[SI-LIST] Re: compare ADS dispersion with H-Spice W-element.
- From: "D G" <dgun@xxxxxxxxxx>
- To: si-list@xxxxxxxxxxxxx
- Date: Sat, 21 Dec 2002 01:17:51 +0800
Ron,
I'm not a theoretician either (at least, that's not what I get paid
for ;). Most of my experience is with low-loss substrates and very
low signal density. ADS and Momentum work well in those cases. I
don't do much with FR4 other than DC and baseband, so I can't really
comment. Many of my coworkers have had success designing high-frequency FR4
and Rogers PCBs with ADS and Momentum so it must
work well enough.
The ADS dispersion models are based on published models. If
Kuznetsov has published papers comparing his model to these other
models, then that would be a good basis to go by. Otherwise, the
best thing is probably to compare measured results with simulation
for your application.
- Daniel
From: Ron Miller <rmiller@xxxxxxxxxxx>
> Hi Dan
>
> Just want to touch base with you since you are using ADS too.
>
> My co-worker Dmitri Kuznetsov invented and owns the W-element
used
> in Hspice. He is convinced that the W-element provides better
fidelity
> than the models for transmission lines in ADS.
>
> Would you care to comment? I am not a deep enough theoritician
to
> be able to make a comparison.
>
> Ron Miller
>
>
> -----Original Message-----
> From: D G [mailto:dgun@xxxxxxxxxx]
> Sent: Monday, December 16, 2002 12:09 PM
> To: si-list@xxxxxxxxxxxxx
> Subject: [SI-LIST] Re: Dispersion
>
>
>
> Jeff,
>
> I thought I answered your question about ADS. I can add a
little more
> detail. The documentation is at:
>
> http://eesof.tm.agilent.com/docs/
>
> Microstrip lines include dispersion. You can select from three
dispersion
> models. However, some of the short components (like bends) do
not include
> dispersion.
>
> Striplines do not include dispersion (as far as I can tell).
>
> PCB multicoupled lines do not include dispersion.
>
> Momentum simulation includes dispersion.
>
> One trick you can use to simulate permittivity change with
frequency is to
> use a frequency-dependent equation for Er in the substrate
models. I've
> done this in the past. Something like
"Er=er_nom+freq*disp_factor" would
> work if the change in Er is linearly dependent on frequency.
You can do the
> same with loss tangent (TanD).
>
> - Daniel
>
> From: "Loyer, Jeff" <jeff.loyer@xxxxxxxxx>
> > OOOOOOOOOH! Now I'm getting somewhere, I think. Attached
(for Bryce,
> anway. Anyone else who would like the data will have to e-mail
me
> individually since it won't get through the reflector) you will
find the
> data which started this whole fiasco (or at least some of it).
In it, I am
> VNA'ing a long (~14") microstrip trace (actually 1/2 of a
differential
> pair), and comparing those measurements with simulations
(Hspice, using
> W-element for the T-line).
> >
> > What I noticed was that measured S21 phase derivative
varied with
> frequency. See "1tz_S21P" sheet.
> >
> > In it (top graph) you will note that the simulation and
measurement phases
> line up at the beginning and ends, but not in the middle.
Trying to explain
> this was my original intent. Below S21 phase, I've also plotted
a
> calculated Tp. For the measurement, this clearly varies with
frequency (Tp
> gets less at higher F).
> >
> > Some things I would tend to conclude at this point are:
> > 1) Yes, I'm seeing dispersion, as defined by Bryce (and his
"official
> definition" seems to be an accurate, concise definition,
based on what I've
> read and heard).
> > 2) It is possible to have losses without dispersion, at
least in
> simulations. On page "1tz", it's clear that S21
magnitude is varying with
> frequency as a result of losses, both in measurements and
simulations. But,
> there is no dispersion apparent in the simulations.
> > 3) Dispersion is causing my Tp for this 14" line to
vary from 2.35nS (at
> low F) to 2.15nS (at 6GHz).
> >
> > So, of my original 2 questions, the 2nd one is answered
(how dispersion
> would show itself in measurements). The first question - can
Hspice or ADS
> simulate dispersion effects - remains open.
> >
> > P.S.
> > I'm not sure there exists enough disk space in the world
for a "book" this
> forum might produce :-)
> >
> > Thanks again for everyone's help...
