I just discovered that I don't know how to close Nyquist plots with poles on the jw-axis. Can you explain that again?
i.e. How do I determine which way to close the Nyquist plots if poles or zeros exist on the jw-axis?

Tim Chung, 25 Nov 02

When faced with the case where the open-loop transfer function (i.e. P(jw)C(jw)) has poles/zeros on the imaginary axis, the Nyquist path must be indented to avoid the singularity.

Generally, the contour is indented to the right-half plane by using a semicircle of arbitrarily small radius, e, about the pole or zero, as in the example below for a pole at the origin

.
Courtesy of UMich and CMU, "Control Tutorial for Matlab"

Then, when you're sketching the Nyquist plot out, you need to evaluate the frequency response at values close to the pole/zero. In other words, for a pole at s₀ on the imaginary axis, you need to examine the behavior of the response at s₀+e and also at s₀-e.