US3825739A

US3825739A - High speed fluidic devices

Info

Publication number: US3825739A
Application number: US00072902A
Authority: US
Inventors: E Richards; W Depperman
Original assignee: Martin Marietta Corp
Current assignee: Martin Marietta Corp
Priority date: 1966-05-02
Filing date: 1970-09-16
Publication date: 1974-07-23
Anticipated expiration: 1991-07-23

Abstract

This invention relates to a binary accumulator stage consisting of pure fluid bistable and OR-NOR elements. The stage is joined with other similar stages to form a binary accumulator. Each stage sums an input value, a carry-in value supplied by a previous stage, and an addend value present in the addend register. An output signal and a carry-out signal are generated by each stage, the carry-out signal being fed to the succeeding stage as a carry-in signal.

Description

t d tslf ent 9 I 9' 82 J 9 ichards-6061.1 y.-, 1 1 i4si ii1 23,-1974 HIGH-SPEED FLUIDIC DEVICES [75] Inventors: Edward F. Richards; Warren B. Y

- Depperman, both of Winterpark,

OTHER PUBLICATIONS Fabrication Techniques, Fluid Amplifier State of the Art, General Electric Co., Schenectady, N.Y., Vol.

['73] Assignee: Martin-Marietta Corporation, New [Langley et al., Modular Pneumatic Logic Package,"

York, NY. f v v IBM Technical Disclosure Bulletin, Vol. 6, No. 3, Oc- {221 Filed? 16! 1970 I. tober 1963,10- 3-.4. g [2 PP ,9 Y 1 Priritary"EacamineP-Lawrence R. Franklin Related Application Data 1 gomgy, Agent, or FirmJulian C. Renfro, Esq.; Gay [62] Division Of'SCT. N0. 546,935, May 2, 1966, Pat. No.

' r g e [57] ABSTRACT "I in ntion relate's.to' a accumulator tage [5.1] Int. e G06d 1/10 Econsisting of pure fluid bistable and ORNOR 1 [58] Field of r 235/201 PF; 1137/ 's ments. The stage'is joined with other similar stages to form a binary accumulator. Each stage sums an input 1 Rsfmnces Cited value, a carry-in value supplied by a previous stage, 1 1 UNITEDSTATES PATENTS I and an addend value present in the addend register. 3,182,676 5/1965 Bauer ..'..l iii/ 1.5 An tp t signal and a y-out ig a are generated "3,-l90,554 6/1965 G'ehring et al.

v 235/201 PF by each stage, the carry-out signal being fed to the 3,226,023 1965. Horton... 235/201 PF succeeding stage as a carry-in signal. 3,286,086 11/1966 Bauer .3235/20-1 PF 3,350,009 10/1967 Rose 235/201 PF 1 g 1 7 Clams, 13 Drawing Flgllms .Aoozuo] f I, sum toeic CARFRIYFLOGI'C PAIENTED 3.825.739

I saw '2 or a FIG. 2

PATENTEDJUL23l974 SHEEI 3 OF 8 OUTPUT CARRY OUT CARRY LOGIC ADDEND LOGIC FIG. 4

ADD SHIFT RESET INPUT ARY CUMULATOR GE PICAL OUTIPUT 2 2 2 2| g I l a5\ 34 as -3 1 QARRY OUT ADDW SHIFT:

RESET:

I I l FIG. 3

PAIENTEnmzamu 325.139

SHEET 5 OF 8 FIG. 7

PAIENTED M2 3.825.739 sum 70F a FIG. IO

FIG. ll

' HIGHSPEED FLUIDIC-DEVICES This is a division of application Ser. No. 546,935,v

filed May 2, 1966, now US. Pat. No. 3,550,604.

computation, and control. We intend also that this'term willernbrace such systems and devices presently known as pure fluid systemsand devices, fluid amplifiers, fluid transistors, fluid power systems andfluid jet systems and devices; t I

'In the past'a'number of so-called pure fluid devices have been proposed, but these have been characterized by their relatively large size and slow operating rates,

and for-these and other reasons, such prior art fluid devices were notsuit'able for complex logic applications.

These earlier devices have typically involved utilizing a'stream of fluid under pressure, and at leasttwo receivers, into one or the other of which receivers the stream of fluid can be caused to flow, with control meansrbeing positioned so as to deflect such stream into the desired receiver. Such devices have been the subject of numerous patentsand many publications,

.with their proposed uses by industry becoming larger each'successive year. Howevenall known prior art devices of this type are either analog devices or digital Coanda effect-devices, and their operating rates have beenvery muchslower than thecomparable electronic devices. As a result, itqhas not heretofore been feasible. to perform the. complex functions normally associated with electronic logic by the use of fluidic devices.

As isknown, a network of logic elements for sequencing, computing and control system applications must complete a data cycle within the time interval defined by the intended use. Therefore, thenumber' of logic elements that can be effectively utilized in a given applicationdepends upon the operatingrate. in other words, the number of logic elements defines the numerical accuracy and functional capability of the digital device, so therefore the operating rate of logic elements is of vital importance in that it establishes directly the capabilityof the digital .mechanization. lt-is known that in a fluid device, the operating rate is directly related to size, and although the velocity of propagationv by fluid signals is limited, by employing very small element ite adeq ate response times can be obtained.

Whereasprior art fluid logic devices have been large and slow, suitable only for primitive logic networks, we

have dramatically extended the utility of the fluid logic technology by making our logic elements small and fast acting, so'as to make them highly useful incomplex logic devices.

a turbulent flow that substantial entrainment. of ,ad- I jacent-fluid can occur, but advantageously, the laminar flow emanating from our nozzle is caused to turn to turbulent flow before the confines of the cavity are passed,

which means that a desirable entrainment of the fluid from the cavity can take place, thus reducing the pressure therein andholding the jet of fluid in'the desired position. e

More specifically, our novel'concept makes possible .a new class of miniaturized fluid logic devices that significantly'can operate on initially laminar jets.

In our elements, the novel interaction area geometry is used to induce desired bistable or monostable ope ratio'n without dependence on sidewall attachment. In this class of device, miniaturized sizes can be em.- ployed, where laminar power jets are encountered.

Laminar power jets interact with boundary fluid I through viscous forces only, so there is no appreciable momentum exchange. As a result, there isi'nsuffrcient entrainment of fluid as is necessary to develop an adequate pressure gradient by proximity to a sidewall.

However, the necessary pressure gradient is established in thisclass by internal feedback, and/or selectively inducing turbulence in a portion of the jet.

In other words, our novel concept makes possible a new class of miniaturized fluid logic devices that can operate on initially laminar jets, and the fact that turbulent nozzle flow is precluded in small sizes due to the high ratio of surface to cross section in small channels is overcome by virtue of the fact that our novel cavity -configuration desirably induces turbulence in the initially. laminar jet at a location close to the nozzle exit.

