Fluidic logic
I recently made a blog post about the analogy between fluid flow in pipes and analog electric circuits. Of course, I’m not the first to think of this. It’s called the hydraulic analogy. But it’s not just analog circuits for which this analogy carries over, the analogy works for digital electronics as well. That is you, can do logic gates hydraulically. This is called fluidics and there already happens to be designs for how to construct hydraulic amplifiers. With hydraulic amplifiers, you can do logical operations.
How does it work? A very small stream of fluid deflects a very large stream. First I’ll show you how to do a NOT gate, then an AND gate and together NAND.
NOT gate:Simply take the left outgoing tube as your output. Done.
AND gate:To do an AND gate, take the output of the first amplifier, the right side, and feed it into the large stream of the second amplifier. Now the second amplifier will only have a stream in the right hand side if both control streams are on. BOOM! An AND gate.
Together, you can make a NAND gate and NAND is universal for logic which basically means you can make a computer if you can make a NAND gate.
The RepRap project was how I first learned about 3d printing. It has the remarkable mission, in that I’m remarking about it, of printing all of the parts to make another printer. Wow. Well, in principle, all of the circuits can be replaced with pneumatic ones. And also, you know, it might make a cool theme for a sci-fi flick. Just imagine a 3d printer printing with a background of hydraulic valves, pumps and hoses. Generate a pressure difference with steam and you’ve made a steam punk wet dream.
Now I’m not suggesting that we start running our 3d printers this way. The point is, if push came to shove, 3d printers could print out much more of their parts than they currently do without the need to resort to printing new materials.
In honor of the RepRap project, I designed a printable version of the fluidic amplifier. You can find it on thingiverse here. You can test that it works using your mouth.
Instructions:Just print it out twice and hold it together, tape it together, screw them together or glue them together. Doesn’t matter how you do it. Just make sure they stay together. To test them, you only need to hold them together with your hands and blow through the power port and one of the output ports. I just tested it and between 70-90% of the air went through the right output port. Considering that I was holding them together with my hands and using my mouth, that’s pretty freakin’ good. 🙂
I should also point out that this thing was already patented in 1974.
My Maker Blog 
Gratz! Your blog is what I’ve been looking for: I knew other people must be looking into this but I couldn’t really find anyone. Whilst I am interested in fluidic computation, I’m very interested in being able to use 3D printed devices to manage fluid flow in complex devices. I’d like other people to throw ideas around with on this if you’re interested?
Awesome! Is it just computation or do you plan on applying it in some way? I was thinking that the big advantage that fluidics has is that it naturally lends itself to controlling mechanical motion. Perhaps a fluidic stepper motor would be cool. Check this out: https://docs.google.com/viewer?url=www.google.com/patents/US3523488.pdf
Or better yet, a fluidic arduino! That might not actually be so hard. Probably the simplest case is to make a fluidic Turing machine which can have a state machine with an order of 10 or so states. More generally, if you had a circuit diagram of an electronic microcontroller, you could translate that into circuits for a fluidic controller, perhaps even write a script that automatically figures out the 3d print from a circuit diagram…
So you’re going more towards the hydraulics side of things as well as the fluidic computation then – interesting. My interests are leaning more towards the use of 3D printing to surplant current fluid handling techniques. It’ll allow the use of some seriously complex internal geometries and also, in time, will be much cheaper than trying to make these kind of things from the likes of CNC’d acrylic. Your solid state Tesla is something I’ve been thinking about trying for a while… left a 3D version I found as a comment on that page.
On my side, I’ve put together a 3D printed pump head with external parts to serve as the check valves, piston seal and piston. Next step is to move towards using dirt cheap external parts instead of expensive checks and piston bodies (need to find something cheap to serve as a very accurate piston) and after that… try and print all the components into a single device so it works right off the print bed!
Basically, the same as what you’re looking at doing but for a completely different end game.
Why are you using a vortex based element? This is an especially rare element in fluidics I suspect this might be due to it having a bad frequency response. Though the patent says it’s a good power amplifier which is nice.
You really should add some tubing connectors so you can ‘wire’ it up to other elements.
I’m actually wondering if you can use a 3d printer to print out an entire circuit in one piece? If you have thousands of fluidic elements, it would suck to have to “wire” or “tube” them all together.
It should be possible to print out an entire circuit in one piece, but it is best to make a breadboard of your circuit first to test to make sure that it works.
Well, you’re both right:
1) Bread boarding the ideas for a complex circuit can really help nix any flaws in your logic before you end up paying for a full print only to excitedly test it and find it doesn’t work… I’ve put together some isolated components for this purpose already and I’m designing more as we speak.
2) You definitely can print entire circuits but I’m currently eyeing up SLA and not FDM for this process as I think that the reliability in producing small internal channels will be much better.
…https://plus.google.com/u/0/102141588501443507606/posts/QW6M7vmd3fB
😀
n
Wow cool! How did you make the one way valve? Does it having a ball?
It doesn’t – I am thinking about a ball-seat one-way but I’ve seen them no seal at low flow rates and I think getting a perfect seal against a 3D printed surface likely won’t happen. It’s an elastomer valve – externally sourced but I’m hoping these seals will soon be printable too!
