International Business Machines Corporation

Lawrence G Heller

John Atkinson Fifield

Lewis M Terman

William R Griffin

Edward J Nowak

Wilbur David Pricer

Minh H Tong

Lloyd Andre Walls

Kenneth E Beilstein, Jr

Harish N Kotecha

Satya N Chakravarti

James M White

Savvas G Chamberlain

Henry J Geipel, Jr

Hsing-San Lee

Donald A Soderman

Yen S Yee

Ronald R Silverman

Dominic P Spampinato

Harry J Jones

Norbert G Vogl, Jr

Charles H Stapper

Andres G Fortino

Thomas V Harroun

Peter N Horowitz

A self biased differential amplifier has a switching point accurately set according to a reference voltage. DC hysteresis is provided, by a circuit internal to the differential amplifier. The amplifier has an input circuit having first and second series connected transistors, wherein the beta ratio of these first and second transistors is changed by enabling an additional transistor of a hysteresis circuit according to an output state of the differential amplifier. When the output state is 'high', the switching point is decreased in order that temporary small drops (due to noise or glitches) in the input signal are ignored. Conversely, when the output state is 'low', the switching point is increased in order that temporary small increases in the input signal are ignored.

Self biased differential amplifier with hysteresis

A novel precharge circuit is provided for dynamic CMOS logic circuits which are immune to leakage currents and reduce overall power consumption. The circuit comprises a precharge transistor for precharging a node to a high voltage level indicting a first logic state during a standby mode. Thereafter, during an active mode, the node may or may not be discharged by connected logic circuitry. If the node is discharged, then an additional transistor is provided to inhibit the precharging of the already charged node during a subsequent standby mode. Similarly, if the node is not discharged, a small keeper transistor is provided to keep the node at a fully precharged level.

Leak tolerant low power dynamic circuits

Disclosed is a Data Valid/Finish circuit element, an integrated circuit using the element, and a method of using the element. The circuit element, which may be incorporated in high speed, digital integrated circuit chips, has an input for receiving input from a data stream, and outputs. One of the outputs is an output true for generating a logical '1' when the input is a logical '1'. The other output is an output complementary means for generating a logical '1' when the input is a logical '0'. The system logically combines the outputs through an output finish/clock for receiving and combining the outputs of the output true and the output complementary. This generates a logical signal when the input from the data stream is either a logical '0' or a logical '1'.

Circuit element on a single ended interconnection for generating a

A self-timed tri-state driver circuit for a dual-rail differential input and single-ended output is disclosed. The circuit generates a tri-state mode in response to an Output Enable (OE) input pulsing low. The OE signal input is driven high to place the driver circuit into a ready state. The circuit is maintained in a tri-state mode until data appears at the inputs. Once a data signal is received after the tri-state circuit is in the ready state, the output immediately outputs this signal. Therefore, the output of the driver is self-timed from the arrival of the data.

Self timed driver

A logic system including a first logic block for providing first differential outputs; a second logic block, identical to the first logic block, for providing second differential outputs; a fault detecting device, coupled to the first logic block, for detecting a fault in the first differential outputs; and a selecting device, coupled to the first and second logic blocks and to the fault detecting device, for selecting an output of one of the first and second logic blocks depending on whether the fault detecting device detects a fault.

Fault tolerant logic system

A system for testing a differential logic network is provided which includes a differential exclusive OR circuit having a plurality of inputs for receiving complementary signals from the differential logic network and first and second output terminals and means, e.g., a conventional exclusive OR circuit, for determining the voltage difference between the first and second output terminals to indicate the presence or absence of a fault or error in the differential logic network under test.