Several other desirable effects are also'brought about, with the net result being 'to provide fluid logic devices in a smallness of size heretofore considered impractical. The aforementioned cavity is. disposed between the control-port and a receiver, whichcavity includes an upstream edge and a downstreamedge. The relationship of the upstream edge with respect to the jet is such that the control cavityis in effect isolated from the control port except when switching'pressure is applied thereto. That is to say, when the jet is flowing past the upstream and downstream edges, the cavity portion of the chamber is in effect isolated, with the entrainment of fluid being such that a negative pressure is created in the cavity, which lowered pressure causes the jet to stay in the selected position. i

Upon the arrival of the control signal at the control port, this causes the jet to move away from the upstream edge or point, thus to allow flow to take place from the control port into the previously evacuated cavity; This of course serves to dissipate the lowered pressure therein andthereby to cause the jet to tend to move away from the'cavity. Switching thereafter takes place very rapidly, such as within 20-microseconds.

By-virtue of the fact that we can'achieve highly satisfactory performance with only a very small supply nozzle, such as a nozzle of a width dimension of 0.004 inch,

We have created a new fluid element configuration I utilizing a novel cavity past which a jet of fluid from a nozzle can'flow. Because .of the unusual properties of this cavity configuration, laminar nozzle-flow can be used and the device made much smaller than was ever previously possible. As is known, it is only by virtue of tional advantages, such as the fact that each logic plane it is possible'for us to resort to printed circuit tech niques in the creation of our elements, and to employ copper foils that are 0.004 inch thick. As will therefore be seen, our nozzles are 0.004 inch on a side, .or have an aspect ratio of 1. This is of course to be contrasted with the smallest nozzle heretofore known, which was some 0.014 deep and 0.008 wide.

Our small nozzle makes possible a number of addi- 3 can be very thin and thereby make possible the stacking of dozens or even hundreds of logic planes into a single logic device. Whereas the aforementioned prior art nozzle that was 0.014 deep could be made by the use of say seven foils of 2-mil stock, we can make selfcontained logic planes that are each 0.004 inch thick, utilizing a single foil thickness for each logic plane, thus of course circumventing the alignment and registration problems that accompany devices whose elements are made of a plurality of layers.

It is significant to note that the application of a control signal to the control port brings about switching of the jet to the other receiver in a very rapid manner by v virtue of the fact that the distance from the control port to a receiver can be very short. This is to be contrasted with the prior art configurations, in which the nozzle was thought to be necessarily large in order to obtain the initially turbulent flow required in fluidic elements.

Accordingly, it is a principal object of this invention to provide miniature high speed fluid logic elements that can be quite successfully employed in sophisticated, yet very small logic networks. Further, since the operating rate in accordance with our invention allows the use of a large number of elements in a given size package, it is another object of this invention to provide high density element packages and fluid logic that can be produced at very low cost.

A further objectof this invention is to provide fluid logic elements with practical circuit characteristics such as adequate fan-in, fan-out capability, low power consumption, and reliable operation over a wide range of supply pressuresand loading characteristics.

Our invention provides a substantial improvement over Coanda effect devices in that through the use of our novel control cavity configuration, initial turbulence in the fluid jet is manifestly not necessary.-Our

cavity enhances stability of the fluid jet with respect to the selected receiver in a most significant manner, and because wall attachment is not required, our logic device is operable over a wide range of supply pressures. This is to be contrasted with Coanda effect device, whose attachment point undesirably varies with supply pressure aswell as other inputparameters.

As will therefore beseen, in a fluidic device in accordance with our invention, a source is employed'from which a fluid jet issues, and control port means are provided for switching the fluid jet between'one or the other of two alternative receivers. Asearlier men-- tioned, means define a novelcontrol cavity between the control port means and a receiver, which cavity serves to generate a low pressure with respect to the jet so as to hold same in a stable position with respect to the respective receiver. This cavity includes an upstream edge contacted by-the jet in such a manner that the control cavity is isolated from the controlports except when switching pressure is applied thereto. Switching pressure applied at the control port serves to move the fluid jet away from such upstream edge and produce a comparatively high pressure within the cavity so as to effect a switching of the jet to the other receiver in a very short time, such as in 20 microseconds.

to hold the fluid jet in a stable position with respect to a selected receiver can be utilized in a number of configurations, such as one in which a pair of cavities is used in symmetrical fashion, with each cavity being associated with one of two receivers, thus to form a fluidic flip-flop device or a pulse relay. As an alternative, however, a single cavity can be used to form a fluidic OR-NOR logic gate, for example.

In each of these instances it will be noted that the use of our novel cavity concept for developing the pressure gradient that holds the fluid jet in a selected position amounts to a substantial improvement over the prior art devices wherein a wall was employed for holding the jet in the selected position. This is of course because a high operating rate simply cannot be obtained in the large sizes required by wall attachment type devices. Further, by virtue of the fact that in accordance with this invention no wall is used to which the jet is to attach, switching can be brought about in an extremely rapid manner by merely causing the jet to move away from a cavity-defining position. It should be noted that our so-called wall-less concept possesses quite accept- .able stability standards inasmuch as low pressure is developed in even a single cavity tending to hold the stream of fluid in the desired location. When a pair of cavities is used, the cavity opposite the jet in any in- .stance serves to increase the fluid pressure on the jet and thus operates with the low pressure to hold the jet in the desired position.

These and other objects, features, and advantages will be more apparent from a study of the drawings in which:

FIG. 1 is a perspective type view of a logic device in accordance with our invention, illustrating the logic matrix in conjunction with top and bottom manifold plates secured in position;

-FIG. 2 is an exploded view of several logic circuit planes of the type used in the logic device of FIG. I, with the interconnecting fluid passages between elements. being indicated;

FIG. 3 is a block diagram of an exemplary fluid circuit in accordance with this invention, in this instance of description;

FIG. 7 represents a device along the lines of FIG. 6 but possessing the configuration of a fluidic flip-flop, presented to a scale much larger than that of the actual device;

FIG. 8 is a perspective view of the device shown in FIG. 6;

FIG59 is a perspective view of the device shown in FIG. 7;

FIG. 10 represents a single-cavity version of our novel concept, this particular device being a fluidic OR-NOR logic gate;

FIG. 11 is another symmetrical cavity arrangement in accordance with our invention, this device being in the form of a load-controlled fluidic pulse relay;

FIG. 12 is a perspective view of the device shown in FIG, 10; and

FIG. .13isa perspective viewof the device shown in FIG. 11.. I l I V v logic matrix in accordance with this invention may Referring to FIG. 1, logic device 10 is there depicted, I

' comprises a comparatively large number of logic circuit planes which together form the logic of this device. In this instance, the device is a six-bit binary accumulator,such as could be an integral building block of a digital computing mechanism. A similar assembly could contain logic fora sequencer or other digital control device. i

As will be apparent from FIG. 1, the INPUTS or augend 14 as well as the fluid power source 15 are disposed in bottom manifold plate 11, with the OUTPUTS 16 as well as the sequencing control ports'17, 18 and 19 being disposed in the top manifold plate 12. All of these inputs and outputs could be provided by-controllable (manually or automatically) interface devices or could be a consequence of othersignals where this device acts as a subassembly of a more complex digital mechanization,

The fluid power source 15 is a continuous pressure with a flow capacity adequate for the device, with prescomprise dozens or even hundreds of logic planes, with it also being understood that the five planes shown herein are typical planes that maybe juxtaposed in the illustrated relation in apre-established location in a typical logic device. Each circuit plane is preferably surized gases such as, for example, air, nitrogen, or the like being usable.

