Title:
ACOUSTIC DELAY LINE MEMORY SYSTEM
United States Patent 3668662


Abstract:
An information storage system utilizing an acoustic medium having its surface subdivided into a plurality of individually isolated acoustic transmission channels upon which a plurality of transmit and receive transducers are selectively placed. Each respective transmit transducer is electrically excited by a signal containing information from at least a preselected one of a plurality of data sources in order to cause a resultant acoustic stress wave to be propagated upon the surface of the medium within the associated acoustic transmission channel until the resultant stress wave is converted by an associated receive transducer into a plurality of charges which, in turn, are applied to a data source which, on command, either selects a new data to be stored or the old data that was circulated to be reconstituted into a transmit format to be recirculated within the associated acoustic transmission channel of the system.



Inventors:
Zimmerman, Robert L. (Northridge, CA)
Schweitzer, Bernard P. (Los Angeles, CA)
Garvin, Hugh L. (Malibu, CA)
Pedinoff, Melvin E. (Canoga Park, CA)
Waldner, Michael (Woodland Hills, CA)
Application Number:
05/082229
Publication Date:
06/06/1972
Filing Date:
10/20/1970
Assignee:
Hughes Aircraft Company (Culver City, CA)
Primary Class:
Other Classes:
333/150, 365/76, 365/77, 365/157, 365/189.08, 365/219, 365/233.1
International Classes:
G11C21/02; H03H9/42; (IPC1-7): G11C21/00
Field of Search:
340/173RC,173R 333
View Patent Images:
US Patent References:
3394355Information storage timing arrangement1968-07-23Suwkowski



Primary Examiner:
Fears, Terrell W.
Claims:
1. A recirculating memory system comprising:

2. A recirculating memory system comprising:

3. The system of claim 2 wherein each of said plurality of transmission channels further includes:

4. The system of claim 2 wherein:

5. The system of claim 2 wherein:

6. The system of claim 2 wherein:

7. The system of claim 6 wherein:

8. The system of claim 6 wherein:

9. The system of claim 2 wherein in relation to each of said transmission channels:

10. The system of claim 9 wherein in relation to each of said transmission channels:

11. The system of claim 10 wherein in each of said transmission channels:

12. The system of claim 10 wherein in each of said transmission channels:

13. The system of claim 2 wherein in relation to each of said plurality of transmission channels:

14. The system of claim 13 wherein each of said transmission channel:

15. The system of claim 13 wherein in each of said transmission channels:

16. The system of claim 2 further including a summer coupled between each pair of said plurality of second transducer means to enable each pair of said plurality of second transducer means to form a signal channel with the corresponding pair of said plurality of first transducer means wherein:

17. The system of claim 16 wherein:

18. The system of claim 17 further including in each signal channel:

19. The system of claim 18 wherein:

Description:
BACKGROUND OF THE INVENTION

1. Field of the Invention

This invention relates to memory systems and particularly to a multichannel acoustic delay line memory system.

2. Description of the Prior Art

Many types of memory systems are presently utilized to store information. Some types of memory systems are magnetic core memories, magnetic drums, magnetic tapes, mercury delay lines and electrostatic storage tubes. Other types of currently used memory systems include flip-flop circuitry, and piezoelectric and non-piezoelectric acoustic media. Disadvantages of the aforementioned systems include bulk, excessive cost in relation to the amount of information stored, a relatively low information storage capacity, relatively low data rates, and a relatively long access time for extracting stored information.

More specifically, the widely used types of memory systems which store information in a receptive medium use the bulk of the medium to store that information. These types of acoustic memory systems use bulk waves to propagate a signal in the medium. The bulk waves spread throughout the entire medium and are not amenable to the physical separation of the medium into channels isolated from each other. As a result, no presently known acoustic memory system possesses all of the relative advantages of: compactness, low power consumption per bit, low cost per bit, high bit density, high data rate, and short access time.

SUMMARY OF THE INVENTION

Briefly, Applicants have provided an acoustic delay line memory system wherein a plurality of transmit and receive transducers selectively placed in a plurality of individually isolated acoustic channels on the surface of a medium selectively operate to delay data from a plurality of data sources and return the delayed data to the plurality of data sources which may be respectively commanded to select either new data to be stored or the delayed old data to be circulated within the associated acoustic channels of the system.

It is therefore an object of this invention to provide an improved acoustic delay line memory system.

Another object of this invention is to provide a compact, multi-channel acoustic memory system having low power consumption per bit, low cost per bit, high bit storage density, high bit rate and a short access time.

Another object of this invention is to provide an acoustic memory system for selectively storing in a plurality of isolated channels data which may be in the form of a plurality of amplitude or phase coded carrier frequencies, or a plurality of coded digital signals.

Another object of this invention is to provide a multi-channel acoustic memory system wherein each isolated channel can be utilized to circulate a plurality of different non-interacting digital trains therethrough.

Another object of this invention is to provide a multi-channel acoustic memory system wherein a plurality of non-interacting coded carrier frequencies can be utilized in each isolated channel.

A further object of this invention is to provide a multi-channel acoustic memory system wherein a plurality of non-interacting complementary digital codes can be utilized in each isolated channel.

BRIEF DESCRIPTION OF THE DRAWINGS

These and other objects, features and advantages of the invention, as well as the invention itself, will become more apparent to those skilled in the art in the light of the following detailed description taken in consideration with the accompanying drawings wherein like reference numerals indicate like or corresponding parts throughout the several views wherein:

FIG. 1 is a schematic block diagram of an acoustic delay line memory system in accordance with a first embodiment of this invention.

FIG. 2 illustrates a top view of one type of acoustic channel which may be used in the embodiment of FIG. 1.

FIG. 3 illustrates a side view of a second type of acoustic channel which may be used in the embodiment of FIG. 1.

FIG. 4 is a schematic circuit and block diagram of one type of data source which may be used in the embodiment of FIG. 1.

FIG. 5 is a schematic circuit and block diagram of another type of data source which may be used in the embodiment of FIG. 1.

FIG. 6 is a schematic circuit and block diagram of an acoustic delay line memory system in accordance with a second embodiment of this invention.

FIG. 7 is a schematic circuit and block diagram of an acoustic delay line memory system in accordance with a third embodiment of this invention.

FIG. 8 is a schematic block diagram of an acoustic delay line memory system in accordance with a fourth embodiment of this invention.

FIG. 9 is a schematic circuit diagram of one of the adjacent channels of FIG. 8.

FIG. 10 illustrates the A-code and auto-correlation function of the A-code.

FIG. 11 illustrates graphs of the auto-correlation functions of the A and B codes generated in the embodiment of FIG. 8.

FIG. 12 illustrates waveforms used in explaining the operation of the embodiment of FIG. 8.

FIG. 13 is a schematic block diagram of an acoustic delay line memory system in accordance with a fifth embodiment of this invention.

FIG. 14 illustrates auto and cross correlation products of complementary and non-interacting complementary codes.

FIG. 15 illustrates a means for reducing the data access time, which could be applied to any of the embodiments presented.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

Referring now to the drawings, FIG. 1 illustrates a schematic block diagram of an acoustic delay line memory system in accordance with a first embodiment of this invention. The upper surface of a receptive acoustic medium 20 is shown divided into 10 signal storage paths or channels 21 through 30 to permit a separate isolated transmission through each channel of a different acoustic energy wave containing signal information to be stored in that channel. It should be understood that more or fewer than ten channels can be utilized on the upper surface of the medium 20. Adjacent channels on the upper surface of the medium 20 are isolated from each other by channel isolators 31 through 39. The acoustic surface wave in each channel spreads as it leaves the associated transducer. The angle of spreading depends upon the width of the transducer as well as the acoustic wavelength in the medium. The spreading of the acoustic beam follows the laws of defraction wherein the sine of the angle of divergence is proportional to the ratio of the wavelength to the transducer width. The beam width increases proportional to the product of the distance from the transducer and the sine of the angle of divergence. Depending on the distance from the transducer the beam will spread such that a portion of the acoustic energy could overlap into the adjacent acoustic channel. The beam spread may be limited by bounding the acoustic channel by a channel isolator or by using the inherent material characteristics of the acoustic medium to provide self channeling.

One type of channel isolator that may be used is a dissipative material, such as apiezon wax or silastic rubber, which is deposited axially between adjacent acoustic channels. A dissipative material absorbs any spreading acoustic energy in any given channel which tends to spread outside of that channel and is incident upon the dissipative isolator. Another type of channel isolator that may be used is a non-dissipative material which has a higher acoustic propagation velocity than that of the medium 20. The material with the higher acoustic propagation velocity tends to cause the portion of the acoustic wave that impinges onto it to travel faster than the portion of the wave that travels within the desired channel. The acoustic energy within the spreading channel therefore tends to bend toward the center of the channel. The acoustic energy is therefore conserved since it is deflected into the channel wherein it can carry signal information. A third means for providing isolation between channels is to use the directional properties of the material used as the acoustic medium to provide inherent isolation. Some types of acoustic materials, such as lithium niobate and bismuth germanium oxide, have a self-focusing or guiding property which permits channel isolation to be obtained without adding physical isolators. The diffraction of the acoustic wave is dependent upon the physical characteristics of the medium. For an isotropic medium the diffraction angle will be related to the ratio of the acoustic wavelength to the transducer width as previously stated. However, an anisotropic material like a piezoelectric crystal affects the acoustic wave and either tends to make the spreading either worse than or better than that obtained for the isotropic case. Materials whose acoustic velocity decreases when a wave is propagated off of the crystal axis when compared with the on axis velocity will have a tendency to cause the diffraction angle to become smaller and therefore the beam will spread less and tend to confine itself to a channel. Bismuth germanium oxide and lithium niobate have axes which exhibit the characteristics described. Quartz, however, has a velocity characteristic which is opposite to materials like bismuth germanium oxide and lithium niobate in that the velocity of propagation increases when an acoustic beam propagates off axis. This characteristic results in an increase in the diffraction angle and therefore the beam becomes more divergent than it would be in the isotropic case. The basis for the above is that a wave will propagate normal to the reciprocal velocity surface which further explains the phenomenom described herein. Isolation between channels may also be attained by using an axially positioned groove between adjacent channels as a boundary for the acoustic wave. The acoustic wave which impinges upon the groove wall will be prevented from spreading into adjacent channels and therefore the wave will be confined to one channel. Furthermore, additional attenuation of the portion of the wave which impinges upon the wall of the groove may be attained when the groove is filled with an absorbing material. Acoustic energy which would tend to propagate across the groove thus will be attenuated upon interacting with the dissipative medium which fills the groove.

The channels 21 through 30, respectively, contain interdigital transmit transducers 41 through 50 located at one end of the medium 20 and respectively contain interdigital receive transducers 51 through 60 located at the opposite end of the medium 20. The medium 20 may be composed of any type of piezoelectric material, such as quartz, or may be composed of any type of non-piezoelectric material, such as glass or sapphire. When the medium 20 is composed of a piezoelectric material the transmit and receive transducers in each channel may be any suitable type of metallic conductor, such as, for example, aluminum, which is placed on the surface of the medium 20 by, for example, photolithographic techniques, which are well known in the art, or by any other suitable method.

A top view of channel 2 2 is illustrated in FIG. 2 which shows metallic interdigital transmit and receive transducers 42A and 52A, respectively, on the upper surface of the medium 20 and isolated from adjacent channels by isolators 31 and 32. The transducers 42A and 52A are each shown to contain five fingers 68, although a greater or smaller number of fingers can be utilized. The spacing between an adjacent pair of fingers in a transducer is designed to be equal to one-fourth of a wavelength (λ/4) of the maximum intended input signal frequency to the channel. An input electrical signal containing the information to be stored is applied between input terminals 61 and 62 of the transmit transducer 42A. The transducer 42A converts the input electrical signal into a mechanical force in the form of a stress wave which moves in the channel 22 through the medium 20. The stressed piezoelectric material develops alternating moving charges which move with the stress wave. The receive transducer 52A receives and converts the alternating moving charges to an electrical signal which appears across its output terminals 64 and 66.

FIG. 3 illustrates a side view of one of the channels when a non-piezoelectric material, such as glass, is used for the medium 20. Layered transmit and receive transducers 42B and 52B are utilized at opposite ends of the upper surface of the non-piezoelectric material. A suitable metal in the form of interdigital thin film fingered devices 63 and 65, similar in shape and deposited in a manner similar to the transducers 42A and 52A, respectively, of FIG. 2, are deposited on the upper surface of the medium 20. Over each of these devices 63 and 65 a thin film piezoelectric material 67 is deposited. In the operation of the channel shown in FIG. 3 the conversion from a stress wave into charges does not take place until the stress wave reaches the layered receive transducer 52B.

Throughout the remaining part of this description a piezoelectric material, such as quartz, will be used with the interdigital transducers to explain all of the embodiments presented. It should, however, be understood that the teachings of this invention in all of the embodiments presented permit the use of the piezoelectric medium with the interdigital transducer or the non-piezoelectric medium with the layered transducers, or by other means suitable for the excitation of an acoustic signal in the medium.

Returning to FIG. 1, it is further shown that one of the input terminals of each of the transmit transducers 41 through 50 and one of the output terminals of each of the receive transducers 51 through 60 is grounded. The ungrounded terminals of the transmit transducers 41 through 50 are respectively coupled to data sources 71 through 80 for respectively receiving different input electrical signals. The excitations of the transducers 41 through 50 by the different input electrical signals from the respective data sources 71 through 80 cause acoustic stress waves containing different information to be stored to propagate down the channels 21 through 30 to the receive transducers 51 through 60, respectively. The polarity of the input electrical signal to each transmit transducer determines the polarity of the charge on the resultant acoustic stress wave. There is is a delay involved in each channel which causes each channel to act as a delay line. The delay involved between the time that a signal excitation is applied to a transmit transducer to create the acoustic stress wave and charges and the time that the corresponding receive transducer reproduces the signal information is dependent upon the type of medium used and the physical distance between corresponding fingers 68 on corresponding transmit and receive transducers. If quartz is the piezoelectric material used as the medium 20, the acoustic wave information in each channel propagates at the approximate rate of one-eighth of an inch per microsecond (1/8 in./μ sec.)

Amplifiers 81 through 90 are respectively coupled between the ungrounded terminals of the receive transducers 51 through 60 and the data sources 71 through 80 in order to amplify the signal information from the receive transducers 51 through 60 to usable levels before respectively applying them to the data sources 71 through 80. Each data source of FIG. 1 either reconstitutes the input amplified signal information into a transmit format for subsequent reapplication to its corresponding transmit transducer, or upon command, selects new data to be placed in a transmit format for subsequent circulation. Each of the data sources 71 through 80 is basically identical in structure and operation, except that each data source processes different signal information in the form of a plurality of amplitude or phase-coded carrier signals, which, however, may or may not operate at the same carrier frequency. One type of data source for the embodiment of FIG. 1 is illustrated in FIG. 4 and will now be explained.

FIG. 4 illustrates a data source for developing signal information in the form of an amplitude-coded carrier frequency signal. In this type of amplitude coding, a carrier frequency burst 93 of a preselected time duration represents a logical one state ("1"), while the absence of a carrier burst 95 for the same preselected time duration represents a logical zero state ("0"). This type of amplitude-coded carrier frequency signal is applied to a corresponding transmit transducer recovered by a corresponding receive transducer after a preselected delay in the channel and amplified by a corresponding amplifier in the manner previously discussed. The output signal from the corresponding amplifier is applied to a demodulator 97, which may be any suitable amplitude detector for converting amplitude-coded input data into digital information in order to develop at its output logical "1"s and "0"s containing the desired channel information.

The digital information from the demodulator 97 is applied via a test point 98, from which the old data may be monitored, to a new-old data recirculate gate 99. In an old data recirculate mode of operation, the gate 99 causes old data to be recirculated through itself and its associated channel and amplifier back to itself. In a new data insertion mode of operation, the gate 99 allows new data to be introduced and then recirculated.

During the old data recirculate operation when no new data is to be inserted, a "1" state signal is applied from some control device such as a computer (not shown) to the upper input or a NOR gate 107, via recirculate terminal 103, to assure that the NOR gate 107 only applied a "0" to the lower input of an OR/NOR gate 109 when only old data is to be recirculated. The OR/NOR gate 109 may basically be an OR gate, or any other suitable type of logic circuit, for providing complementary (logically noninverted and inverted) outputs at its respective output terminals 117 and 118. The terminals 117 and 118 are also the output terminals of the gate 99. The digital data at the terminal 118 has the same phase relationship with the old data output at the test point 98 during the old data recirculate mode of operation or with the new data being applied to the terminal 101 during the new data recirculate mode of operation. The "1" at the terminal 103 is also logically inverted by a NAND gate 111, which applies the resultant "0" to the lower input of a NOR gate 113. The digital information from the demodulator 97 is applied via the test point 98 to the upper input of the NOR gate 113, which acts as a logic inverter, since its lower input is a "0". The logically inverted old data digital information from the NOR gate 113 is applied to the upper input of the OR/NOR gate 109. Since a "0" is being applied to the lower input of the OR/NOR gate 109 at this time, the OR/NOR gate 109 logically inverts the inverted digital information applied to its upper input, thereby causing the digital information at the output terminal 118 to be in phase with the digital information at the test point 98.

During the new data insertion operation, new data to be stored and a "0" state are applied from the aforementioned control device (not shown) to the respective lower and upper inputs of the NOR gate 107 via a new data terminal 101 and the recirculate terminal, respectively, to cause the NOR gate 107 to act as a logic inverter of the new data being applied to its lower input. The "0" state at the terminal 103 is also logically inverted by the NAND gate 111 before being applied to the lower input of the NOR gate 113 to assure that the NOR gate 113 only applied a "0" to the upper input of the OR/NOR gate 109 during the new data insertion operation, thereby preventing old data from passing through the NOR gate 113 while new data is being inserted.

The data from output terminal 118 of the recirculate gate 99 is then applied to a data terminal 119 in a pulse controlled oscillator 120. The digital data at the data terminal 119 may be utilized to monitor the channel operation and is also utilized to supply a potential through a resistor 121 to the collector of a NPN transistor 123. The transistor 123 has its base and emitter electrodes respectively coupled through resistors 125 and 127 to ground. A crystal oscillator 129 provides a carrier signal at a preselected frequency to the base of the transistor 123. Whenever a "0" is present at the data terminal 119, the transistor 123 is cut off. During the time that a "1" is present at the terminal 119, the carrier frequency signal from the oscillator 129 is amplified by the transistor 123 and applied from the collector of the transistor 123 through a resistor 131 to the corresponding transmit transducer associated therewith. The output of the pulse-controlled oscillator 115 is therefore composed of combinations of carrier bursts 93 which represent logical "1" states and lack of carrier bursts 95 which represent logical "0" states, with a carrier burst being defined as a sequence of sine waves of a preselected duration and at the carrier frequency, which appear at the collector of the transistor 123 during the time that a "1" is present at the terminal 119.

FIG. 5 illustrates a different type of data source which is used for developing signal information in the form of a phase-coded carrier frequency signal. In this type of phase-coding a carrier frequency burst 133 of a preselected phase relationship and duration represents a logical "1" state, whereas a carrier frequency burst 135 shifted in phase by 180° from the carrier frequency burst 133 and of the same duration represents a logical "0" state. This type of phase-coded carrier frequency signal is also applied to a corresponding transmit transducer, recovered by a corresponding receive transducer after a preselected delay in the channel and amplified by a corresponding amplifier in the manner previously discussed. The output signal from the corresponding amplifier is applied to a demodulator 137, which may be any suitable phase detector for converting phase-coded input data into digital information in order to develop at its output logical "1"s and "0" s containing the desired channel information.

The digital information from the demodulator 137 is applied via a test point 98A to a new-old data recirculate gate 99A, similar in structure and operation to the gate 99 in FIG. 4, with the outputs of the gate 99A being taken from terminals 117A and 118A, which respectively correspond to the terminals 117 and 118 in FIG. 4. The digital data information from the terminals 117A and 118A is then used to control the operation of a pulse-controlled oscillator 141, with the data information at the terminals 117A and 118A being respectively applied to gates 143 and 145 in the oscillator 141. A paraphase amplifier 147, consisting of an NPN transistor 149 having its collector coupled to the gate 143 and also through a resistor 151 to a source of positive potential (+V), its emitter coupled to the gate 145 and also through a resistor 153 to ground, and its base coupled to a crystal oscillator 155 and also through a resistor 157 to ground, is responsive to a carrier frequency signal from the oscillator 155 for developing and applying paraphase carrier frequency signals to the gates 143 and 145.

In operation, when the output of the recirculate gate 99A respectively develops logical "0" and "1" states at the terminals 117A and 118A, the gate 143 is closed to prevent the inverted carrier frequency signal at the collector of the transistor 149 from passing therethrough, while the gate 145 is opened to allow the non-inverted carrier frequency signal at the emitter of the transistor 149 to pass therethrough and through a resistor 159 to the corresponding transmit transducer associated therewith. In a similar manner, when the recirculate gate 99A respectively develops logical "1" and "0" states at the terminals 117A and 118A, the gate 145 is closed, while the gate 143 is opened to allow the inverted carrier frequency signal at the collector of the transistor 149 to pass therethrough and through a resistor 161 to the corresponding transmit transducer associated therewith. The output of the pulse-controlled oscillator 141 is therefore composed of combinations of phase-shifted carrier bursts 133 and 135 which respectively represent logical "1" and "0" states.

In the previous discussion of the circuitry and operation of the embodiment of FIG. 1, it has been shown that a plurality of isolated acoustic channels on the upper surface of an acoustic medium 20 can be utilized in conjunction with external circuitry to respectively store information. It should further be understood that the lower surface (not shown) of the medium 20 can also be utilized in conjunction with additional external circuitry to further increase the information storage capacity of the embodiment of FIG. 1, as well as that of each of the embodiments to be subsequently presented.

FIG. 6 illustrates a schematic circuit and block diagram of an acoustic delay line memory system in accordance with a second embodiment of this invention. Channel isolators 201 and 203 limit the width of an acoustic channel 205 in a medium (not shown) such as the medium 20 in FIG. 1. Although only the channel 205 is shown, it should be understood that the teachings in regard to this second embodiment apply to a plurality of acoustic channels on either or both of the upper and lower surfaces of the medium utilized.

In the embodiment of FIG. 6, a plurality of transmit transducers 211, 212, . . . NT are disposed on one end of the channel 205, while a plurality of receive transducers 221, 222, . . . NR are disposed on the opposite end of the channel 205. Each of the transmit transducers 211, 212, . . . NT is designed to be responsive to a different frequency by controlling the physical spacing between its adjacent fingers, as discussed in relation to the fingers 68 in FIG. 2. For example, the transducer 212 is responsive to a higher frequency signal than the transducer 211, since adjacent fingers 231 of the transducer 212 are physically spaced closer to each other than adjacent fingers 233 of the transducer 211. The frequency responsive characteristic of any given one of the transmit transducer 211, 212 . . . NT is sufficiently displaced from that of any of the others so that the given transmit transducer does not interact with the operation of any of the remaining transmit transducers. Furthermore, the receive transducers 221, 222, . . . NR are designed to have the same physical and frequency response characteristics as the transmit transducers 211, 212, . . . NT, respectively, in order to enable N number of signal channels to operate within the acoustic channel 205 without interacting with each other.

In operation, the transmit transducers 211, 212, . . . NT are respectively excited by different coded carrier frequency signals respectively applied from data sources 241, 242, . . . NDS, similar to the data source of either FIG. 4 or FIG. 5. Each of the data sources 241, 242, . . . NDS operates at a different preselected frequency from the others. In response to the signals from the data sources 241, 242, . . . NDS, stress waves from the transmit transducers 211, 212, . . . NT respectively propagate along the surface of the channel 205 until they are respectively received by the receive transducers 221, 222, . . . NR, which in turn respectively reproduce the signal information contained in the stress waves. The reproduced signal information at the outputs of the receive transducers 221, 222, . . . NR, are respectively amplified by amplifiers 251, 252, . . . NA before being returned to the data sources 241, 242, . . . NDS, respectively.

If the number N of signal channels in each acoustic channel were five and 10 acoustic channels were on each of the upper and lower surfaces of an acoustic medium, the medium, when combined with sufficient external circuitry, would contain 100 different signal channels for storing information. Furthermore, if each signal channel is designed to store 5,000 bits, a total of 500,000 bits of information could be stored in the one acoustic medium.

FIG. 7 is a schematic circuit and block diagram of a acoustic delay line memory system in accordance with a third embodiment of this invention. In this embodiment a wideband transmit transducer 301 and a wideband receive transducer 303, positioned at opposite ends of a channel 305 and isolated from other channels 307 in an acoustic medium 309 by channel isolators 311 and 313, are each composed of coded segments 315, 317, and 319. Each of the segments 315, 317, and 319 is physically arranged to have a sufficiently different frequency response characteristic in order to enable each segment to operate without interacting with the remaining ones of the segments 315, 317, and 319, in the manner described in relation to FIGS. 2 and 6. Data sources 321, 323, and 325 respectively develop different coded carrier frequency signals at frequencies to which the segments 315, 317 and 319 are respectively responsive. The output signals from the data sources 321, 323 and 325 are applied to a frequency summer 327, which may be a resistor network, for developing one composite signal containing all of the signal information from the data sources 321, 323, and 325.

In operation, this composite signal is applied to the transmit transducer 301. The segments 315, 317, and 319 of the transducer 301 selectively respond to the component signals of the composite signal to respectively cause acoustic stress waves to propagate to the receive transducer 303. The segments 315, 317, and 319 of the receive transducer 303 respectively reproduce and recombine the component signals into a composite signal, which is then amplified by a wideband amplifier 324 before being applied to the data sources 321, 323, and 325. The data sources 321, 323, and 325 are similar to the data source of either FIG. 4 or FIG. 5. However, each of the demodulators (not shown, but similar to either demodulator 97 in FIG. 4 or demodulator 137 in FIG. 5) of the data sources 321, 323, and 325 must be selectively responsive to only a preselected one of the component signals of the composite signal in order to enable each of the data sources 321, 323, and 325 to reprocess only that coded carrier frequency signal which is associated therewith.

FIG. 8 is a schematic block diagram of an acoustic delay line memory system in accordance with a fourth embodiment of this invention. While only two isolated acoustic channels 401 and 403 are illustrated in FIG. 8, it should be understood that additional channels can be utilized on the upper and lower surfaces of an acoustic medium 405 in conformance with the teachings of the invention. This fourth embodiment uses a pure digital approach for signal excitation, as contrasted with the carrier frequency method described in the first three embodiments. In this pure digital approach, special interdigital transmit transducers 407 and 409 are respectively disposed on one end of the acoustic channels 401 and 403, and special interdigital receive transducers 411 and 413 are respectively disposed at the opposite end of the acoustic channels 401 and 403. The transducers 407 and 411 are each physically arranged to develop an A code, while the transducers 409 and 413 are each physically arranged to develop a B code. The A and B codes from a complementary code pair of the type described by Marcel Golay in his article, "Complementary Series," on pp. 82-87 of IRE Transactions on Information Theory, dated April, 1961. In Golay's article a complementary code is defined as a pair of binary sequences of length N, whose elements are +1 and -1, and whose autocorrelation functions are such that the sum of the corresponding terms of the autocorrelation functions is identically zero, except for the center terms, whose sum is equal to 2N. The autocorrelation function of a binary sequence of length N (Code A) will be discussed later in conjunction with FIGS. 9 and 10.

Complementary or push-pull pulses 415 and 417, or their inversions, are respectively supplied from a data source 419 to the L and U input terminals of each of the transmit transducers 407 and 409, which have their respective M input terminals connected to ground. The simultaneous pulses 415 and 417 represent the development of a digital "1" from the data source 419, while the simultaneous inversions of the pulses 415 and 417 represent a digital "0". The digital information supplied by the data source 419 to the transducers 407 and 409 is composed of sequences of digital "1"s and "0"s. The structure and operation of the channel 401 in FIG. 8 will now be developed more fully by referring to FIG. 9.

FIG. 9 discloses that the transmit transducer 407 must have 2N+1 fingers to generate a code of a certain number (N) of bits when the transducer is impulsed by the pulses 415 and 417. For purposes of explanation, the A-code will be chosen to be equal to the eight-bit sequence (where N = 8) of -1, -1, -1, +1, -1, -1, +1, and -1. Therefore, in order to comply with Golay's requirements for complementary codes, the B-code must be equal to the eight-bit sequence of -1, -1, -1, +1, +1, +1, -1 and +1. In order to mechanize the transducer 407 to be A-coded, the transducer 407 must have a sequence of 17 fingers 421 through 437, since 2N+1 = 17 when N = 8. To generate the aforementioned A-code, the fingers 422, 424, 426, 430, 432, and 436 are all coupled to the U input terminal; the fingers 421, 423, 425, 427, 429, 431, 433, 435 and 437 are all coupled to the M input terminal; and the fingers 428 and 434 are coupled to the L input terminal.

The fingers 421 through 437 selectively form N (eight in this illustration) sequentially aligned, basic, three-finger, elemental, interdigital transducers which comprise slots or groups of fingers consisting of: 421 through 423, 423 through 425, 425 through 427, 427 through 429, 429 through 431, 431 through 433, 433 through 435, and 435 through 437. Each three finger transducer group has a bandwidth approximately equal to 1/T where T is the time taken by the acoustic wave in travelling across an elemental transducer. The nature of the ensuing processing permits a number of basic three-finger transducer groups to operate as if they were independent when excited in parallel with a single pulse. Furthermore, by having the transducer 407 composed of a number of basic three-finger transducer groups, the power into the transducer 407 can be increased in proportion to the number of basic three-finger transducer groups without reducing the bandwidth. The coded transducer 407 is thus a wideband device regardless of the number of basic three-finger transducer groups used in its construction. The receive transducer 411 has fingers 441 through 457 which respectively correspond in design and operation to the fingers 421 through 437 of the transmit transducer 407. The output from the transducer 411 is taken from its U and L terminals, with its M terminal remaining unconnected.

In operation, upon the application of the pulses 417 and 415 to the U and L input terminals of the transducer 407, the aforementioned basic three-finger transducer groups of 421--423, 423--425, 425--427, 427--429, 429--431, 431--433, 433--435 and 435--437 develop corresponding stress waves accompanied by respective charge dipoles of polarity -1, -1, -1, +1, -1, -1, +1, and -1. For example, each of the basic three-finger transducer groups of 421--423, 423--425, 425--427 429--431, 431--433 and 435--437 develops a dipole of polarity -1 when the negative voltage pulse 417 is applied to the fingers 422, 424, 426, 430, 432, and 436, since the fingers 421 and 423, 423 and 425, 425 and 427, 429 and 431, 431 and 433, and 435 and 437 are all coupled to ground. On the other hand, each of the basic three-finger transducer groups of 427--429 and 433--435 develops a dipole of polarity +1 when the positive voltage pulse 415 is applied to the fingers 428 and 434, since the fingers 427 and 429, and 433 and 435 are all coupled to ground. It should also be understood that when the data source 419 of FIG. 8 develops a logical "0" state output the inversions of the pulses 415 and 417 are respectively applied to the U and L input terminals of the transducer 407, resulting in the development therefrom of the dipole sequence of polarities +1, +1, +1, -1, +1, +1, -1, and +1, while the inversions of the pulses 415 and 417 are respectively applied to the U and L input terminals of the transducer 409, resulting in the development therefrom of the dipole sequence of polarities +1, +1, +1, -1, -1, -1, +1, and -1. It is important to note that the relationship between Golary's complementary codes is unchanged when the signs of all of the terms of the complementary codes are changed. As a result, for further explanation of this fourth embodiment, only the relative polarities shown by the pulses 415 and 417 will be considered.

The dipole wave of polarities -1, -1, -1, +1, -1, -1, +1 and -1, which was developed by the transducer 407 in response to the respective application of the pulses 417 and 415 to the U and L input terminals of the transducer 407, propagates along the channel 401 until it reaches and is processed by the receive transducer 411. As specified previously, the transducer 411 is also A-coded. As the A-coded dipole wave (-1, -1, -1, +1, -1, -1, +1, and -1) enters the A-coded receive transducer 411, an autocorrelation process of multiplication and addition takes place, as shown in FIG. 10. For example, as the eighth term (-1) of the charge waveform moves into the first group or slot 1 of the transducer 411, a one-term product (-1 × -1) of +1 results, as shown in the column designated "Sum of Products." In like manner, as the eighth term (-1) of the charge waveform moves into slot 2, the seventh term (+1) moves into slot 1, thereby developing the products -1 × -1 and -1 × +1, whose sum is equal to zero, and so on. When the eighth term of the charge waveform moves into slot 8, eight products are developed whose sum ((-1)2 + (+1)2 + (-1)2 + (-1)2 + (+1)2 + ( -1)2 + (-1)2 + (-1)2 = 8) is called the mainlobe of the A-coded autocorrelation function. Subsequently, the charge waveform starts leaving the slots. The movement of the A-coded charge waveform into and exit from the A-coded receive transducer 411 sequentially develops the terms 1, 0, 1, 0, 3, 0, -1, 8, -1, 0, 3, 0, 1, 0, 1 of the autocorrelation function of the A-code, as shown in the column designated "Sum of Products" in FIG. 10, with the 8 representing the mainlobe and the remaining terms representing the sidelobes.

Returning now to FIG. 8, the aforementioned sequential terms of the autocorrelation function of the A-code appear between the U and L terminals of the transducer 411 and are illustrated by the waveform 461 in FIG. 11. It should be recalled at this time that it was specified that the A and B codes were a complementary code pair of the type described by Golay in his aforementioned article. It should also be recalled that the B-code was required to be -1, -1, -1, +1, +1, +1, -1, and +1, since the A code was given as -1, -1, -1, +1, -1, -1, +1, and -1. The sequentially developed terms of the autocorrelation function of the B-code are found to be -1, 0, -1, 0, -3, 0, 1, 8, 1, 0, -3, 0, -1, 0 and -1, in the same manner as those of the A-code were found. These terms of the autocorrelation function of the B-code appear between the U and L terminals of the transducer 413 and are illustrated the waveform 463 in FIG. 11.

The U and L terminals of the transducer 411 are parallel coupled to the U and L terminals of the transducer 413 and are also coupled across a resistor or summer 465, which has one end grounded. The resistor 465 sums the corresponding terms of the autocorrelation functions of the A and B codes to develop the waveform 467, which is show in FIGS. 8 and 11.

A comparison of each of the terms of the waveforms 461 and 463 discloses that the sum of each of the corresponding terms of the autocorrelation functions of the A and B codes is equal to zero, with the exception of the middle term which is equal to 2N, or +16 when N=8. As a result, the A and B codes have been found to satisfy Golay's requirements for a complementary code pair. It should be noted at this time that each term of the waveforms 461, 463 and 467 is a Ricker pulse. The Ricker pulse of the waveform 467 is amplified by an amplifier 469 before being returned to the data source 419.

To better explain the overall operation of FIG. 8 reference should also be made to the waveforms of FIG. 12. The operation of FIG. 8 has been previously discussed in relation to only one bit of digital information (pulses 415 and 417) being stored in the signal channel composed of the acoustic channels 401 and 403. However, it is more desirable to store a plurality of bits in the signal channel of FIG. 8 by having the data source 419 sequentially pulse the transmit transducers 407 and 409 with a pair of push-pull binary sequences of logical "1"s and "0"s containing the information to be stored. The faster the transducers 407 and 409 are pulsed, the more data can be stored in the signal channel of FIG. 8. In actual operation the data source 419 supplies a pair of complementary or push-pull data streams to the transmit transducers 407 and 409. The interval between bits in each data stream is such that the Ricker pulses developed across the summer or resistor 465 are as close together as possible without obliterating the digital information contained in each Ricker pulse. If the rate (data rate) at which the data is supplied to the signal channel of FIG. 8 is increased such that adjacent Ricker pulses overlap each other by one-fourth, as shown in the waveform 471, the digital information contained in each Ricker pulse is not interferred with or obliterated. The waveform 471 shows a sequence of overlapping Ricker pulses whose sidelobes overlap and whose mainlobes have the logical states of 1, 1, 0, 1 and 0. As shown in the waveform 471, this lack of interference between adjacent Ricker pulses is due to the fact that the digital information in each Ricker pulse is contained in the mainlobe of that Ricker pulse, which occupies approximately the middle third of the Ricker pulse. The sequence of Ricker pulses in the waveform 471 is amplified (if required) by the amplifier 469 before being applied to the D input of a D-flip-flop 473. The flip-flop 473 has auxiliary clock pulses, shown by the waveform 475, applied to its C (clock) input from an auxiliary channel.

This auxiliary channel which is just used for the synchronization of all the data channels, looks identical to the dual acoustic channels and circuitry shown in FIG. 8, but in operation only a continuous stream of new data logical "1"s (derived from the system clock pulses) is transmitted therethrough in order to provide the auxiliary clock pulses from the output of its amplifier, which is similar to the amplifier 469 in FIG. 8. A change of temperature, for example, may cause a variation in the time delay involved in the propagation of data through the signal channels. The auxiliary clock pulses experience the same time delay in the auxiliary channel as any of the data in the other channels since they are all located on the surface of the same medium. Therefore, the auxiliary clock pulses remain in the same phase relationship with the recirculating data, regardless of temperature changes or other environmental, etc. changes which may affect the propagation time delay. As a result, these auxiliary clock pulses are used to synchronously sample the data at the input of the data source in each channel in order to compensate for any variations in the associated channel time delay due to temperature variations, etc.

More specifically, with reference to the data source 419 of FIG. 8, the auxiliary clock pulses from the auxiliary channel are applied to the C input of the flip-flop 473 to synchronously sample the center portion or mainlobe of each of the Ricker pulses (1, 1, 0, 1, and 0) of the waveform 471 applied to the flip-flop 473. By sampling only the center portion of each Ricker pulse, the digital information contained therein is retrieved. The D-flip-flop 473 transfers whatever data information was at its D input at each auxiliary clock pulse time of the waveform 475 to its Q-output. The time between the auxiliary clock pulse (waveform 475) and the system clock pulse (waveform 481) is the effective residence time of the data in flip-flop 473. The effective residence time in the flip-flop 473 is determined by the delay time of the acoustic medium, which is the length of time that it takes that bit to travel through the medium in the channel (401 and/or 403). Variations in the delay time of the acoustic medium, which may be caused by temperature changes or by other changes in environmental conditions, are accommodated by the time interval between the system clock pulse (waveform 481) and the auxiliary clock pulse (waveform 475). The sum of the effective residence time in the flip-flop 473, the delay time in the medium, and the relatively fixed delays inherent in the amplifier 469 and data source 419 is a constant delay period for the channel, which therefore compensates for changes in environmental conditions. The maximum effective residence time that can be achieved in the flip-flop 473 is equal to one bit time or interval. For example, if the operating frequency in FIG. 8 were 25 MHz, the bit time or interval would be 40 nsec. Therefore, if a temperature increase causes the channel delay time to increase, the data into the flip-flop 473 would spend less effective residence time in the flip-flop 473 to compensate for the increase in the channel delay time.

The flip-flop 473 also performs a bit-stretching function by stretching each bit of information present at its D input at each auxiliary clock pulse time over a full bit period. The sequence of stretched bits, as shown by the waveform 477, is applied from the Q output of the flip-flop 473 to the D input of another D-flip-flop 479. System clock pulses, shown in the waveform 481, are applied from a conventional system clock generator (not shown) to the C input of this D-flip-flop 479. In response to the system clock pulses applied to its C input, the D-flip-flop 479 delays the sequence of stretched bits applied to its D-input sufficiently to make the sequence synchronous with the system clock. The synchronized output of the flip-flop 479 is taken from the Q output of the flip-flop 479, and is shown by the waveform 483. A comparison of the waveforms 477, 481 and 483 discloses that the waveform 483 is the delayed inversion of the waveform 477 and is synchronized with the system clock pulses of the waveform 481.

The waveform 483 from the Q output of the flip-flop 479 is applied via a test point 98B, from which the waveform 483 may be monitored, to a new-old data recirculate gate 99B, which is similar in structure and operation to the gate 99 of FIG. 4. Output terminals 117B and 118B of the gate 99B provide complementary outputs from the gate 99B in the same manner that the respective terminals 117 and 118 of the gate 99 provided complementary outputs from the gate 99. Assume at this time that only the old data from the flip-flop 479 is to be recirculated. As a result, the complementary waveforms 487 and 488 will be respectively developed at the output terminals 117B and 118B of the gate 99B. The waveform 487 at the terminal 117B is applied to the lower input of an OR gate 489 and to the lower input of a NOR gate 491. The waveform 488 at the terminal 118B is applied to the lower input of a NOR gate 495 and to the lower input of an OR gate 497. The gates 489, 491, 495 and 497 function together as a three level driver 499 for developing the coded pair of push-pull binary sequences of pulses supplied to the transducers 407 and 409. Gate clock pulses shown in the waveform 501 are supplied to the upper input of each of the gates 495, 489, 491, and 497. These gate clock pulses of the waveform 501 are derived from and operate at the same frequency as the system clock pulses of the waveform 481. However, these gate clock pulses are phased with respect to the system clock pulses by any suitable phase shifter (not shown) so as to compensate for delays in the channels and circuitry shown in FIG. 8. Furthermore, the duty cycle of these gate clock pulses is shortened in a conventional manner by any suitable circuit (not shown) such that the negative going portion of each gate clock pulse has a duration less than one-half of the bit time or interpulse period of each gate clock pulse. In this embodiment, the duration of the negative going portion of each gate clock pulse determines the duration of each driving pulse, such as the pulse 415 or 417.

If the distance between adjacent fingers in, for example, the transducer 407 is represented by Δ (delta), the distance that an elemental transducer, for example, 421-423 encompasses is equal to 2Δ. Each time that the transducers 407 and 409 are pulsed by either a logical "0" or "1" from the data source 419, a Ricker pulse is subsequently developed at the output of the summer 465. The length of this Ricker pulse is 4Δ/V, where V is the acoustic velocity in the acoustic medium used. Since the sidelobes of adjacent Ricker pulses were specified to overlap by approximately one-third in order to increase the bit density, the bit time in the signal channel of FIG. 8 would be equal to two-thirds of the length of a Ricker pulse (2/3 of 4Δ/V or 8Δ/3V).

When quartz is used as the medium, the acoustic wave travels through the quartz medium at an approximate velocity (V) of one-eighth inch per microsecond. If a 64 μ sec long delay line is desired, the distance between corresponding fingers of, for example, the transmit transducer 407 and the receive transducer 411 must be approximately 8 inches. The rate at which data is applied from the data source 419 to the signal channel of FIG. 8 is determined by the system clock pulse (waveform 481) rate. If the system clock pulse rate (F) were 25 MHz, the interpulse period of the system clock or bit time would be equal to 1/F or 40 n sec. With a bit time of 40 n sec, about 1,600 bits of digital information could be stored in the signal channel of FIG. 8. Furthermore, the distance between adjacent fingers of, for example, the transducer 407 would be equal to 3V/8F or about three sixteen-hundredths inch.

The duration of each drive pulse (see waveforms 513 and 525) from the data source 419 may typically be three-eighths of the interpulse period of the system clock pulse (waveform 481). When the acoustic medium is quartz and the system clock pulse rate is 25 MHz, each of the drive pulses in the waveforms 513 and 525 would have a duration of about 15 n sec. In addition, the center to center spacing of adjacent transducer fingers would correspond to a propagation time of approximately 15 n sec.

The waveform 503 is developed at the output of the NOR gate 495 upon the simultaneous application of the waveforms 488 and 501 to the inputs of the NOR gate 495, and the waveform 505 is developed at the output of the OR gate 489 upon the simultaneous application of the waveforms 487 and 501 to the inputs of the OR gate 489. The waveforms 503 and 505 are respectively applied through resistors 507 and 509 to a summing point 511, where these waveforms are summed to provide the three level driving voltages shown in the waveform 513. These driving voltages in the waveform 513 are used as the voltage drive to the L-terminals of the transducers 407 and 409.

The waveform 515 is developed at the output of the NOR gate 491 upon the simultaneous application of the waveforms 487 and 501 to the inputs of the NOR gate 491, and the waveform 517 is developed at the doutput of the OR gate 497 upon the simultaneous application of the waveforms 488 and 501 to the inputs of the OR gate 497. The waveforms 515 and 517 are respectively applied through resistors 519 and 521 to a summing point 523, where these waveforms are summed to provide the three level driving voltages shown in the waveform 525. These driving voltages in the waveform 525 are used as the voltage drive to the U-terminals of the transducers 407 and 409.

Upon being driven by the sequences of driving pulses in the respective waveforms 513 and 525, the transducers 407 and 409 cause sequences of stress waves and charges to propagate through the respective channels 401 and 403 to the receive transducers 411 and 413, whose outputs are summed in the summer or resistor 465 to develop the sequences of Ricker pulses shown in the waveform 471. Thus, every two adjacent acoustic channels on either surface of the medium used, in conjunction with an associated amplifier and an associated data source, form one signal storage channel for recirculating digital information.

