Path: news.mathworks.com!not-for-mail
From: "Steve " <steveDEL.bachmeierDEL@yahoo.com>
Newsgroups: comp.soft-sys.matlab
Subject: not converging
Date: Sat, 1 Sep 2007 01:10:33 +0000 (UTC)
Organization: The MathWorks, Inc.
Lines: 24
Message-ID: <fbae69$5ut$1@fred.mathworks.com>
Reply-To: "Steve " <steveDEL.bachmeierDEL@yahoo.com>
NNTP-Posting-Host: webapp-06-blr.mathworks.com
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
X-Trace: fred.mathworks.com 1188609033 6109 172.30.248.36 (1 Sep 2007 01:10:33 GMT)
X-Complaints-To: news@mathworks.com
NNTP-Posting-Date: Sat, 1 Sep 2007 01:10:33 +0000 (UTC)
X-Newsreader: MATLAB Central Newsreader 1034320
Xref: news.mathworks.com comp.soft-sys.matlab:426544



Hello,

I have a relatively complex Simulink model.  I decided to 
control the outputs using a simple proportional gain.  When 
I set the gain to zero, I get the same results as I did 
before adding the gains.  However, when I set the gain to 
where I need it to be (15,000), then I get "not converging" 
error.  In both cases, (as well as gains up to about 40), I 
used ODE15s with a min step of 1 and a max step of 10, and 
an error tolerance of 1e-09.  I'm curious as to why the 
gain would cause the system to not converge?

I thought that maybe at the first time step, the error was 
large due to my initial conditions, which made my 
proportional control react heavily due to the gain, which 
blew the hole system up.  But when I set my initial 
conditions equal to the setpoint that I'm trying to control 
the outputs to, the same thing happened.

Why is this, and is there a way around it besides reducing 
my minimum step size or relaxing my tolerance?

Thanks,
Steve