- The INPUT 14 is so configured as to allow a series of fluid pressures applied as a binary coded number. For example, the number 3 could be applied at the INPUT 14 by pressurizing the 2 and 2 ports, with zero pressure simultaneously being maintained at the- 2 2 2,

and 2 ports. Similarly, the number l5 could be applied to the INPUT 14 by pressurizing the 2, 2, 2 and 2 ports, with no pressure applied at the 2 and 2 control ports. I i

The top manifold plate 12 contains the ADD control port 17, SHIFT control port 18, and RESET control port 19, as mentioned earlier, which ports are used to sequence the logic device- Here momentary pressure pulses areselectively applied which control the logic operations. On the side of the top manifold plate, the OUTPUT set 16 is located, with this set servingto transmit the result of the logic device operation to a readout or. to a subsequent similar logic device. In other words, this output can be used in controlling and/or sequencing mechanisms or applied to another similar logic device to perform a more complex computing function. Each port of theOUTPUTset 16 is selectively pressurized as a consequence of the logic device operation, and as an example, this output set could represent a binary coded number. For instance, the number 29 would be presented by having pressure at 2, 2 2*,and 2, with zero pressure at ports 2 and 2 i The hole 21 in the middle of the top manifold plate extends down .through'thelogic matrix to provide a chamber which is used for venting the internal logic devices. The outer boundaries of the logic matrix are similarly used to vent the logic devices, such as to the atmosphere or a lowpressure plenum.

Turning to FIG. 2, it will be noted that in this exploded view of a typical fluidic logic assembly in'accordance with our invention consists of five exemplary circuit planes I through V containing our novel logic elements and their interconnections, arrayed in the approximate relationship. that they are disposed in the logic matrix. It is of course to be understood that the made of copper foil, because of the comparative ease with which known etching techniques may be employed tocrea'te logic elements in accordance with this inventionfAlso, copper foils are preferred from the standpoint of manufacture, for a desired number of foils may be satisfactorily secured together in a preestablished relationship either by clamping, screwing, or suitable bonding techniques. However, it is within the contemplation of our invention to use foils of brass, copper-brass alloy, or even stainless steel if such be desired, withthese of course being used in the thickness desired.

As will be apparent, each logic plane is prepared in accordance with a pre-established standard and basic'ally involves the use of three types of interconnections; power supplies, vents, and signal passages, with it also being evident that several logic elements may be disposed in each logic plane. These elements will be discussed in detail hereinafter, and it should at this point suffice to say that the elements used on the planes 'of FIG. 2 are either flip-flops of the type appearing in FIGS. 7 and 9, orelse OR-NOR gates of the type appearing in FIGS. 10 and 12..The element configurations are designed so that the metal foil after etching does not fall apart. This is made possible by the fact that stepping between elements often takes place between two or more adjacent planes, with the arrange- 35 ment being such that appropriate communication among the elements is conveniently made possible. As will also be apparent, the logic pattern in each instance'is arranged circularly around the common vent 21 to facilitate simultaneous venting on the inside as well as on the outside of the stack. Power supply ports -22 through 27 are disposed adjacent the elements, or

perhaps more accurately, the elements are disposed in such a manner as to partake of the fluidpressure available at power supplyports. For example, elements P &

Q are disposed in Plane 1 connected to

ports

22 and 23,

element R is disposed in Plane II connected to port 26, etc. The power supply ports will be noted to continue through all five illustrated planes. As will be understood, the Fluid Power Source 15 of FIG. 1 is connected to the supply ports'by appropriate power supply interconnections and passages disposed in bottom manifold plate 11, and in a similar manner, the input ports, output ports and control ports are connected to the matrix 13 formed of the standardized circuit planes by means of other appropriate passages disposed in the manifold plates. By standardized it is meant inthis context that the power supply ports, the central vent, and the indexing holes are in a pre-established location, and that the material size is the same.

FIG. 2 also reveals how fluid logic elements can be interconnected in complex sequencing and computing circuits. Each of the logic elements employed in this matrix must operate at a high enough rate so that the composite speed of the complete logic device is suitable for the intended application. In addition, a number i of the logic elements must be capable of farming out to a multiplicity of other elements and simultaneously.

being controlled by a fan-in from a multiplicity of elements. Interconnections among logic elements can be made either in a horizontal plane or by transferring vertically from stack to stack, the decision in each instance largely dependingon good design practice. Since as is apparent, at large number of discrete digital logic elements is required to mechanize digital devices in most practical applications, this imposes a requirement in addition to the other characteristics of the logic circuits, which is that they must be suitable for low cost fabrication techniques such as by certain photo etch processes.

Our invention will be explained by next referring to FIG. 3, which is a block diagram of a logic device, in this example a six-bit binary accumulator employing stages 31 through 36. As indicated earlier and shown here, the input to this device is a binary coded set, and the three controls, ADD, SHIFT and RESET, are used to sequence its operation. Similarly, the output is another binary coded set. Associated with each binary bit there are'blocks containing the CARRY LOGIC, the SUM LOGIC and the ADDEND LOGIC, as indicated in FIG. 4. In this example, the binary input set is added in a parallel fashion to the existing binary coded num ber on the ADDEND LOGIC and the result transferred to the SUM LOGIC. Whenever necessary as the consequence of the addition operation, digital logic is employed'to transmit a carry to the next higher ordered bit. This-addition operation is controlled by a momentary' removal of the fluid signal tothe ADD control channel. This binary coded set is then transmitted and- /or displayed at the OUTPUT ports.

To prepare the device fo the next add operation, the binary coded number of the SUM LOGIC is transferred to the ADDEND, is this instance by momentary removal of the pressure signal at the SHIFT control port. In, this case, element of the SUM LOGIC, and elements P, Q and R of the ADDEND LOGIC cooperate to perform the function of a shift register, where the state of the first bistable element 0 is shifted to the second bistable element R without changing the condition of element 0. The result of this automatic, or manually applied sequence of ADD and SHIFT commands is to accumulatively add thebinary codednumber applied as the INPUT set. The INPUT set can of course be changed-at any time, as for example from a bina'ry'function generator or an interface associated with the application of this device. v

The RESET control port can be used-atinitiation of the operation and/or intermittently to reset each bit of the device to zero. The OUTPUT is continuously presented and presents the binary number existing at any point in time in the SUM LOGIC.

