Abstract: A dynamic redundancy memory includes a redundancy control module and an ECC module. The ECC detects bit errors, stores the addresses of the error bits, counts the bit errors. If the number of errors exceeds a threshold, the ECC identifies the address as a suspect bit and sends a suspect bit signal to the redundancy control module, which determines whether the suspect bit address is already stored in a redundancy element. If already stored, the element is marked bad, the address of the suspect bit is replaced with a new redundant address and the suspect bit address is stored in a good unused element. The ECC determines whether the error occurrences at the address exceeds a bit error rate threshold. If the error rate threshold is exceeded, the ECC identifies the address as suspect bit and sends the suspect bit signal to the redundancy control module.

Abstract: A dynamic redundancy memory includes a redundancy control module and an ECC module. The ECC detects bit errors, stores the addresses of the error bits, counts the bit errors. If the number of errors exceeds a threshold, the ECC identifies the address as a suspect bit and sends a suspect bit signal to the redundancy control module, which determines whether the suspect bit address is already stored in a redundancy element. If already stored, the element is marked bad, the address of the suspect bit is replaced with a new redundant address and the suspect bit address is stored in a good unused element. The ECC determines whether the error occurrences at the address exceeds a bit error rate threshold. If the error rate threshold is exceeded, the ECC identifies the address as suspect bit and sends the suspect bit signal to the redundancy control module.

Abstract: A memory device, a memory system, and corresponding methods are provided. The memory device includes a non-volatile random access memory. The non-volatile memory includes a suspect bit register configured to store addresses of bits that are determined to have had errors. The non-volatile memory further includes a bad bit register configured to store addresses of bits that both (i) appeared in the suspect bit register due to a first error and (ii) are determined to have had a second error. Hence, the memory device overcomes the aforementioned intrinsic write-error-rate by identifying the bad bits so they can be fused out, thus avoiding errors during use of the non-volatile random access memory.

Abstract: A memory device, a memory system, and corresponding methods are provided. The memory device includes a non-volatile random access memory. The non-volatile memory includes a suspect bit register configured to store addresses of bits that are determined to have had errors. The non-volatile memory further includes a bad bit register configured to store addresses of bits that both (i) appeared in the suspect bit register due to a first error and (ii) are determined to have had a second error. Hence, the memory device overcomes the aforementioned intrinsic write-error-rate by identifying the bad bits so they can be fused out, thus avoiding errors during use of the non-volatile random access memory.

Abstract: Magnetic tunnel junction antifuse devices are protected from degradation caused by programming voltage drop across the gates of unselected magnetic tunnel junction antifuses by connecting said magnetic tunnel junctions serially with a first field effect transistor and a second field effect transistor, the first field effect transistor having its gate connected to a positive supply voltage while the gate of the second field effect transistor is switchably connected to a programming voltage, such that when the second field effect transistor of a selected magnetic tunnel junction is switched to direct the programming voltage to program the selected magnetic tunnel junction an unswitched magnetic tunnel junction and the second field effect transistor do not experience a voltage drop across the gates thereof sufficient to degrade.

Abstract: A method includes applying a first voltage to a source of a first p-channel FET connected in series with a second p-channel FET, applying a second voltage, lower than the first voltage, to a source of a third p-channel FET connected in series with a fourth p-channel FET, applying a third voltage, lower than the first and second voltages, to a source of a second n-channel FET connected in series with a first n-channel FET, drains of the second p-channel FET, the fourth p-channel FET, and the first n-channel FET connect at a connection point including an output terminal for outputting an output signal, and outputting one of the first voltage, the second voltage, and the third voltage from the output terminal based on input signals inputted to corresponding gates of the first p-channel FET, the third p-channel FET, the fourth p-channel FET, and the second n-channel FET.

Abstract: Improved spin hall MRAM designs are provided that enable writing of all of the bits along a given word line together using a separate spin hall wire for each MTJ. In one aspect, a magnetic memory cell includes: a spin hall wire exclusive to the magnetic memory cell; an MTJ disposed on the spin hall wire, wherein the MTJ includes a fixed magnetic layer separated from a free magnetic layer by a tunnel barrier; and a pair of selection transistors connected to opposite ends of the spin hall wire. An MRAM device and method for operation thereof are also provided.

Abstract: Improved spin hall MRAM designs are provided that enable writing of all of the bits along a given word line together using a separate spin hall wire for each MTJ. In one aspect, a magnetic memory cell includes: a spin hall wire exclusive to the magnetic memory cell; an MTJ disposed on the spin hall wire, wherein the MTJ includes a fixed magnetic layer separated from a free magnetic layer by a tunnel barrier; and a pair of selection transistors connected to opposite ends of the spin hall wire. An MRAM device and method for operation thereof are also provided.

Abstract: Techniques for improving the security of nonvolatile memory such as magnetic random access memory (MRAM) are provided. In one aspect, a method of operating a nonvolatile memory chip is provided. The method includes: overwriting data stored on the nonvolatile memory chip automatically upon the nonvolatile memory chip being powered on. For example, all bits in the nonvolatile memory chip can be written to either i) a predetermined data state (e.g., a logic 1 or a logic 0) or ii) a random data state. A system is also provided that includes: a nonvolatile memory chip; and a writing circuit configured to overwrite data stored on the nonvolatile memory chip automatically upon the nonvolatile memory chip being powered on.

Abstract: A memory device, a memory system, and corresponding methods are provided. The memory device includes a non-volatile random access memory. The non-volatile memory includes a suspect bit register configured to store addresses of bits that are determined to have had errors. The non-volatile memory further includes a bad bit register configured to store addresses of bits that both (i) appeared in the suspect bit register due to a first error and (ii) are determined to have had a second error. Hence, the memory device overcomes the aforementioned intrinsic write-error-rate by identifying the bad bits so they can be fused out, thus avoiding errors during use of the non-volatile random access memory.

Abstract: A memory device, a memory system, and corresponding methods are provided. The memory device includes a non-volatile random access memory. The non-volatile memory includes a suspect bit register configured to store addresses of bits that are determined to have had errors. The non-volatile memory further includes a bad bit register configured to store addresses of bits that both (i) appeared in the suspect bit register due to a first error and (ii) are determined to have had a second error. Hence, the memory device overcomes the aforementioned intrinsic write-error-rate by identifying the bad bits so they can be fused out, thus avoiding errors during use of the non-volatile random access memory.

Abstract: A method includes applying a first voltage to a source of a first p-channel FET connected in series with a second p-channel FET, applying a second voltage, lower than the first voltage, to a source of a third p-channel FET connected in series with a fourth p-channel FET, applying a third voltage, lower than the first and second voltages, to a source of a second n-channel FET connected in series with a first n-channel FET, drains of the second p-channel FET, the fourth p-channel FET, and the first n-channel FET connect at a connection point including an output terminal for outputting an output signal, and outputting one of the first voltage, the second voltage, and the third voltage from the output terminal based on input signals inputted to corresponding gates of the first p-channel FET, the third p-channel FET, the fourth p-channel FET, and the second n-channel FET.

Abstract: Techniques for improving the security of nonvolatile memory such as magnetic random access memory (MRAM) are provided. In one aspect, a method of operating a nonvolatile memory chip is provided. The method includes: overwriting data stored on the nonvolatile memory chip automatically upon the nonvolatile memory chip being powered on. For example, all bits in the nonvolatile memory chip can be written to either i) a predetermined data state (e.g., a logic 1 or a logic 0) or ii) a random data state. A system is also provided that includes: a nonvolatile memory chip; and a writing circuit configured to overwrite data stored on the nonvolatile memory chip automatically upon the nonvolatile memory chip being powered on.

Abstract: A level shifter and dynamic random-access memory that includes a first output terminal and a second output terminal. A first voltage or a third voltage is outputted from the first output terminal. A second voltage or a fourth voltage is outputted from the second output terminal. The second voltage is lower than the first voltage. The third voltage is lower than the first voltage and higher than the second voltage. The fourth voltage is lower than the first voltage and higher than the third voltage.

Abstract: According to one aspect, a method for adaptive error correction in a memory system includes reading data from a memory array of a non-volatile memory device in the memory system. Error correcting logic checks the data for at least one error condition stored in the memory array. Based on determining that the at least one error condition exists, a write-back indicator is asserted by the error correcting logic to request correction of the at least one error condition, where the write-back indicator is a discrete signal sent to a memory controller, and the at least one non-volatile memory device asserting the write-back indicator extends cycle timing monitored by the memory controller while the write-back indicator is asserted. Based on determining that the at least one error condition does not exist, accesses of the memory array continue without asserting the write-back indicator.

Abstract: Improved STT MRAM source line configurations are provided. In one aspect, a STT MRAM array includes: a plurality of cells including magnetic tunnel junctions in series with field effect transistors; a plurality of word lines perpendicular to a plurality of bit lines; a plurality of source line segments spanning m+1 of the bit lines, wherein m of the bit lines include regular array bit lines, and wherein at least one other of the bit lines includes an extra bit line that is connected to the source line segments such that the source line segments span the regular array bit lines and the extra bit line. An STT MRAM device and a method for operating an STT MRAM device are also provided.

Abstract: A bit line clamp voltage generator circuit for a Spin Torque Transfer Magnetoresistive Random Access Memory is provided. The circuit includes a negative channel Field Effects Transistor having a source, a drain, and a gate, the gate being connected to the drain. The circuit further includes a resistor Rs having a first end connected to a first voltage and a second end connected to the source. The circuit also includes a resistor Rd having a first end connected to a second voltage and a second end connected to the drain to form an output node for outputting a bit line clamp voltage.

Abstract: A method for memory management includes streaming bits to a memory buffer on a memory device using a write data channel that optimizes a speed of writing to the memory devices. The bits are written to non-volatile memory cells in the memory device at a first speed, using a bi-directional bus. Bits are read from the memory device over a read channel to provide reads at a second speed that is slower than the first speed, using the bi-directional bus.

Abstract: Improved STT MRAM midpoint reference cell configurations are provided. In one aspect, a STT MRAM midpoint reference cell includes: a plurality of word lines having at least one write reference word line and at least one read reference word line; a plurality of bit lines perpendicular to the word lines; at least one source line perpendicular to the bit lines; at least one first magnetic tunnel junction in series with i) a first field effect transistor gated by the write reference word line and ii) a second field effect transistor gated by the read reference word line; and at least one second magnetic tunnel junction in series with iii) a third field effect transistor gated by the write reference word line and iv) a fourth field effect transistor gated by the read reference word line. A method of operating a STT MRAM midpoint reference cell is also provided.

Abstract: A memory includes multiple non-volatile memory devices, each having multiple nonvolatile memory cells. A write controller is configured to stream bits to the memory devices using a write data channel that optimizes a speed of writing to the memory devices to provide writes at a first speed. A read controller is configured to read bits from the memory devices, at a second speed slower than the first speed, using a read channel. A bi-directional bus that both the write controller and the self-referenced read controller share to access the plurality of non-volatile memory devices.