System and method for adaptive self-referencing read of memory devices
By applying an adaptive voltage step in the memory array, adjusting the number of step steps and voltage level according to the number of switching events and distribution statistical assumptions of memory cells, the problem of low data reading efficiency in the prior art is solved, and more efficient and Accurate data reading.
Patent Information
- Application Number
- CN202210719405.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2021-06-30
- Filing Date
- 2022-06-23
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2042-06-23
AI Technical Summary
In the prior art, when reading data in a threshold type memory device, there are problems of low efficiency and poor step count, resulting in inefficient data reading process.
By applying an adaptive voltage step in the memory array, the number of step steps and voltage levels are adjusted according to the number of handover events and distribution statistical assumptions of the memory cells to achieve more efficient data reading.
Improves the efficiency and accuracy of data reading, reduces the use of step number and voltage levels, and reduces the data reading time and error rate.
Smart Images

Figure CN115547380B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to systems and methods for adaptive self-referencing read of memory devices. Background Art
[0002] This section is intended to introduce the reader to various aspects of the technology that may be related to various aspects of the technology described and / or claimed below. It is believed that this discussion helps to provide the reader with background information to promote a better understanding of various aspects of the present disclosure. Therefore, it should be understood that these statements should be read in this light, rather than as an admission of the prior art.
[0003] In general, a computing system includes a processing circuit system, such as one or more processors or other suitable components; and a memory device, such as a chip or integrated circuit. One or more memory devices may be used on a memory module, such as a dual in-line memory module (DIMM) to store data accessible to the processing circuit system. For example, based on user input to the computing system, the processing circuit may request the memory module to retrieve data corresponding to the user input from its memory device. In some examples, the retrieved data may include firmware, or instructions that can be executed by the processing circuit system to perform an operation, and / or may include data to be used as input to the operation. In addition, in some cases, data output from the operation may be stored in the memory to enable subsequent retrieval of the data from the memory.
[0004] Some of the memory devices include memory cells that can be accessed by turning on a transistor that couples the memory cell (e.g., a capacitor) with a word line or a bit line. In contrast, a threshold-type memory device includes a memory device that is accessed by providing a voltage across the memory cell, where a data value is stored based on a threshold voltage of the memory cell. For example, the data value may be based on whether a threshold voltage of the memory cell is exceeded, and in response to the voltage provided across the memory cell, the memory cell conducts current. The stored data value may be changed, for example, by applying a voltage sufficient to change the threshold voltage of the memory cell. One example of a threshold-type memory cell may be a cross-point memory cell.
[0005] For threshold type memory, word lines and bit lines are used to transmit selection signals to corresponding memory cells. The selection signal may include a signal characterized by a voltage level used to store data in or retrieve data from the memory cell. The word lines and bit lines may be coupled to the selection signal source by a driver. The memory cells may be organized into one or more layers of memory cells, such as layers defined between overlapping word lines and bit lines. These layers may be referred to as stacks (e.g., memory stacks). Various combinations of word lines, bit lines, and / or decoders may be referenced for specific memory operations using addresses (e.g., memory addresses). The address may indicate which memory cell will be selected using a combination of signals from word lines, bit lines, and / or decoders, and the specific value of the address may be based on the address range of the memory device. As may be understood, some memory cells at some intersections of word lines and bit lines may be farther away from the driver than other memory cells at other intersections. In order to cause memory cells at various distances to be turned on from the corresponding decoders, the memory device may utilize one or more step voltages. It may be advantageous to improve the reading technology for threshold type memory. Summary of the invention
[0006] One aspect of the present application is directed to a memory device, comprising: a memory array comprising a plurality of memory cells; a control circuit operatively coupled to the memory array, the control circuit being configured to: receive a read request for data; apply a first voltage to the memory array based on the read request, wherein a first set of memory cells among the plurality of memory cells are configured to be read as storing a first logic value based on the first voltage; count a total number of the plurality of memory cells that have been switched to a valid read state based on the first voltage; apply a second voltage to the memory array based on the total number, wherein a second set of memory cells included in the plurality of memory cells are configured to be read as storing the first logic value based on the second voltage; and return the data based on at least bits stored in the first set of memory cells and the second set of memory cells.