Also me I was looking exactly for a blog of this kind. I have started making a LEGO based microfluidic kit with valves,resistors, connectors, capacitors, colored fluid indicators (equivalent of LEDs) that work on air as fluid. You can find a first instructable on how to build these on “instructables”. Others will follow soon. Just search for microfluidic LEGO bricks. To my great surprise – I have found that circuits made of several tens of these components simply “wired” together with pressure-fit tubes work very well once you have found the right values of channel impedances. I feared that tube connections would slow down everything to the point making it no longer work. Instead even the more complex circuits time it worked very well. My valves use a silicone membrane bonded between two acylic pieces with CNC milled channels (as small as 0.2 mm wide/deep) and its design follows concepts borrowed from the scientific literature of microfluidics (I am researcher).
I want to stress that I do this purely as my hobby and I have made these fluidic LEGO bricks with materials and equipment available to any DIYer willing to spend around 2000 euros for manufacturing equipment/materials. (I use a home built CNC mill)
Some circuits that successfully worked were: ring oscillator, master-slave flip-flop, 3 bit multiplexer, pneumatic touch switch, logic gates. I build these from individual fluidic LEGO valves, resistors and capacitors. Then i set out to make integrated circuits.
At the moment I managed to make one working master-slave flip flop on an ordianry 2×4 LEGO brick, as well as a 3 bit multiplexer..although the latter is on a somewhat larger base 3 x 3 cm). The only drawback is that at the moment I need to assemble each brick individually and to make a kit you need many bricks. I feel though that the advantage of assembling bricks instead of building them as one monolitic piece is that you can have flexible (e.g. soft silicone membrane incorporated) and have more freedom in choosing the materials. ..e.g. using transparent acrylics milled with CNC allows you to see inside your components and see what the circuits are doing, which is cool by itself, especially if some parts of the circuit are filled with colored food dye. In principle you could automate the assembly process by home-building as “assembly line” …should not be to difficult conceptually.
Finally I wanted to use LEGOs as a base because I soon recognized that designing and fabricating fluid logic integrated circuits without simulation tools easily can lead in wrong designed impedances/capacitances so that you are not shure wether they will work. Using bricks you can first assure yourself to get a “feeling” for the critical parameters. Then once you have them right you can go and integrate them to make a LEGO fluidic IC…which as I said did sucessfully for a master-slave flip flop….but still a lot to discover ..and a long way to go for a simple “turing machine” or other “fluidino” . You will need a programmble pneumatic memory (some kind of RAM or at least ROM). Also I am asking myself why normal people – not crazy guys like me – should be intersted in a fluidic kit to build fluidic logic circuits or fluidic amplifiers or other fluidic stuff as a hobby. People using arduino are using it for fun stuff – usually blinking or making sound or driving some robot – or for useful appliances for home use, not so much for scientific equipment or for teaching. The same should be for a fluidic kit. It should be able to make “cool stuff”..that people want to build as gadgets or should enable to make new robots that you cannot do with electronics electromechanics/motors…maybe in the emerging field of soft-robotics – check out mikey77’s “soft-robot arm and gripper” instructable” or maybe pushing LEGO company to adopt pneumatic logics for extending mindstorm capabilities…and things like that. The “who cares” question is presently my ony doubt, since I have personally verified tha it can be made and that it works surprisingly well.
Is 0.2 mm the diameter of your smallest toolbit or the resolution of the machine? If it’s just your smallest toolbit, then you ought to try making some tiny fluidic amplifiers. If you can make fluidic elements that small, you should easily be able to make fluidic elements with low power consumption, acoustic frequency response, and high sensitivity.
So you should be able to make a nice sound amplifier, which should take care of “making a sound.” Then again this is probably small enough to make pretty any traditional planar fluidic element, so you should be able to make things like light sensors, proximity sensors, sonars, densitomers, angular rate sensors, and accelerometers, which should pretty much take care of driving a robot.
Turing machines are doable in fluidics, Univac built a proof of concept 4 bit completely fluidic computer and there’s a patent out there for implementing Stephen Wolfram’s Rule 30 and Conway’s game of life in a fountain:
Click to access viewer
Fountains are nice. A fluidic kit for making complicated fountains would be nice. At the very least it’d be a more fun way to learn about digital logic than wiring up a breadboard.
Fluidic elements for legos would also be nice, one could easily make some simple fluidically controlled underwater ‘robots.’ One might also make lego pneumatic cylinders able to cheaply ‘Servo’ to a commanded position. Should be as easy as the addition of a fluidic absolute linear encoder, which is basically a thin strip with a long triangular hole in it between a jet of fluid, a control circuit, a power amplifier, and a variable orifice for providing the control input. One could make a pretty nice pneumatic lego excavator this way.
