Charge Transfer Binary Counter

This invention relates to charge transfer devices and, more particularly, to change coupled devices (CCDs) for performing a binary counting function.

The recent emergence of charge coupled technology has brought with it the advent of shift register and memory devices now well known in the art. In order to fabricate complete systems, other circuit functions are frequently utilized. By way of illustration, such supplementary functions often include logical AND and OR, binary counting, anad signal filtering. Advantageously, if all of the circuits performing the various functions of the system are charge coupled devices, the manufacture of the system is simplified. A shift register and an AND gate, for example, could be fabricated on a single chip by well-known integrated circuit technology. Moreover, interface problems, such as impedance matching and loading due to stray capacitance, would be alleviated.

The invention is a charge transfer binary counter capable of counting the number (from 0 to 2.sup.n - 1) of charge packets in a particular time interval. The charge packets might, for example, be generated from signal pulses in a PCM system and the time interval might constitute a "word" or frame in PCM terminology. In an illustrative embodiment the binary counter comprises a charge storage medium in which is formed at least one charge storage "Q-cell" adapted to quantize each signal pulse into a packet of charge Q. In tandem with the Q-cells are n one-bit shift registers each of which, except perhaps the last, comprises first and second charge storage cells connected in series. Adjacent cells of the same shift register are separated from one another by a threshold potential barrier of magnitude V.sub.T which permits only partial transfer of charge thereacross. On the other hand, adjacent cells of consecutive shift registers are separated from one another by a smaller potential barrier of magnitude V.sub.B which permits virtually complete charge transfer thereacross. In a preferred embodiment, V.sub.T is approximately equal to 1.5 V.sub.B if the Q-cells are of area A and at least the first cells of each shift register each have an area equal to 2A. Adjacent cells of the device are connected to opposite phases of a two-phase clock. A feedback path is established between adjacent cells of the same shift register in order to clear residual charge from the first cell whenever charge is transferred across its threshold barrier. At the end of the particular time interval (e.g., word or frame), however, a quasi-steady state is reached; that is, charge transfer ceases until a read operation is performed or the next word arrives. In the quasi-steady state, charge packets residing in the first cells of the shift registers correspond to the binary equivalent of the number of signal pulses in the word.

The invention, together with its various features and advantages, can be easily understood from the following more detailed description taken in conjunction with the accompanying drawing, in which:

FIG. 1 is a plan view of a binary counter in accordance with an illustrative embodiment of my invention;

FIG. 2 is a schematic plan view of the binary counter of FIG. 1 in which the electrodes have been omitted for convenience and the barrier lattice configuration is shown by solid lines; and

FIG. 3A shows schematically the quantizer and the first shift register of FIG. 2, whereas FIGS. 3B-F show the surface potential of the various cells of FIG. 3A at different times during the clock cycle. It has been assumed for convenience that the voltage drop across the insulating layer (oxide) of the storage medium is negligible so that the surface potential(s) is nearly equal to applied voltage.

Turning now to FIG. 1, there is shown a plan view of a binary counter in accordance with an illustrative embodiment of the invention. The counter 10 comprises a storage medium 10.1 illustratively a p.sup.--type semi-conuctor substrate such as silicon on which is formed an insulative layer (not shown) typically thermally grown silicon dioxide. In the substrate are a plurality of rectangular charge storage cells, the boundaries of which are designated by dashed and dot-dashed lines. These boundaries, termed a barrier lattice, are illustratively established in the substrate by ion implantation or diffusion of stripes of immobile charge (i.e., impurity centers) in the fashion taught in U.S. Pat. No. 3,789,267 issued on Jan. 29, 1974, to R. H. Krambeck and myself.

The barrier lattice is composed of stripes of charge with three different potential heights: (1) chanstop barriers (dashed lines) which are designed to prevent charge transport across them. The object of chanstop barriers, as described in U.S. Pat. No. 3,728,161, issued on Apr. 17, 1973, is to eliminate spurious inversion of the surface of a semiconductor integrated circuit chip due to capacitive coupling between metallization and/or field oxide in the semiconductor substrate. If such coupling were strong enough to invert the surface of the semiconductor, current might leak between adjacent devices or might even short elements of a single device; (2) transfer barriers (dot-dashed lines) which have a height typical of an n-channel device; i.e., the application of most positive clock voltage to the barrier region should permit complete transfer of charge. Each transfer barrier is asymmetrically positioned with respect to the center of the overlaying electrode in order to cause charge to flow in a predictable direction; and (3) threshold barriers (dot-dashed double lines) which have a height such that charge is transferred thereacross only if two packets of charge have been transferred into the preceding cell. In addition, the boundaries of diffusion zones for diodes, IGFETs, and the like are shown by dotted lines.

The three types of potential barriers in the lattice define a quantizer 12 and three (n = 3) one-bit shift registers 14, 16 and 18 capable of counting the number of charge packets in a particular time interval up to a maximum of 2.sup.n - 1 = 7. In binary notation the outputs of the shift registers 14, 16 and 18 correspond to the powers of two as follows: 2.sup.0, 2.sup.1 and 2.sup.2, respectively.

The quantizer 12 comprises at least one charge storage cell (hereinafter a "Q-cell") of area A having a charge carrying capacity Q. Thus, each signal pulse to be counted produces a quantum of charge Q in the Q-cell. For reasons discussed later, the quantizer of FIG. 1 comprises first and second Q-cells 12.1 and 12.2 each having an area A and arranged in series relationship. Transfer barriers are located at 12.4 between the first and second Q-cells 12.1 and 12.2, respectively, and at 12.5 between the second Q-cell 12.2 and the first shift register 14. A transfer barrier may also be located at 12.3 if the input device 11 is a charge storage cell of a preceding CCD device. However, if the input device 11 is a diode, as shown, then the interface at 12.3 would designate one edge of an n.sup.+-diffusion zone. On the other hand, chanstop barriers are located at 12.7 and 12.8. In addition, a pair of electrodes 12.9 and 12.10 overlay the first and second Q-cells 12.1 and 12.2, respectively, and are connected to opposite phases of two-phase clock means 100.

In a similar fashion the first shift register 14 comprises first and second charge storage cells 14.1 and 14.2 arranged in series relationship. The transfer barrier 12.5 mentioned previously is located between the first cell 14.1 of shift register 14 and the second Q-cell 12.2 of the quantizer 12. A threshold barrier is located at 14.3 between the first and second cells 14.1 and 14.2. Overlaying cells 14.1 and 14.2 are a pair of electrodes 14.7 and 14.8, respectively, connected to opposite phases of clock means 100.

As described more fully hereinafter, in order to clear residual charge from the first cell 14.1 in response to charge transferred across threshold barrier 14.3, a feedback path is established between the first and second cells 14.1 and 14.2. Illustratively, this path includes a conventional grounded source IGFET 17 having its gate electrode 17.1 in contact with an n.sup.+-diffusion gate zone 14.6 which is partly overlapped by electrode 14.8. The drain electrode 17.2 actuates a dump gate 15 coupled to the first cell 14.1.

The dump gate 15 comprises a charge storage cell 15.1 adjacent the first cell 14.1 and includes a transfer barrier 14.5 at the interface therebetween. An electrode 15.3 overlays cell 15.1 and is connected to the drain 17.2 of IGFET 17. Charge transferred into the second cell 14.2 of first shift register 14 causes a voltage to be applied to dump gate electrode 15.3 via IGFET 17 and, in turn, causes residual charge to be drained from the first cell 14.1 into dump gate cell 15.1. In order to remove the residual charge from the dump gate, an IGFET diode 25 is utilized as described hereinafter.

In addition, a charge-sensing IGFET has its gate electrode 19.1 connected to the n.sup.+-diffusion zone 15.2 which extends under electrode 15.3. The drain of IGFET 19 is connected to a load schematically designated by resistor 20. The voltage at output terminal 20.1, when read at a prescribed time such as the end of a word, corresponds to the first digit of a three-digit binary number. That is, the presence of a voltage at terminal 20.1 corresponds to binary digit 2.sup.0.

In the interest of brevity a detailed description of the second and third shift registers 16 and 18, and their respective feedback paths, will not be given. It is to be understood that the structure of these components, as well as the circuit connections, are with one exception substantially identical to those previously described with respect to the first shift register 14. The exception is that the third shift register, and in general the n.sup.th shift register, need not include a second cell inasmuch as no charge can be transferred across the last threshold barrier if the maximum number of signal pulses in any given word is 2.sup.n - 1. Accordingly, the n.sup.th shift register also need not include a feedback path of the type described.

As mentioned previously, in the quasi-steady state, such as the end of a word, charge residing in the first cell of each shift register is the binary equivalent of the number of pulses in the word. In order to obtain the desired count, a read control 110 is connected to each dump gate. A clock cycle counter 120 is connected to clock means 100, and, after a predetermined number of clock cycles, activates the read control 110. The latter applies a suitable voltage to the dump gate electrodes effective to transfer charge, if any, from the first cells of the shift registers into the dump gate cells. Thereupon, voltages appear at the drains of IGFETs 19, 21 and 23 (i.e., at output terminals 20.1, 30.1 and 40.1) depending on which of the first cells contained charge. IGFET diodes 25, 27 and 29 have their sources connected to the diffusion zones adjacent the dump gate cells of the first, second and third shift registers, respectively, in order to provide a path for the charge in the dump gate cells to flow to the substrate (ground). In conventional fashion, each IGFET diode has its gate and drain shorted to one another and connected to a bias source 130. Thus, reading the presence of charge in the first cell of each shift register also clears these cells in preparation for the next counting operation (e.g., the arrival of the next word). The read and clear operations are described more fully hereinafter.

Assuming that the duration of a clock cycle is equal to the time spacing of adjacent pulse positions, the read operation, performed under the control of counter 120 and read control 110, occurs after an integral number of clock cycles if an even number of Q-cells are used in quantizer 12, but occurs after an integral number of half-cycles if an odd number of Q-cells are used. The precise number of clock cycles after which each read operation is performed depends on design considerations such as the maximum number of pulses in a word, which in turn determines the number n of shift registers required in the counter. In any event, however, the outputs 20.1, 30.1 and 40.1 correspond in binary notation to powers of two as follows: 2.sup.0, 2.sup.1 and 2.sup.2, respectively. Thus, output voltages read on terminals 20.1 and 30.1 but none on terminal 40.1 corresponds to the binary number 110 which equals decimal 3; i.e., the word detected had three pulses.

Operation of the invention will now be described with reference to the schematic drawing of FIGS. 2 and 3. In the interests of clarity of illustration the electrodes have been omitted and the barrier lattice has been shown by solid lines. Corresponding components have been given identical numerical designations in FIGS. 1 and 2. Although the clock means is not shown, it is to be understood that adjacent cells of each shift register are connected to opposite clock phases as in FIG. 1. Note the last shift register 18' has only a first cell 18.1; no second cell or feedback path is required as discussed above. In addition, understanding of the operation of my invention will be facilitated by assuming a preferred embodiment in which the Q-cells of the quantizer each have an area equal to A, the cells of the shift registers have an area equal to 2A, the threshold barriers V.sub.T are equal to 1.5 V.sub.B approximately, and the voltage swing of the clock voltage is equal to V.sub.B, approximately (FIG. 3A).

Assume that the input signal is a PCM word having a maximum of seven pulse slots, three of which contain signal pulses. During phase one of the clock cycle, the first signal pulse is applied to quantizer 12 which generates a packet of charge Q in the first Q-cell. By two-phase clocking the packet Q is transferred across transfer barrier 12.4 into the second Q-cell and then across transfer barrier 12.5 into the first cell 14.1 of the first shift register 14. This first packet is temporarily inhibited from further transfer by the threshold barrier 14.3; i.e., due to the larger area (2A) of cell 14.1, the charge Q only half fills the cell. More specifically, an amount of charge Q in a cell of area A decreases the surface potential thereof by an amount equal to V.sub.B, whereas the same amount of charge Q in a cell of area 2A will decrease the surface potential by only 0.5 V.sub.B. Consequently, during phase two when the threshold barrier is reduced from 1.5 V.sub.B to 0.5 V.sub.B, which is equal to the surface potential of cell 14.1, no charge is transferred into cell 14.2 (FIG. 3C).

Subsequently, the second signal pulse is applied to quantizer 12 which generates another packet of charge Q in the first Q-cell. By two-phase clocking this second packet of charge Q is similarly transferred through the second Q-cell into the first cell 14.1 where, it will be remembered, the first packet of charge Q is waiting. Charge in the amount of 2Q is now present in the first cell 14.1. This charge decreases the surface potential of the first cell 14.1 by approximately V.sub.B (FIG. 3D). Consequently, during phase two of the clock cycle, when the surface potential under the second cell 14.2 of the first shift register increases by an amount equal to V.sub.B (the clock voltage swing), the effective threshold barrier is reduced from 1.5 V.sub.B to 0.5 V.sub.B. Because the surface potential in the first cell 14.1 is smaller by 0.5 V.sub.B than the reduced threshold barrier, half of the 2Q charge (i.e., Q) will transfer into the second cell 14.2 (FIG. 3E) and subsequently into cell 16.1 (FIG. 3F). Therefore, the second cell 14.2 receives charge only after two packets of charge are transferred into the first cell 14.1.

When this process is completed, the first cell 14.1 of the first shift register 14 is still half full; i.e., it contains an amount of charge Q (FIG. 3F). This excess charge is drained off before the next pulse arrives at the first cell 14.1 by means of a feedback path which includes the dump gate 15 and the IGFET inverter 17 (FIG. 2). Briefly, the IGFET inverter 17 is open-circuited by the presence of charge in the second cell 14.2. The resulting increase in its drain voltage opens the dump gate 15. Excess charge in the first cells of the other shift registers, except the last one 18', is cleared in a similar fashion.

