Re: [SI-LIST] : Package Parasitics Modelling

gedlund@us.ibm.com
Mon, 29 Mar 1999 10:55:05 -0600

What about a TSSOP package like memory drivers come in? I've found that a
lumped RLC model still works pretty well for these kinds of packages (i.e.,
I get good lab correlation) - and the package effects are certainly not
negligible. I haven't done very extensive power distribution modeling for
TSSOPs, but my guess is that an RLC matrix would do the job. So I wouldn't
go so far as to say that lumped RLC models are a thing of the past.

I would also argue that an inductance matrix is the appropriate model for
bond wires in just about any package that uses them. But Larry's right -
once you get onto the substrate in something like a laminate BGA package
you need transmission line analysis. Sometimes what you really need is
coupled transmission line analysis where the cross-section varies as you
move from the bond pad to the via. This is a difficult structure to
accurately represent in a package model.

Greg Edlund
Advisory Engineer, Critical Net Analysis
IBM
3650 Hwy. 52 N, Dept. HDC
Rochester, MN 55901
gedlund@us.ibm.com

Larry Smith <ldsmith@lisboa.eng.sun.com> on 03/29/99 10:15:11 AM

Please respond to si-list@silab.eng.sun.com

To: si-list@silab.eng.sun.com
cc: (bcc: Gregory R Edlund/Rochester/IBM)
Subject: Re: [SI-LIST] : Package Parasitics Modelling

Daniel - I agree. R-L-C models worked well for packages that had
no ground or power planes because the parasitics really were similar
to lumped parameters. But in modern packages with power and ground
planes, the package parasitics are definitely distributed like
transmission lines. If the package parasitics are important enough
to be considered in a design problem, then a lumped (single) R-L-C model
does not have enough detail to give you the answer you are looking for.

regards,
Larry Smith
Sun Microsystems

> From: Daniel.Adar@ecitele.com
> X-Lotus-FromDomain: ECI TELECOM
> To: si-list@silab.eng.sun.com
> Date: Mon, 29 Mar 1999 15:26:35 +0300
> Subject: [SI-LIST] : Package Parasitics Modelling
> Mime-Version: 1.0
> Content-Disposition: inline
>
> I made some spice simulations of package parasitics effect.
> I compared the RLC model of IBIS with an ideal Transmission line (with
> zo=sqrt(L/C) and Td=sqrt(L*C).
> In both cases I drove the package model (RLC or T line) by a pulse with
various
> rise time, Tr.
> I'd like to share with you the results and conclusions:
> 1. For very small parasitics (whereTd<<Tr) the RLC gave similar results
to those
> of the T line, but there was no need in any model since the effect was
> negligible.
> 2. For larger parasitics there is a significant difference between the
two
> models.
>
> My conclusion is that referring the package parasitics simply as R,L and
C is
> incorrect.
> It should always be modelled as a T line or skipped for small values!
>
> Does anyone like to comment on that?
>
>
>
>
> Daniel Adar
> e-mail: Daniel.Adar@ecitele.com
>
>
>
> **** To unsubscribe from si-list: send e-mail to
majordomo@silab.eng.sun.com. 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
majordomo@silab.eng.sun.com. 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 majordomo@silab.eng.sun.com. 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 ****