From: Eric Bogatin ([email protected])
Date: Wed Dec 22 1999 - 13:15:25 PST
Since no Ansoft tech folks responded to this last note about their solver, I
thought I would throw in my 20 milli cents worth of comments. Jim Foppiano
mentioned in his initial post that he uses the Ansoft field solver to
extract frequency dependent lossy line models for HSPICE. The Ansoft tool
Jim used for the 2D analysis of his transmission lines was Ansoft's Maxwell
2D Extractor, bundled in the SI2D suite. It is a finite element based field
solver.
When you select L and C analysis, it assumes all the currents are surface
currents, and the conductors are perfect. It calculates the "high frequency"
inductance and capacitance, skin depth limited.
When you select the impedance option, it meshes the inside of the
conductors, based on the frequency selected, and solves the Helmholtz,
frequency dependent equation. This allows it to output the frequency
dependent R and L, with the capacitance constant over frequency. In you
select the freq sweep mode, it will use the highest frequency selected to
mesh the conductor and calc the R and L values from DC to the upper limit.
It uses these numbers to calculate the coefficients that HSPICE needs for
the W element when outputting a W element model.
You can also plot the L and R values with frequency to see the details.
Of course, for transmission line analysis, when the structure is uniform in
the Z direction, a 2D solver is the preferred tool that balances accuracy
and computation time. It assumes a TEM mode for the fields, hence no Z
dependence on the currents.
Hope this helps to clarify what is going on.
--eric
Eric Bogatin
BOGATIN ENTERPRISES
Training for Signal Integrity and Interconnect Design
26235 W. 110th Terr.
Olathe, KS 66061
v: 913-393-1305
f: 913-393-1306
pager: 888-775-1138
e: [email protected]
web: <http://www.bogatinenterprises.com/>
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of Qu Pingyu
> Sent: Tuesday, December 21, 1999 7:28 PM
> To: '[email protected]'
> Subject: RE: [SI-LIST] : W element correlation with measured data
>
>
> Hello:
>
> I'm not sure which Ansoft software you are talking about. If you refers to
> Quick 3D, it will mesh the conductor to the skin depth if the material was
> specified other than perfect conductor as Jim mentioned. However,
> a constant
> current distribution across the cross section will be assumed. Ansoft
> engineer told me that there is another Ansoft product called Ansoft EM 3D
> field solver, which can handle the uneven current distribution in
> very thin
> conductor. I guess this could give us a more accurate result for
> those cases
> when thickness of conductor is close to skin depth.
>
> Regards
>
> Qu Pingyu
>
> Institute of Microelectronics
> 11 Science Park Road, Singapore
>
> DID: 65-7705413
> Fax: 65-7745747
>
>
> > -----Original Message-----
> > From: [email protected] [SMTP:[email protected]]
> > Sent: Wednesday, 22 December 1999 2:59
> > To: [email protected]
> > Subject: RE: [SI-LIST] : W element correlation with measured data
> >
> > Vadim,
> >
> > You are correct of you are doing a discrete solution for L.
> > In this case it treats the conductors as perfect, all currents
> > and therefore all fields are outside the conductor. But if
> > you specify a material property such as copper on a conductor
> > than the solver will mesh inside the conductor to the skin
> > depth of the frequency you specify. In other words it will
> > add the inductance due to the magnetic fields inside the
> > conductor to the inductance due to the magnetic fields
> > outside of the conductor. It will also compute R at the
> > skin depth frequency specified. C is treated both at DC
> > and high frequency the same, all fields are outside the
> > conductor.
> > Regards,
> >
> > Jim
> >
> > -----Original Message-----
> > From: Heyfitch, Vadim [mailto:[email protected]]
> > Sent: Thursday, December 16, 1999 6:30 PM
> > To: '[email protected]'
> > Subject: RE: [SI-LIST] : W element correlation with measured data
> >
> >
> >
> > Jim,
> >
> > It's my understanding that Ansoft's Maxwell does not creat any mesh
> > into the conductor bulk (or skin) when you are extracting AC values.
> > Instead
> > it only creates the surfice mesh. Therefore there is no point in
> > specifying
> > 500 MHz instead of 100 MHz just to make sure that the is enough mesh
> > layers
> > within the skin depth. Please correct me if this is a
> misconception on my
> > part.
> >
> > Vadim
> >
> > > -----Original Message-----
> > > From: [email protected] [mailto:[email protected]]
> > > Sent: Thursday, December 16, 1999 7:18 AM
> > > To: [email protected]
> > > Subject: RE: [SI-LIST] : W element correlation with measured data
> > >
> > >
> > > All,
> > >
> > > I want to warn everyone that hspice is extremely sensitive to
> > > time step.
> > > The time step I used in the .sp file produced the results shown.
> > > If you vary time step you can get different answers. Also the
> > > meshing in Ansoft is sensitive to geometry. I specified frequencies
> > > of 100 MHz for the cable geometry and 500 MHz for the stripline
> > > geometry in order to avoid problems in skin depth meshing.
> > > Ansoft as well as any FE solver at some point may not be able to
> > > handle high Aspect ratios (thin wide conductors).
> > > Even a round conductor can start to look like a plane at sufficiently
> > > high a frequency in terms of skin depth.
> > > Regards,
> > >
> > > Jim
> > >
> > >
> > > Jim Foppiano
> > > Signal Integrity
> > > [email protected]
> > > (512)-723-3914
> > >
> > >
> > > **** To unsubscribe from si-list: send e-mail to
> > > [email protected]. In the BODY of message put:
> > > UNSUBSCRIBE si-list, for more help, put HELP.
> > > si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
> > > ****
> > >
> > >
>
> **** To unsubscribe from si-list: send e-mail to
> [email protected]. In the BODY of message put:
> UNSUBSCRIBE si-list, for more help, put HELP.
> si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
> ****
>
**** To unsubscribe from si-list: send e-mail to [email protected]. In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
****
This archive was generated by hypermail 2b29 : Tue Feb 29 2000 - 11:39:12 PST