You don't have to be Bill Gates. A few years back I built a small In-draft (sucks air thru) tunnel, about the size of the Wright Brothers tunnel, to investigate a number of different things. Total cost (for the tunnel, not instrumentation) was around $200. Test section dimensions are 20"Lx9"Hx14"W, max velocity in test section to date is 120 feet per second. If you can build your average RC balsa model you can build a wind tunnel [its more cabinet making than model building].

Dr. Selig's work requires very precise data because of the nature of his work [comparing very small differences in Lift/Drag data for airfoil shapes].

The tunnel I built was originally designed for studying model stability [you can actually fly the model in pitch and yaw in the test section].

The top image is a view looking down the throat [12" span Katana aerobatic model], the bottom image [is not out of focus] the model CG is at a critical location [in this case about 40% of the mean aerodynamic chord] were the model becomes divergent in pitch. Meaning your out of control, in this case the model would oscillate wildly in pitch and eventually slam into the ceiling or floor of the test section.

The difficult part of wind tunnel work is data gathering. For a little money and a LOT of time I built a Lift/Drag balance using two very simple but very sensitive load cells. The first thing I did was to perform some Lift/Drag tests on well documented shapes; Sphere, Square cornered Flat Plate, Cone etc. Once things were well calibrated and I had confidence in my test section velocity measurements I was able to get REPEATABLE results within 1-2% of other published tunnel results.