Air-Powered Computer: A Revolutionary Design

[wolram]

can anyone recall a computer that instead of electricity
used air, i know logic circuits can be air powered and
I'm sure i read about this computer. the designer made
air bend around right angles somehow.

 

[MathematicalPhysicist]

can anyone recall a computer that instead of electricity
used air, i know logic circuits can be air powered and
I'm sure i read about this computer. the designer made
air bend around right angles somehow.

the only thing i can think of is using the motion of the air flow to generate electricity but it still uses electricity. (like on hidraulic power plants).

 

[wolram]

LQG, several machines i have worked used air logic AND,OR,NOT etc,
these control systems are used in hazardous conditions, where an
electrical discharge could cause an explosion, I'm sure someone took
the idea of air logic and created a working computer.

 

[MathematicalPhysicist]

LQG, several machines i have worked used air logic AND,OR,NOT etc,
these control systems are used in hazardous conditions, where an
electrical discharge could cause an explosion, I'm sure someone took
the idea of air logic and created a working computer.

ah, you are referring to boolean operators.

 

[Kenneth Mann]

Another possibility

Search through the subjrct "Fluidics". It was very popular several years ago, and I believe is still in being used. It might be useful to you.

 

[Njorl]

Search through the subjrct "Fluidics". It was very popular several years ago, and I believe is still in being used. It might be useful to you.

Have you read "Cryptonomicon"? In it, a mathematical savant and pipe organ enthusiast makes a computer out of pipe organ pipes and mercury. It made quite a racket.

Njorl

 

[mechie]

Wolram,
Clue us in then ... if you already know of air logic in control systems, how complex must a logic scheme be before you class it as a 'computer'?

Could you tell us more of the system(s) you worked on; is it fluidics (Coanda, jet destruction etc.) or clunky solenoid valve-type stuff? I ask this as I'm searching for (and failing to find any) info on Plessey fluidics (circa 1966-9).

 

[Averagesupernova]

An automatic transmissionis pretty close to air logic except that is uses fluid instead of air. The decision of when to shift is arrived at based on a number of inputs including engine load (vacuum), input shaft speed, output shaft speed, probably several others I am not currently thinking of.

 

[wolram]

If i remember correctly, the air computer relied on "boundary air flow", a jet of
air can be bent 90% by a second pulls of air, the first air jet will retain its modified
path due to this boundary air flow until it is acted on again, something akin to a
transistor, i have searched for more information on the subject, as I am sure a working
air based computer was built and tested for the mining industry.

 

[mechie]

Wolram
I have been searching for info but with little success !
Typical links worth looking at are ...
http://jnaudin.free.fr/html/coanda.htm
http://fluid.power.net/techbriefs/hanghzau/4_33.pdf
http://www.iitb.ac.in/~insight/issues/new/vol4iss2/fluidics.htm

The effect you describe (Coanda or 'wall attachment') allows for functions such as flip-flops, it provides the 'memory' required. Other gates requiring no memory seem to use fluid jet-destruction or some form of interference between power and signal fluid streams.
Push buttons and indicators have been produced for I/O, along with proximity detectors, liquid level sensors, solenoid input devices, pressure switches, etc.

As fluidic logic can include AND/NAND, OR/NOR and latches (I saw a spec sheet for a single module 'half-adder'), surely a computer will be nothing more than a collection of these devices?

I would love to know more about such a computing device, its circuitry, operation, programming, etc. though !
Could it date back to the late sixties ? it appears that fluidics in general industrial control died out at about the mid seventies; I know of two local Universities that have dumped their fluidic training gear so they can't be expecting a re-surgence of interest

 

[wolram]

Thanks Mechie ,
i suppose with the advances in electronics little use can be found for
alternatives, (Coanda or 'wall attachment') is an interesting property
of air, What about radio active environments or such extremes that
would destroy electronics?

 

[mechie]

The bit of info I've gathered so far seems to suggest that fluidics came along just too late for its own good - the mid sixties. Solid state integrated circuits were just starting to be developed for a 'mass market' so logic was rapidly becoming cheaper and easier with integrated circuits than their predecessors, thermionic valves or discrete transistors. Bear in mind that all these systems were very power-hungry.