Turning now to FIG. 5, this is a schematic of the logic ineach bit of the accumulator. Element 0 is the SUM flip-flop, element 0 the OUTPUT flip flop, and element R is the ADDEND flip-flop. Element R' of course appeared in FIG. 2, and the precise operation of this result on element 0 and O, and transmitting a CARRY (and the other) bistable devices will be set forth in connection with FIGS. 7 and 9. Similarly, the OR-NOR elements of FIGS. 2 and 5 will be discussed in detail in connection with FIGS. 10 and 12. It should be noted from FIG. 5 that removal of the SHIFT signal to the control ports of OR-NOR elements P and Q results in the transfer of the state of element 0 to element R. Similarly, removal of an ADD signal to the control ports of OR-NOR elements M and N results in adding the ADDEND, CARRY IN, and INPUT, displayingthe OUT through element F as required. Monostable elements A through G of FIG. 5 represent the CARRY LOGIC, elements H through 0 the SUM LOGIC, and elements P through T the ADDEND LOGIC.

In the device shown in FIG. I, assume that signals are normally present at the ADD CONTROL PORT 17 and the SHIFT CONTROL PORT 18, and that both the ADDEND LOGIC and SUM LOGIC of FIG. 5 have been reset to O. In this initially assumed condition, all OUTPUT signals 16 will be zero. Now assume that an INPUT signal 14 is applied to the first bit or the 2 INPUT port.

In order to add this number, which represents 1 in this case, to the number in the ADDEND LOGIC, which is now 0, one must momentarily remove the ADD control signal at port 17. This process sets the elements O and O in the SUM LOGIC in the first bit to the 1 channel which indicates 2 or 1. The signal to the ADD control port 17 is then reapplied and the number I in the SUM LOGIC of the first bit is then shifted to the ADDEND LOGIC of the first bit. This is accomplished-by momentarily removing the SHIFT control signal 1 8 which sets the ADDEND flip-flop R of the first bit from the 0 channel to the 1 channel. This in conjunction with the 2 input set signal generates a CARRY OUT signal from bit 1 which is applied to bit 2 as a CARRY IN. The second add operation will then add the number existing in the ADDEND LOGIC, in this case 2 or 1, to the number existing in the INPUT, again in this case 2 or 1. This is accomplished by again momentarily interrupting the signal at the ADD control port which results in setting the elements 0 and O in the SUM LOGIC of the second bit from the 0 channel to the 1 channel, and simultaneously setting elements 0 and O of the SUM LOGIC of the first bit from the 1 channel back to the 0 channel. This results in changing the condition of the OUTPUT signals 16 from a 2 state to a 2 state. This is equivalent to saying that the sum of 2 2 =2 or 2.

The exact mechanics of how this is accomplished may be seen by referring to FIG. 5 and it is described in the following discussion.

In this situation assumed for the purpose of explanation, let FIG. 5 represent the second bit of the 6 bit binary accumulator. Because all input signals except the 2 signal are absent, the input to bit 2 will be in the 0 state, thus explaining the 0 at the INPUT near the bottom of FIG. 5. As a result of the first addition and shift, a CARRY OUT was generated from bit 1, thus explaining the 1 condition existing here at the CARRY IN of bit 2. The ADDEND flip-flop R of bit 2 was previously set to the 0 channel by the initial premise of the problem. Now, in order to explain how the second addition process sums these three signals, the INPUT 0, the CARRY IN 1, and the ADDEND 0 to provide a 2 OUTPUT, the following discussion is presented.

As previously indicated, it is to be realized that the arrangement of OR-NOR gates and bistable elements shown in FIG. 5 represents one bit of an exemplary logic device designed to receive, shift and add binary logic signals. Each element of FIG. 5 is connected to the fluid power source 15 of FIG. 1 and each element is designed to perform a discrete logic function in a manner described in detail hereinafter. As will be apparent, the control ports of monostable logic elements A andB receive the CARRY IN signals and the control I port of monostable element G. g

. SHIFT pulses supplied through port 18 of FIG. 1 are i ports of elementsC and Bare designed toreceive the INPUT signalfThe control port of elementgD is designed to receive the output from the receivers of elements A and C when the control ports of latter elements are receiving no signaland such elements are therefore inthe on condition. Similarly, the outputs of elements DI: and B. are directed to the control ports of elements E and -F,*respectively, as well as to the control directed to the control ports of elements P and-Q of the ADDENDILOGIC of FIG. 5, with these elements also on occasion receivinginputs on

lines

51 and 52 from bistable element O, the presence of a signal on a given line depending of course upon the state of latter element. ADD pulsessupplied to port 17 of FIG. 1 are reelements aredirectly responsible for switching bistable element 0. One of the outputs of related bistable element represents the OUTPUT 16.

Asearlier mentioned, the CARRY INto the CARRY LOGIC is' from the CARRY OUT from the preceding stage, and conversely the CARRY OUT depicted in FIG. 5 becomes the CARRY IN of the succeeding stage. 7 An. input ,present in the .CARRY LOGIC can be transferred to the OUTPUT 16 by means of operating the SHIFTand'ADD control signalsthrough

ports

18 and 17 of FIG. 1. It should be noted that a change in the state of the CARRY LOGIC can be brought about without affecting the state of the output. Operating the vice are performed in binary logic fashiomand proviso that then state" 6r; element I will bedecided by the state of element T of the ADDEND LOGIC.

Assuming element T ison, element J is turned off and therefore no signal will exist on line 66, thus allowing the state of element K-to depend upon the state of element 11. Since element 11 was turnedoff by element G, there will be no signal in channel 72 to element K, so element K will be on and therefore element L will be off. Assuming no pressure is provid'ed at this moment by the ADD input port, element M'will be on and will provide a signal in line 68 to the control port of bistable "element 0, causing it to have an output in thel channel ceived by a control port of elements M and N, which port of element Q. Since element K is now on, element N is off, thereby allowing element 0 as just mentioned to provide a signal on line 52 that will turn" element Q off. The signal and thereby provide a signal on line 52 to the control on line 68 further flips element 0 to the OUTPUT 16 SHIFT and ADD input control signals, the new input .will be added to the "quantity already present in the SUM LOGIC block, All the logic functions in this de--.

sions are made to. carry the necessary-signals to the next bit of the device.

Referring to the by control port 14,0f FIG. 1 to either a lora 0, and as just reiterated, the CARRY IN represents the I or T the 0 CARRY OUT from thepr evious bit. In thisexample, the-INPUT is a 0, so element C is in the on position, and a pressure signal flows through line 40 to a control port of monostable element D. This of course causes a switching of element D to the off condition during the continuation of the pressure signal, thus preventing at this time the transmission of a signal to the control ports of elements E and G. I In this case-,a l is present in the. CARRY lN,.and this turns off element A, but this is irrelevant in this instance insofar as element Dis concerned, for whether or not there is a signalfrom A to element D on line 41 will not change the'state of element D when as here it has already been switched to the'off position by the sig nal from C. -:f' t

CARRY Loorc; the INPUT is set positionlAs will be seen, we have performed the binary addition of three signals, the, ADDEND a 0, the

CARRY IN a Land the INPUT a 0, with the result j being a 1. Since thiswas assumed to be the second binary bit .in the system, this 1 appears at the OUTPUT 16 as a signal to the 2 output. This OUTPUT represents the value 2, which is the sum of the INPUT 2 or 1, and the value that was in theSUM LOGIC, also a 2 tion one must momentarily interrupt the SHIFT control signal 18, whichessentially transfersthe number exist- "ing at the-OUTPUT'16 of the SUM LOGIC into the The CARRY IN signal is also senton line 42 to'a con 'trol port of element B and turns it-off, thus' in this instance making the presence of anINPUT at the control port of element B irrelevant.

