[SI-LIST] Re: SpecctraQuest Convergence Issue
- From: "C. Kumar" <kumarchi@xxxxxxxxx>
- To: James_R_Jones@xxxxxxxx, si-list@xxxxxxxxxxxxx
- Date: Mon, 2 Dec 2002 14:50:32 -0800 (PST)
First you have to understand what type of convergence problem you are getting.
The most common type is "DC convergence" . Generally, howeve the more
troublesome type is Transient convergence problems.
To get a handle on these problems it is best to run the simulator stand alone
in a command line window on the circuit file. In Specctraquest the simulator is
"tlsim"and the top level spice file is "main.spc".
Many of the following observations are common to most circuit simulators
1. DC convergence:
1.1 The most common type occurs in large circuits ( 1000's of nodes, circuit
file containing huge passive package and connector models). Contrary to what
the simulator says, in most cases this is not a convergence problem at all but
actually result of numerical error caused by huge ill conditioned dc matrix.
The classic brute way to force a solution here is to install a small
conductance to ground at every node. In tlsim you can do that by using the
command line option
MinNodeToGndCond
1.2 Another popular source is inconsistent tv curves or leaky clamp curves in
IBIS models. For example if you have a leaky clamp (i.e clamp draining a small
current at the dc operating points) and your circuit is unterminated, you
certainly will have a convergence problem because conservation is violated! If
the leak is "small" the MinNodeToGndCond will solve it. If your problems is due
to tv curves, you can delete these curves from the local ibis_models.inc file.
Alternatively you can globally turn of all the tv curves at once using a
command line option (I believe)
BhvrTvOff OR BhvrModelsTvOff
1.3 In rare instances there may be a "legitimate" convergence problem due to
non-linearity in IBIS models. You can try command line options like
BhvrDcModifyNewtonOff
2. Transient Convergence
These are generally difficult to debug. But in many instances you can make
educated guesses.
2.1 You can turn on variable time step in tlsim using
VariableTstepOn
This is slow in tlsim but usually will solve transient convergence problem. But
the circuit may actually have a real problem and you may find the solution
unsatisfactory.
2.2 By intelligent poking and guessing you can solve the problem by adding
small shunt or floating capacitors. This "method" will work most of the time.
(You can brute force this by adding a shunt capacitance (similar to shunt
conductance), but this will not help you in debugging complex problems). While
debugging, it is good to remember that tlsim automatically paritition's the
transient circuit into many small circuits when it encounters a transmission
line.
2.3 You can also try command line option
BhvrTransientModifyNewtonOff
Hope this helps.
- kumar
James_R_Jones@xxxxxxxx wrote:
Gurus,
I was interested in learning more about the convergence issues in
SpecctraQuest. I get quite a few of these when I simulate, with severity
ranging from "Serious Convergence Error", to "jagged" waveforms, to no
convergence problem at all. Does anyone have a good understanding of what
causes these convergence issues and how to prevent them?
A couple of theories that I have heard so far include:
1. Spectral content resonating back and forth along transmission line
segments that is aliased back as noise causing the "jagged' waveforms.
2. Simulator is confused, particularly during design link, by reference
designators and makes 'mistakes' in handling topology.
Has anyone done work in this area to determine the cause of these
convergence issues? Any help is appreciated.
Thanks in advance,
James R. Jones
ESG Signal Integrity
Dell Computer Corporation
(512) 725-1809
------------------------------------------------------------------
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
---------------------------------
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now
------------------------------------------------------------------
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:
