Minimum Pin Memory Device

The present invention is directed to a technique and circuitry for minimizing the number of electrical input terminals to a monolithic integrated electrical structure, which structure includes memory capabilities. More specifically, the present invention is directed to the merging of the functions associated with the operation of such a monolithic integrated electrical device so as to permit the reduction of electrical conductors and/or pins which connect the device in a system so as to, in turn, permit a substantially larger number of such devices to be placed in a given area.

Substantial research effort is continually expended by manufacturers of monolithic integrated circuits to substantially reduce the physical size and volume of monolithic integrated circuits in order to provide both increased circuit density and decreased cost, whereby the cost per bit of storage in high access speed static memories constructed utilizing such devices is comparable to the cost per bit of storage provided by relatively low access speed dynamic memories such as, for example, magnetic discs, drums, tapes, and the like. Even though integrated circuits (IC's) and large-scale integrated circuits (LSI's) have increased the density of electronic circuits per se substantially, the overall size and volume of the package thereof is still much too large to achieve the same relative packing density by high-speed static memories as achieved by relative low-speed dynamic magnetic memories. In the present state of the art, connections to these circuits are generally made by bonding leads to pads which are fixed on the chip supporting the integrated circuit and by bonding the leads to package tabs. The package houses the integrated circuit and the tabs project through the package. The tabs are electrically connected to conductors, which conductors are bonded to a circuit board and are routed on the circuit board to interconnect the package IC in its circuit environment. Each conductor occupies an area upon the board. The area occupied by the conductors detracts from the area occupied, or capable of being occupied, by an IC device, and therefore substantially reduces the density of the electronics on the circuit board. A reduction in the number of conductors on the board will, in turn, increase the area onto which additional IC devices may be positioned. An increase in IC devices in a particular area will of course increase the density of the IC's. Increased density is particularly important when dealing with memory devices in that the greater the density, the greater the number of memory bits that can be feasibly designed into a system having fixed physical constraints on size.

The present invention deals with the problem of increasing memory density by substantially decreasing the number of conductors and associated pins that are needed to interface to an integrated circuit type memory device.

In the preferred embodiment of the present invention, coding techniques are utilized for simultaneously merging two or more functions, which functions are directed to a monolithic integrated structure containing a memory means, and wherein at least one of the merged functions is digital in character.

A clock coding technique is preferably utilized for simultaneously providing a clocking function and a memory select function. Formating of signals applied to a bi-directional function line in conjunction with the level of the signal on the function line at the time the memory is selected controls the mode of operation of the memory means.

More specifically, the preferred embodiment is implemented utilizing a monolithic integrated structure which is comprised of, a memory element and a means for detecting a change in an applied synchronizing signal for enabling or for disabling the memory element. In addition, there is provided a means for selecting the mode of memory operation in response to a function signal at the time of enabling the memory element. Merging of the functions and formating of the functions provides a minimum electrical contact memory device.

From the foregoing, it can be seen that it is a primary object of the present invention to provide a minimum electrical contact device.

It is a further object of the present invention to merge the functions carried over electrical contacts in order to reduce the number of electrical contacts.

It is another object of the present invention to provide a device which is amiable to high-density packaging.

It is a further object of the present invention to accomplish all of the above objects at a minimum cost with an acceptable compromise in device speed.

These and other objects of the present invention will become more apparent when taken in conjunction with the following description and the attached drawings, wherein like characters indicate like parts, and which drawings form a part of the present disclosure.

FIG. 1 is a perspective view illustrating a typical monolithic memory device embodying the present inventive concept;

FIG. 2 is a block diagram illustrating the preferred embodiment of the present invention;

FIGS. 3A and 3B illustrate signals embodying one coding technique, which technique may be utilized with the preferred embodiment of FIG. 2;

FIG. 4 illustrates a decoding device which may be used to decode the signals of FIGS. 3A and 3B;

FIGS. 5A to 5F illustrate waveforms useful in understanding the operation of the preferred embodiment of FIG. 2; and

FIG. 6 illustrates, in block diagram form, an alternate embodiment for a portion of the preferred embodiment illustrated in FIG. 2.

Referring now to FIG. 1, there is illustrated therein a minimum electrical contact monolithic structure 10 containing monolithic integrated electronic components having a total of four electrical input pins (contacts) labeled CK, F, V and Gnd. A memory device 12, along with associated selection and enabling circuitry, is integrated into the structure 10. The selection and enabling circuitry includes an integral power supply 14 responsive to the voltage applied between pins V and Gnd for providing the voltage levels needed to operate the integrated electronics, a clock generator means 16, decoder means 18, gate means 20, mode selector means 22, address shift register 24, a shift register 26 and a gate 28. In the preferred operating environment of the present invention, the monolithic structure 10 is plugged into electrical contact with a circuit board by means of the input pins.

Referring now to FIG. 2, the preferred embodiment of the invention is illustrated in block diagram form. The pins labeled V and Gnd are electrically connected to the input of an internal power supply 14. The supply 14 provides the voltages, for example V.sub.O thru V.sub.N, that are required to operate the integrated electronics. Power supplies which may be integrated into the monolithic structure 10 are well-known in the integrated circuit art and further description of the detailed construction thereof is not deemed necessary in order to insure a full and complete understanding of all aspects of the present invention. Such details of construction and operation may be found in Japanese Patent Application No. 146230/75, entitled "Voltage Multiplier," by G. C. Lockwood, which application is assigned to NCR Corporation the assignee of the present application. The referenced Japanese patent application was laid open for public inspection on July 22, 1976, and bears the Kokai (laid open) No. 83443/76. The pin CK is operatively connected to clock generator 16, which generator converts an input synchronizing signal for example, a clock signal, received on the CK pin into, internal clocking signals, for example, clocking signals .phi..sub.O and .phi..sub.N. A decoder circuit 18, to be hereinafter described in detail, is operatively connected to the clock generator 16 for decoding a coded signal that is applied to the CK pin. The decoder circuit 18 provides a memory select signal MS, which signal changes state as a function of a code component contained in the coded signal that is applied to the CK pin.

Memory device 12 may be a charge coupled type memory device of the type described in the article "CCD Memories" by W. S. Kosonocky and D. J. Saver in Electronic Design 6, Mar. 15, 1976, pages V-520 to V-534. The memory device receives the memory select signal on its (MS) input and responds by being either activated or inhibited, depending on the state of the MS signal. As is common with memory devices, device 12 also has a plurality of address inputs labeled O to N, a DATA IN terminal, a READ/WRITE (R/W) mode selection terminal for instructing the memory to either read-out data or to write-in data, a DATA OUT terminal, and one or more clocking terminals CR.sub.O, CR.sub.n. A memory select signal applied to terminal (MS) enables the memory device such that data may be read out of or written into memory. The absence of a memory select signal will inhibit the memory thereby preventing data from being either read out of or written into the memory device.

A gate means 20, which may be a NAND gate, receives as gate control signals the memory or chip select signal MS from decoder 18, plus a mode selector signal, which signal is present at the Q output of a mode selector 22, and the signal from the Q.sub.n output of a shift register 26. The gate means 20 also receives as the gated input the signal present at the DATA OUT terminal of the memory device. The output of gate means 20 is connected to the function pin F and will pass signals from the DATA OUT terminal of memory device 12 to the F pin when the gating signals to the NAND gate 20 are all at a high level, and will inhibit the passage of signals when one or more of the gating signals are of a low level. For purposes of this description, a low level is denoted as a "0" and a high level as a "1."

The mode selector means 22, which may be a well-known D-type flip-flop, is connected at its D input to the function pin F and at its output Q to the R/W mode terminal by means of an OR gate 28. The clock input CK of mode selector 22 is connected to receive the internal clock signal .phi..sub.O. A clock enable input CK enb. is also provided, for either enabling or disabling flip-flop 22 regardless of the level of the clock signal .phi..sub.O appearing at the flip-flop terminal CK. The circle at the CK enb. input denotes that the flip-flop is enabled with a low level signal and disabled with a high level signal. Such being the case, the flip-flop is disabled when the memory is selected, that is when the signal MS is high.

As used in the preferred embodiment, a D-type flip-flop is one which transfers the signal level present on its D-input to its Q output upon the occurrence of a negative transition of the clock signal present on its CK input, and which maintains the signal level at the Q output until the next occurrence of a negative transition of the clock signal. A negative transition occurs when the clock signal changes from a high state to a low state.

An address shift register 24 having 0 to N storage locations or "cells" has its respective cells connected to the corresponding terminals O to N on the memory device 12. A D-input terminal to the shift register 24 is connected to the function pin F for entering address data into the shift register. A clocking terminal CK is connected to receive the internal clocking signal .phi..sub.O for clocking the address data through the address shift register. The shift register 24 is also provided with a clock enable input CK enb. 1, which input is connected to the MS line to the NAND gate 20 and to the SET and D inputs of the shift register 26. The clock signal .phi..sub.O is also applied to the CK input of shift register 26. The full (count) output of shift register 26, labeled Q.sub.n, is directed to a clock enabling input CK enb. 2 on the address shift register 24. In addition, the Q.sub.n output from shift register 26 is connected to the other input of the OR gate 28. The register 26 is selected to step thru (shift) the signal on its D input in response to the clock signal .phi..sub.O. The number of stages in the shift register 26 is selected to correspond to the number of bits contained in the address signal. A SET signal will initialize register 26 such that all the register stages are set to a count of 1. As the signal MS on the D input to register 26 is clocked thru the register by the clock signal, zeros are propagated through the length of the register. When the register is full, it outputs a "0" on its Q.sub.n output and disables the address shift register 24 thru the input CK enb. 2. In the preferred embodiment, four bits are used for address information, therefore register 26 is a four bit register.

Referring now to FIGS. 3A and 3B, the coding technique used in the preferred embodiment of the invention is to decrease the pulse width of the clock signal applied to the CK pin in order to effect selection of the memory device. In FIG. 3A, the CK signal is depicted as having a pulse width of M and a period of T.sub.1 between the leading edge of each succeeding pulse. If the memory device is, as in the preferred embodiment, a charge coupled type device or other type of volatile memory device, then the clock signal is used to refresh or to regenerate the memory, but the memory device is not selected when the clock signal of FIG. 3A is present. When memory device 12 is to be selected, the width of the pulse M of the clock signal is reduced to N as is shown in FIG. 3B. In addition, the rate or frequency of the narrowed pulse N clock signal can be increased in order to speed up the operation of the memory. The one limitation that exists with this particular coding technique as regarding the increase in rate is that the period T.sub.2 must not be less than the pulse width M, otherwise erroneous memory selection will be effected unless additional detection circuitry is utilized.

The clock signal CK thus carries two pieces of informtion in coded form, one is the synchronizing (clock) rate and the other is the selection or non-selection of the memory device. Although in this preferred embodiment there is shown a pulse width technique for coding the CK signal, it will be obvious to those persons skilled in the art that other types of coding techniques such as, for example, amplitude and/or frequency coding could also be used with equal success.

A suitable decoding circuit for decoding the coded clock signal CK to provide the memory select signal MS is shown in detail in FIG. 4. In FIG. 4 there is provided a delay network 30 comprised of six serially connected inverters 32. The total time delay provided by the serially connected inverters is equal to

where M and N are the pulse widths of the clock signal shown in FIGS. 3A and 3B, respectively. The input to the delay network 30 is the externally applied clock signal CK. The delayed output from the delay network 30 is directed to the CK input of a D type flip-flop 34. The D input of flip-flop 34 is connected to the CK input to decoder 18. The Q output from flip-flop 34 is the memory select signal MS, whereas, the Q output therefrom is the memory select signal MS in the inverted form.

Referencing the waveforms shown in FIGS. 5A to 5F, in conjunction with the preferred embodiment shown in FIG. 2, the operation of the device and the techniques used will now be described in detail. FIGS. 5A and 5B described the recirculate mode for the case when a volatile memory such as a CCD memory is used for the memory device 12. The recirculate mode also corresponds to the non-select mode in that the memory is not being selected. The CK signal is shown having a pulse width M and a period equal to T.sub.1. The period T.sub.1 is chosen to provide adequate rejuvenation of the memory device with the use of minimum power. The signal level on the function pin F shown in FIG. 5B can be at any level due to the fact that the select signal MS does not effect selection of the memory device.

In FIGS. 5C and 5D, the READ MODE of operation is depicted. The signal present on the function pin F carries three major segments of instruction; the first segment is mode select, the second is the address and the third is the data. Mode selection is determined merely by the level of the signal present on the pin F and, in turn, at the D input of the mode selector flip-flop 22 when the memory select signal MS is generated. The flip-flop 22 is latched onto the level present at its Q output by the memory select signal MS because it disables the flip-flop from further toggling in that the MS signal is applied to the flip-flop's CK enb. input. A reference to FIG. 5C shows the READ MODE being selected in response to the signal on the function pin F being at a low level just prior to the generation of the memory select signal MS. The clock signal CK is decreased in pulse width to N, and the period of that particular portion of the clock signal is decreased to T.sub.2. The first period during which the pulse width is changed is labeled MS denoting memory select. In the READ MODE the Q output is high and the output of the OR gate is also high. A high level signal on the R/W mode terminal of memory device selects the READ MODE of memory operation. A low level signal on the R/W mode terminal selects the WRITE MODE of memory operation.