[0007] Another aspect of the present application is directed to a method, comprising: receiving a read request for data at a control circuit included in a memory device; applying a first voltage to a memory array of the memory device based on the read request, wherein a first set of memory cells among a plurality of memory cells of the memory array are configured to be read as storing a first logic value based on the first voltage; counting a total number of the plurality of memory cells that have been switched to a valid read state based on the first voltage; applying a second voltage to the memory array based on the total number, wherein a second set of memory cells among the plurality of memory cells are configured to be read as storing the first logic value based on the second voltage; and returning the data based on at least bits stored in the first set of memory cells and the second set of memory cells.
[0008] Another aspect of the present application is directed to a memory device, comprising: a memory array comprising a plurality of memory cells; a control circuit operatively coupled to the memory array, the control circuit being configured to: receive a read request for data; apply a first voltage to the memory array based on the read request, wherein a first set of memory cells among the plurality of memory cells are configured to be read as storing a first logic value based on the first voltage; count a total number of the plurality of memory cells that have been switched to a valid read state based on the first voltage; determine whether the total number is within a proximity value range, within a final step value range, or a combination thereof; and apply one or more voltage steps if the total number is within the proximity value range and a single voltage step if the total number is within the final step value range to return the data. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] Various aspects of the present disclosure may be better understood after reading the following detailed description and referring to the accompanying drawings, in which:
[0010] Figure 1 is a simplified block diagram illustrating certain features of a memory device including a memory array of memory cells according to an embodiment of the present disclosure;
[0011] Figure 2 It is an explanation of the embodiments according to the present disclosure Figure 1 A schematic side view of a portion of a memory array;
[0012] Figure 3 According to an embodiment of the present disclosure, the memory array is divided into a plurality of partitions. Figure 1 A block diagram of an embodiment of a memory device;
[0013] Figure 4is a block diagram of user data mode encoding according to an embodiment of the present disclosure;
[0014] Figure 5 The description of the embodiments of the present disclosure is suitable for reading Figure 1 A block diagram of two ramped shapes of data of a memory array;
[0015] Figure 6 According to an embodiment of the present disclosure, Figure 1 a graph showing a bell curve distribution of bit data in a memory array;
[0016] Figure 7 According to an embodiment of the present disclosure, it is suitable for adaptively reading the stored Figure 1 A flowchart of a process for storing data in a memory array;
[0017] Figure 8 According to the embodiment of the present disclosure, Figure 1 A graph of electrical voltage steps for reading data from a memory array;
[0018] Fig. 9 According to the description of the embodiment of the present disclosure Figure 1 A timing diagram of fixed latency data transfers of certain systems that may include fixed latency communications interfacing with a memory device of the present invention;
[0019] Fig.10 is a timing diagram illustrating a typical case read step level and a worst case read step level according to an embodiment of the present disclosure;
[0020] Fig.11 is a timing diagram of an embodiment of a typical case read step level and a worst case read step level that can use a ready busy signal according to an embodiment of the present disclosure;
[0021] Fig.12 The invention is described in accordance with the embodiments of the present disclosure. Figure 1 A block diagram of a partition architecture of a memory device of FIG. 1 ; and
[0022] Fig.13 is a flow chart of a process that may be used to adaptively determine a time interval for a read step according to an embodiment of the present disclosure. DETAILED DESCRIPTION
[0023] One or more specific embodiments will be described below. In order to provide a brief description of these embodiments, not all features of the actual implementation will be described in this specification. It should be understood that in the development of any such actual implementation, as in any engineering or design project, many implementation-specific decisions must be made to achieve the specific goals of the developer, such as consistency with system-related and enterprise-related constraints, which may vary from one implementation to another. In addition, it should be understood that such development efforts may be complex and time-consuming, but will be a routine of design, construction, and manufacturing undertaken by those skilled in the art who benefit from this disclosure.
[0024] The memory typically includes an array of memory cells, wherein each memory cell is coupled between at least two access lines. For example, the memory cells may be coupled to access lines, such as bit lines and word lines. Each access line may be coupled to a large number of memory cells. To select a memory cell, one or more drivers may provide a selection signal (e.g., a voltage and / or current) on the access line to access the storage capacity of the memory cell. By applying a voltage and / or current to the corresponding access line, the memory cell may be accessed so that data is written to the memory cell and / or data is read from the memory cell.
[0025] In some memories, the memory cells in the array may be organized into stacks of memory cells. A stack of memory cells may be a single plane of memory cells disposed between word line layers and bit line layers. The array may be a stack of stacks that includes any number of stacks of memory cells (e.g., 1 stack, 2 stacks, 4 stacks, any number of stacks) as different array layers.
[0026] In some embodiments, a logic state 1 (e.g., a set (SET) state of a memory cell, which may also be referred to as a set cell or bit) may correspond to a set of threshold voltages that are lower than a set of threshold voltages associated with a logic state 0 (e.g., a reset (RESET) state of a memory cell, which may also be referred to as a reset cell or bit). Thus, a set cell may be read using a lower voltage than a reset cell. An adaptive read technique may include applying a voltage (e.g., a read voltage) to a memory array that is configured to activate a group of memory cells containing encoded user data. In some embodiments, the read voltage may have a constant rate of increase with respect to time, for example, exhibiting a ramp shape. In other embodiments, the read voltage may have an incremental step shape, such that a first voltage is applied in a first time period, followed by a second, different voltage being applied in a second time period, and so on.
[0027] The applied voltage may initiate a series of switching events by activating groups of memory cells storing encoded user data. A switching event may be attributed to a memory cell that turns on (e.g., conducts a significant amount of current) when the applied voltage across the memory cell exceeds its threshold voltage (VTH). Thus, initiating a series of switching events in response to an increasing read voltage may be analogous to identifying memory cells in ascending order in terms of their threshold voltage values.
[0028] The adaptive reading techniques described herein may derive a better number of steps, a better voltage for a step (e.g., a starting voltage), and / or a better duration for each step. In certain embodiments, statistical assumptions based on set cell distributions, such as minimum, mean, σ, and the like, may be used to derive a step shape that may start from a better voltage and apply the adaptively derived number of steps. Additionally or alternatively, the step duration may also be adaptively derived based on, for example, the number of cells that are turned on at a given step. For example, if there are memory cells that are still turned on when compared to a step with fewer memory cells turning on a given voltage, then the step may be turned on for a longer duration.
[0029] In practice, embodiments described herein may be used to adaptively adjust the number of steps to be used during a read operation. In one example, a lookup table (LUT) may be used to determine the zones or regions to be used for the steps to follow. That is, the first zone or region may determine the next zone or region, and so on, until the number of steps is adaptively derived. Similarly, the voltage level may be determined. By adaptively deriving the total number of steps and / or adaptively determining the voltage to use for each step, the techniques described herein may more efficiently read data from certain memory devices.
[0030] Turning now to the accompanying drawings, Figure 1 1 is a block diagram of a portion of a memory device 100. The memory device 100 may be any suitable form of memory, such as non-volatile memory (e.g., cross-point memory) and / or volatile memory. The memory device 100 may include one or more memory cells 102, one or more bit lines 104 (e.g., 104-0, 104-1, 104-2, 104-3), one or more word lines 106 (e.g., 106-0, 106-1, 106-2, 106-3), one or more word line decoders 108 (e.g., word line decoding circuitry), and one or more bit line decoders 110 (e.g., bit line decoding circuitry). The memory cells 102, the bit lines 104, the word lines 106, the word line decoders 108, and the bit line decoders 110 may form a memory array 112.
[0031] Each of the memory cells 102 may include a selector and / or a storage element. When the voltage across the selector of the corresponding memory cell reaches a threshold, the storage element may be accessed to read a data value from the storage element and / or write a data value to the storage element. In some embodiments, each of the memory cells 102 may not include a separate selector and storage element, and have a configuration such that the memory cell still functions as a selector and storage element (e.g., may include using materials that behave similarly to both the selector material and the storage element material). For ease of discussion, the bit lines 104, word lines 106, word line decoders 108, and bit line decoders 110 may be discussed. Figure 1 , but these designations are non-limiting. The scope of the present disclosure should be understood to cover memory cells 102 coupled to multiple access lines and accessed through corresponding decoders, wherein the access lines can be used to store data in the memory cells and read data from the memory cells 102. In addition, the memory device 100 may include other circuitry, such as biasing circuitry configured to bias the bit lines 104 or word lines 106 in corresponding directions. For example, the bit lines 104 can be biased with positive biasing circuitry, while the word lines 106 can be biased with negative biasing circuitry.
