[python-control-discuss] Nyquist-Plots

[Erik F. <er...@er...>]

Hi Everyone,

while designing a controller based on measured frequency response data I
noticed some strange results by in control.nyquist_plot().

Looking at the code I noticed that it first calculates the magnitude and
the phase of the system response and then converts it back to the x and
y components.

I have modified control.nyquist_plot() to directly use the complex
frequency response for plotting:

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Index: src/freqplot.py
===================================================================
--- src/freqplot.py	(revision 293)
+++ src/freqplot.py	(working copy)
@@ -227,14 +227,12 @@
             #TODO: Add MIMO nyquist plots.
             raise NotImplementedError("Nyquist is currently only
implemented for SISO systems.")
         else:
-            # Get the magnitude and phase of the system
-            mag_tmp, phase_tmp, omega = sys.freqresp(omega)
-            mag = np.squeeze(mag_tmp)
-            phase = np.squeeze(phase_tmp)
+            # Get the system response as complex numbers
+            z = sys.evalfr(omega)[0,0,:]

             # Compute the primary curve
-            x = sp.multiply(mag, sp.cos(phase));
-            y = sp.multiply(mag, sp.sin(phase));
+            x = np.real(z)
+            y = np.imag(z)

             if (Plot):
                 # Plot the primary curve and mirror image
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

With this patch my nyquist plots look ok. What do you think of this change?

Best regards
Erik Fleischer

-- 
Andrew Lewis:
  "Wenn Sie für einen Dienst nichts bezahlen, sind Sie offenbar nicht
   Kundin oder Kunde, sondern die Ware, die verkauft wird."