Something else for you nerds
 

This is the first phase of a followup to the fork study I started back in March.

I had measured the resonance frequencies of the forks on several of my bikes. I did this by mounting a high-frequency accelerometer (my phone) on the fork, lifting the front wheel off the ground, then thumping the top of the wheel with a rubber-coated mallet. The results can be seen in this picture:

http://www.theworld.com/~muller/pics...Comparison.gif

One feature which jumps out is the complexity of the Peugeot PFN10's spectrum. Higher frequenceis are present, but what vibration mode would cause that? Are the fork blades bending in two or more modes, or is the fork exhibiting some sort of torsional vibration?


I had known that the fork blades weren't aligned. Tightening the QR skewer always pulled the wheel over to one side, and bolts mounted to the insides of the DO's were neither aligned nor even parallel. So at some point I took the fork to Peter Mooney and asked him to straighten it. (I am not sure I can feel any improvement but I had not ridden it enough prior to that to evaluate it. It sure rides well now.)

I always wondered if poor alignment might have contributed to the unusual resonance. So I repeated the experiment and obtained a very different result. Here are comparisons of the data and spectra, pre- and post-alignment. The difference is astounding. All those high frequencies in the original spectrum are gone.

http://www.theworld.com/~muller/pics...eugeotFork.gif

http://www.theworld.com/~muller/pics...requencies.gif

Why should alignment have produced such a difference? One conjecture is this. When an impact hits the fork the blades bend elastically. If they are pre-stressed differently, one up and forward and the other down and back, they could build up bending-induced stress at different rates and begin oscillating up/forward-down/back out of phase. The mode is like how you swing your arms in opposite directions when walking. When the arms bend opposite directions it swings the wheel back and forth. The frequency is higher than the basic fork resonance because the moment of inertia about the steering axis is less than that of the whole fork.

To test this idea I repeated the experiment but thumped the bike at two different places. I thumped the front of the wheel so as to rotate the steering axis, and then again on the side of the fork to make the entire fork resonate laterally with minimal steering rotation. I repeated these thumps several times to make sure the results caused characteristic vibrations.

The spectrum from a thump rotating the steering shows a distinct match to the high frequencies in the original spectrum.

http://www.theworld.com/~muller/pics...iesWFront1.gif

The spectrum from a thump on the side of the fork is quite different:

http://www.theworld.com/~muller/pics...ciesWSide2.gif

Finally, to demonstrate that the results are repeatable I superimposed the three wheel-front thumps:

http://www.theworld.com/~muller/pics...elAllThree.gif

Conclusion: Mis-aligned DO's on the fork create higher frequencies in the response to bumps. The behavior seems to be that the two blades do not respond identically to fore and aft deformation, resulting in steering oscillation. It is not clear whether this can be felt on the road. However it does suggest that fork alignment is good.

Thank you for reaffirming that I am not a nerd. I didn't get any of that. :lol:

'Cracks' me up thinking what might be the results for my Viscount.

Add the irony Yamaha (with their tuning fork logo) took over distribution of Viscount in the US.

https://cimg1.ibsrv.net/gimg/bikefor...88904a28e5.gif
https://cimg2.ibsrv.net/gimg/bikefor...79171a8214.jpg

Huh?

Join the club.

I think you should take into account the effects of the rim glue curing/hardening after some time after tire installation, as it turns from a liquid towards solid state. The harmonic dampening effects of the glue, especially where they accumulated into the spoke head wells, will be much diminished......::speedy:

Looks to me with your misaligned fork, the vibration waves were cancelling each other out, probably giving you a smoother ride.

Of course handling issues were another matter.

I'd be curious why the Masi flatlines so quickly. Does it ride stiff?

Interesting. [MENTION=190941]jimmuller[/MENTION]: would this be something other members could reproduce with just a phone, an app and a mallet? Like C&V venturing into the big data thing?

Topic aside, those are pretty good FFT plots.

:lol:

If you feel you might have too much free time I have a roof to basement reno you could come and scrape wallpaper or sand floors.

All this information resonates within me.

An idle mind is.........(fill in the blank)......:rolleyes:

Interested why you thumped the wheel with a mallet; could you not just lift the front wheel a set height off the ground then drop it?