[0032] The bit line decoders 110 may be organized in multiple sets of decoders. For example, the memory device 100 may include a first set of bit line decoders 114 (e.g., multiple bit line decoders 110) and / or a second set of bit line decoders 116 (e.g., different sets of multiple bit line decoders 110). Similarly, the word line decoders 108 may also be arranged into multiple sets of word line decoders 108, for example, a first set of word line decoders 118 and / or a second set of word line decoders 120. When selecting a target memory cell 102A from the memory cells 102, the decoders may be used in combination with each other to drive the memory cell 102 (e.g., in pairs and / or pairs on either side of the word line 106 and / or the bit line 104). For example, the bit line decoder 110-3 may operate in conjunction with the bit line decoder 110'-3 and / or in conjunction with the word line decoders 108-0, 108'-0 to select the memory cell 102A. As may be appreciated herein, the decoder circuitry on either end of the word line 106 and / or the bit line 104 may be different.
[0033] Each of the bit lines 104 and / or word lines 106 may be a metal trace disposed in the memory array 112 and formed of a metal such as copper, aluminum, silver, tungsten, etc. Thus, the bit lines 104 and word lines 106 may have a uniform resistance per length and a uniform parasitic capacitance per length so that the resulting parasitic load may increase uniformly per length. It should be noted that the components of the depicted memory device 100 may include additional circuitry not specifically depicted and / or may be disposed in any suitable arrangement. For example, a subset of word line decoders 108 and / or bit line decoders 110 may be disposed on different sides of the memory array 112 and / or on different physical sides of any plane including circuitry.
[0034] The memory device 100 may also include control circuitry 122. The control circuitry 122 may be communicatively coupled to the respective ones of the word line decoder 108 and / or the bit line decoder 110 to perform memory operations, such as by causing decoding circuitry (e.g., a subset of the word line decoder 108 and / or the bit line decoder 110) to generate a select signal (e.g., a select voltage and / or a select current) for selecting a target of a memory cell. In some embodiments, positive and negative voltages may be provided to the target of the memory cell 102 on one or more of the bit lines 104 and / or the word lines 106, respectively. In some embodiments, the decoder circuitry may provide biased electrical pulses (e.g., voltages and / or currents) to the access lines to access the memory cells. The electrical pulses may be rectangular pulses, or in other embodiments, pulses of other shapes may be used. In some embodiments, the voltage provided to the access lines may be a constant voltage.
[0035] Activating the decoder circuit can achieve the goal of delivering an electrical pulse to the memory cell 102 so that the control circuit 122 can access the data storage device of the target memory cell to read from or write to the data storage device. After accessing the target of the memory cell 102, the data stored in the storage medium of the target memory cell can be read or written. Writing to the target memory cell can include changing the data value stored by the target memory cell. As previously discussed, the data value stored by the memory cell can be based on the threshold voltage of the memory cell. In some embodiments, the memory cell can be "set" to have a first threshold voltage, or can be "reset" to have a second threshold voltage. The setting memory cell can have a lower threshold voltage than the resetting memory cell. By setting or resetting the memory cell, different data values can be stored by the memory cell. Reading the target of the memory cell 102 can include determining whether the target memory cell is characterized by a first threshold voltage and / or by a second threshold voltage. In this way, the threshold voltage window can be analyzed to determine the value stored by the target of the memory cell 102. The threshold voltage window may be generated by applying programming pulses having opposite polarity biases to the memory cells 102 (e.g., specifically, writing to the selector device (SD) material of the memory cells) and reading the memory cells 102 (e.g., specifically, reading the voltage stored by the SD material of the memory cells 102) using signals having a given (e.g., known) fixed polarity. In some embodiments, a select input may be received from a host device 128, such as a host processor reading data from the memory device 100 to cause the control circuit 122 to access a particular memory cell 102.
[0036] The control circuit 122 may additionally adaptively read data from the memory cell 102, for example, by applying voltages in a pattern (e.g., steps, ramps) to activate groups of memory cells storing data (e.g., encoded data). The voltages may be applied via the bit line 104 and the word line 106. The applied read voltages may then cause an activation event (e.g., a switching event) that is readable via the word line decoder 108 and the bit line decoder 110. The set cell may be activated at a first voltage threshold (Vth) that is lower than a second Vth of the reset cell. The voltage threshold of the set cell may be included in a first bell curve shape, while the voltage threshold of the reset cell may be included in a second bell curve shape. The first bell curve shape may not overlap with the second bell curve shape. That is, all voltage thresholds in the first bell curve shape may be lower than any of the voltage thresholds in the second voltage shape.
