Fourier analysis with a mechanical computer

Albert Michelson's Harmonic Analyzer:

And I really think you should just leave this one playing all the time.
Let me hear your sinusoidal war cry.
Especially 16:07.

Previously, previously, previously, previously, previously, previously, previously, previously, previously, previously, previously, previously, previously.

Tags: , , , , ,

9 Responses:

  1. phuzz says:

    The mechanism at the top for summing the cosines is genius in it's simplicity.

  2. bbot says:

    Would have been neat to see an analysis of sources of error, of which there's probably many. On page 76-77 of the book, there's a bunch of output plots from each of the sine gears, some of which are visibly irregular, both in period and constant. Why? Gear wear? Varying spring constants on the springs? Warped rocker bars? Chain stretch?

    • Russ says:

      I would guess the largest contributor to visual irregularities would be backlash.

      • bbot says:

        What backlash? The geartrain and cam system is all being driven forward continiously. Backlash in the pen-paper contact? That wouldn't show up as a variable irregularity in just one sine gear.

        What on earth is wrong with rocker arm #19 there? It can't be a problem with the visibly-worn cone gears, otherwise it would happen once every period. The only other source of variable error would have to be the cylinder gear, I guess?

        • Richard says:

          The page-by-page guide to the book answers this starting at about 14:40. Quick summary: the machine is quite sensitive to the friction between the pen and the paper, and the specific irregularity on gear 19 happened because the paper bulged out from the platen at that point, increasing the friction and skewing the results.

          The issue of pen-paper friction was one of their bigger challenges in getting the machine working; he mentions it at several points in the videos.

        • Jeff Bell says:

          Possibly the other nearby arms (16,17,18,20) are not perfectly at zero.

  3. Don Wannasay says:

    This tops the Curta calculator video you posted 2 months back. Yay for complex things from a simpler time (or so I like to think).

  • Previously