Waveform 471 illustrates a one-fourth overlap of the Ricker pulses, whereas the previous discussion assumed a one-third overlap. The specific overlap is not a critical parameter. It should be understood that the illustrations and formulas given are examples and do not constitute limitations on the generality of the various embodiments.

FIG. 13 illustrates a schematic block diagram of an acoustic delay line memory system in accordance with a fifth embodiment of this invention. In this embodiment the acoustic channels 401 and 403 in the acoustic medium 405, the A and B complementary coded transmit transducers 407 and 409, the A and B complementary coded receive transducers 411 and 413, the summer 467, the amplifier 469 and the data source 419 are all identical in structure, arrangement and operation with the circuitry of FIG. 8, and function to develop and process the complementary codes A and B to form a first signal storage channel for recirculating digital information.

Also contained within the acoustic channels 401 and 403 are the respective C and D complementary coded transmit transducers 531 and 533 and the respective C and D complementary coded receive transducers 535 and 537, each of which has U, M and L terminals. The transmit transducers 531 and 533 are similar in structure and operation to the transducer 407 of FIG. 9, while the receive transducers 535 and 537 are similar in structure and operation to the transducer 411 of FIG. 9. The transducers 531 and 533 are respectively positioned a first predetermined distance from the transducers 407 and 409, while the transducers 535 and 537 are respectively positioned a second predetermined distance from the transducers 411 and 413. While the delay time between the A-coded transducers 407 and 411 is always equal to the delay time between the B-coded transducers 409 and 413, and the delay time between the C-coded transducers 531 and 535 is always equal to the delay time between the D-coded transducers 533 and 537, the delay time between the A-coded transducers 407 and 411 does not necessarily have to be equal to the delay time between the C-coded transducers 531 and 535, within the scope of the invention. In conformance with the teachings described in relation to FIGS. 9, 10 and 11, the transducers 531 and 535 are each physically arranged to develop a C code, while the transducers 533 and 537 are each physically arranged to develop a D code. The C and D codes a complementary code pair of the type described by Marcel Golay in his aforementioned article. However, these complementary C and D codes are chosen, as will be explained later, such that after autocorrelation summing, the C and D codes do not interact with the A and B codes even though they respectively share the same acoustic channel 401 and 403.

The U and L terminals of the transducer 535 are parallel coupled to the U and L terminals of the transducer 537 and are also coupled across a resistor or summer 539, which has one end grounded. The resistor 539 sums the corresponding terms of the autocorrelation functions of the C and D codes, in a manner similar to that described in relation to the embodiment of FIG. 8. As a result, the sum of each of the corresponding terms of the autocorrelation functions of the C and D codes is zero, with the exception of the middle term of the sum, which is equal to 2N. The resultant signal output across the resistor 539 is amplified (if required) by an amplifier 541 before being applied to the input of a data source 543. The C and D complementary coded transmit transducers 531 and 533, the C and D complementary coded receive transducers 535 and 537, the summer 539, the amplifier 541 and the data source 543 form a second signal channel, which shares the acoustic channels 401 and 403 with the first signal channel without interacting with the first signal channel. The components 531, 533, 535, 537, 539, 541 and 543 of the second signal channel respectively correspond to the components 407, 409, 411, 413, 467, 469 and 419 of the first signal channel. It will now be shown how two sets of complementary series are mutually non-interacting even though both sets share the same acoustic channels 401 and 403.

Before describing non-interacting complementary codes, it should be recalled that Golay, in his aforementioned article, effectively defined a complementary code as a pair of binary sequences of length N, whose elements are +1 and -1, and whose autocorrelation functions are such that their sum is identically zero, except for the center term, which is equal to 2N. It should also be recalled that the circuitry of FIG. 13 was stated to be mechanized such that the A and B codes form a set or pair of complementary codes (A,B) and the C and D codes form a set or pair of complementary codes (C,D), where each of the sets of complementary codes (A,B) and (C,D) fulfills Golay's requirements for complementary codes. By constructing the complementary series pairs in sets of two, where the sets are mutually non-interacting, two acoustic channels, such as the channels 401 and 403, can be utilized to provide two signal channels; thereby producing a resultant storage system having a factor of two improvement over having just one signal channel in the two acoustic channels.

Non-interacting sets of complementary series pairs, which henceforth will be referred to as non-interacting complementary (N.I.C.) codes, are defined by the fact that the sum of the corresponding terms of the cross-correlation functions of the N.I.C. codes is identically zero in all terms. To illustrate, let the A, B, C and D codes shown in FIG. 13 have the elements as shown in the following:

A = (a1, a2, . . . , an),

B = (b1, b2, . . . , bn),

C = (c1, c2, . . . , cn), and

D = (d1, d2, . . . , dn).

Let AA represent the autocorrelation function of the code A, and BB represent the autocorrelation function of the code B. Now let the autocorrelation terms (derived from the process of multiplication and addition discussed in relation to FIG. 8) of the respective codes A and B be as shown in the following:

AA = (x1, x2, x3, . . . , x2N-1) and

BB = (y1, y2, y3, . . . , y2N-1).

As indicated, the autocorrelation function of the A-code (AA) and the autocorrelation function of the B-code (BB) each have 2N-1 terms. The A and B codes will meet Golay's requirements for a complementary code pair if:

xi + yi = 0, where i represents any positive (1) integer from 1 through 2N-1, except N, and

xN + yN = 2N. (2)

Let AC represent the cross-correlation function of the A and C codes, which is derived by a process of multiplication and addition in the acoustic channel 401 in a manner similar to that shown in relation to FIG. 10 Let BD represent the cross-correlation function of the B and D codes, which is derived by a process of multiplication and addition in the acoustic channel 403 in a manner similar to that shown in relation to FIG. 10. Each of the cross-correlation functions AC and BD, has 2N-1 terms. Now let the cross-correlation terms of AC and BD be as shown in the following:

AC = (q1, q2, q3, . . . , q2N-1) and

BD = (r1, r2, r3, . . . , r2N-1).

Then (A,B) and (C,D) are N.I.C. codes if:

qi + ri = 0, where i = 1,2 . . . 2N-1. (3)

An example of N.I.C. codes will now be given. Let (A1, B1) and (C1, D1) be two sets of complementary codes, where the codes A1, B1, C1 and D1 have the elements as shown in the following:

A1 = (1,1), (4)

b1 = (1,-1),

c1 = (1,-1) and

D1 = (1,1).

The autocorrelation functions of the codes A1, B1, C1 and D1 have the terms as shown in the following:

A1 A1 = (1,2,1),

b1 b1 = (- 1,2-1),

c1 c1 = (-1,2,-1) and

D1 D1 = (1,2,1).

The respective summing of corresponding terms of the autocorrelation functions A1 A1 and B1 B1 and of the autocorrelation functions C1 C1 and D1 D1 discloses that each of the complementary code sets (A1,B1) and (C1,D1) fulfills Golay's requirements in equations (1) and (2) for complementary codes, as shown in the following:

A1 A1 + B1 B1 = (0,4,0), and

C1 C1 + D1 D1 = (0,4,0).

The cross-correlation functions of the complementary code sets (A1,B1) and (C1,D1) are shown in the following:

A1 C1 = (-1,0,1), and

B1 D1 = (1,0,-1).

The sums of the corresponding terms of the cross-correlation functions of the complementary code sets (A1,B1) and (C1,D1) are shown in the following:

A1 C1 + B1 D1 = (0,0,0).

As a result, the code sets (A1,B1) and (C1,D1) are N.I.C. codes since they fulfill the requirement of equation (3) that the sum of the corresponding terms of their cross-correlation functions is identically zero in all terms.