[0037] In some embodiments, statistical assumptions based on cell distribution, such as minimum, mean, σ, and the like, may be used to derive a step shape for applying a read voltage to array 112. The step shape may start from a preferred voltage greater than 0 volts and apply an adaptively derived number of stair steps. Additionally or alternatively, the step duration may also be adaptively derived based on, for example, the number of cells turned on at a given step.
[0038] Figure 2 1 is a diagram illustrating a portion of a memory array 130 according to an embodiment of the present disclosure. Within the memory array 130, memory cells are located at the intersections of certain lines (e.g., orthogonal lines). The memory array 130 may be a cross-point array including word lines 106 (e.g., 106-0, 106-1, ..., 106-N) and bit lines 104 (e.g., 104-0, 104-1, ..., 104-M). A memory cell 102 may be located at each of the intersections of the word lines 106 and the bit lines 104. The memory cell 102 may function in a two-terminal architecture (e.g., with a particular word line 106 and bit line 104 combination acting as an electrode for the memory cell 102).
[0039] Each of the memory cells 102 may be a resistance variable memory cell, such as a resistance random access memory (RRAM) cell, a conductive bridge random access memory (CBRAM) cell, a phase change memory (PCM) cell, and / or a spin transfer torque magnetic random access memory (STT-RAM) cell, among other types of memory cells. Each of the memory cells 102 may include a memory element (e.g., a memory material) and a selector element (e.g., a selector device (SD) material) and / or a material layer that functionally replaces a separate memory element layer and a selector element layer. The selector element (e.g., the SD material) may be disposed between a word line contact (e.g., a layer interface between a corresponding one of the word lines 106 and the memory material) and a bit line contact (e.g., a layer interface between a corresponding one of the bit lines 104 and the selector element) associated with a word line or a bit line forming the memory cell. When a read or write operation is performed on the memory cell, an electrical signal may be transmitted between the word line contact and the bit line contact.
[0040] The selector element may be a diode, a non-ohmic device (NOD), or a chalcogenide switching device, etc., or form a similar underlying cell structure. In some examples, the selector element may include a selector material, a first electrode material, and a second electrode material. The memory element of the memory cell 102 may include a memory portion of the memory cell 102 (e.g., a portion that is programmable to different states). For example, in a resistance variable memory cell 102, the memory element may include a portion of the memory cell having a resistance that is programmable to a specific level corresponding to a specific state in response to an applied programming voltage and / or current pulse. In some embodiments, the memory cell 102 may be characterized as a threshold type memory cell that is selected (e.g., activated) based on a voltage and / or current that crosses a threshold associated with the selector element and / or the memory element. Embodiments are not limited to one or more specific resistance variable materials associated with the storage element of the memory cell 102. For example, the resistance variable material may be a chalcogenide formed of various doped or undoped chalcogenide-based materials. Other examples of resistance variable materials that can be used to form memory elements include dual-state metal oxide materials, giant magnetoresistive materials, and / or various polymer-based resistance variable materials, among others.
[0041] In operation, the memory cell 102 may be programmed by applying a voltage (e.g., a write voltage) across the memory cell 102 via a selected word line 106 and bit line 104. A sensing (e.g., a read) operation may be performed to determine the state of the one or more memory cells 102 by sensing current. For example, in response to a particular voltage applied to a selected one of the bit lines 104 / word lines 106 forming the respective memory cell 102, a current may be sensed on the one or more bit lines 104 / one or more word lines 106 corresponding to the respective memory cell 102.
[0042] As illustrated, the memory array 130 may be arranged in a cross-point memory array architecture (e.g., a three-dimensional (3D) cross-point memory array architecture) that extends in any direction (e.g., x-axis, y-axis, z-axis). The multi-stack cross-point memory array 130 may include a plurality of consecutive memory cells (e.g., 102B, 102C, 102D) disposed between alternating (e.g., interleaved) stacks of word lines 106 and bit lines 104. The number of stacks may be increased or decreased, and should not be limited to the volumes or arrangements depicted. Each of the memory cells 102 may be formed between a word line 106 and a bit line 104 (e.g., between two access lines) such that a respective one of the memory cells 102 may be directly electrically coupled (e.g., electrically coupled in series) to its respective pair of bit lines 104 and word lines 106, and / or formed by electrodes (e.g., contacts) made of respective portions of metal in the respective pair of bit lines 104 and word lines 106. For example, memory array 130 may include a three-dimensional matrix of individually addressable (e.g., randomly accessible) memory cells 102 that may be accessed at a granularity as small as a single storage element and / or a plurality of storage elements for data operations (e.g., sensing and writing). In some cases, memory array 130 may include a plurality of storage elements. Figure 2 The examples shown may include more or fewer bit lines 104, word lines 106, and / or memory cells 102. Each stack may include one or more memory cells 102 aligned in the same plane.
