25

We usually use electrons to communicate data across wires, and sometimes we use light also. Does anyone know of real world example systems that use fluids (including air) to do this, and the advantages vs using wires? When I search for air pressure communications I get TPMS tire pressure monitoring or "how to communicate under pressure" lol. Also, I'm not referring to the systems of tubes that carry messenger capsules with paper inside, but instead to direct digital communications via the modulation of air or fluid pressure through a pipe or hose.

I've heard of using air pressure to communicate analog process data in factories before 4-20mA was invented, but I'm not sure of the details. Also, that's an analog technology and I'm more interested in digital data streams.

I'm sure this type of system would be slow, but could be interesting to study.

Thanks for any info!

Ryan Griggs
  • 2,596
  • 2
  • 30
  • 53
  • 79
    I think that is called "sound" :-) – Axis Jul 04 '19 at 00:36
  • 1
    @axis LOL great answer! To clarify, I'm looking for systems that run pipes instead of wires and force air or fluid through them to communicate digitally. – Ryan Griggs Jul 04 '19 at 00:37
  • 2
    I once used a 4" steel-braided air hose as a (very) leaky waveguide ... Does that count? – brhans Jul 04 '19 at 02:01
  • 2
    ["How is sounds used to transmit data underwater?"](https://dosits.org/people-and-sound/communication/how-is-sounds-used-to-transmit-data-underwater/) – JimmyB Jul 04 '19 at 09:20
  • 1
    FYI [Air is a fluid](https://www.thefreedictionary.com/fluid) – Reversed Engineer Jul 04 '19 at 10:15
  • 1
    The ventilation controls in my previous car (Hyundai Accent CLi, 1997 model) used pipes with pressure or vacuum (not sure) to respond to changes to the dials on the dashboard, presumably opening flaps deeper in, that controlled hot/cold air, opening or closing air to the face or feet or windscreen. – Reversed Engineer Jul 04 '19 at 10:18
  • @daveboltman I forgot about that. My 1979 Caballero used vacuum from the engine through the ac/heat controls to do that. – Ryan Griggs Jul 04 '19 at 12:36
  • @DaveBoltman yep for clarity I updated the question to reflect that air is a fluid. – Ryan Griggs Jul 04 '19 at 12:43
  • 1
    It is used much more than you imagine. I remember some technology course in school, where we used penumatic elements to make logic circuits. Some companies specialize in such components, see for example [this catalog](https://www.parker.com/Literature/Literature%20Files/pneumatic/Literature/Telepneumatic/PCC-4.pdf). This kind of things is very common in trains, construction vehicles, factory machines, ... anything industrial. – dim Jul 04 '19 at 13:37
  • 1
    Pneumatic controls are still being used in the old hospital I'm working at right now. – JMac Jul 04 '19 at 13:56
  • 1
    Related but not quite what you're asking: using mercury to store data. https://en.wikipedia.org/wiki/Delay_line_memory – thosphor Jul 04 '19 at 14:48
  • @JMac can you share any details? How does it work? What is being transmitted? – Ryan Griggs Jul 04 '19 at 15:42
  • @dim that catalog is awesome. – Ryan Griggs Jul 04 '19 at 15:44
  • 11
    You can send a 1 bit message that the dinner's ready with roasted chicken molecules. – init_js Jul 04 '19 at 17:54
  • 4
    @init_js Actually it is a 3-state system - cooking (no smell), processing complete (yummy!), burnt (sometimes accompanied by an audio signal from an automated system attached to the ceiling). – manassehkatz-Moving 2 Codidact Jul 04 '19 at 22:44
  • 1
    A starter's gun seems to fit the bill. A single bit transmitted by air. – WhatRoughBeast Jul 05 '19 at 22:32

13 Answers13

44

Mercury delay line memory used sound pulses in mercury to transfer bits. The advantage of this (compared to electrical signals) is the relatively slow propagation speed of 1 450 m/s while electrical signals travel at speeds above 100 000 000 m/s.

This slow speed was taken advantage of to create memory. An emitter and receiver were connected using a column of mercury. Storing a bit is done by emitting a pulse into the mercury. This pulse will take some time to travel through the mercury to the receiver. Once the pulse reaches the receiver it can be re-emitted again (and again and again...), allowing the bit to be stored until no longer needed. More data can be stored in a single delay line by sending out trains of pulses.

UNIVAC I is a famous example of a computer using this type of memory. It stored 120 bits of data per column.