> >
> > Jeff Loyer
> >
> >
> >
> > -----Original Message-----
> > From: Bolton, Bryce D [mailto:BoltoBD@xxxxxxxxxxxxxxxxxxxxxx]
> > Sent: Friday, December 13, 2002 9:59 AM
> > To: si-list
> > Subject: [SI-LIST] Re: Dispersion
> >
> >
> >
> >
> > Hi Folks,
> >
> > I often enjoy hearing discussions and interpretations about
concepts and
> > practices. It adds a huge amount of understanding to the
theory. My
> > favorite books and mentors show the theory and verbalize it.
> >
> > Given any system H(s), we can derive:
> >
> > - insertion loss, also known as magnitude response.
> > - phase response
> > - group delay (derivative of phase)
> > - dispersion (derivative of group delay) <- official
definition of
> > dispersion.
> >
> > If I were to observe a phase or group delay plot, I would
have an idea of
> > how much spectral component spreading or dispersion would
occur at each
> > frequency. But, my eyes aren't that good at doing
derivatives, so it
> might
> > be easier to simply look at the dispersion plot. So, when
we say
> > "dispersion" we are conceptually talking about
this general spreading,
> that
> > we know exists in the 2nd derivative plot, even though most
of us don't
> make
> > dispersion plots of an H(s) every day. One could talk
about group delay
> or
> > phase delay just as well, and say that they cause
dispersion. It's just
> > another way of talking about the same property. But, if we
are talking
> > about dispersion specifically, then one might want to know
exactly how
> much
> > spectral spreading was introduced by a given trace, etc.
And, one would
> > want to know how to minimize that number to meet a spec.
> >
> > I think it would be quite valuable if we could share some
examples of
> actual
> > measurements--especially with plots and descriptions.
> >
> > Signal integrity is broad, and I often think that this list
would be
> > benefited by an online book that just had everything
clarified. We could
> > just say--look at section 1.2 to find your answer. We
could improve the
> > "book" as a group.
> >
> > Thanks,
> > Bryce
> >
> >
> >
> >
> >
> >
> > -----Original Message-----
> > From: C Deibele [mailto:deibele@xxxxxxx]
> > Sent: December 13, 2002 10:18 AM
> > To: Fred Balistreri
> > Cc: Loyer, Jeff; si-list
> > Subject: [SI-LIST] Re: Dispersion
> >
> >
> >
> > I disagree.
> >
> > Take a trace, with two ports, and put an open circuited
"T" in the
> > middle of it. This is a perfect source of dispersion. The
entire
> > basis for schiffmann phase shifters is easily realizable on
pcb's.
> > Loss has nothing to do with these forms of dispersion.
> >
> > Dispersion means that the phase velocity is different from
the
> > group velocity. or the slope of the curve of omega versus
k
> > (frequency versus wavenumber) is different from the ratio
of omega
> > divided by k.
> >
> > Craig
> >
> >
> > >===== Original Message From Fred Balistreri <fred@xxxxxxxxxxxxx>
=====
> > >Except that in a PCB it is not possible to have
dispersion without loss.
> > >Dispersion without loss is defined only for wave
guides. In a PCB the
> > >resulting change in er carries with it losses in the
dielectric
> > >characterized
> > >by the loss tangent. The losses and resulting change in
er are a
> > >function
> > >of frequency, which results in dispersion and
attentuation. There is
> > >also
> > >additional attenuation with frequency caused by the so
called skin
> > >effect.
> > >I believe the loss in the conductor is often associated
with the
> > >attenuation
> > >constant. While the losses in the dielectric are often
attributed to the
> > >phase constant. The propagation constant is a function
of the two and is
> > >frequency dependent. The degree of change depends on
the materials used.
> > >
> > >Best regards,
> > >
> > >C Deibele wrote:
> > >>
> > >> Jeff,
> > >>
> > >> You bring up some very good points -- The real
definition of dispersion
> > is
> > the
> > >> fact that the phase velocity (omega divided by k)
is different than the
>
> > group
> > >> velocity (d omega / dk). When the phase velocity
is different from the
>
> > group
> > >> velocity, an impulse "disperses" or gets
"fatter" when looked in the
> time
> > >> domain.
> > >>
> > >> In fact, even though I don't like the textbook in
general, Jackson's
> book
> > >> "Classical Electrodynamics" has a great
theoretical treatment of the
> > subject
> > >> that can be fairly well understood even with a
cursory overview.
> > >>
> > >> Dispersion, in general, has absolutely nothing to
do with losses. For
> > >> example, a waveguide (here I mean a pipe, circular
or rectangular), is
> > >> absolutely dispersive. the phase velocity is
different from the group
> > >> velocity.