Random logic error detecting system for differential logic networks

A methodology is provided for reducing an arbitrary Boolean logic expression to static CMOS circuits by the use of a general matrix of P channel devices and N channel devices which are interconnected in accordance with the terms of Boolean logic expressions derived from a truth table. More specifically, from a Boolean expression a sum-of-products expression giving the 1 binary data outputs of a truth table having a 0 input is found. This is accomplished by complementing, or barring, the literals which are a binary 1 when the output is 1 and leaving true or unbarred the literals that are a binary 0. Then each input of a given product term is applied to the control gate of a P channel device, which devices are connected in series with one end tied to a source of potential and the other end of the series circuit connected to an output terminal. Each product term is arranged in parallel with other P channel device series circuits to form one half of a complete logic matrix. Similarly, for the other half of the matrix, a sum-of-products expression giving the binary 0 outputs of a truth table having a binary 1 for an input is found. Each input of a given product term is applied to the control gate of an N channel device, which devices are connected in series with one end tied to a potential reference point, such as ground, and the other end of the series circuit is connected to the output terminal. Each product term is arranged in parallel with other N channel device series circuits.

Methodology for making logic circuits

A differential logic system is provided for a complete logic family which has a first switching circuit that produces a given output signal at a first output node and a second switching circuit that produces a second output signal which is the complement of that of the given output signal at a second output node. First and second clocked devices are connected from the first and second output nodes, respectively, to a voltage source, and first and second inverters are connected to the first and second output nodes, respectively. Additionally, a regenerative circuit may be connected between the first and second output nodes and the voltage source.

Clocked differential cascode voltage switch logic systems

A dense memory is provided which includes a one device random access memory cell using charge fill and spill techniques wherein a potential well under a storage node is filled with charge and the excess charge above a predetermined level is spilled to a diffusion or drain region connected to a sense line through a channel region controlled by pulses on a word line. One bit or two or more bits of information may be stored in the potential well at any given instant of time. Depending upon the value of the increment of voltage applied to the storage node or electrode, a given analog charge packet is stored in a potential well formed under the storage electrode. Information is read by applying a voltage to the word line to turn on the channel region and then stepping down the voltage on the storage electrode in fractional, preferably one half, increments. Charge from a charge packet spilled from the potential well under the storage electrode is detected by a sensing circuit connected to the sense line. To rewrite information into the potential well, the original increment of voltage is applied to the storage node and the sense line is pulled to ground so that the diffusion region acts as a source of charge for the potential well.

Charge-stabilized memory

The invention is the structure and process for making a bucket brigade device which comprises the merger of an MOS capacitor with an MOSFET device to form the charge transfer cell. A first thin N-type region is implanted at a first concentration in a portion of the P-type channel region of an FET device adjacent to the drain diffusion. A second region is implanted with N-type dopant at a second concentration less than the first concentration, adjacent to and continuous with the first implanted region. The N-type concentration in the second region is just sufficient to compensate for the P-type background doping in the channel region. This structure increases the charge transfer efficiency for the cell and reduces its sensitivity of the threshold voltage to the source-drain voltage. The gate for the device has a substantial overlap over the drain and a minimal overlap over the source and the gate to drain capacitance per unit area is maximized by maintaining a uniformly thin oxide layer across the gate region.

Process for making a dual implanted drain extension for bucket brigade

A calibrated sensing system is provided in accordance with the teachings of this invention for sensing charge in a storage medium, such as a storage capacitor, coupled to an access or bit/sense line which compensates for most sources of variability in the storage medium and in the access line. In the system, the unknown charge stored in the storage medium is transferred to a first capacitor or potential well via the access line. A high charge state of the storage medium is written into the storage medium and known fractional packets of charge are prepared therefrom, transferred selectively to a second capacitor or potential well and compared with the unknown charge in the first potential to determine the relative level of the unknown charge that was stored in the storage medium. By selectively using two or more fractional packets of charge multilevel sensing is performed.

Calibrated sensing system

An SPS CCD memory system is provided wherein a single tap, preferably a storage node, on an input serial or shift register is connected to the input of a plurality of parallel shift registers through a fan out circuit and the output of the plurality of parallel shift registers is connected to a single tap, preferably a storage node, on an output serial or shift register through a fan in circuit.

