More followup...
Once upon a time (i.e. last March) I started this hoping to see the shape of tire deformation over a bump. That turns out to be harder than I thought because it happens so fast. I have not given up on that part, actually tried it once more but decided the data wasn't clean enough to draw good conclusions. When weather and time permit I intend to do that again.
[MENTION=381793]gugie[/MENTION], your model is similar to my thoughts. The two fork arms were twisted a little by the skewer but also pulled up or down differently. With different stress pre-load it would not be surprising if the fore & aft movement of the two arms were different, resulting in the wheel wobbling side to side. The wobble would be higher frequency than the main fork oscillation because of the lower moment of inertia and the greater stiffness. The data shows that the wobble happens but it doesn't directly explain how it comes about from a blow radial to the wheel. That's where our thought models come in.
[MENTION=61707]squirtdad[/MENTION], I spend my days renovating houses with high-dielectric rim tape.
Okay, I lie. I don't do that very much. I am a scientist-turned-software-engineer. I could have retired five years ago but I like what I am doing so I keep doing it. Plus I can ride to work when the weather permits.
What can one feel? Good question. Some of my initial measurement were vertical acceleration of the front axle during my commute. I was surprised to see what may have been 10cm wavelength vibration on the seemingly smooth surface of the Minuteman Bikeway. I thought it was the fork, then thought it was the pavement. On my next ride I paid attention to the feel of the handlebar and realized that it vibrated quite a bit. I had just never noticed it before, never paid attention. One could certainly feel vibrations in the range I show here, but it depends on how strong they are amid all the other sensory inputs.
In those graphs the simple oscillations are vertical acceleration (in g's) vs time on the horizontal axis. (I should label the graphs better.) The actual vertical units are not normalized to anything but it doesn't really matter. The horizontal time scale matters only to the extent that one is trying determine actual frequency. IIRC, the phone's sample rate is 409 samples/sec so aliasing isn't an issue.
In the spectrum plots the vertical axis is also unnormalized but the only thing that matters is the relative amounts of any frequency compared to the others. The horizontal scale is frequency, but the actual values must be computed from the sampling interval and the length of the FFT. The important thing you need to know is that the big peak of the main resonance is about 31Hz. The rest you can deduce from the position along the x axis.
Okay, that's enough for now. You are getting sleepy...