If a complementary pair (A,B) is given, equations (1), (2) and (3) can be solved sequentially in i to find an N.I.C. pair (C,D). For example, Golay gives the length 10 complementary code (A3,B3), where:

A3 = (-1,1,-1,1,-1,-1,-1,-1,-1,1,1), and

B3 = (-1,-1,-1,-1,1,-1,-1,1,1,-1).

By following the above procedure, the complementary pair (A3,B3) is found to have the following N.I.C. pair:

C3 = (1,- 1,-1,1,1,-1,1,1,1,1), and

D3 = (1,1,-1,-1,-1,-1,1,-1,1,-1).

A complementary code pair of length N which is not derivable from a shorter complementary code pair is defined by Golay as a kernel. The search for an original kernel is a random process and may be best accomplished with the aid of a computer. Once a kernel is found, a longer complementary code pair can be formed by applying certain algorithms in Golay's aforementioned article to the kernel. In a like manner, once a pair of N.I.C. kernels is found, a longer pair of N.I.C. codes can be formed. For example, the previously given N.I.C. code sets of (A 1,B1) and (C 1,D1) may be expanded in the aforementioned manner to produce the length eight N.I.C. code sets (A,B) and (C,D), where:

A = (-1,-1,-1,+1,-1,-1,+1,-1),

b = (-1,-1,-1,+1,+1,+1,-1,+ 1),

c = (-1,+1,-1,-1,-1,+1,+1,+1), and D = (-1,+1,-1,-1,+1,-1,-1,-1).

The transducers 407 and 411, shown mechanized in FIG. 9, develop this A-code. The transducers 409 and 413, 531 and 535, and 533 and 537 can be also mechanized in a similar fashion to respectively develop the B,C and D codes. Reference should also be made to FIG. 14 in conjunction with the following explanation regarding the A,B,C and D codes and the autocorrelation and cross-correlation functions associated therewith.

The autocorrelation functions of the A,B,C and D codes have the terms as shown in FIG. 14 and in the following:

AA = (1,0,1,0,3,0,-1,8,-1,0,3,0,1,0,1),

bb = (-1,0,-1,0,-3,0,1,8,1,0,-3,0,-1,0,-1),

cc = (-1,0,-1,0,-3,0,1,8,1,0,-3,0,-1,0,-1), and

DD = (1,0,1,0,3,0,-1,8,-1,0,3,0,1,0,1).

The respective summing of corresponding terms of the autocorrelation functions AA and BB and of the autocorrelation functions CC and DD discloses that each of the complementary code sets (A,B) and (C,D) fulfills Golay's requirements in equations (1) and (2) for complementary codes, as shown in FIG. 14 and in the following:

AA + BB = (0,0,0,0,0,0,0,16,0,0,0,0,0,0,0), and

CC + DD = (0,0,0,0,0,0,0,16,0,0,0,0,0,0,0).

The cross-correlation functions of the complementary code sets (A,B) and (C,D) are shown in FIG. 14 and in the following:

AC = (-1,-2,-3,0,1,2,-1,0,1,2,-1,0,3,-2,1), and

BD = (1,2,3,0,-1,-2,1,0,-1,-2,1,0,-3,2,-1).

The sums of the corresponding terms of the cross-correlation functions of the complementary code sets (A,B) and (C,D) are shown in FIG. 14 and in the following:

AC + BD = (0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)

As a result, the code sets (A,B) and (C,D) are N.I.C. codes, since they fulfill the requirement of equation (3) that the sum of the corresponding terms of their cross-correlation functions is identically zero in all terms.

It has therefore been shown in the embodiment of FIG. 13 that by using four transmit and four receive transducers per pair of acoustic channels and designing the transducers to develop N.I.C. codes, each pair of acoustic channels can accommodate a pair of signal channels, thereby producing a factor of two improvement of the embodiment of FIG. 13 over the embodiment of FIG. 8.

In each of the embodiments herein presented, the nominal access time is effectively equal to the time delay through the medium. It should, however, be understood that the actual access time could be reduced, if desired, in any embodiment by interposing additional transducers of the type associated with that embodiment at appropriate locations along the acoustic paths, and providing suitable detection circuitry for these transducers. For example, FIG. 15 illustrates a means for reducing the data access time of the embodiment of FIG. 8. However, the principles presented can be applied to any of the embodiments previously presented.

In FIG. 15, the structure, arrangement and operation of the acoustic channels 401 and 403, the transmit transducers 407 and 409, the receive transducers 411 and 413, the summer or resistor 465, the amplifier 469 and the data source 419 have already been discussed in relation to the embodiment of FIG. 8 and hence will not be further discussed. Receive transducers 601 and 603, respectively identical in structure and operation to the receive transducers 411 and 413, are respectively positioned in the channels 401 and 403 at a first preselected distance from the respective transducers 407 and 409. A summer or resistor 605 is parallel coupled across the output terminals of each of the transducers 601 and 603 to sum the autocorrelation functions of the A and B coded sequences generated by the transducers 407 and 409 in order to develop a first output. The resistor 603 is also coupled between a terminal 607 and ground in order to apply the first output thereto.

Further up the signal channel between the set of receive transducers 601 and 603 and the set of receive transducers 411 and 413, another set of receive transducers 609 and 611 are respectively positioned. The transducers 609 and 611 are also respectively identical in structure and operation to the receive transducers 411 and 413. A summer or resistor 613 is parallel coupled across the output terminals of each of the transducers 609 and 611 to sum the autocorrelation functions of the A and B coded sequences respectively propagating in the channels 401 and 403 in order to develop a second output. The resistor 613 is also coupled between a terminal 615 and ground in order to apply the second output thereto.

The ungrounded side of the resistor 465 is also coupled to a terminal 617 in order to apply the signal channel output thereto. The terminals 607, 615 and 617 may be utilized in conjunction with any suitable circuitry for selectively monitoring the data. For example, the terminals 607, 615 and 617 could be the input terminals of an electronic switch (not shown) which could select any desired output to be applied to circuits similar to the serially coupled amplifier 469 and D-flip-flops 473 and 479 in FIG. 8 for monitoring at a test point similar to the test point 98B in FIG. 8.

If the time delay between the set of transmit transducers 407 and 409 and the set of receive transducers 411 and 413 were 60 M.Sec., the sets of receive transducers 601 and 603, and 609 and 611 could be respectively placed within the signal channel of FIG. 15 in order to provide twenty and forty microsecond access times. This placement of transducers would therefore reduce the access time to one-third, since the desired data in the signals channel could be accessed at any one of three positions. Shorter or longer access times could be realized in the same manner by increasing or decreasing the number of sets of receive transducers between opposite ends of the acoustic paths in any given signal channel.

As in any recirculating delay line mechanization, suitable addressing circuitry must be provided. An appropriate approach would involve the use of a channel selection device, a channel position counter actuated by the system clock pulse, and coincidence circuitry to indicate coincidence between the desired channel position and the actual channel position.

The invention thus provides, as disclosed in the various embodiments, an information storage system utilizing an acoustic medium having its surface subdivided into a plurality of individually isolated acoustic signal channels upon which a plurality of transmit and receive transducers are selectively placed, thereby enabling the system to have a very high data storage capacity.

While the salient features have been illustrated and described with respect to five embodiments, it should be understood that non-binary extensions of the approaches disclosed in these embodiments are within the scope of this invention. For example, although binary coding approaches to delay line utilization are discussed in this specification, the principles set forth therein are not restricted to binary coding. In particular, a multilevel coding scheme or a combination of coding schemes may be utilized, with the scope of this invention, in any of the embodiments herein presented. It should, therefore, be readily apparent to those skilled in the art that modifications can be made within the spirit and scope of the invention as set forth in the appended claims.