Re: [linux-audio-dev] off topic, optical protocols

New Message Reply About this list Date view Thread view Subject view Author view Other groups

Subject: Re: [linux-audio-dev] off topic, optical protocols
From: D. Stimits (stimits_AT_idcomm.com)
Date: Wed Mar 06 2002 - 00:27:03 EET


"Richard C. Burnett" wrote:
>
> > That is one variation. But if the "sphere" is really the "disco ball"
> > type surface...a sphere that has a large number of small flat surfaces
> > that altogether almost are a sphere...then that isn't a problem. Like I
> > said, there are variations on the geometry to do different things. The
> > base shape plus the surface details (such as facets) for the reflective
> > device are half of it, the other half is the base shape of the detection
> > system...e.g., flat or curved. Someone else mentioned interferometers,
> > this is another variation, one can use interference waves for some
> > pretty nice sensitivity. More can be done if the photo detector system
> > is interlocked the way the rods or cones of the eye do: neighboring rods
> > or cones alter the sensitivity of their neighbors, increasing the
> > sensitivity of any otherwise almost imperceptible gradient. One can use
> > a single source of light, and go for a single position for information;
> > or use multiple sources and look for interference fringes.
>
> Yes, I have worked with interferometer's before, and I can say they can be
> extremely over sensitive for some applications. A friend of mine did his
> thesis with one and made the mistake of mounting the whole assembly on
> aluminium. The changes in air pressure and temperature in the room
> GREATLY affected the phase differences and required him to redesign on a
> platform of lesser thermal expansion (I think he used like 2 inch thick
> steel). So your housing for this system might be VERY dependent on
> temperature and pressue.
>
> Also, if you are going to be detecting 'levels' of light hitting the
> detectors, then you are going to need to understand the output power of
> the light source, its spectral makeup, and the sensitivity of the
> detectors. I have seen power plots and they are not simple gaussian
> distributions, some have lobes that could give you some really interesting
> issues.
>
> With faceted surfaces you might not get the resolution you are looking
> for, because changes in location are not going to be that great on sound
> waves, and discrete flat reflectors over that range I would think would be
> hard to do for quite a bit of ability. The sphere idea still sounds
> better (or curved surface really). I can forsee an incredible tolerance
> issue with all the parts.

All of this depends on what is to be accomplished, and the actual
digital data that is produced can easily be ignored if changes are not
some minimum value.

>
> > In the case of the variation that is similar to the rods or cones of the
> > eye, with one detector able to alter the gain of another, it will not be
> > a problem. They would be interlocked such that one and only one wins the
> > war on gain, a kind of hysteresis/extreme case of rod/cone structure. In
> > the case of facets, there is no spreading perceptible in such a short
> > distance.
>
> So this part of it is analog, because you are not dealing with complete
> values. While you could set it for a threshold of gain, what happens if
> two get the same value because the light is at the edge. At some point
> you are going to need to compare bits, and the circuitry to do that is
> going to probably have lots of variance due to part tolerances. IE, like
> a large group of comparitors. There would probably be a time delay with
> the comparison.

No, it is not analog. The idea is that a sensor is marked as a 1 or a 0,
and that no two sensors sitting side by side are allowed to be 1. The
gain modifications result in amplification being so high on one unit,
and so low on a neighbor unit, that there is no in-between value. All
digital chips are analog if you want to deny that argument, all of them
are basically such an arrangement of competition for state 1 or state 0.
There is no attempt to detect just how much gradient there is between
two sensors, there is only an attempt to make the brightest location a 1
and all else that is less a 0. If two locations wish to be a 1 and they
sit next to each other, they will alter their gains until one of them
snaps to maximum and one to minimum. More variation is possible by
altering whether this occurs on all directions or just in some single
axis along the array.

>
> > True, but less than analog means. There is no reverse EMF to dampen it
> > like a dynamic coil detector, no capacitive charge to add repulsion like
> > in a condensor, no piezo electric crystal to resist movement. Regular
> > microphones suffer from the holding method just like you mention above,
> > plus they suffer from the movement detection also altering it and
> > distorting it. This eliminates all but the effect of the photons hitting
> > the surface...and I doubt the force in question is detectable even in
> > the most serious of settings. Now if you really wanted to have some fun,
> > and it isn't just a microphone you are interested in, you can create a
> > vacuum packed container, and a superconductive magnetic suspension
> > system, and literally float the reflecting device by levitating it (it
> > would be overly sensitive, and dampening vibrations would be a problem;
> > however, having a direct digital output, it is possible to implement
> > fourier based filtering). By playing with geometry, the scheme can be
> > made to isolate movement in particular directions...vector components of
> > the central reflector. This latter part is why I am most interested in
> > it, it could become a single point detector capable of knowing the
> > direction of the force being applied to it; with fourier and the right
> > geometries, it could detect the direction of the force being applied by
> > different tones.
>
> Outside of the superconductor example :) You would need something to hold
> your moving part at a center location so it had freedom to move. It has
> to have some force to bring it to a center position. This is no different
> than a piezo-electic holding its position. (Unless I don't understand
> what you mean). Because some force needs to hold it. If you don't have a
> force, then how will it be able to retain location. Also, this 'thing'
> that holds it will be frequency dependent, and react (vibrate) with
> different resonant frequencies modulating the affect on the moving part
> itself.

