Sandisk Technologies Inc.

Chen Chen

YenLung Li

Min Peng

Hua-Ling Cynthia Hsu

Wanfang Tsai

Masaaki Higashitani

Tien-Chien Kuo

Bo Liu

Jongmin Park

Siddarth Naga Murty Bassa

Mitsuyuki Watanabe

For a non-volatile memory that uses hard bit and soft bit data in error correction operations, to reduce the amount of soft bit data that needs to be transferred from a memory to the controller and improve memory system performance, the soft bit data can be compressed before transfer. After the soft bit data is read and stored into the internal data latches associated with the sense amplifiers, it is compressed within these internal data latches. The compressed soft bit data can then be transferred to the transfer data latches of a cache buffer, where the compressed soft bit data can be consolidated and transferred out over an input-output interface. Within the input-output interface, the compressed data can be reshuffled to put into logical user data order if needed.

On-the-fly multiplexing scheme for compressed soft bit data in non-volatile memories

Use of data latches for compression of soft bit data in non-volatile memories

In a non-volatile memory circuit, performance is improved by converting data between a serial format, for transfer on and off of the memory circuit, and a parallel format, for transfer to and from the memory latches used for read and writing data into the memory array of the memory circuit. The memory array is split into M+N divisions, but transferred with a degree of parallelism of M, allowing M words of data to be transferred in parallel at a fixed transfer rate while allowing for up to N bad columns in a transfer. In the write path, a column skipping mechanism is used when converting words of write data into a parallel format. In the read path, a set of (M+N) to 1 multiplexers is used to align the word of read data so that read data can be transferred at a fixed rate and without any added latency.

Centralized fixed rate serializer and deserializer for bad column management in non-volatile memory

Apparatuses and techniques are described for transferring data out of a memory device with low latency. Data can be stored in data transfer latches for NAND strings arranged in columns in divisions of a block. Data can be output from the data transfer latches for different columns in different divisions in each transfer. For example, the data output can include data from an nth column in some divisions and an n+1column in other divisions. This avoids outputting unwanted data at the start of a data transfer. The data from the data transfer latches is output to a data pipeline and then to a set of control latch circuits. The data can be clocked out from a last control latch circuit of the set in a desired division order by use of separate multiplexer control signals for the control latch circuits.

Low latency data transfer

Apparatuses and techniques are described for accessing redundant columns of data in a memory device. To facilitate scaling of a memory device and reduce a clock rate used to access latches of the redundant columns in program and read operations, one or more first-in, first out (FIFO) buffers are provided to output data to, and receive data from, the latches. The FIFO buffers act as an interface between a controller and the latches, and exchange data with the controller at a relatively high clock rate, and exchange data with the latches of the redundant columns at a slower clock rate. During a read operation, the FIFO can prefetch read data from one or more columns and store it until it is needed to replace the data of a defective primary column.

Asynchronous FIFO buffer for redundant columns in memory device

In non-volatile memory circuit, the area devoted to the cache buffer of the read and write circuitry is reduced through the sharing of data latches. In an array structure where memory cells are connected along bit lines, and the bit lines organized into columns, each of the columns has an associated set of data latches, including one or more data latches for each bit line of the column. Data is transferred in and out of the read and write circuit on a data bus, where data is transferred between the data latches and the data bus through a set of transfers latches. The area used by the latch structure is reduced by sharing the transfer latches of the read and write circuitry between the data latches of multiple columns.

Non-volatile memory with reduced data cache buffer

In a non-volatile memory circuit, read and write performance is improved by increasing the transfer rate of data through the cache buffer during read and write operations. In an array structure where memory cells are connected along bit lines, and the bit lines organized into columns, pairs of data words are stored interleaved on the bit lines of a pair of columns. Data is transferred in and out of the read and write circuit on an internal bus structure, where part of the transfer of one word stored on a pair of columns can overlap with part of the transfer of another word, accelerating transfer times for both read and write.

Non-volatile memory with fast data cache transfer scheme

A memory circuit includes an array subdivided into multiple divisions, each connectable to a corresponding set of access circuitry. A serializer/deserializer circuit is connected to a data bus and the access circuitry to convert data between a (word-wise) serial format on the bus and (multi-word) parallel format for the access circuitry. Column redundancy circuitry is connect to the serializer/deserializer circuit to provide defective column information about the array. In converting data from a serial to a parallel format, the serializer/deserializer circuit skips words of the data in the parallel format based on the defective column information indicating that the location corresponds to a defective column. In converting data from a parallel to a serial format the serializer/deserializer circuit skips words of the data in the parallel format based on the defective column information indicating that the location corresponds to a defective column.

Centralized variable rate serializer and deserializer for bad column management

A circuit for counting in an N-bit string a number of bits M, having a first binary value includes N latch circuits in a daisy chain where each latch circuit has a tag bit that controls each to be either in a no-pass or pass state. Initially the tag bits are set according to the bits of the N-bit string where the first binary value corresponds to a no-pass state. A clock signal having a pulse train is run through the daisy chain to 'interrogate' any no-pass latch circuits. It races right through any pass latch circuit. However, for a no-pass latch circuit, a leading pulse while being blocked also resets after a pulse period the tag bit from 'no-pass' to &#x201c;pass&#x201d; state to allow subsequent pulses to pass. After all no-pass latch circuits have been reset, M is given by the number of missing pulses from the pulse train.

Growing community of inventors

Mountain View, CA, United States of America

Chen Chen