Re: Make your own linear scales

Hi,

Here's a crazy idea I've thought about but never pursued beyond a few
experiments to see if it is feasable:

Take two long cylinders (like a double-barreled shotgun). In one
cylinder place fixed ultrasonic transmitter (speaker) at one end,
have a movable ultrasonic receiver (microphone) that can move the
lenght of the tube, linked to an axis.

Drive the transmitter with a 40 kHz signal; this will launch sound
waves down the tube. The tube acts as waveguide so there is no
attenuation of the sound down its lenght.

Pick up the received signal from the microphone and compare it
against the transmitted signal. At 40kHz, the wave-lenght is
about .025". As you move the microphone, a phase detector (receiver
vs. transmitter) would show a 360 degrees phase shifht for every .025"
you move. Resolve the phase detector output to about 1.5 degrees (not
hard to do) and you can read a 0.0001" movement.

Now for the other tube. Have the speaker and microphone fixed at the
ends a known distance apart. Do the same with the signals as for the
first tube. This is the reference channel.

The speed of sound changes with temperature, humidity and barometric
pressure; probably other things as well. These changes will produce a
phase shift in the reference channel. Servo the transmitter frequency
(up or down) to keep the reference phase equal to zero (give or take
a degree). This cancels the change in the speed of sound.

The "encoder" reads 40,000 times a second. Signal averaging and
bandwidth techinques can cancel noise very effectively.
This "encoder" would be perfectly linear and have an accuracy equal
to the resolution and independent of its lenght. You could also
change its "lines per inch" by simply shifting its operating
frequency or the phase detector resolution.

So, what do you think. Crazy idea?

Mariss

--- In CAD_CAM_EDM_DRO@y..., beer@s... wrote:
> > I've always thought the "spherosyn" could be "home-shop-able"
> > (precision ball bearings in a tube, read capacitively sin/cos)
> >
> > Also have thought about the old "wire" tape recorders...
> > What about recording a sine wave (signal generator created), and
> > reading it with a std. recorder head. Then looking for a way to
have
> > TRUE metal tape. Iron oxide?
>
> I've thought about both of these as well, and every time I get
excited
> about it, I go out to the shop, grab something or other, and shake
off
> a whole pile of magnetized chips.
>
> Frankly, I don't understand just what it is in the cutting or
milling
> process that magnetizes previously non-magnetized steel.
>
> Anyway, the sight of all those filings clinging to things puts me
off
> any magnetic solution.
>
>
> On the rotary front, though, I heard from guy a few years ago who
told
> me of his setup. I'm still not completely clear in my own mind
that it
> works, lacking the geometric skills to prove it one way or the
other,
> but the solution is more than a little intriguing.
>
> There are basically two problems with a pulley system. The first is
> slippage. While this can be reduced with a enough spring tension,
it is
> hard to eliminate. Also, there is some chance of wear over time as
a
> result of that tension.
>
> The second is the problem of getting the pulley exactly the right
size,
> that pesky PI entering into the pulley diameter equation.
>
> This fellow's solution ( and I apologize for not giving him the
credit
> he is due, as I cannot find the original message ) is to stretch a
> length of shim stock. He used .008" stock. A pulley close to the
> right size then rides along this length of shim stock and the
friction
> between the two turns the pulley.
>
> The width of the stock gives a wide bearing area, and so no wear is
> likely.
>
> The pulley is tensioned against the "tape" ( shim stock ) by a pair
of
> bearings to either side.
>
> -- crude ascii art --
>
>
> ----/.\--\*/--/.\ ---
>
> where
>
> --- is the shim stock
> . is a bearing, about 3/4" OD, 1/4" ID, and the width of the
tape.
> * is the pulley
>
> Things are aligned so that the tape rides OVER the first bearing,
> slightly above the nominal centerline of the tape, UNDER the pulley
> then OVER the last bearing.
>
> The OVER distance is slight, perhaps even zero ( I'll have to think
> about that as I can't remember )
>
> The UNDER distance of the pulley is similarly small, but not zero,
and
> is adjustable ( as is one end of the tape, so that as the UNDER
distance
> is adjusted, suitable tension can be reapplied )
>
> The pulley is machined slightly smaller than what common sense would
> suggest. ( On a perfectly flat tape, then, the readings would be
high.)
>
> Here's the magic and the important point. As a result of the small
> "loop" formed by the two bearings and the UNDER distance, the actual
> distance the pulley travels is greater than the distance between
the
> ends of the tape.
>
> By adjusting the magnitude of the UNDER distance, one can zero out
the
> error in the pulley diameter.
>
> All in all, a clever solution, it seems, assuming the geometry
actually
> works.
>
> Alan
>
> --
>
> Alan Rothenbush | The Spartans do not ask the number
of the
> Academic Computing Services | enemy, only where they are.
> Simon Fraser University |
> Burnaby, B.C., Canada | Agix of
Sparta