Is the PulseIn() accurate when working with measurements 10-500uS? I checked with an arduino using a crystal and had the same readings.
Here's a picture with a larger inductor just to show that the circuit works. http://i55.tinypic.com/fvwc2d.jpg
I initially thought that the pulseIn() value would be a tad small since the square wave is slightly shaped like a trapezoid. Just testing a few inductors, it seems that the inaccuracy is proportional to the pulseIn() reading, and not a added in constant due to the trapezoid signal.
Measured square wave frequency is about 1/20uS = 5.0khz.
Calculated sinusoidal frequeny 1/21.8 = 4.6khz
Arduino is reporting on the square wave 1/(2*35uS) = 2.86khz
Hauled out the scope from the 80s that RARELY works. So the measured could be off just because the scope hasn't been calibrated, but it's pretty good.