front entrance of
your building?
What ever your answer to that may be... none of that will matter to the
circuit on the board... because everything that matters to your circuit
being measured or referenced to that exact Ground Net that one so wishesto
measure... and this may be surprising to some but it is not measured withis
respect to the flag pole in your parking lot. :-)
The concept of Ground bounce is an interesting one... many seem to get
fixated on it and can't seem to shake it off...
Power and Ground is a fully coupled system...it's like marriage... If
your spouse is happy then you will be happy... if your spouse is sad...
then you will be sad... you can't separate the two. So if your Ground
that bad... then it is all but likely that the problem will show up inyour
Power rail... so rest assured... if you model the PDN correctly... youneed
should see Ground problems in your PDN analysis... meaning there is no
for anyone to go and probe the Ground with respect to the flag pole.family
Now, is there some value to "model" (not to be confused with "probe") the
Ground separately? Yes! If you are in mid-design stage... and want to
get a check on/quantitative feel for the Ground net integrity for better
design optimization... then sure there are tools to model the Ground
parasitics, separate from the Power and Signal parasitics. One such
of tools are called Quasi-static EM Field Solvers. We, here atwithout
Mentor/Siemens has one.. it's called *HyperLynx Fast 3D Solver*... Ansys
has one, called Q3D.... I believe Cadence has one too in their PowerSI
family of tools. Some in the Academia offer free ones too, I believe...
These Quasi-static field solver tools will solve the quasi-static form of
the Maxwell's Eqns which conveniently decouples the electric and magnetic
field equations such that you can solve the associated E and H fields for
any free standing single conductor all by itself.... this means you can
solve the RLC parasitics of a single Ground conductor all by itself
having the need for a second conductor like Power or Signalinterconnects.
This is very convenient when *partial element* values are needed fordesign
purposes. But do note that this comes with an understanding that yourthe
interconnect is electrically short enough that quasi-static assumptions
would hold true.
Some asked about using a third conductor as a PEC reference to measure
Ground parasitics. That would be a requirement, if you were using 2.5Dmodes
hybrid EM field solvers that were mentioned earlier such as Ansys SIWave
and Cadence PowerSI.... we have one here at Mentor/Siemens...and it is
called *HyperLynx Hybrid Solver* (give naming credit to our marketing
gods...) Since these hybrid EM tools solve the full-wave propagating
of the transmission structures in the DUT... it requires you to have at aparasitics*
minimum, a 2 conductor system to measure and capture the *loop
of the interconnect. So if you want to measure the Ground parasitics,you
would need a second conductor... so the PEC reference serves thatpurpose.
But one thing to be cautious here is that by adding this externalreference
conductor... you have essentially changed your physical structure. Youradded
real design does not have this PEC metal... but you have artifically
this, for the sake of Ground parasitic extraction in your EM tool. So aartifically
great care must be taken to ensure that this PEC metal doesn't
load up your system too much that it negatively impacts the accuracy ofin
your model. But at the end of the day when this model is actually used
your transient circuit simulation... all you will care about is thevoltage
measurements between Signal and Ground.... and the voltage measurementsis
between Power and Ground.... but it is highly unlikely you will be
concerned with the voltage from the Ground to the PEC conductor. That
all but meaningless in my view in the interest of your circuitsimulation.
than
OK, I think it's time for me to stop and change my mask for the day...
If you own any of the tools above, contact your favorite AEs at your
favorite tool vendor to get the good help you need... if you have any
questions about our HyperLynx Suite of tools, feel free to reach me
offline.... unless you know a different Mentor AE that you like better
me... ;-)might
Stay Healthy.
Regards,
Chase
___________________________________________________
Signal and Power Integrity | Mentor, A Siemens Business | Boston, USA
+1-508-565-8080
yun_chase@xxxxxxxxxx
https://support.sw.siemens.com/
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
On Behalf Of Tramp
Sent: Wednesday, May 20, 2020 6:02 AM
To: wang1.1@xxxxxxx
Cc: si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: how to use s-parameter to simulate power-aware SI
issues.
Interesting topic...
So it seems there are at least two ways of separating ground
characteristics from that of power or signals:
Way#1: extracting RLC network of ground using SIWave (or XtractIM, etc).
Way#2: set up ports on ground and reference them to PEC.
This begs one question: is the RLC representation obtained in way#1
actually based on way#2?
My suspicion is that in principle, it is. Otherwise, the RLC network
extracted based on empirical equations would be way too rough.
In addition, to obtain the accurate characteristics of ground, there
be some special skills in terms of port setup, etc., or the results maynot
be good enough.
Best Regards,
Tramp
On Thu, May 14, 2020 at 6:48 PM TangH <wang1.1@xxxxxxx> wrote:
Hello everyone,pin.
Recently i have a study about PI+SI co-simulation. I have extracted
the PDN and got the 3 ports S-parameter file. The port 1 for cpu
power, port 2 for DDR power and port 3 for VRM. So in the schematic
diagram (using ansys
designer) I connected port 1 and 2 to corresponding ibis-model's power
When the simulation is done. I noticed that level '1' is much worse
than level '0'. In my schematic diagram design, each port is only one
pin visible and is is ref GND by default, and the ibis-model ref GND
too.It means all port's GND and ibis-model's GND is short, the current
does'nt go through the ground plane any more.
What should i do if i want to probe the ground's voltage in each port,
and see the impact of level '0'.
Thanks
Tangh
------------------------------------------------------------------
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
------------------------------------------------------------------
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
