[ibis-macro] Re: Test Cases for IBIS Next (or whatever we decide to call it)

  • From: "Walter Katz" <wkatz@xxxxxxxxxx>
  • To: <gedlund@xxxxxxxxxx>, <ibis-macro@xxxxxxxxxxxxx>
  • Date: Tue, 28 Aug 2012 17:52:33 -0400 (EDT)

Greg,

 

I think the contents of the subckt are not important. I would start off
with just package. The problem statement is:

 

1.       A .ibs file

2.       A list of subckts, they can be empty, just the ports of the
.subckt are important, and any parameters that you would like to pass into
them

3.       And most importantly, how you want the subckts applies to each
"channel"

4.       The subckts should be the ones that you currently generate for
your tool (they can be sNp).

5.       They can be singled ended or differential

6.       They can be coupled or uncoupled

7.       There can be just power subckts, or power coupled with signal

8.       The point is not what variations you can think of, but what
variations you need to solve your engineering questions (if you are the
user, or the problem your customer needs to solve based on the models you
as an IC vendor deliver.

 

I am enclosing two examples of a real .ibs file from an IC Vendor X. Names
have been redacted. 

 

In by_pin.ibs the following information on each pin:

| Diff_Package pkg_{corner}.mod pkg_{corner} L=0.01755

Says to use pkg_typ in pkg_typ.mod, pkg_min in pkg_min.mod, or pkg_max in
pkg_max.mod for typ, min, max corners respectively. The subckts are:

 

.subckt pkg_typ padP padM pinP pinM L=.017

W1 padP padM 0 pinP pinM  N=2 RLGCMODEL=rlgc L=L

.model mname=rlgc .

.ends pkg_typ

 

In by_model.ibs the similar information is associated with the model.

 

Walter

 

From: ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Gregory R Edlund
Sent: Tuesday, August 28, 2012 4:19 PM
To: ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] Test Cases for IBIS Next (or whatever we decide to
call it)

 

OK, guys.  What would the EDA vendors like to see as examples?  Both
silicon and package?  Or just package?  AMI or garden-variety IBIS?

Here are the variations I can come up with for a flip-chip BGA package:

1. single-ended
2. differential
3. uncoupled
4. coupled
5. T-line
6. vias
7. signal
8. power
9. s-parameters
10. lumped elements

Let me know if there is anything I'm missing.

Greg Edlund
Senior Engineer
Signal Integrity and System Timing
IBM Systems & Technology Group
3605 Hwy. 52 N  Bldg 050-3
Rochester, MN 55901

Other related posts: