Texas Instruments Corporation

George B Jamison

Stephen Wayne Spriggs

Tam Minh Tran

Bryan David Sheffield

Suresh Balasubramanian

Vikas K Agrawal

David Barry Scott

Hugh Thomas Mair

Xiaowei Deng

Shyh-Horng Yang

David J Toops

Rashmi Sachan

Luan A Dang

M Dwayne Ward

Santhosh Narayanaswamy

Mohan Mishra

Suresh Balasuramanian

Methods, apparatus, systems and articles of manufacture are disclosed that improve performance while reading memory. The method includes initializing an output of a of a sensing circuit to be a first logic high value, obtaining, from the memory, a first current corresponding to a memory bit stored in the memory, replicating the first current, determining whether the replicated first current is greater than a second current, and in response to determining that the replicated first current is greater than the second current, generating a second logic high value at the output of the sensing circuit.

Methods and apparatus to improve performance while reading a one-time-programmable memory

Methods, apparatus, systems and articles of manufacture are disclosed to improve performance while reading a one-time programmable memory. An example apparatus includes: a voltage boost circuit including a first output, a second output, a first input configured to be coupled to a controller, a second input coupled to a first output of a decoder, a third input coupled to a second output of the decoder; and a multiplexer including a first input coupled to the first output of the voltage boost circuit, a second input coupled to the second output of the voltage boost circuit, a third input coupled to an array of memory, and an output coupled to a sensing circuit.

Methods and apparatus to improve performance while reading a one-time programmable memory

A technique to enhance performance and reduce silicon area for a TCAM system which includes a plurality of CAM blocks that are organized into at least one rectangular array having rows each having a plurality of CAM blocks, a group of TCAM cells and associated read/write bit lines connecting the group of CAM cells to search bit lines. Each TCAM cell in the TCAM architecture includes a pair of memory elements that is connected to a pair of associated compare circuits such that the interconnections between the memory elements and the compare circuits are substantially vertical in active MOS layers and substantially horizontal in the metal layers to facilitate sharing of adjacent cells thereby providing reduced silicon area and a short aspect ratio.

High performance and reduced area architecture for a fully parallel search of a TCAM cell

A low power E-fuse repair methodology substantially removes system latency during memory and/or E-fuse farm module power-down in a device that employs E-fuse farm technology. The method maintains power to the repair registers and minimal control logic in the memories, while all other circuitry can be either placed in a low power data retention mode, or completely powered off. There is no need to rescan the repair data from the E-fuse farm after one or more memories are powered back up. This provides dynamic power savings since there is no longer any need to idle the system to reload repair data. Since the E-fuse farm can be powered down after initial system power-up and repair data is loaded into the memories, there is also a significant leakage power savings.

Low-power SRAM E-fuse repair methodology

The memory array contains two layers representing word lines of different rows. Each row contains multiple bit cells sharing the same word line. The two layers are stacked one on top of another to form a high density memory array.

High density memory array

An embodiment of a ultra low-power data retention latch circuit involves a slave latch SL that concurrently latches the same data that is loaded into a main circuit (such as a main latch ML) during normal operation. When the circuit enters a low power (data retention) mode, power (VCC) to the main latch ML is removed and the slave latch SL retains the most recent data (retained data SA, SA-). When power is being restored to the main latch ML, the slave latch's retained data SA, SA- is quickly restored to the main latch ML through what constitute Set and Reset inputs SAR, SAR- of the ML. This arrangement ensures that data restoration is much quicker than conventional arrangements that require the output data path DATA- to be stabilized before power is re-applied to the main latch. Further, there is no need to wait for power to the ML to be stable before restoring data from the SL to the ML, providing an increase in data restoration speed over conventional data retention latches. Using retained data SA, SA- (as mirrored in SAR, SAR-) to control the Set and Reset inputs prevents data contention in the main latch ML. Moreover, compared to known arrangements, the arrangement provides minimal loading on the DATA, DATA- output paths (driving only N, N), thus not compromising speed on the data path (DATAIN . . . DATA/DATA-) through the main latch during normal operation.

Ultra low-power data retention latch

The method for powering down a circuit for a data retention mode includes: changing a supply voltage node from an active power voltage level to an inactive power level; coupling a source of a P channel device to the supply voltage node; providing a retaining power supply voltage level to a back gate of the P channel device; changing a drain voltage of the P channel device to a reference voltage level, wherein the reference voltage level is different from the retaining power supply voltage level; and changing a gate voltage of the P channel device to the reference voltage level.

Leakage current reduction method

According to one embodiment of the invention, a memory circuit operable to assume a standby mode is provided. A memory circuit includes a transistor comprising a gate and a bulk. The bulk is at a retention voltage level. The memory circuit also includes a first node and a second node that are coupled to each other by the transistor. The first node is operable to assume a higher voltage level than the second node in response to an initiation of the standby mode. The memory circuit also includes a third node coupled to the gate of the transistor. The third node is operable to assume a voltage approximately equal to the retention voltage in response to an initiation of the standby mode. The transistor is operable to reduce any direct current flow between the first node and the second node in response to a rise in voltage at the third node to a voltage approximately equal to the retention voltage.

Method and system for power conservation in memory devices

A memory device ( ) includes a memory array ( ) having storage units ( ) arranged in a plurality of rows ( ). A row decoder ( ) receives address information and determines which of the plurality of rows ( ) to enable. According to the determined row ( ), a row selector ( ) drives the storage units ( ) associated with the determined row ( ) to provide their outputs onto respective bitlines ( ) for identification by a bitline sensor ( ). If the received address information indicates an out of range address that does not identify any of the plurality of rows ( ) of the memory array ( ), an out of range decoder ( ) provides such determination to drive an out of range selector ( ) to enable storage units ( ) arranged in a single row ( ) of a bitline driver ( ). Outputs from the storage units ( ) are applied to the respective bitlines ( ) during an out of range address occurrence to prevent the bitlines ( ) from being placed in an undesirable floating state.

Memory device and method for handling out of range addresses

A memory timing architecture which very accurately tracks the read and write timing of a memory over a wide range of array sizes, with separate read and write timing circuits. The read reset circuitry uses a plurality of dummy cells to gauge the time necessary to complete the read operation, while the write reset uses a single dummy cell to gauge the time necessary to complete the write operation. These circuits provide for a more accurately-timed feedback signal, which allows for increased speed while at the same time reducing power consumption and heat buildup.

Self-timed memory reset circuitry

The write port circuits of a static memory cell includes a first conditional conduction path between a first output of the latch and ground active if and only if both a word line input and a write data true bit line input receive active signals. The write port circuit includes a second conditional conduction path between a second output of the latch and ground active if and only if both the word line and a write data complement bit line receive active signals. The first and second conditional conduction paths may be formed by a series connection of the source-drain paths of two transistors. In each conditional conduction path the gate of a first transistor receives a corresponding column signal and the gate of a second transistor is connected to the word line. The first and second transistors for each conduction path may be N-channel MOS transistors formed in a single N-type region. The first and second transistors forming the conditional conduction paths may be in either order. The word line transistors may bee shared between bit line transistors of a single memory cell or of memory cells in plural contiguous adjacent columns. The memory cells may include a plurality of write ports with this inventive write port circuit used for each write port.

Growing community of inventors

Murphy, TX, United States of America

George B Jamison