Okay, time for more explanation. That first post was short to avoid driving away all you tldr'ers.:D There is a lot to respond to, and I'm sure I'll miss some. In no particular order...

In that original study I made no attempt to match the impulse strengths or the system damping. The Masi does appear to decay away faster but I wouldn't read too much into that. Not counting the PFN10, the only real observation you make is that the frequency is slightly lower for the Grandis with the clincher wheel even though you'd think it might have a stiffer fork and thus be higher. The Peugeot on the other hand is so different it begs for further investigation.

I suspect neither the sewup glue nor the wheel has any effect on this. After all, the primary frequency response of the fork is about 31Hz and the higher frequency stuff is more than double that. It would feel to the hand like buzzing and may be a cause of fatigue, but probably does not contribute to wheel oscillation.

I used a mallet on the tire instead of dropping the wheel on the floor because I wanted to remove the effect of the bike bouncing off the floor at the start. The mallet is more controlled. But just bouncing the tire might have worked too. Anyone can do this with a bike and a cellphone app. But I analyzed the data with a program I wrote to let me step through it quickly, compute the FFT over selected intervals, and plot the result. No, I am not interested in helping out with house renovations.:eek:

Now for further explanation. The original PFN10 resonance was so squiggly compared to the other bikes' decaying harmonic resonance that something was clearly amiss. A plot such as this:
http://www.theworld.com/~muller/pics...requencies.gif
is oscillation "strength" on the vertical axis and frequency on the horizontal axis, low frequency to the left, high frequency to the right. The labels "64", 128", etc do not indicate the frequency but rather, the number of data points across the FFT output. The large peak at about data point 40 computes out to be about 31Hz. The presence of higher frequency content is what made the original look so squiggly.

An oscillation involves elastic bending and has a frequency determined by the stiffness and the mass. When two different frequencies are present it indicates that two different kinds of bending are happening. Clearly one type is the fork bending such that the wheel moves forward and up then backward and down. All the bikes did that about the same. The PFN10 was clearly moving in some other way too originally, but aligning the fork fixed it. That just begs for an explanation! Of course it might have been due to the bar tape dielectric constant, but I doubt it.

I wondered where this thread went or what you had concluded after time, and I'm glad you've resurrected the subject as I find it fascinating! It does make sense that a misaligned fork would give a squiggly frequency reading. The fact that the (presumably aligned) other forks gave very similar readings or reading patterns is cool to see as it creates a reference of sorts with which to compare other forks or tires etc.

Have you thought about carrying this test out on the rear wheel? The front makes sense as there is no triangulation and thus the fork is able to flex and give pronounced readings. The rear triangle seems like it would produce much smaller/tighter frequency readings--ones that would require re-scaling of the vertical axis to see more clearly. Compare and contrast front and rear readings with which frame/bike you like riding (vs not like as much) or which frame/bike has characteristic x or y or z, and see if there exists any correlation, loose or not.

Jim, just two cents from a somewhat nerdy pastor/freewheel medic in snow-bound NH (and it's only November :notamused:):

This needs to be repeated on a control course, set up over northern New England winter frost heaves. Same rider on different bikes/forks, but with the same wheelset (if possible). And then repeated with different weight riders. Of course speed and variations in tracking the control course need to be accounted and factored into final charting. And to add more complexity to the entire engineering investigation, accelerometers (i.e. smart phones) should be strapped to each fork blade and each seat stay.

I'm curious if you have some idea about what range of frequencies a person might be able to actually feel. I imagine the answer depends a lot on the displacement, and whether it's growing or shrinking with time with the various inputs that happen while riding (i.e. repeated periodic whacks with the hammer) or whether the reaction is in some other axis and one feels him/herself reacting to these things.

That resonated with my inner nerd. :) Might you have pictures of your experimental setup? How did you calibrate your mallet thumps?

This outdoes your 'wheel as a structure under tension' monologue.

Top

Piano Man
 

Back in the day a LBS that employed a master wheelbuilder would have to have a in-tune stand up piano to ensure the spokes were correctly tensioned to keep the groove.
With all due respect to Doc Spectrometer I seriously doubt no amount of twisting forks could correct a B flat laced turd into a a C sharp laced wil-o-the-wisp.