A signal generation circuit includes: a clock circuit configured to receive a flag signal and generate a clock signal; a control circuit configured to generate a control circuit; and a generation circuit connected to both the clock circuit and control circuit and configured to receive the clock signal, the control signal, and the flag signal and generate a target signal, wherein when the flag signal changes from a first level to a second level, the target signal changes from a third level to a fourth level, and after the target signal is maintained at the fourth level for a target duration, the target signal changes from the fourth level to the third level; and the generation circuit is further configured to determine the target duration according to the clock signal and control signal.

A failure bit count (FBC) circuit for memory array is provided. The memory array includes pages each having plural sectors and a redundancy column. The FBC circuit includes FBC units, in which each FBC unit is respectively coupled to each sector for providing a failure bit count current; a redundancy FBC unit coupled to the redundancy column and provides a redundancy current; a switch having a first end and a second end capable of being switched to couple to one of outputs of the FBC units to receive the failure bit count current from one of the FBC units; a comparator having a first input end that receives a reference current, and a second input end that receives a measurement current obtained by adding the failure measurement current and the redundancy current, and an output end outputting a judge signal to indicate a number of failure bits for each sector.

Methods and apparatuses with counter-based reading are described. A memory cells of a codeword are accessed and respective voltages are generated. A reference voltage is generated and a logic state of each memory cell is determined based on the reference voltage and the respective generated cell voltage. The reference voltage is modified until a count of memory cells determined to be in a predefined logic state with respect to the last modified reference voltage value meets a criterium. In some embodiments the criterium may be an exact match between the memory cells count and an expected number of memory cells in the predefined logic state. In other embodiments, an error correction (ECC) algorithm may be applied while the difference between the count of cells in the predefined logic state and the expected number of cells in that state does not exceed a detection or correction power of the ECC.

Methods and apparatuses with counter-based reading are described. A memory cells of a codeword are accessed and respective voltages are generated. A reference voltage is generated and a logic state of each memory cell is determined based on the reference voltage and the respective generated cell voltage. The reference voltage is modified until a count of memory cells determined to be in a predefined logic state with respect to the last modified reference voltage value meets a criterium. In some embodiments the criterium may be an exact match between the memory cells count and an expected number of memory cells in the predefined logic state. In other embodiments, an error correction (ECC) algorithm may be applied while the difference between the count of cells in the predefined logic state and the expected number of cells in that state does not exceed a detection or correction power of the ECC.

Technology is disclosed for detecting latent defects in non-volatile storage systems. Prior to writing data, a stress voltage is applied to SGS transistors in a 3D memory structure. After applying the stress voltage, the Vt of the SGS transistors are tested to determine whether they meet a criterion. The criterion may be whether a Vt distribution of the SGS transistors falls within an allowed range. If the criterion is not met, then a sub-block mode may be enabled. In the sub-block mode, data is not written to memory cells in a sub-block that contains SGS transistors whose Vt does not meet the criterion. Hence, the possibility of data loss due to defective SGS transistors is avoided. However, in the sub-block mode, data is written to memory cells in a sub-block that does not contain SGS transistors whose Vt does not meet the criterion. Hence, data capacity is preserved.

A memory device includes a content addressable memory (CAM) block storing a plurality of stored search keys. The memory device further includes control logic that determines a first number of memory cells in at least one string of the CAM block storing one of the plurality of stored search keys, the first number of memory cells storing a first logical value, and stores a calculated parity value representing the first number of memory cells in a page cache associated with the CAM block. The control logic further reads stored parity data from one or more memory cells in the at least one string, the one or more memory cells connected to one or more additional wordlines in the CAM block, and compares the calculated parity value to the stored parity data to determine whether an error is present in the one of the plurality of stored search keys in the CAM block.

A memory device includes a content addressable memory (CAM) block storing a plurality of stored search keys. The memory device further includes control logic that determines a first number of memory cells in at least one string of the CAM block storing one of the plurality of stored search keys, the first number of memory cells storing a first logical value, and stores a calculated parity value representing the first number of memory cells in a page cache associated with the CAM block. The control logic further reads stored parity data from one or more memory cells in the at least one string, the one or more memory cells connected to one or more additional wordlines in the CAM block, and compares the calculated parity value to the stored parity data to determine whether an error is present in the one of the plurality of stored search keys in the CAM block.

Methods, systems, and devices for read threshold adjustment techniques for memory are described. A memory device may read a codeword from a memory array of the memory device using a read threshold having a first value. The memory device may increment one or more counters of the memory device based on reading the codeword. The counter may indicate a quantity of bits of the codeword that correspond to a first logic value. The memory device may detect an error, such as an uncorrectable error, in the codeword based on reading the codeword. The memory device may adjust the read threshold from the first value to the second value based on the quantity of bits indicated by the counter. The memory device may read the codeword using the read threshold having the second value.

Methods, systems, and devices for read threshold adjustment techniques for memory are described. A memory device may read a codeword from a memory array of the memory device using a read threshold having a first value. The memory device may increment one or more counters of the memory device based on reading the codeword. The counter may indicate a quantity of bits of the codeword that correspond to a first logic value. The memory device may detect an error, such as an uncorrectable error, in the codeword based on reading the codeword. The memory device may adjust the read threshold from the first value to the second value based on the quantity of bits indicated by the counter. The memory device may read the codeword using the read threshold having the second value.

Methods for improving timing in memory devices are disclosed. A method may include sampling a data signal according to a clock signal to obtain a data sample; sampling the data signal according to an advanced clock signal to obtain an advanced data sample; and sampling the data signal according to a delayed clock signal to obtain a delayed data sample. The method may also include comparing the data sample with the advanced data sample and the delayed data sample and performing an action based on the comparison. The action may include selecting a data sample, selecting a clock signal and/or adjusting a clock signal. Associated devices and systems are also disclosed.