Abstract: A method for correcting the duty cycle of a clock signal uses two-dual-slope integrators with two comparators; each comparator is connected to both integrators and configured to include a “dead band” when the input pulse duty cycle is at or near 50%. One comparator detects when duty cycle is high and the other comparator detects when the duty cycle is low. When the duty cycle is within the “dead band” range, neither comparator goes valid. This provides an analog filter where the output comparators will not instantaneously switch between opposite duty cycle correction states. When the duty cycle is greater or less than 50%, the integrated voltages on the two integrators move in opposite directions producing twice the signal magnitude on differential inputs of the comparators, as compared with using a single integrator architecture.

Abstract: A method for correcting the duty cycle of a clock signal uses two-dual-slope integrators with two comparators; each comparator is connected to both integrators and configured to include a “dead band” when the input pulse duty cycle is at or near 50%. One comparator detects when duty cycle is high and the other comparator detects when the duty cycle is low. When the duty cycle is within the “dead band” range, neither comparator goes valid. This provides an analog filter where the output comparators will not instantaneously switch between opposite duty cycle correction states. When the duty cycle is greater or less than 50%, the integrated voltages on the two integrators move in opposite directions producing twice the signal magnitude on differential inputs of the comparators, as compared with using a single integrator architecture.

Abstract: A technique for reducing parity bit-widths for check bit and syndrome generation through the use of additional check bits to increase the number of minimum weighted codes in the Hamming Code H-Matrix. The technique of the present invention may be implemented while adding no additional correction/detection capability, in order to reduce the number of data bits that are used for each check bit/syndrome generation and to reduce the width of the parity generating circuitry.

Abstract: A technique for reducing parity bit-widths for check bit and syndrome generation through the use of additional check bits to increase the number of minimum weighted codes in the Hamming Code H-Matrix. The technique of the present invention may be implemented while adding no additional correction/detection capability, in order to reduce the number of data bits that are used for each check bit/syndrome generation and to reduce the width of the parity generating circuitry.

Abstract: An ECC circuit and method for an integrated circuit memory allows a user to enter a test mode and select a specific location to force a known failure on any memory chip, whether it is fully functional or partially functional. Additional circuitry is placed in the data path where existing buffers and drivers are already located, minimizing any additional speed loss or area penalty required to implement the forced data failure. In a first general method, a logic zero is forced onto a selected data line at a given time. In a second general method, a logic one is forced onto a selected data line at a given time.

Abstract: A circuit and technique for reducing parity bit-widths for check bit and syndrome generation is implemented through the use of additional check bits to increase the number of minimum weighted codes in the Hamming Code H-Matrix. The circuit and technique of the present invention may be implemented while adding no additional correction/detection capability, in order to reduce the number of data bits that are used for each check bit/syndrome generation and to reduce the width of the parity generating circuitry.

Abstract: A circuit and technique for reducing parity bit-widths for check bit and syndrome generation is implemented through the use of additional check bits to increase the number of minimum weighted codes in the Hamming Code H-Matrix. The circuit and technique of the present invention may be implemented while adding no additional correction/detection capability, in order to reduce the number of data bits that are used for each check bit/syndrome generation and to reduce the width of the parity generating circuitry.

Abstract: A technique for reducing parity bit-widths for check bit and syndrome generation through the use of additional check bits to increase the number of minimum weighted codes in the Hamming Code H-Matrix. The technique of the present invention may be implemented while adding no additional correction/detection capability, in order to reduce the number of data bits that are used for each check bit/syndrome generation and to reduce the width of the parity generating circuitry.

Abstract: A data inversion register technique for integrated circuit memory testing in which data input signals are selectively inverted in a predetermined pattern to maximize the probability of identifying failures during testing. In accordance with the technique of the present invention, on predetermined input/outputs (I/Os,) data inputs may be inverted to create a desired test pattern (such as data stripes) which are “worst case” for I/O circuitry or column stripes which are “worst case” for memory arrays. A circuit in accordance with the technique of the present invention then matches the pattern for the data out path, inverting the appropriate data outputs to obtain the expected tester data. In this way, the test mode is transparent to any memory tester.

Abstract: Using the tail level referencing for an inverter stage immediately following a differential amplifier provides trip point tracking with the variations in magnitude of the output level swings on the differential amplifier stage output over the operating range of the circuit. When the tail voltage increases and the VOL of the differential stage increases, the trip point of the receiving inverter also increases. When the tail voltage decreases and the VOL of the differential amplifier goes lower, the trip point of the inverter decreases. An additional benefit is provided by the tail connection to the inverter. Faster switching of current from the right side to the left side of the differential amplifier occurs due to the tail node voltage being raised momentarily by a transistor in the inverter stage when the input of the inverter stage transitions high.

Abstract: A clock distribution tree for an integrated circuit memory includes a set of data drivers, a corresponding set of input buffers coupled to the data drivers, a first clock distribution tree coupled to the data drivers, and a second clock distribution tree coupled to the input buffers, wherein the first and second clock distribution tree are substantially matched and mirrored distribution trees. The line width of the first clock distribution tree is substantially the same as the line width of the second clock distribution tree. The line spacing of the first clock distribution tree is substantially the same as the line spacing of the second clock distribution tree. Numerous topologies for the first and second clock distribution trees can be accommodated, as long as they are matched and mirrored. Valid times for the integrated circuit memory are maximized and data and clock skew is minimized.