As will now be apparent 0 (zero) signals exist on lines 43 and as a result of B and D being in the off condition, so element G'will be in the on positionand will provide a pressure signal on line 64 to the control ports of elements 11 and I of theSUM LOGIC, turning them off. Since element I is nowoff, there will be no signal on line 65 connected to the control port of element J,

ADDEND LOGIC. This is accomplished by'the AD- DEND LOGIC block of FIG. 4 which involves elements.

B through T of FIG. 5.. I v

.In orderto actually seehow this is accomplished in thephysical circuit, refer back to'the exploded view of FIG. 2, which-is relatable to a portion of FIG. 5. Since the output ofbit 2isnowin the l channelas theresult of the previous operation, a signal will be present on line52, which appears in plane I of FIG. 2 as well as in FIG. 5. This of course is a control input to monostable element O which holds this element in the off condition or 0 state. A shift control is normally present in channel 53, and as seen-in plane II, this channel branches into channels 54 and 55(not shown in FIG. 5 Channels 54 and 55 intersect channels56 and 57, respectively, with these latter channels being control inputs to elements P and Q respectively. As long as the shift control signal is present, both, of thelatter. elements will be held in the off condition. However, when .the shift pulse is interrupted, element P will turn on since as was noted, no

signal is present in line 51 from the zero channel of element 0. Element 0, however, will remain off because of the signal that is present in channel 52. When ele-" ment P turnson, its output is appliedthrough logic plane II into channel 58 of logic plane III and back to channel 59 of logic plane II, which is the control port of bistable element R OF latter plane. This sets element R tothe channel 1 condition. Since element R is a bistable element, it will remain in this condition 'even 7 11. y when the signal from channel 59 is terminated. This signal is terminatedby'reapplication of the shift" pulse,

which again'turns element P off. Element R has now been switched to the 1 channel and the signal from the l channelis transferred through logic plane III to logic plane IV, where via line 71 it feeds the control port of element T. Contemporaneously, since element R was switched from the channel to the 1 channel, the signal in the 0 channel was removed, which removes the sig- 5 to be carried out. For example, a total of say logic in line 51 to element P and that the signal in channel 52 has therefore been removed. This results when ele ment 0, due to successive addition processes, has been switched from the 1 channel back to the 0 channel.

. Now, when the shift control signal is terminated, element P will remain off due to the signal in channel 51, but element 0 will turn on since the signal in channel 52 is no longer present. When this occurs, the output from element Q travels through logic plane II, through channel 60 of logic plane III, back up to channel 61 of logic plane II, back down to channel 62 of logic plane Ill, and up to channel 63'of logic plane II, which is a control port of element R. This switches element R from the 1 channel back to the 0 channel and causes a corresponding change of state of the remaining ele ments driven by element R. Again, when the shift control signal is reapplied, element R will now remain in the 0 channel since it is bistable, even thoughthe shift control signal again turns off element O. This completes the detailed discussion of FIG. 2.

Under the initially-assumed conditions, element F was not generating aCARRY OUT. I-Iowever,.when R is flipped t0 the b As will be apparent from this exemplary circuit, element F will provide a l at the CARRY OUT if any two of the CARRY IN, INPUT, or ADDEND LOGIC represent a- 1. However, if only one, or none, of these isa 1, then element F is off, and a 0 is provided;

The foregoing description of a typical logic mechanization was presented in order to illustrate the way in which our novel fluidic elements can be employed in complex'integrated fluidic circuits. By using our novel elements in a counter circuit in which simple interconnections were possible, a packaging density of approximately 500 elements per cubic inch .was" obtained, whereas in a computer circuit made in accordance with these principles, a number of interconnecting planes in the nature of Plane III of FIG. 2 were required, and the packaging density fell to 200 elements per cubic inch.

As will be apparent to those skilled in the art, by effective design, certain logic devices have packaging densities of 1,000 fluidic logic elements per cubic inch or even higher can be built. As a result of the use of our principles, substantial improvement in overall operat-' ing rate can be effected relative to existing fluidic techniques.

Turning now to a detailed description of our novel fluid element configurations, and referring to FIGS. 6 and 7, it will be noted that nozzle is designed to provide a stream of fluid or jet that flows from a suitable source into control chamber 81, which chamber is principally defined by

arcuate sidewalls

82 and 83. The dimension of the channel associated with source 80 can be quite small, such as 0.004 inch wide. As will be more apparent hereinafter, these

sidewalls

82 and 83 in effect define

cavities

84 and 85 that serve in a highly advantageous manner to hold the stream of fluid from 80 flowing into receiver 86 or receiver 87, the particular receiver depending upon the direction of the most recent signal received in the control section 88 of the device.

' As will be noted, the

cavities

84 and 85 are of welldefined configuration, which start from a lower edge or point adjacent the control section 88. This is upstream edge or point 90 in the case of cavity 84, and upstream edge or point 91 in the case of cavity 85. It is most significant to note that when the stream of fluid from source 80 is flowing into receiver 86, for example, the stream flows closely adjacent upstream edge 90 whereas when the flow offluid is into receiver 87, the stream of fluid flows closely adjacent upstream edge 91. As will be obvious from the configuration exemplified in the figures of drawing shown herein, when the stream of fluid is closely adjacent a point or edge, the respective cavity is in effect isolated from the rest of the chamber, and most significantly such cavity is isolated from the control port area. At such times as the jet flows closely adjacent one of these well-defined cavities, a comparatively low pressure is developed therein, as will be described more fully hereinafter, thus holding the jet in a desired position and eliminating the need for utilization of the Coanda wall attachment phenomena, as inprior art devices.

Control parts

92 and 93 are provided on opposite sides of the device as shown in FIG. 1 and arranged to open into the control section 88 of the device. A proper signal at control port 92 will be sufficient to rapidly switch the stream of fluid or jet from receiver 86 over to receiver 87,-and conversely, a proper signal at port 93 will be sufficient to rapidly switch the jet from receiver 87 over to receiver 86. Exemplary FIG. 6 therefore represents a form of fluidic flip-flop.

Referring to FIG. 7 wherein a typical flow configuration is illustrated, it is to be understood that a lowpressure region is created in one or the other of the well-defined

cavities

84 or 85 as a result of entrainment of cavity fluid by the jet when it is adjacent a particular cavity. The jet, which is initially laminar, seals off the cavity at

upstream edge

90 or 91 as well as the respective downstream edge94 or 95 so that a small amount of entrainment creates a substantial decrease in pressure that is suitable for holding the jet very stably in that position. In this figure, the flow is into receiver 87, and two small arrows indicate the flow of fluid from cavity 85 as a result of this entrainment.