Well, for the joystick sample, which is nowhere near as sensitive as a
microphone, this is trivial. For an optical microphone, take something
like an existing dynamic microphone, and simply remove the coil, and you
have a start. This is trivial for creating a demo technology, maybe far
from trivial to push it to the technological limits.

>
> > The chip I was speaking of before that did not exist way back when is an
> > optical detector array. But someone else mentioned interferometers, and
> > the use of interference patterns can greatly reduce the amount of
> > detectors required; one would no longer detect a single point location,
> > but instead rate the interference pattern on a series of detectors. If
> > you really want to know how sensitive such a thing can be, look up RLG,
> > ring laser gyroscopes. The main limitation of how sensitive they can
> > become is a two-part question: 1, as two coherent light sources are
> > mixed and approach each other in phase, they tend to lock on to the
> > phase of the other and you lose the phase difference, and 2, at any
> > given tendency for closely phased coherent light streams to merge, a
> > shorter wavelength still provides a finer measure. Using the newer blue
> > laser diodes versus infrared you can gain a lot of ability to detect
> > smaller variations. There is even a security type system based on
> > interference patterns produced by the difference in path length of a
> > single laser source being put there two separate fiber optic tubes
> > planted under the ground. What happens is the pair or more of fiber
> > tubes are just different strands of a multi-strand fiber optic cable,
> > placed under the ground maybe 5 feet or so, and run in a circle around
> > an installation; someone walking anywhere near, or cars driving in the
> > area, distort that cable's shape through ground vibrations, causing the
> > relative length of the two strands to change...the interference pattern
> > then changes, which is detected, and alarms go off. So interference
> > patterns can make the sensitivity part a whole new game, detecting
> > distances roughly in the neighborhood of the wavelength of the light
> > being used.
>
> How many detectors does the array have? I guess you could use also like a
> CCD type array. I know they have those.

I believe it is about 10 thousand per square inch. What I'm looking at
does not need to be anywhere near as sensitive as a CCD array.

>
> On the fiber intrusion detection, a kid I went to school with had built
> something similar with fiber inside a block of plastic. He would use it
> to measure water pressure in the ocean, because as he lowered it, he had
> pieced of plastic going perpendicular to the fiber, it would create
> microbends and change the signal at the end (the level). Very similar.

Yes, and I know of similar items that have been built, it has become a
popular theme. Several of the space structures have optical systems
along all the structural members to determine hull stress, along with
many different carbon-fiber structures used in aerospace. The original
article was interesting, because it was published by Chinese engineers
and then put into public American journals...the original was a
description of how Chinese radar facilities were protected by it. From
there the variations are countless.

>
> > In any case, I had two things in mind, and a microphone is only one of
> > them. The optical equivalent for joystick control probably requires only
> > the equivalent of 8 to 11 bit to be useful; 12 bits would be extremely
> > fine. And in no case am I doing this for anything other than fun, so I
> > don't take the limitations very seriously.
>
> Do you mean 8 to 11 sensors or 256 to .... sensors when you mention bits?

8 to 11 bits. The number of sensors, if you do not use an interferometer
method, would require 256 to howevermany sensors. Use of interference
patterns drastically reduces the required number of sensors. The chip I
am thinking of contains about 10k sensors per inch. But I am just
playing around with this for fun, and I won't bother doing anything with
it if I can't use my optical input on my sound card. Someone mentioned a
chip to do A/D and optical encoding, this is what I am looking for.
Hopefully though the optical encoding section is not isolated and does
not *require* me to use analog A/D features, I could toss those features
out and not care (except possibly for creating a mic preamp).

>
> I think your project is very cool :) I hope my questions are helpful.

Definitely a fun topic for me too.

D. Stimits, stimits_AT_idcomm.com

>
> Rick
>
> > D. Stimits, stimits_AT_idcomm.com
> >
> > >
> > > Rick
> > > +------------------------+-----------------------+
> > > | T a l i t y | +------+ |
> > > +------------------------+ +----+-+ | |
> > > | Richard Burnett | +-+ | |
> > > | Senior Design Engineer +---+ +----+ |
> > > | burnett_AT_tality.com | | |
> > > | | | |
> > > | Phone: 919.380.3014 | |
> > > | Fax: 919.380.3903 | | |
> > > +------------------------------------------------+
> >
>
> +------------------------+-----------------------+
> | T a l i t y | +------+ |
> +------------------------+ +----+-+ | |
> | Richard Burnett | +-+ | |
> | Senior Design Engineer +---+ +----+ |
> | burnett_AT_tality.com | | |
> | | | |
> | Phone: 919.380.3014 | |
> | Fax: 919.380.3903 | | |
> +------------------------------------------------+


New Message Reply About this list Date view Thread view Subject view Author view Other groups

This archive was generated by hypermail 2b28 : Wed Mar 06 2002 - 00:14:35 EET