Abstract: Using the tail level referencing for an inverter stage immediately following a differential amplifier provides trip point tracking with the variations in magnitude of the output level swings on the differential amplifier stage output over the operating range of the circuit. When the tail voltage increases and the VOL of the differential stage increases, the trip point of the receiving inverter also increases. When the tail voltage decreases and the VOL of the differential amplifier goes lower, the trip point of the inverter decreases. An additional benefit is provided by the tail connection to the inverter. Faster switching of current from the right side to the left side of the differential amplifier occurs due to the tail node voltage being raised momentarily by a transistor in the inverter stage when the input of the inverter stage transitions high.

Abstract: An enhanced DRAM contains embedded row registers in the form of latches. The row registers are adjacent to the DRAM array, and when the DRAM comprises a group of subarrays, the row registers are located between DRAM subarrays. When used as on-chip cache, these registers hold frequently accessed data. This data corresponds to data stored in the DRAM at a particular address. When an address is supplied to the DRAM, it is compared to the address of the data stored in the cache. If the addresses are the same, then the cache data is read at SRAM speeds. The DRAM is decoupled from this read. The DRAM also remains idle during this cache read unless the system opts to precharge or refresh the DRAM. Refresh or precharge occur concurrently with the cache read. If the addresses are not the same, then the DRAM is accessed and the embedded register is reloaded with the data at that new DRAM address.

Abstract: A high-speed, static random access memory (SRAM) compatible, high availability memory array and method employing synchronous dynamic random access memory (DRAM) in conjunction with a data cache and separate data read and write registers and tag blocks. The inclusion of separate data read and write registers allows the device to effectively operate at a cycle time limited only by the DRAM subarray cycle time. Further, the inclusion of two tag blocks allows one to be accessed with an externally supplied address and the other to be accessed with a write-back address, thus eliminating the requirement for a single tag to execute two read-modify write cycles in one DRAM cycle time.

Abstract: A data inversion register technique for integrated circuit memory testing in which data input signals are selectively inverted in a predetermined pattern to maximize the probability of identifying failures during testing. In accordance with the technique of the present invention, on predetermined input/outputs (I/Os,) data inputs may be inverted to create a desired test pattern (such as data stripes) which are “worst case” for I/O circuitry or column stripes which are “worst case” for memory arrays. A circuit in accordance with the technique of the present invention then matches the pattern for the data out path, inverting the appropriate data outputs to obtain the expected tester data. In this way, the test mode is transparent to any memory tester.

Abstract: A technique for reducing parity bit-widths for check bit and syndrome generation through the use of additional check bits to increase the number of minimum weighted codes in the Hamming Code H-Matrix. The technique of the present invention may be implemented while adding no additional correction/detection capability, in order to reduce the number of data bits that are used for each check bit/syndrome generation and to reduce the width of the parity generating circuitry.

Abstract: A high-speed, static random access memory (SRAM) compatible, high availability memory array and method employing synchronous dynamic random access memory (DRAM) in conjunction with a data cache and separate data read and write registers and tag blocks. The inclusion of separate data read and write registers allows the device to effectively operate at a cycle time limited only by the DRAM subarray cycle time. Further, the inclusion of two tag blocks allows one to be accessed with an externally supplied address and the other to be accessed with a write-back address, thus eliminating the requirement for a single tag to execute two read-modify write cycles in one DRAM cycle time.

Abstract: An ECC circuit and method for an integrated circuit memory allows a user to enter a test mode and select a specific location to force a known failure on any memory chip, whether it is fully functional or partially functional. Additional circuitry is placed in the data path where existing buffers and drivers are already located, minimizing any additional speed loss or area penalty required to implement the forced data failure. In a first general method, a logic zero is forced onto a selected data line at a given time. In a second general method, a logic one is forced onto a selected data line at a given time.

Abstract: A clock distribution tree for an integrated circuit memory includes a set of data drivers, a corresponding set of input buffers coupled to the data drivers, a first clock distribution tree coupled to the data drivers, and a second clock distribution tree coupled to the input buffers, wherein the first and second clock distribution tree are substantially matched and mirrored distribution trees. The line width of the first clock distribution tree is substantially the same as the line width of the second clock distribution tree. The line spacing of the first clock distribution tree is substantially the same as the line spacing of the second clock distribution tree. Numerous topologies for the first and second clock distribution trees can be accommodated, as long as they are matched and mirrored. Valid times for the integrated circuit memory are maximized and data and clock skew is minimized.

Abstract: A data inversion register technique for integrated circuit memory testing in which data input signals are selectively inverted in a predetermined pattern to maximize the probability of identifying failures during testing. In accordance with the technique of the present invention, on predetermined input/outputs (I/Os,) data inputs may be inverted to create a desired test pattern (such as data stripes) which are “worst case” for I/O circuitry or column stripes which are “worst case” for memory arrays. A circuit in accordance with the technique of the present invention then matches the pattern for the data out path, inverting the appropriate data outputs to obtain the expected tester data. In this way, the test mode is transparent to any memory tester.