During the next half clock cycle, the packet of charge Q in the second cell 14.2 is tranferred into the first cell 16.1 of the second shift register where it awaits the arrival of another charge packet. This counting process continues as subsequent charge packets are generated by the quantizer in response to applied signal pulses in the PCM word. Thus, a third charge packet is propagated along the device in the same manner as the first and second charge packets until it reaches the first cell 14.1 where the threshold barrier 14.3 causes it to await the arrival of the next charge pacekt, if any. Inasmuch as it has been assumed that the word contained only three signal pulses, a quasi-steady state is reached (i.e., no further charge propagation occurs). At the end of the word, which may be sensed by counting clock cycles (viz., counter 120, FIG. 1), the read control activates all of the dump gates so that voltages appear on output terminals 20.1 and 30.1 but no voltage appears on terminal 40.1. Thus, the binary output is 110 which equals decimal 3, the number of pulses in the PCM word. After the reading operation is completed, the first cell of each shift register is cleared so that the counter is in condition to count the number of pulses in the next word to arrive.

In a similar fashion, any number of charge packets from 0 to 7 in a given word can be counted by the structure of FIG. 2. In general, a larger number of charge packets, up to a maximum of 2.sup.n - 1, can be counted by utilizing a plurality n of shift registers of the type described.

The reading and clearing operations are described more particularly as follows. The read control 110 includes a plurality of IGFETs, e.g., 111, 111.1 and 111.2, which have their drains connected to source 130 of bias voltage V.sub.B, their sources connected to separate dump gate electrodes and their gates connected to the clock cycle counter 120. In practice the IGFETs of the read control may be fabricated on the same chip as the binary counter. Depending on whether a clearing or reading function is to be performed, the clock cycle counter 120 applies one of two voltages V.sub.R = V.sub.a or V.sub.R = V.sub.b > V.sub.a to the gates of the IGFETs of read control 110. Consider IGFET 111.2 for illustrative purposes. For clearing V.sub.R = V.sub.a so that the conductance of IGFET 111.2 has a value g.sub.R = g.sub.a << g.sub.I where g.sub.I is the conductance of the IGFET 17, for example, when no charge is present in cell 14.2. Consequently, the potential on the dump gate 15 is nearly zero and no clearing occurs. When, however, a quantum of charge Q is transferred across threshold barrier 14.3 into cell 14.2, the IGFET 17 is cut-off and the potential on the dump gate 15 is approximately V.sub.B so that clearing of cell 14.1 is accomplished. For reading, on the other hand, in the quasi-steady state no charge is present in cell 14.1 so that IGFET 17 is conducting and has a conductance equal to g.sub.I. To read, the clock cycle counter 120 sets V.sub.R = V.sub.b > V.sub.a so that the conductance of IGFET 111.2 now has a value g.sub.R = g.sub.b >>g.sub.I. Hence, the potential on the dump gate 15 is approximately V.sub.B and charge in cell 14.1 flows through the channel under the dump gate onto the gate of IGFET 19 producing a signal at output terminal 20.1. In a similar fashion the first cells of the other shift registers are cleared and read.

It is to be understood that the above-described arrangements are merely illustrative of the many possible specific embodiments which can be devised to represent application of the principles of the invention. Numerous and varied other arrangements can be devised in accordance with these principles by those skilled in the art without departing from the spirit and scope of the invention. In particular, although it has been pointed out that the last shift register need not include a "second cell" or a feedback path, it should be noted that in some cases, where the pulse count is high enough, 2Q charge will reach the last cell and Q will be transferred over the last threshold barrier. Thus, it may be desirable to dissipate this transferred charge by coupling a diode, for example, to the last cell of the last shift register.

In addition, in order to provide a path from the read control 110 through the electrode of the last dump gate (e.g., 45 of FIG. 2) to ground, an IGFET 112 has its drain connected to that electrode, its source grounded and its gate connected to the bias source (i.e., V.sub.B). Thus, IGFET 112 is always conducting, but functions substantially in the same manner as inverters 17 and 37 in providing continuity to ground.

Moreover, it will be appreciated that the invention, when viewed from a slightly different standpoint, can be considered as comprising a quantizer 12, a plurality n of threshold cells (14.1, 16.1 and 18.1) coupled in tandem with one another each of which has a transfer barrier (12.5, 14.4 and 16.4) at its input and a threshold barrier (14.3, 16.3 and 18.3) at its output. Feedback means (storage cells 14.2 and 16.2, IGFET inverters 17 and 37, and dump gates 15, 35 and 45) are responsive to charge transferred across each threshold barrier for removing residual charge in the threshold cells.