Aaganrmu
  • 530
  • 4
  • 8
  • This is really cool. I would love to see this in operation. – Ryan Griggs Jul 04 '19 at 12:40
  • 1
    [The Register](https://www.theregister.co.uk/2013/06/28/wilkes_centenary_mercury_memory/) has some more historical details and a picture. Alan Turing wanted to use gin instead of mercury! – Ben C Jul 05 '19 at 18:24
  • 1
    Yep. In the early 70s I worked with a guy (Walt Helvig) who had previously worked on a computer with delay line memory. Semiconductor memories were just being invented, though, and the acoustic delay line quickly lost traction. – Hot Licks Jul 06 '19 at 11:20
37

It's used in oil drilling operations. Telemetry data from the drill head is transmitted as sound waves that propagate through the coolant.

Data rates are abysmal (~10 bits per second) because a lot of frequency spreading is needed to get a signal that can be separated from the noise of the actual drilling.

It is part of Measurement While Drilling (MWD) techniques, and often referred to as mud pulse communications (the signal is transmitted by modulating the fluids which are generically called 'mud')

rolinger
  • 215
  • 1
  • 5
Simon Richter
  • 12,031
  • 1
  • 23
  • 49
  • How do they generate the sound waves? Do they use electronics like a speaker/ultrasonic transmitter or is it done mechanically and powered by the spinning drill? – Ryan Griggs Jul 04 '19 at 12:39
  • 4
    Interesting answer. This seems an utterly obnoxious application for sound transmission, for the noise reason you mentioned; I really wonder why they can't use some kind of electronic transmission, seeing as there is a mechanical link to the surface anyway. Of course, acoustics make a lot of sense though for underwater applications where there's _no_ connection (submarines/AUVs), because neither light nor radio gets very far through water. – leftaroundabout Jul 04 '19 at 13:41
  • 4
    @leftaroundabout Maybe because the attachment of multiple, flexible, drill rods in a dirty, high vibration environment to get the drillhead deeper and deeper is not conducive to having stable, clean electrical contacts. You'd also have to have electrical channels running inside the drill rod which alongside the coolant which sounds like a PITA and would require dedicated drill rods to use with your telemetry system. – DKNguyen Jul 04 '19 at 20:30
  • 2
    What's the power level required for this? How is the telemetry equipment powered? – TLW Jul 05 '19 at 01:49
  • 2
    The coolant (the mud) is forced through the system by volumetric pumps, so the flow is _near_ constant. The "mud pulses" are generated by creating an temporary obstruction on the flow path, which in turn generate a pressure increase, then decrease when the obstruction is removed. Modulating the obstruction activation allows to create a pressure signal detectable from surface. The device which generate the obstruction (piston or rotating), is powered electrically, either by it's own batteries (most often), or sometimes by an embedded electricity generating turbine. – Hoki Jul 05 '19 at 15:41
  • 2
    And here is a quick overview of the 3 main types of pressure pulses: [Mud Pulse telemetries](https://qph.fs.quoracdn.net/main-qimg-3952d9228682338f22b50cb7d74c2b6b) – Hoki Jul 05 '19 at 15:48
24

Read the Wikipedia entry on "Fluidics". You will find that a fluidic digital computer, named FLODAC, was built in 1964. You will also see descriptions of fluid versions of logic gates. These types of components were very useful in applications where electromagnetic interference and/or radiation levels were too high for electronics.

Dancrumb
  • 245
  • 1
  • 2
  • 9
Barry
  • 15,733
  • 1
  • 26
  • 28
  • 9
    The tragedy is that this physics was known to the Victorians. In some alternative reality Babbage met the right person, and the digital computer was born in the 1850s. – nigel222 Jul 04 '19 at 12:32
  • @nigel222 no way. You don’t have the fan out with fluid gates. It requires too much power to switch the state of each gate. You really need the transistor for that and you don’t have the transistor without quantum mechanics, which was developed in the early 20th century. – user110971 Jul 04 '19 at 15:47
  • 3
    @user110971 the wiki article cited above (https://en.wikipedia.org/wiki/Fluidics#Amplifiers) suggests otherwise, along with the existence of FLODAC. – nigel222 Jul 04 '19 at 16:03
  • 9
    @user110971 the existence of vacuum tube computers implies you don't need transistors to have something useful. – mbrig Jul 04 '19 at 16:53
  • 2
    In 2009, Rhee and Burns fabricated a [pneumatic (micro)processor](https://www.newscientist.com/article/dn17730-new-microprocessor-runs-on-thin-air/). – Eric Towers Jul 04 '19 at 20:51
15

Yes there are a few cases but I'm not sure you'd be happy with the answers.

Before electronics was invented, large pipe organs used small lead tubes to carry the signal from the console to the relevant pipe. The system is called "Tubular pneumatic action" Each key on the manuals requires its own tube, and each "stop" required a tube as well.