The decoder 18 senses the change in the pulse width of the clock signal and outputs the memory select signal to the (MS) terminal of memory device 12. Those signals present on the function pin F are then directed to the D input of the shift register 24 and to the DATA IN terminal of memory device 12. The signals on pin F are formated such that the next following N bits corresponding to the desired first address bits A.sub.O to A.sub.N will be loaded into shift register 24 at the clock rate .phi..sub.O. Shift register 24 is enabled by receipt of the memory select signal MS at its clock enable terminal CK enb. 1. When the address bits have been completely loaded into the address register 24, the shift register 26 disables shift register 24 by the application of a low level signal to the CK enb. 2 input of shift register 24 and the data corresponding to the loaded address in memory device 12 is serially fed to the DATA OUT terminal of the memory device 12 in synchronism with the subsequent .phi..sub.O pulses. The NAND gate 20 receives a high level output from the Q output of the mode selector 22, along with the high level signal from the Q.sub.n output of the shift register (when the address signal has been completely loaded into the address shift register 24), and, in addition, the high level memory select signal MS. Because the memory device 12 has been selected, as previously described, the NAND gate 20 is enabled and passes the serial data DO(O) to DO(n) (FIG. 5D) from the DATA OUT terminal of memory device 12 to the function terminal F.

In FIGS. 5E to 5F the WRITE MODE of operation is depicted. The mode selector 22 is toggled to output a low level signal at its Q output in response to the high level signal present at its D input from function pin F. The occurrence of a memory select signal MS disables the flip-flop latching the mode latch in the WRITE MODE. When the WRITE MODE is selected, NAND gate 20 is disabled due to the low level of the signal present at the Q output of the mode selector. FIG. 5E shows the clock CK changing to the memory select MS pulse width, which causes the generation of the memory select signal MS by the decoder 18. The address register 24 is again enabled by the presence of the memory select signal at its CK enb. 1 input, and the register 26 will be initialized by the memory select signal on its SET input. The write address is then directed to the function pin F and entered into the address register at its D input. Although the Q output of the mode selector is low, indicating a WRITE MODE of operation, the other input to the OR gate 28 is at a high level which causes the output of the OR gate to remain high. A high input on the R/W mode terminal maintains the memory device in the READ MODE of operation. The memory is held in the READ MODE as the write address is loaded into the address shift register in order to prevent the writing of address bits into associated memory locations as data bits. This would occur if the memory were not held in the READ MODE because the address bits also appear at the memory's DATA IN terminal at the same time as they appear on the D input of register 24. Once the address bits are loaded, shift register 26 will output a low level signal on its Q.sub.n output, which signal will appear at the other input to OR gate 28 causing the output of OR gate 28 to go low. With the output of OR gate 28 low, the memory device will be switched into the WRITE MODE. In addition, when the Q.sub.n output goes low the address shift register 24 is disabled. The data bits D.sub.1 (O) to D.sub.1 (n) are then placed on the function pin and written into memory at the address specified by the outputs O to N of register 24. After the data is written into memory, a change in clock width back to M will cause the memory select signal to de-select the memory device.

FIG. 6 illustrates an alternate embodiment of the invention wherein the memory select signal MS is not generated from the decoding of a coded synchronizing signal but, instead, is generated external of the monolithic structure and applied to an additional pin labeled MSp. The pin MSp is connected to the (MS) terminal of memory device 12, an input of NAND gate 20, the CK enb. input of flip-flop 22, the CK enb. 1 input of the address shift register 24, and the SET and D input of shift register 26.

The clock signal applied to the pin CK remains directed to the clock generator 16, with the clock outputs .phi..sub.O through .phi..sub.n directed to the inputs CR.sub.O through CR.sub.n of the memory device 12.

Although a specific preferred embodiment of the invention has been shown in detail, it will be understood by those skilled in the art that the foregoing and other changes in form and details may be made therein without departing from the spirit and scope of the invention. For example, while the preferred embodiment has been described as utilizing a charge coupled type of memory device, it will be immediately obvious to those skilled in the art, following a full understanding and appreciation of the novel aspects of the present invention, that such coding technique is applicable to other well-known monolithic integrated circuit structures such as, for example, bubble type memories, shift registers, random access memories, read only memories and the like.