Only problem is getting a flow source for Legos…
I can answer this question. I’ve been interested in ‘soft robotics’ for a long time with goal of making cheap inflatable robots. The cost is in the pneumatic controls – and the maze of tubes you need to control anything (see ‘Otherlab’ and their soft robot boxers). I think it’s feasible to use microfluidic ideas to shrink the controls down, eliminating components such as pneumatic proportional valves. I don’t have the time or right technical background to drive this but I’m hoping to sponsor a student to work on this idea. I’ve stared for example at the work Dr. Grover did on latching multiplexer valves and wondered if it could be developed for this application.
very interesting what you say. I am actually making exactly these types of “Grover” valves and logic circuits. As I said I have made some “single valve” LEGO bricks (transistor equivalent) as well as some integrated circuits such as grover did, (e.g. a master-slave flip-flop on a 2×4 LEGO brick worked well) with the only difference that I do them at home with a CNC mill instead of using regular photolithography, and second difference i make the microfluidic chips supported/embedded in lego bricks so that you can more easily handle them and connect tubes simply by sticking them into the milled conical access holes (see my “instructable” on “microfluidic bricks” for one of my first results…just a resistor on a LEGO…as I said meanwhile i have made progress to single valves and integrated circuits bricks) . It’s like building circuits with electronic components, just you do it on a LEGO base instead of a electronic breadboard. If you like we can exchange more information and collaborate…..e.g. send me a “private e-mail” via my instructables page.
My first aim was to show that it is possible to make a micropneumatic kit that mimics electronic component circuits.
I found that it is and that connection tubes do not give much troubles , except for introducing some short delay.
My second thought was the same as yours: is it possible to use such pneumatic logic circuits to control soft-robots and to eliminate much of the electromechanical pneumatic controls ? There are two points that need to be answered i think: 1) is the flow rate of air through microvalves high enough to control soft robot “muscles” rapidly enough ? 2) is the pneumatic logic circuit fast enough to process pneumati control pulses fast enough to control “useful” soft-robot movements ?
I have not yet experimented much on these topics. Concerning point 1) probably the answer could be yes if the soft robot has small air-consumption.
0.2 mm is the smallest diameter of my toolbit. Very interesting suggestions. Could you post a link to documentation of the Univac 4 bit fluidic computer ? My fluidic elements are based on membrane valves following design of PDMS microfluidic logic circuits of Matthies & Grover later used by Minsoung Rhee (Mark Burns Group) to build integrated pneumatic “microcprocessors”… actually “just” a shift-register to perform a serial/parallel conersion of pneumatic input pulses to control a microfluidic chip with in principle unlimited number of valves (but its operating at few tens of Hz so its slow) http://pubs.rsc.org/en/Content/ArticleLanding/2009/LC/b904354c. Other – somewhat faster – interesting integrated pneumatic logic circuits were recently made using PDMS pressure driven membrane valves using a modified “Quake valve” design http://pubs.rsc.org/en/Content/ArticleLanding/2012/LC/c2lc21155. In particular the latter are even smaller than mine, made by using photolithographic tecniques, but not readily accessible for hobbyists. So what I am trying is provide a easier way for hobbyist to make fluidic logic circuits, starting from “discrete component circuits” to “integrated circuits” such as the LS74 family of ICs. If someone manages to “mass produce” and sell standard logic or analog pneumatic fluidic ICs then everybody could indeed enter the field of fluidics quite easily and it could also lower the faabrication cost of professional complex microfluidic chips that presently often require tens of expensive electromechanical “off-chip” valves. I will try to go from CNCing every single component to molding or casting. I think either CNC or 3D printing will only be useful to make prototypes or molds for making fluidic kits, not to produce the kits themselfs. It drives me crazy allready making some tens of fluidic elements.
I also thinking about some “artistic” applications such as “fluid sculptures” http://www.youtube.com/watch?v=xx3NUlmjLw0. with pneumatic control. Fountains are also nice, but probably need to be small if such small valves or circuits are used, but that could be OK for a gadget. I know lots of sensors / sound amplifiers can be build with the valveless “old-timer” (but still powerful) flow bases fluidic circuits that you mention. Maybe I will explore building some of these in addition to the valve-based logic circuits I am making and see if they can be made working together. Pneumatic servo LEGO piston is also a nice idea.
Source of fluid/compressed air/vacuum ? Yes that appears to me the main problem. It is also for pneumatic robots or pneumatix exoskeletons. I am using a small (5 x5 x 5 cm) electrical motor driven pump, but still I need a big reservoir vacuum container or balloon for pressure otherwise the pump needs to be allways “on” and vacuuum/pressure are too much “pulsating”. So I was thinking about using CO2 cartridges, but it seems to me they will not last too long, given the still relatively large “air consumption” even of valve-based fluid logic…at least mine. It would be nice the be able to build “complementary logic” fluidic circuits with P and N-channel type FET valve analogues where air is consumed only during switching and not allways as is with transistor-resistor logic. Other idea could be to rely on “automobile style” fuel explosion based ways to generate gas as proposed recently by Whitesided for pneumatic soft-robots http://www.nature.com/news/explosive-power-makes-silicone-robot-jump-1.12402. Many ideas…too many..to explore. Let’s see.
Do you think you could make something like this: http://www.thingiverse.com/thing:111017 ?
It’s a fluid acoustic amplifier from this patent: http://www.google.com/patents/US3666273
yes its possible to make with CNC. I don’t know however if the surface machined by CNC will be sufficiently smooth.