I have found some 'sales pitch' references to fluidics being ideal for hash environments such as nuclear reactor cores or jet engine exhausts (being powered by the exhaust gas flow) but so far I can't think how often a controller could be useful inside a reactor!
Maybe if you fear an EMP attack you could construct fluidic based military defence system (tageting calculators etc.) but as the EMP would wipe out the electricity distribution system you would have to consider a suitable energy source - water powered?
I still find this a fascinating technology even if it does appear to be a real 'fringe' system, the Universities seem to have abandoned it, with the possible exception of Bombay (http://www.iitb.ac.in/~insight/issues/new/vol4iss2/fluidics.htm ), even they sound despondent about its future!

Bring back the 'Good Old Days', Radio Caroline, BBC Radiophonic Workshop and all ?

 

[wolram]

Mechie,
Bring back the 'Good Old Days', Radio Caroline, BBC Radiophonic Workshop and all ?
-------------------------------------------------------------------------------
Those were the days my friend i wished they would never end, oops,
You seem well informed Mechie, i have a passing interest in the science,
gained when using a pulsed air jet to identify the correct orientation
of a product used by our better halves, it was only 99% accurate and
no other solution on the horizon.
I imagined that air circuits could be "etched" out of plastic and layered
just like PCBs, i am unsure how mutch air would be needed to power
the beastie, but maybe air cushion shoes could pump enought
ghetto blaster? no its my air powered computer.

 

[manu_phm]

Hi!
There have been a lot of technologies which could not keep up in pace
with moore's law, and so they died. Air logic is one, another one being
magnetic bubble logic. Fascinating in themselves with rich physics, they
could not really turn into real products with a advnatge over other
technologies.

We wanted to produce combinatorial logic circuits at remote places in the world
cut off from many technological advances for supporting local projects. See http://fab.cba.mit.edu/

Tons of stuff exists on air based logic. Here are few references
books(you need a good lib to find them) :
- design theory of fluidic components - kishner/katz
- Fluid amplifiers - Kishner

proceedings
- Fluidics quaterly
- IFAC symposium on fluidcs.
- First fluidcis conference
and so on.

These devices do not scale with shrinking size cause of low reynolds number flows
in smaller channels, which are very laminar and thus none of the nonlinearities,
most high reynolds number devices depend on, work for Newtoninan fluids.

What might find exciting is how to shrink these devices. some interesting papers

- http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=PRLTAO000092000009094501000001&idtype=cvips&gifs=yesA
Microfluidic Rectifier: Anisotropic Flow Resistance at Low Reynolds Numbers

- Microfluidic Large-Scale Integration
http://www.sciencemag.org/cgi/content/abstract/298/5593/580

drop me a email if you need more pointers :)
m.

 

[wolram]

Manu_phm
Many thanks, a heck of a lot has been done, i will enjoy reading
the articles at the links you provided.

 

[wolram]

A subscription is needed for one link, and my PC can not
read the second ********

 

[manu_phm]

Hi!
Sorry, I did nto realize the subscription. Here are links to the first paper
without the details
http://physics.ucsd.edu/~groisman/windows/win9.htm
http://physics.ucsd.edu/~groisman/windows/win8.htm

alternate link to the second paper. You need acrobat pdf reader
http://thebigone.caltech.edu/quake/publications/scienceoct02.pdf

cheers
m.

 

[wolram]

Well maybe speed is not everything, a fluidic system could
provide both computational and muscular power for an android,
is that not the way biology works

 

[manu_phm]

Yes, That is exactly the point. Being able to actuate and compute with
no energy transformations. Also the whole system can be build on the
same substrate.

 

[mechie]

Moore's Law - some randomisation.

There have been a lot of technologies which could not keep up in pace
with moore's law, and so they died. ... ... ...

Would it be fair to say that transport, power generation and ball-point pens are "technologies" which could not keep pace with moore's law ?
Remember this 'law' was nothing but a mid-60's observation until the press blew it out of all proportion (as the press will do with everything they 'report').

Maybe fluidics could still be waiting for it's turn in the spotlight, much the same as Boolean algebra had to wait for a century or so before we realized how useful it could be?

OK - got to go, the straight jacket is waiting for me

 

[mechie]

Fluidic computer ?

Wolram, could this be something like what you were origionally asking about ?

http://americanhistory.si.edu/scienceservice/057001.htm

This looks like it could be described as a computing device ...
I wish I could get my hands on this wee beastie.

 

[wolram]

Mechie
Sorry for delay in replying, that device is a neat bit of kit, but not the one
i am thinking of, i am almost certain that an air powered comp was prototyped
for use in coal mines, i may have seen it in the," Focus ", mag some years ago
Thanks for link.