Re: enhancing Bode functionality

[A S Hodel]

There's certainly room to refine and/or improve performance of these  
functions from when my students and I wrote them 7 or 8 years ago.   
Some things that have changed since then.  For example, user-defined  
types (in C++) are a possibility.  The "object-oriented" design in the  
octave functions predates MATLAB's release by a year, but is kludgy in  
the sense that, instead of
        PC = P*C   % MATLAB syntax
we must write
          PC = sysmult(P,C);  % do the multiplication
          sysout(PC);               % display the system

For reasons that made sense at the time I allowed a system to contain  
both continuous and discrete states; in retrospect I'm not sure that  
was smart.

When the Octave controls toolbox was implemented there were neither  
cell arrays (MATLAB) nor lists (octave), so I was kind of stuck with  
SISO frequency response treatment from a data-structure and complexity  
standpoint.  Whether it's desirable for
        bode(sys)
to generate umpteen plots for an MIMO system vs requiring a channel to  
be specified as in octave
        bode(sys,[],output_number,input_number)
is open to question.  For MIMO systems octave currently plots  the  
maximum singular value (induced two-norm gain) of the MIMO system.   
This may be relevant to those using the hinfsyn function (and its  
related functions) that Kai Mueller kindly
updated several years ago after my initial implementation based on the  
DGKF paper.   For more information you can take a look at
@inproceedings{hodtenetal96,
        author = {A. S. Hodel and R. B. Tenison and D. A. Clem and J.  
E. Ingram},
        title = {The Octave Control Systems Toolbox: A ${\rm  
MATLAB^{tm}}$-like {CACSD} Environment},
        booktitle = {Proceedings of the 1996 {IEEE} International  
Symposium on Computer-Aided Control System Design},
        year = 1996,
        pages = {386--391},
        publisher = {{IEEE} Control Systems Society},
        address = {Dearborn, Mich.},
        month = {Sept. 15-18}
}
(The session also includes several papers related to SciLab, a similar  
project that I think was initially developed in Europe.)

The discussion of displaying frequency data in Hertz vs rad/s sec is  
trivial (scale frequency data by 2*pi and plot).  For such a purpose,  
it's worthwhile to use the advice given immediately below and replot as  
desired.  The dB - rad/s - degrees units for bode plots used in Octave  
is pretty standard in most controls textbooks (although Ziemer, Tranter  
and Fanin's signals and systems  book works in both rad/s and Hz).

On Monday, November 25, 2002, at 03:49 PM, Julian DeMarchi wrote:

> Hmmm.  The syntax is more or less
>
>         [mag,phase,w] = bode(sys)
>
> and this function will not generate any plots unless
> you don't specify the left-side parameters.  So, rather
> than write separate functions, why not simply use bode()
> as above, and plot as a second step?
>
>         [mag,phase,w] = bode(sys);
>         semilogx(w, [mag phase]);
>
> Or something of that nature.
>
> Note, Matlab [mag,phase] arguments are MIMO-ised
> (see  
> http://www.mathworks.com/access/helpdesk/help/toolbox/control/ref/ 
> bode.shtml)
>
> I'm not sure that Octave works the same way.
> Also, the example in the link uses the function tf()
> which for Octave would be tf2sys().
>
> - JD
>
> -----Original Message-----
> From: Steven Pavao [mailto:address@hidden
> Sent: Monday, November 25, 2002 3:20 PM
> To: address@hidden
> Subject: enhancing Bode functionality
>
>
> Hi,
>
> I'm interested in improved bode functions for Octave,
> so I'm doing a little work to try and get the
> improvements I'm looking for.  Maybe some others have
> also done some work, or would like to help me along in
> my effort.
>
> First, I've split the bode magnitude and phase
> portions into two functions I'm calling bmag and
> bphase.  This lets me get around a bug I was informed
> of - one where plotting multiple curves on multiple
> plots on a page does not work.  So now I can get
> multiple curves onto either a magnitude or phase plot.
>  Cool.  I'm really most interested in the magnitude in
> general, though I would like to inspect the phase as
> well.  Splitting them out just made sense for my
> purposes, and for geting around the bug.
>
> Now, I really want to plot the x axis in Hz, not
> radians.  It's just the right thing for my my
> purposes.  I have not done the work to do this.  I
> wonder if the best approach is to just change what is
> sent to gnuplot, although this may not suffice for the
> feature I mention in the next paragraph...
>
> The other annoyance is the auto-scaling of the axes.
> Many times when plotting, I want to compare multiple
> curves, say, based on a varying coefficient(s).  So, I
> don't want it to autoscale the axes in that case.  In
> that case, I'd like to expose the axes "limits" to the
> command line.  Of course, for the magnitude plot, I'd
> like to specify these limits in dB and Hz.  If no
> limits were specified, then autoscaling based on the
> most recently plotted curve would be fine.
>
> I would appeciate any tips or help.  I'm willing to do
> some work and share my results.  I know this bode
> functionality could be much more useful to my learning
> process if I could just accomplish this small handful
> of improvements, and others could probably benefit,
> too.
>
> best,
> Steve Pavao
>
>
> __________________________________________________
> Do you Yahoo!?
> Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
> http://mailplus.yahoo.com
>
>
>
> -------------------------------------------------------------
> Octave is freely available under the terms of the GNU GPL.
>
> Octave's home on the web:  http://www.octave.org
> How to fund new projects:  http://www.octave.org/funding.html
> Subscription information:  http://www.octave.org/archive.html
> -------------------------------------------------------------
>
>
>
>
> -------------------------------------------------------------
> Octave is freely available under the terms of the GNU GPL.
>
> Octave's home on the web:  http://www.octave.org
> How to fund new projects:  http://www.octave.org/funding.html
> Subscription information:  http://www.octave.org/archive.html
> -------------------------------------------------------------
A. S. Hodel, Assoc. Prof, Dept. Elect & Comp Eng, Auburn University, AL  
36849-5201
(334) 844-1854 200 Broun Hall address@hidden  
http://ww.eng.auburn.edu/~scotte



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------