Serial-parallel-serial CCD memory system with fan out and fan in circuits

A method of making a transistor array includes forming a plurality of gate electrodes insulated from a semiconductor substrate having an impurity of a given conductivity, introducing a first impurity having a conductivity opposite to that of the given conductivity into a given region of the substrate which is adjacent to an edge of each of the gate electrodes, introducing a second impurity having the given conductivity into given regions adjacent to selected gate electrodes, the second impurity having a significantly higher diffusivity than that of the first impurity in the semiconductor substrate, and driving the second impurity along the surface of the semiconductor substrate to form in the substrate under each of the selected gate electrodes a region having a concentration of impurity of the given conductivity higher than that of the semiconductor substrate. The transistor array may be used, e.g. to make a read only memory (ROM) by connecting appropriate current sensing means to each of the given regions to indicate the presence or absence of the higher diffusivity impurity when a predetermined voltage is applied to the gate electrodes. In one embodiment, the semiconductor substrate is made of P type conductivity, the first impurity is arsenic which produces N type conductivity regions and the second impurity is boron, which produces P type conductivity. Since boron has a higher diffusivity than arsenic, after they are driven by heating the boron impurity produces a high threshold region under the gate electrode when introduced in sufficiently high concentration.

Method of making a transistor array

A voltage-to-charge transducer utilizing a capacitor circuit is provided to generate whole or fractional charge packets at spaced apart locations in a semi-conductor substrate. Each of a pair of conductive lines connected to opposite sides of the capacitor circuit is selectively precharged to a predetermined voltage value by means of one of a pair of charge transfer devices coupled to the input of first and second charge coupled device (CCD) storage systems. Means are also provided for applying voltages of first and second magnitudes to each of the pair of conductive lines. By precharging one of the pair of conductive lines to the predetermined voltage value and by altering the voltage on the other of the pair of conductive lines from the first magnitude to the second magnitude, a first charge packet is formed in a well of the first CCD storage system. To inject a duplicate of the first packet into a well of the second CCD storage system, similar voltages are established across the capacitor circuit relative to the second CCD storage system. To inject into the second CCD storage system a charge packet which is a fraction of the first packet, the similar voltages differ only in that the voltage difference between the first and second magnitudes is altered to a fraction equal to the fraction of the desired fractional charge packet. By precharging the conductive lines coupled to the capacitor circuit, parasitic effects are neutralized or eliminated in the transducer.

Voltage-to-charge transducer

Storage systems are provided with memory cells made of devices having different voltage thresholds for storing information permanently or semipermanently. The devices are arranged adjacent to each other and communicating with a diffusion region in a semiconductor substrate. Information is sensed by detecting the charge transferred from a selected cell to the diffusion region. In an embodiment of the invention, a P-type substrate has an N+ diffusion region formed therein with a plurality of adjacent and parallelly arranged word lines insulated from the substrate and disposed adjacent to the N+ diffusion region. A P+ region, preferably implanted into the substrate, is disposed under selected segments of the word lines to provide devices having a first or high threshold voltage magnitude. The remaining devices which are not associated with a P-30 region have a second or low threshold voltage magnitude. By applying a voltage of the same magnitude to each of the word lines, potential wells are formed which are filled by charge or carriers from the diffusion region. Since the potential wells associated with the high threshold devices hold less charge than do the low threshold devices, a charge or voltage sensing circuit connected to the N+ diffusion region is used to detect the amount of charge flowing between the wells and the diffusion region to thus identify the high and low threshold devices when the voltage on the selected word line is decreased. By eliminating the P+ regions and storing charge, e.g., electrons, at selected locations under the word lines in dual insulating layers, the cells may be electrically programmable.

Growing community of inventors

South Burlington, VT, United States of America

Lawrence G Heller