When the organist presses the key they either vent the tube to atmosphere or to a vacuum or pressure source, depending on the specific organ. That propagates up the tube to the base of the pipe where it opens a valve to blow air into the relevant pipe.

This is a truly digital system, the signal is either present or not, it just maybe isn't quite what you thought.

The second case is a little more computer-y. Back in the very early days of home computers, radio stations would sometimes broadcast game code over the radio as audio. If you didn't have cords to connect your computer to the radio or to a tape recorder, you could use a microphone. The very first old school modems were designed in such a way that you would set a regular phone on them and the modem had a speaker and microphone on it, instead of plugging into the line directly

Andrew Crews
  • 151
  • 2
  • 1
    Those are excellent examples, thanks! I remember those old modems with rubber cups where you would place a phone handset! The pipe organ is also a really cool example. Didn't know about that. – Ryan Griggs Jul 04 '19 at 07:56
  • 1
    In US the _first_ modems from Ma Bell were wired-in, and leased (only) at high monthly rates, but (until the Carterfone decision) _competitive_ devices had to use 'acoustic coupling' as you describe, and those were the first ones most people could afford – dave_thompson_085 Jul 04 '19 at 17:04
14

I remember it as fluidic logic. Here's an old cover from Scientific American: enter image description here

Channels were formed in plastic and streams of air or fluid were used to "switch" the "circuits".

Elliot Alderson
  • 31,192
  • 5
  • 29
  • 67
  • 5
    I think old automatic transmissions (automobiles) used fluidic logic also. – Marla Jul 04 '19 at 01:59
  • 2
    Why is "switch" in scare quotes? The flows really are switched from one channel to another. – Pete Kirkham Jul 04 '19 at 11:47
  • 4
    @PeteKirkham I agree with you about the verb "switched" but I felt that the terms "switch" and "circuit" had been so strongly associated with electrical elements that I wanted to emphasize the analogous (same?) functions in the fluidic context. – Elliot Alderson Jul 04 '19 at 12:56
12

I'm surprised that nobody has mentioned the acoustically coupled modem, though admittedly that's a bit of computer technology dating from before I knew what "computer" meant. Anyway, you can see one in operation here.

nigel222
  • 641
  • 5
  • 10
9

Before LED-based remote controls became common, many were ultrasonic.

Some of the transmitters were even entirely mechanical, requiring no batteries.

https://www.youtube.com/watch?v=PlgSuaIHYsY

Matt Timmermans
  • 1,358
  • 6
  • 7
  • Good one! I used to work on Zenith TVs that used those mechanical remote controls - basically just like a tiny ultrasonic xylophone. – Edward Jul 05 '19 at 17:32
7

Look up "air logic" or "pneumatic computers".

I doubt you will find many digital examples (if you find any at all) because to do anything at all requires too much hardware which is impractical with non-solid state, non-miniaturized technologies. Analog does a lot more with less hardware. The main advantage of digital is flexibility and programmability but most of that doesn't matter if your machine is too unreliable to run.

Example: A single full adder (which is just 1-bit) needs a couple dozen transistors. You can make an op-amp for the same number of transistors but it can do useful addition and a lot more. If these were vacuum tubes or pneumatics instead of transistors, it's a no brainer whether you would go digital or analog unless you're someone crazy like the US military.

Would you rather run 8 digital pipes to signal 256 different values? Or just a single analog pipe?

DKNguyen
  • 54,733
  • 4
  • 67
  • 153
  • I was actually looking for examples where fluidics were used as a means of data *transmission* in the same way wires and fiber optics are used. I wasn't really looking for fluidics data *processing* (but it's totally awesome anyway!) just examples of transmitting data through this type of medium. I.e. Air or fluid data buses. – Ryan Griggs Jul 04 '19 at 03:25
  • 1
    They're essentially the the same thing though. The reason you have standards like ethernet is because electronics can be mass manufactured and the same components being re-used and interfacing with each other. With a fluidic computer, each system is probably (I'm guessing) completely custom built. When it's custom built there's no standards since re-use and interchangeability is not as big an issue. After all, that fluiduc computer has to send and receive signals to other things in the plant somehow (as well as within itself just like any computer). It just won't have a name like "ethernet". – DKNguyen Jul 04 '19 at 03:31
  • 2
    Sort of. As an example, we use light (fiber optics) mostly for data transmission between digital electronic systems. We don't do much with optical gates and optical data processing: that work is done by electronics. That's the difference I was trying to explain. I'm looking for fluidics media to interface between standard electronic digital systems, just as light is used in fiber optic cables. – Ryan Griggs Jul 04 '19 at 03:32
  • @RyanGriggs Well, the reason we don't do optical data processing is because we haven't figured it out. If we could we would and there is research going on in that area. – DKNguyen Jul 04 '19 at 03:33
  • 1
    Absolutely. I agree that digital light processing would be awesome. But I was simply trying to explain that I'm looking for transmission media, not processing capability. – Ryan Griggs Jul 04 '19 at 03:35
  • @RyanGriggs I'm saying the transmission media is inherent in any factory that uses a fluidic control system. It just won't have specific name due to non-standardization. All you have to do is find an example of a factory control system that has a fluidic computer https://www.sciencedirect.com/science/article/pii/S1474667017675098/pdf – DKNguyen Jul 04 '19 at 03:44
  • "*Would you rather run 8 digital pipes to signal 256 different values? Or just a single analog pipe?*", isn't this 99.99% dependent on the noise level at wherever you are implementing it? – Harry Svensson Jul 04 '19 at 04:19
  • @HarrySvensson Probably. 256 feels like it's pushing it. A better example might be 2 to 4 digital pipes for 4 or 16 different levels. – DKNguyen Jul 04 '19 at 04:22
5

