Points of Inflection

**These original Kirt's Cogitations™ may be reproduced (no more than
5, please) provided proper credit is given to me, Kirt Blattenberger.**

Please click * here*
to return to the Table of Contents.

Cog·i·ta·tion [koj-i-tey'-shun] – noun: Concerted thought or
reflection; meditation; contemplation.

Kirt [kert] – proper noun: RF Cafe webmaster.

When reading technical articles, I very often see the authors incorrectly refer to a certain point on a curve as being the inflection point. That was the case in an article I read today that dealt with open-loop polar modulation in EDGE amplifiers. There exists an unambiguous definition of an inflection point, and all engineers were taught it in school. Pardon me if this seems trivial or picayune, but the purpose of the magazine articles is to teach, so if this factoid can eliminate the error in future articles, then it will have accomplished its objective. Here is a brief review of what an inflection point is, and, equally important, what an inflection point is not.

An inflection point is the point at which the second derivative equals zero. Accordingly, it is the point where a curve changes from concave up to concave down. A curved region is concave up if all the points in that region lie above a line tangent to it (in the + y-axis direction). A curved region is concave down if all the points in that region lie below a line tangent to it (in the - y-axis direction). The small graphic that accompanies this factoid illustrates all of these concepts (curves not plotted to scale). Another way of stating it is that directed lines perpendicular to the tangent lines have a downward direction for downward concavity, while directed lines perpendicular to the tangent lines have an upward direction for upward concavity. Note that not all curves have an inflection point, while some curves have many.

A familiar inflection point, and quite likely the first one introduced to each of us in calculus class, is the one that occurs at the axis crossing in a cubic curve: y = x

The same situation is true for the half cycle of the tangent curve (blue). The second derivative of tan(x) is d

Now, here is where a lot of authors go wrong. Many of the curves we plot look like the red plot on the graph. AM/PM conversion, power versus frequency in a bandpass filter, and component Q-factor plots are a few examples. This plot happens to be a parabola whereby y=x

So there you have it. Let us resolve to never again refer to a local maximum or minimum (where the first derivative equals zero) on a curve as an inflection point.