The cavities are also employed to quickly induce turbulence in an initially laminar jet. This is necessary since laminar jets do not entrain fluid effectively enough to create the desired low-pressure region in the cavity. The transition to turbulence is effected by an initial disturbance from the respective upstream edge,

in this instance. edge 91,

13 transverse fluid interactions during flow acrosscavity 85,,andthe sonic reflections from the respective downstream edge, in this instance edge 95. These effects combine to induce turbulence in the initially laminar jet at a point betweenthe upstream and downstream edges of the cavity, as illustrated, and it is the turbulent portion of the jet that entrains fluid from cavity 85 to create a low-pressure region in the portion of the chamber defined by the arcuate wall 83 and the stream of fluid when such stream is adjacent this cavity. In other words, when the jet or stream '18 on the right hand side of the chamber as viewed'in FIG. 7, it seals off the cavity 85 and thus isolates this low pressure region fromjthe control section 88 of the device. This advantageously prevents dilution of this low pressure region by flow through the control port and enhances the stability of the element as well as elimi nating undesirable interactions with other elements connected to the control zone by means of attachment to controlports 92 or 93 as the case may be.

It should be noted that the stability of the jet that is achieved as a result of the low pressure created in cavity 85 when the stream of fluid is disposed on the righthand side of the chamber is not derived entirely by virtue of cavity 85, fora desirable high-pressure regionis created on the opposite side of the stream of fluid as a result of the feedback flow injector 96 operating in conjunction with opposite cavity84. The immediately foregoing statement is based upon the fact that the feedback flow injector 96 returns a portion of the power jet along the arcuate upper and side boundaries of chamber 81 to react against the jet in thevicinity of itsupstream edge, as is depicted'by some eight small arrows in FIG. 7. In addition to thisreaction, the rein jected velocity head is converted tov static pressure which acts against the entire jet to-help maintain ,iton the selectedside. Thus it will be seen that whether the stream of fluid is flowing into receiver 86 or receiver 87, the feedback flow injector 96.will serve in concert with the cavity thatat that instant is farther from the stream'of fluid to create a positivepress'urebuildup serving withthe low pressure created in the cavity neartion switching of the jet over the-receiver 86, this being accomplished in 20 microsecondsorso rather than reest the jet to hold such stream offluidin the selected position.

It is not to be assumed tion ofhigh and low pressures serving to hold the stream of fluid in the desired position that switching is difficult to bring about, for in reality the converse is true, switching is quite easily'an d very rapidly brought about in accordance with our novel design. Significantly, the pressure required to direct the jet from one receiver to the other issubs'tantially lower than the pressure recovered in the receiver, which of course detines the gain of'the element and establishes the fan-out capability of our device. F an-out of course refers to the;

number of downstream logic elements that can be controlled from'one element. Assuming the stream of fluid in the right-hand position as illustrated in the'bistable device depicted in.

FIG. 7, upon-the application of a control pressure at control port 9 3,such acts on the jet upstream of edge 91 and establishes an initial deflection. This serves to open a low resistance path between edge 91 andthe jet control cavity 85, allowing the control signal to flow rapidly into the low-pressure region and distributing the switching pressure along the entire span of the ca'vity. The result of this is a" nearly instantaneous snap acthat because of this combina quiring' one or more milliseconds in. accordance with prior art devices wherein the control pressure had to walk along the length of an attachment wall'in order to bring about switching of the jet away from that wall to the alternate position. Since the switching control forces act on the jet upstream of the cavity, switching is accomplished in accordance with our invention without the necessity of having to overcome the stabilizing forces.

The fluidic flip-flop illustrated in FIG. 7 shouldlbe noted to bear a resemblance to the illustrative device depicted in FIG. 6, but differs in that it utilizes bleed ports or

channels

98 and 99 connected to the left and

right receivers

86 and 87, respectively. As will be seen,

' these bleed channels serve to enhance stability of the element with respect to output impedance changes, increase pressure recovery, permit stableoperation into a blocked load condition, and provide signal isolation in circuit interconnections.

Assuming the jet is entering the right-hand receiver 87 of FIG. 7, 'if, the load impedance of this receiver is increased, pressure would rise throughout the righthand receiver if ableed channel were not present, and

eventually a pressure value would be reachedthat would cause the jet to separate from downstream edge 95, which would unseal the low-pressure region existing in cavity 85 and allow receiver pressure to feed backinto'this region. This of course would serve to raisethe pressureon the right-hand side of the jet and undesirably cause'switching.

When in accordance with the embodiment shown in FIG. :7 bleed channels are incorporated, the pressure at point'95 is prevented by channel 99 from rising to a level.'-that would cause separation of the jet fromthis point and the jetremains stable on the right-hand side jacent left-hand receiver 86, a similar advantageous operating' characteristic is obtained when the fluid jet is disposed to flowinto the opposite receiver.

As another point with respect to this novel configuration, when the jet is flowing into say the right-hand receiver, the bleed channel of the left-hand receiver acts as a vent for any signals that may be impressed on the left-hand receiver, thus tending to provide a desirable isolation of the high-pressureregion from such signals.

- A similar advantage is also obtained of course when the jet is flowing into the left-hand receiver, by virtue of the location of the right-hand bleed channel. Althoughwe are-not to be limited to certain foil thicknesses, we generally prefer the use of comparatively thin foils, of a thickness of 0.010 inch or less, and highly satisfactory results have been obtained by using 0.004 inch foils. An approximate relationship between foil thickness and preferred cavity size may be deduced from FIGS. 8 and 9, which of course correspond .to FIGS. 6 and 7 respectively. Like FIGS. 6 and 7, FIGS.

trating the chamber relationships, and do not purport to show as FIG. 2 somewhat did, the inlet and outlet connections or orifices associated with the control ports and the receiver s. These orifices are of course to be understood to be connected to appropriate channels in the same plane, or in adjacent planes.

As revealed in FIG. '9, if the nozzle 80 is presumed to be of a width dimension D,the chamber dimension from the nozzle outlet to the feedback flow injector 96 maybe timesD, the width of a cavity'in the direction of flow may be 6.5 D, and the depth of the cavity 2 D. These dimensions are quite satisfactory over a wide range of source pressures, but we of course are not to be limited to same, for the chamber configuration can be varied somewhat if pre-established source pressures are to be used.

Turning now to FIG. 10, we illustrate in detail an OR- NORdevice in accordance with our invention. In this device, a geometric bias exists, so that an output is provided in NOR receiver 107 at all times except when a control signalis present. When any combination of one .or more control inputs is applied to the control port 113, the device will switch so that an OR output is provided in left-hand receiver 106. Therefore, this defice is monostable and can satisfactorily perform the afore v8 and 9 are primarily provided for the purpose of illussion at the nozzle eXit,-the upstream edge or point 110,

the downstreamedge or point- 114, and thecomplex flow interactions in the jet control chamber 101 is essentially the same manner as that described earlier for the symmetrical jet control chamber. The transition to turbulent flow is depicted in, FIG. 10.1-lowever, in this embodiment, the jet is biased, as previously mentioned,

so that it always establishes itself in the right-hand receiver 107 when a control signal is not present. This biasing is accomplished by rotating the power nozzle 100 through a small angle from the centerline of the device so that it pointslmore towa'rd'the right-hand receiver 107 than to receiver 106. The .jet remains 'in this position as a result of its momentum and the force developed along its left-hand side due to the high-pressure region resulting from the action of the feedback flow injector described earlier. In this case, provision is not made for a sealed low-pressure cavity on the right-hand side of the jet since the forces developed by the highpressure region on the left-hand'side of the jet and-the jet's momentum are adequateto normally hold the jet in the right-hand receiver. In this instance, the righthand bleed channel 119 serves the same purpose as in the corresponding description of the fluidic flip-flop; it enhances stability of the device with respect to output impedance fluctuations, increases pressure recovery, and permits stable operation into a blocked load. The left-hand bleed channel 118 provides isolation from signals that may be impressed on the left-hand receiver 106.

assume, as illustrated, that a signal is applied to the 'control port 113. This signal will raise the pressure on mentum and the force resulting from the high-pressure region on the left-hand side of the jet due to the feedback flow injector 116, and the jet will switch to the left-hand receiver 106.

The jet is held in this position by a combination of two forces. These are the force resulting from a high pressure region on its right-hand side due to the control signal, and a force resulting from a low pressure region developed in thesealed cavity 104 disposed between

points

110 and 114 located on the left-hand side of the jet. The reasons for establishment of this low pressure region were of course discussed in detail earlier.

In this state, the left-hand bleed channel 118 now provides the required stability with respect to output impedance variations, resulting in increased pressure recovery and stable operation with blocked receiver outputs. Again, the right-hand bleed channel 119 serves to isolate the jet control chamber from spurious signals that may be impressed on the righthand receiver 107. However, in this case, the right-hand bleed channel serves another very important function. As shown in FIG. 10, the beginning of channel 119 is located upstream of the tip of the feedback flow injector 116. This allows the flow from the feedback flow injector 116 to be dumped out the right-hand bleed 119 during the illustrated conditions, which prevents establishment of a high-pressure region due to feedback flow injection on the right-hand side of the jet. This enhances return switching to the right-hand receiver when the control signal is removed, thereby increasing frequency response.

Another bleed channel, channel 112, is used in this embodiment, and is located opposite to the control port 113. This channel is-identical to the left-hand controlport described earlier in the discussion of the fluidic flip-flop. Although it is not used as a control input in the OR-NOR logic gate, channel 112 serves a twovalue. This in turn permits switching to the left-hand receiver with a lower magnitude control signal than would be the case if this bleed channel were not used.

This feature then provides increased gain and frequency response.

In addition, bleed channel 112 enhances return switching from the left-hand to the right-hand receiver. To illustrate this, assume the jet is switched to the lefthand receiver and then the control signal is removed.

As the control signal is removed, the power jet deflects toward the right because of the angle of the nozzle, and breaks-the seal-between it and point 110. This opens the low pressure region 104 on the left-hand side of the jetto the bleed channel-112, which provides a source As to the details of switching the OR-NOR device, I

of flow to fill the low-pressure cavity 104 and raises the pressure in this cavity to ambient level. This results in easier and more rapid return of the jet to the right-hand receiver than if bleed channel 112 were not used. This feature also tends to increase frequency response.

As is therefore to be seen, the arcuate cavity 104, in a manner analogous to that of the bistable device, facillitates the transition to turbulent flow and isolates the resultant low-pressure region from both downstream and upstream effects, thus to enhance pressure recovery. The very fast response of our device is of course .power source. Also, the output'receiver in which the jet is established is controlled by a pressure differential applied across the receiver-outputs rather than by conventional control ports. 1

Consider first operation of the device when the nozzle is powered. Assuming the jet is set to the right-hand receiver l 27,=a turbulent jet is developed due to the combinedinfluence .of disturbances that occur at the nozzle exit, point 131, point 135, and the complex flow interactions inthe jet control chamber 125 in a manner similar to that described previously. A low-pressure region will be developed on the right-hand side of the jet, and a high-pressureregion on its left-hand side. in exactly the same manner as these pressures are developed in the basic jet control chamber. If the jet is established in the left-hand receiver 126, the pressure forces acting across itare of course reversed. The rightand lefthand bleed

channels

139 and 138 serve the same purpose as the corresponding bleed channels of the fluidic flip-flop. I v

The unique feature of this embodiment is the manner in which the jet is directed toward a certain preselected receiver. Consider the-device when there is no pressure applied to. the power nozzle, and apply a pressure differential across the output receivers withthe pressure applied to rightfhand receiver 127 greater than that applied to left-hand receiver 126;This pressure differential will establish aflow pattern in the device such that 'flowtravels down the right hand receiver, across the jet control chamber from right to left, and up through the left-hand receiver. Of course, some flow also escapes through the vent channels, but asubstantialamount is passed acrossjthe jet control chamber from right to left.

Now consider what occurs when a pulse'is applied to the power nozzle 120. Initially a lowenergy jet begins to issue from nozzle exit and the circulating flow in the jet control chamber deflectsthis jet to the left. As the power jet builds in intensity it developsa low-pressure region in left cavity 124 and a high-pressure region on its right .in a manner described previously, and the forces resulting from these pressures hold the jet stably in left-hand receiver 126. The jet will remain in this position until the pulse at the power nozzle is terminated. Oncethepowet pulse is terminated, a controlpressure from the load device.during the off portion of the nozzle power pulse.

As is therefore to be seen, we have provided a new, useful and unobviou's contribution to the fluidic art, in the form of 'a fluid jet device utilizing elements with 'unique chamber configurations. At least'one cavity in accordance with this invention is disposed in such chamber, between the control port and receiver sections of the element, and by generating low pressure on one side of the fluid jet flowing into a receiver, and a high pressure on the other side, this novel cavity brings about the stable maintenance of the jet in a desired position in the chamber until such time as a control signal is applied to the control portsection.

This inventioncan manifestly be utilized in monostable as well as bistable forms, and makes possible the utilization of elements so small that only laminar flows can emanate from the fluid nozzles of the elements.

However, by virtue of thefact that ournovel'cavity plays a vital role in bringing about the turbulent flow desired in the element chamber, elements with comparatively large jet nozzles are not required, thus making possible the construction of fluidic devices of a small size not heretofore thought possible, withmuch higher switching speed.

Six or so of such elements may be disposed on a single logic' plane or foil smaller than one inch on aside, and by the use of appropriate interconnection and stacking techniques, dozens if not hundreds of such foils can be arrayed into a highly versatile device in which element density of several hundred per cubic inch is obtained. i I I As mentioned earlier, the logic planes are designed to conform to a standard layout. With this layout each logic element is placed in one of six orso correspond ing positions in-all planes containing logic elements, so that when these planes-are stacked, a vertical column differential can be re-established across the output receivers; Suppose that inthe next case the pressure a'pplied to the left-hand receiver is greater than that applied to the right-hand re'ceivfenThis will establish a flowin the jet control chamber'from left to right. When a pulse is again applied to the power nozzle, this left-toright flow circulation will direct the jet to the righthand receiver 127,'where it will remain until the power pulse is terminated. Therefore, this device transmits a power pulse to a particular output receiver, such receiver being selected by a pressure. differential supplied of elements is created. These columns of elements are arranged in a circular fashion around a common center vent 21 extending through the central area of the stack, which arrangement permits each column of elements to be powered from a common vertical supply passage formed from aligned ports, 22, 23, 24, etc. It also permits short, direct,and properly oriented interconnections from element column to element column. In addition, element bleed passages on the outside of the element ring'can be ported directly to the edge of the stack, while vents on'the inside of the element ring are ported directly to the center vent. eliminating the crossing of bleed and signal passages and simplifying circuit design.

As will be understood by those skilled in this art, a given device in accordance with our invention may involve some degree of repetition of plane design, and for example, one device of planes used some 40 different plane designs constructed to the aforementioned standard layout principles. However as the device involvedbecomes more complex, there generally is less repetition of plane design in a device.

As will be apparent, by utilizing the aforementioned standard configuration principles, several different basic logic planes can be made in large numbers utilizing certain etching techniques. These logic planes are typically made of foil as previously mentioned, and because each foil is an inch or less on a side, such logic planes can be made in large sheets that are thereafter separated to form the-individual logic planes. Quite ob- 7 plane contain an element, for as was noted in connection with logic plane Ill in FIG. 2, some of the planes may contain only interconnecting passages. Further, not all the logic planes of a given fluidic device need be of the same thickness, for if such be warranted, thicker or thinner planes than a standard thickness may be incorporated in a given fluidic device. Further, it is not necessary that the logic planes constituting a fluidic device be bond'ed together, for other techniques such as screwing the planes together may be employed. Furthermor e the planes need not be metallic material, for in some instances thin planes of dimensionally stable plastic or even glass may be employed.

With regard to the recovery obtained by our elef ments, we have found that a pressure recovery of percent is typical, with better values than this being obtained on many occasions. Flow recovery under certain conditions may actually exceed nozzle flow. As to pressure gain, which may be defined as recovered pressure divided by the pressure required to switch the element, values from 5 to 10 can be expected, depending on element design. Flow gain, as defined in an analogous manner, may also range from 5 to 10. These values when multiplied together represent power gain, which therefore may range from 25 to 100.

We claim:

l. A fluidic binary accumulator stage comprising:

a. fluidic means for entering a carry-in signal from a previous stage;

b. fluidic means for enteringan input signal;

0. fluidic carry logic means for summing said input signal and said carry-in signal into a first sum and for providing a partial carry-out signal;

d. a fluidic addend register including a first fluidic memory means which contains an addend value;

e. fluidic sum logic means for summing said first sum and said addend value into a second sum;

f. a second fluidic memory means;

g. a fluidic OR-NOR logic circuit means responsive to said second sum for setting said-second memory means;

b. means to receive an add signal for actuating said OR-NOR logic circuit means;

i. a feed back circuit from said second memory means to said addend register for entering said second sum into said first memory means, said feedback circuit including fluidic gating means;

j. means to receive a shift signal for actuating said gating means; and

k. a fluidic carry-out logic circuit means responsive to said partial carry-out signal and an output of said addend register to provide a complete carry-out signal for a subsequent accumulator stage.

2. A fluidic binary accumulator stage comprising:

a. fluidic means for entering a carry-in signal from a previous stage;

b. fluidic means for entering an input signal;

c. fluidic carry logic means for summing said input signal and said carry-in signal into a first sum and for providing a partial carry-out signal;

d. a fluidic addend register including a first fluidic bistable flip-flop which contains an addend value;

c. fluidic sum logic means for summing said first sum and said addend value into a second sum;

g. a fluidic OR-NOR logic circuit means responsive to said second sum for setting said second bistable flip-flop;

h. means to receive an add signal for actuating said OR-NOR logic circuit means;

i. afeedback circuit from said second flip-flop to said addend register for entering said second sum into said first bistable flip-flop, said feedback circuit including fluidic gating means;

j. means to receive a shift signal for actuating said gating means; and

3. A fluidic binary accumulator stage as defined in claim 2 in combination with at least one other accumulator stage of substantially identical construction thereby forming a multi-bit accumulator.

4. A fluidic binary accumulator stage as defined in claim 2, including a fluidic output stage coupled to said fluidic OR-NOR circuit means and responsive to its output.

5. A fluidic binary accumulator stage as defined in claim 4 in combination with at least one other accumulator stage of substantially identical construction thereby forming a multi-bit accumulator.

6.."A fluidic binary accumulator stage as defined in claim 4, wherein said fluidic output stage is a third fluthereby forming a multi-bit accumulator.

Claims

1. A fluidic binary accumulator stage comprising: a. fluidic means for entering a carry-in signal from a previous stage; b. fluidic means for entering an input signal; c. fluidic carry logic means for summing said input signal and said carry-in signal into a first sum and for providing a partial carry-out signal; d. a fluidic addend register including a first fluidic memory means which contains an addend value; e. fluidic sum logic means for summing said first sum and said addend value into a second sum; f. a second fluidic memory means; g. a fluidic OR-NOR logic circuit means responsive to said second sum for setting said second memory means; h. means to receive an add signal for actuating said OR-NOR logic circuit means; i. a feed back circuit from said second memory means to said addend register for entering said second sum into said first memory means, said feedback circuit including fluidic gating means; j. means to receive a shift signal for actuating said gating means; and k. a fluidic carry-out logic circuit meanS responsive to said partial carry-out signal and an output of said addend register to provide a complete carry-out signal for a subsequent accumulator stage.

2. A fluidic binary accumulator stage comprising: a. fluidic means for entering a carry-in signal from a previous stage; b. fluidic means for entering an input signal; c. fluidic carry logic means for summing said input signal and said carry-in signal into a first sum and for providing a partial carry-out signal; d. a fluidic addend register including a first fluidic bistable flip-flop which contains an addend value; e. fluidic sum logic means for summing said first sum and said addend value into a second sum; f. a second fluidic bistable flip-flop; g. a fluidic OR-NOR logic circuit means responsive to said second sum for setting said second bistable flip-flop; h. means to receive an add signal for actuating said OR-NOR logic circuit means; i. a feedback circuit from said second flip-flop to said addend register for entering said second sum into said first bistable flip-flop, said feedback circuit including fluidic gating means; j. means to receive a shift signal for actuating said gating means; and k. a fluidic carry-out logic circuit means responsive to said partial carry-out signal and an output of said addend register to provide a complete carry-out signal for a subsequent accumulator stage.

6. A fluidic binary accumulator stage as defined in claim 4, wherein said fluidic output stage is a third fluidic bistable flip-flop.

7. A fluidic binary accumulator stage as defined in claim 6 in combination with at least one other accumulator stage of substantially identical construction thereby forming a multi-bit accumulator.