Here are three things not yet mentioned:

Audio programming of sports watch

I once owned a sports watch (Polar RS 100) which featured configuration via sound. Much like the already mentioned acoustic cup modem, software on a computer would encode the setting information as sounds which could then be sent to the watch by putting the watch into a receive mode and putting earbuds under the watch. The computer would send the sound and the watch would receive and apply the settings. It had the advantage that it was inexpensive and didn't require an electrical connection.

Railroad switch control and signaling

Another system is the Bianchi and Servettaz hydraulic interlock used in lever frames for Italian railroads in the 19th century. The same hydraulic mechanism that operated a rail switch also controlled the signals to tell approaching trains which direction the switch was set, so one could use it as an example of a digital hydraulic signal.

Pneumatic thermostats

Pneumatic thermostats use air as the medium to transmit a control signal to an actuator. Essentially, it's a one-bit digital transmission system.

Edward
  • 2,698
  • 1
  • 18
  • 23
  • I've seen pneumatic thermostats. The watch is cool... What data rates did it achieve? – Ryan Griggs Jul 04 '19 at 22:18
  • I’m not sure about the watch data rate, but you can read the owners manual here. That might be in there somewhere. https://support.polar.com/e_manuals/RS100/Polar_RS100_user_manual_English.pdf – Edward Jul 04 '19 at 23:05
  • Similar to your watch, some LG fridges have a diagnostic mode that emits data over sound waves. – Jacob Krall Jul 05 '19 at 04:21
3

2012 and later Furby toys communicate with each other and a tablet/smartphone app using audio signals modulated with the other chatting and squeaky noises they make; you can hear this as a kind of background hiss:

https://en.wikipedia.org/wiki/Furby#2012_Furbies

Someone has done some reverse engineering:

https://github.com/iafan/Hacksby

Goodlife77
  • 31
  • 1
2

A few years ago there was a case where a security researcher concluded that his BIOS had been compromised by a virus which was transmitted using PC speakers and microphones. I think it's pretty universally accepted that he was incorrect in that case (and that it would be basically impossible to cause an initial infection via this vector, unless there was already some backdoor in which case it's not really worth the effort).

It did however encourage a group at Fraunhofer FKIE to test the feasibility of this data transfer method, where they managed to get a transfer rate of ~20 bit/s over about 20 metres line of sight.

llama
  • 548
  • 3
  • 9
  • I remember reading about this. Obviously as you say, the initial infection vector would require a backdoor to already be in place. However it would let infected computers discover each other and possibly update or sync data. Also it would be cool to try sharing data between smartphones with this. You could even broadcast data like urls etc at events to anyone Iistening with a special app with no need to be connected to the venue's wifi or cellular data. – Ryan Griggs Jul 04 '19 at 22:17
0

Advertizers are using ultrasonic (or nearly sonic) beacons that are picked up by smartphone microphones to track people. I can imagine that such a beacon is emitting some kind of FSK data.

Examples:

filo
  • 8,801
  • 1
  • 25
  • 46
0

Backhoes use pneumatic controls. It's more analog than digital Any pipe organ uses air controls, it's pretty close to on/off And some cars used a vacuum system to control vents. The carburetor & intake manifold of the engine have a low pressure zone. So tap into that and install a hose and you have a source of 'vacuum' that is used to control A/C and heating. You can't get, obviously, more than about 14 psi, and only for small devices, and only for occasional actuations not continuous flow.

Woody
  • 1