> > >>
> > >> Regarding loss -- when the loss varies w.r.t.
frequency, this causes
> > >> dispersion. While the loss variance may be
accounted for in any myriad
> > of
> > >> techniques, the end result is dispersion.
> > >>
> > >> In fact, I've designed lots of equalizers to rid
systems of dispersive
> > >> properties.
> > >>
> > >> I agree, if a material is lossy, it is dispersive.
This is easy to
> > measure
> > in
> > >> the laboratory. Take a short piece of coax and
put in a *great* square
>
> > wave,
> > >> and measure the rise time on a scope. Now, take a
100 meter section of
>
> > coax
> > >> and insert it in place of the short piece. The
rise time measurement
> > will
> > be
> > >> *much* worse. This is an effect of dispersion.
> > >>
> > >> In essence, lossy implies dispersive. but
dispersive does _not_ imply
> > lossy.
> > >> A perfectly conducting waveguide is a great
example of this property.
> > >>
> > >> And yes, knobbing is perfectly legal. While this
isn't the perfect
> > definition
> > >> for measuring the dispersion, one can see the
phase dispersive
> qualities.
> >
> > The
> > >> magnitude is also very important. So, one has to
consider the
> bandwidth
> > of
> > >> the source -- and relate that back to the
measurement.
> > >>
> > >> So, if one corrects the magnitude to be flat, and
corrects the
> "knobbed"
> > phase
> > >> to be flat, the system is absolutely
non-dispersive.
> > >>
> > >> Craig
> > >>
> > >> >===== Original Message From "Loyer,
Jeff" <jeff.loyer@xxxxxxxxx>
=====
> > >> >This part of the thread (discussion of
dispersion) began when I asked
> > the
> > >> question below. It seems that we are back to the
original question.
> > >> >
> > >> >ORIGINAL QUESTION:
> > >> >When you use the term "dispersive",
are you talking about losses
> > (resistive,
> > >> skin effect, dielectric), or about differences in
phase velocities
> (page
> > 170
> > >> of Pozar's book)? I've heard others refer to loss
effects as
> dispersive
> > and
> > >> have had
> > >> >confusion as a result. Are both uses of the
term "dispersive"
> correct?
> > >> >
> > >> >The explanation of how to measure dispersion
(S21 magnitude) implies
> you
> > >> believe "dispersion" and what I would
have termed "effects of conductor
> > and
> > >> dielectric losses" are the same. I have
trouble with that, since
> > stripline
> > >> insertion loss
> > >> >magnitude definitely varies with F, and that
effect is explained
> without
> > >> dispersion. I believe "dispersion" is a
separate effect than conductor
> > and
> > >> dielectric losses. The only tie between them that
I've heard of is
> that
> > Steve
> > >> Corey (who I am
> > >> >loath to contradict) stated "if a
material is dispersive, it is also
> > lossy".
> > >> It may be that the converse holds (if a material
is lossy, it is also
> > >> dispersive), but I believe the 2 effects are
separate (even if one
> can't
> > occur
> > >> without the other).
> > >> >Maybe Steve would clarify this?
> > >> >
> > >> >I couldn't follow the explanation of
"knobbing" electrical delays
> until
> > S21
> > >> phase is flat. Is that legal? ;-)
> > >> >
> > >> >Jeff Loyer
> > >>
> > >> Craig Deibele
> > >> Spallation Neutron Source
> > >> 701 Scarboro Road
> > >> Room 301 MS 6473
> > >> Oak Ridge, TN 37830
> > >> mailto:deibele@xxxxxxx
> > >> office: +1 865.574.1969 cell: +1 865.719.4381
fax: +1 865.241.6739
> > >>
> > >>
------------------------------------------------------------------
> > >> 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 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
> > >>
> > >
> > >--
> > >Fred Balistreri
> > >fred@xxxxxxxxxxxxx
> > >
> > >http://www.apsimtech.com
> >
> > Craig Deibele
> > Spallation Neutron Source
> > 701 Scarboro Road
> > Room 301 MS 6473
> > Oak Ridge, TN 37830
> > mailto:deibele@xxxxxxx
> > office: +1 865.574.1969 cell: +1 865.719.4381 fax: +1
865.241.6739
--
Daniel
ZZZ-dgun-ZZZ-@xxxxxxxxxxxxxxx
(remove the Z-'s to reply--they're what I do when I read spam)
--
__________________________________________________________
Sign-up for your own FREE Personalized E-mail at Mail.com
http://www.mail.com/?sr=signup
Meet Singles
http://corp.mail.com/lavalife
------------------------------------------------------------------
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 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:
