Re: [Phys-L] spinner data -- experimentation 101

[John Denker <jsd@av8n.com>]

On 09/18/2017 04:38 PM, bernard cleyet wrote:

> > OK, I finally found the audio data.
> > https://pastebin.com/raw/neb8qPg5

Thanks!

The basic data is plotted and fitted here:
  https://www.av8n.com/physics/img48/fidget-spinner-freq.png
  https://www.av8n.com/physics/img48/fidget-spinner-period.png

> > IF ….   no one is interested in photogating.   I’ll do it.  

It would be really nice for somebody to redo the experiment.
The data cited above has some nasty glitches in it, as
you can see by fitting and then subtracting.

For the period data, the residuals are here:
  https://www.av8n.com/physics/img48/fidget-spinner-period-residuals.png

There is an outside chance that the glitches are real, perhaps
due to grit in the bearings ... but I would be willing to bet
they are bogus, due to problems in the analysis, perhaps in
the FFT routine.  This is why you want to record the raw
signal.  Don't be in a hurry to do the FFT.

This makes a good "experimentation 101" lesson.  Do a preliminary
experiment, then analyze it as soon as possible.
 *) Maybe you need to redo the theory
   -- power law instead of exponential.
 *) Maybe you need to redo the experiment
   -- photogate instead of audio;
   -- raw period instead of FFT;
   -- free air without nearby clutter;
   -- et cetera.

Also:  Do not accept excuses like "the differential equation is
too hard to solve".  First of all, the relevant equation is not
hard to solve ... and secondly, even if it were hard to do analytically,
you could do it numerically.  The importance of this becomes even
clearer when you realize that without a theory you can't do a good
fit, and without some kind of fit you can't look at the residuals,
and therefore would never notice the nasty glitches.