Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: cavendish expt



Things have progressed a bit since 1797.

Here's a state-of-the-art G apparatus:
http://www.aps.org/meet/APR00/baps/vpr/layp11-03.html

1) It's a thing of beauty. Clever things like this
make me fall in love with physics all over again.

Striking improvements over the 1797 version include
-- For heaven's sake don't make the measurement at
DC. Generate an AC signal and use synchronous
detection. Ever hear of 1/f noise?
-- Don't let the fiber twist. Use feedback to
track out any incipient twist.
-- Don't use a dumbbell rotor. Use a shape that's
easy to calibrate. The dumbbell would give you
a bigger signal, but there's more to life than
getting the biggest possible signal, as Larry W.
mentioned recently in another thread. G is
small, but it's not so small that you're starving
for signal-to-noise ratio. Trade off some noise
(random error) to reduce systematic error.

=============

2) It wouldn't be too hard to make a simplified knock-off
of this apparatus. To simplify things, you could
-- Replace the air bearings with wheel bearings from
ye olde bicycle shop.
-- Don't bother to pull a vacuum on it.
-- Watch the rotor with a home-made encoder:
Take a disk a couple of inches in diameter and
stick a Gray code on its perimeter.
Watch it with a small telescope -> webcam -> computer.
(Put a fine-pitch Gray code onto self-adhesive
mailing-label stock using an ink-jet printer,
then stick it onto the edge of the disk.)
-- Replace the tungsten wire with fishline.
-- Drive the top of the fiber with a stepping motor,
so you don't need an encoder for the top at all.
Steppers are wonderful things. See e.g.
http://www.google.com/search?q=stepping-motor+OR+stepper-motor+PC

This might make an amusing term project for some student.
Or maybe two or three students, since various subsystems
could be attacked separately.