Re: another ebd modeling question

From: Jongho Kang <jhkang@hynix.com>
Date: Thu Apr 26 2001 - 00:55:21 PDT

Kris,

Did you figure out what configuration of L and C is used when Len=0 and
L and C is specified ?
I'm in the same situation as you was. I'm trying to describe C-L-C lumped
circuit in the EBD.
If you figured out , please let me know.

Thanks,
Jongho Kang

----- Original Message -----
From: Kris Young <kyoung@phaseiv.com>
To: Mike LaBonte <mike@labonte.com>
Cc: <ibis-users@eda.org>
Sent: Friday, April 20, 2001 11:34 PM
Subject: RE: another ebd modeling question

Mike,

I have been using the PSPICE distributed transmission line model, and you're
right, it's very convenient to plug in the values from the EBD file. I was
very relieved when I figured that out.

My question concerns the special case in the EBD file when LEN=0.
Supposedly this requires the use of a lumped element. For instance, when
LEN=0 and R=0.003, it is specifying a series 3 mOhm resistor. In the file
I'm simulating, there are a few cases where LEN=0, and L and C are
specified. I'm trying to figure out what configuration of L and C is
specified by such a description.

Kris Young
Phase IV Systems, Inc.
Huntsville, AL

-----Original Message-----
From: Mike LaBonte [mailto:mike@labonte.com]
Sent: Friday, April 20, 2001 5:39 AM
To: Kris Young
Cc: ibis-users@eda.org
Subject: Re: another ebd modeling question

Kris,

Why not use the PSPICE transmission line model? It is a true distributed
transmission line model, so you don't have to be concerned about how
to do the lumping. Also, it accepts per-length RLGC values. For example:

T3 one_end 0 other_end 0 LEN=1 C=3.1p L=1.05p

PSPICE accepts the notion that LEN is not in any particular units,
same as EBD. Of course you can add R= and G= for lossy lines, too.

That said, if you still need to create a lumped model, try splitting
the series elements in two to make it symmetric. For example, use
two inductors, each with half the inductance. This way a signal going
one way comes out identical to the same signal going the other way.
You can leave the middle series elements unsplit.

Mike LaBonte

Kris Young wrote:
>
> I have one more question for the IBIS gurus out there. I'm using Pspice
to
> model a signal path from an EBD file. I've figured out everything except
> what to do when the length is zero and there's an L and a C. When I look
at
> the classic lumped lossy line (taking R=0 and G=inf), the series L comes
> first and then the parallel C. However, when I look at the classic
lossless
> case, the parallel C comes first, and then the series L.
>
> Since the impedances of these configurations differ, which one should I
use.
> I can only guess, but it should be clearer than that, right? Would the
> lossy case be the one to use, even if there's no R?
>
> Thanks again,
>
> Kris Young

 
Received on Thu Apr 26 00:54:55 2001

This archive was generated by hypermail 2.1.8 : Fri Jun 03 2011 - 09:53:47 PDT