[0043] Figure 3 1 is a block diagram of an embodiment of a memory device 200 in which a memory array 112 has been partitioned into a plurality of partitions 202 (e.g., partitions 202-a, 202-b ... 202-n). Each partition 202 may include associated local control circuitry 204 (e.g., local control circuitry 204-a, 204-b ... 204-n). In some embodiments, the local control circuitry 204 may be included in or interfaced with the control circuitry 122. Thus, each partition 202 may operate independently of the other partitions 202, which may enable parallel reading and writing of the memory array 112.
[0044] In some embodiments, the memory array 112 is a 3DXP memory array, and each individual partition 220 is a 1 Gigabyte partition. The memory device 200 may include 16 1 Gigabyte partitions. In some examples, the memory in each partition 202 may be accessed at a granularity of 16 bytes, so 26 bits of memory address information are provided to the memory array 110. In addition, four bits may be used to provide a partition identification in this example. It should be noted that the specific partition size, number of partitions, and bits for command and address operations described above are provided only as examples, and different partition sizes, number of partitions, and command / address bits may be used in other embodiments.
[0045] In some embodiments, the data stored in the memory array 112 may be encoded, for example, by adding certain encoding bits. Encoding the data bits may enable faster reading, as further described below. Figure 4 , illustrates an embodiment of a user data pattern 400 that may be used by the adaptive read techniques described herein. The user data pattern 400 illustrates unencoded user data 402 that may subsequently be encoded as user data 404. The unencoded user data 402 may be referred to as an input vector in some cases. The encoded user data 404 may include additional bits (e.g., b1 to b4). The additional bits may be referred to as toggle bits and may indicate the state of the user data, as described below.
[0046] Adaptive reading may include an encoding technique for generating encoded user data having a weight (e.g., the number of bits having a logic state 1 out of the total number of bits in the user data) within a predetermined interval. In some embodiments, the interval is 50% to (50+50 / k)%, where k is a predetermined factor as further described below. In some cases, the interval is expressed as [50%, (50+50 / k)%. For example, when k is equal to 4, the interval may be 50% to 62.5% (e.g., [50%, 62.5%]), 40% to 70%. A different weight other than 50% may be used as a lower limit for the interval. Figure 4 The illustrations in refer to 50% as the lower limit of the interval for a more concise description of the depicted features; however, other alternatives and different variations are contemplated and are within the scope of the present disclosure.
[0047] For example, user data 402 is shown as having 16 bits (eg, a1 to a16). In the case where k is equal to 4, the predetermined interval satisfied for the encoded user data may be [50%, 62.5%]. Figure 44 shows various forms of the encoded user data 404 when k=4. The encoding technique may add k number of flip bits (e.g., b1 to b4 when k=4) to the user data 402 (e.g., a1 to a16) to generate the encoded user data 404. In addition, the original user data pattern may be segmented into k number of parts (e.g., four parts or segments when k=4). For example, the first part may include bits a1 to a4. The first part may be associated with the first flip bit b1. The second part may include bits a5 to a8. The second part may be associated with the second flip bit b2. The third part may include bits a9 to a12. The third part may be associated with the third flip bit b3. The fourth part may include bits a13 to a16. The fourth part may be associated with the fourth flip bit b4. In some embodiments, the initial values of b1 to b4 correspond to a logic state 1 (e.g., 1111 of the encoded user data pattern 406). The logic state 1 in the flip bit may indicate that the corresponding part of the original user data is not inverted. Conversely, a logic state of 0 in a toggle bit may indicate that the corresponding portion of the original user data is inverted.
[0048] As described above, the adaptive reading techniques described herein may determine the weight of the encoded user data pattern 404 as a percentage (e.g., adding a logic 1 bit and dividing the sum by the total number of unencoded bits). For example, the encoded user data 560-a has a weight of 25% (e.g., 4 bits with a logic state of 1 among the 16 bits in the user data), which does not satisfy the predetermined interval of [50%, 62.5%] when k=4. In addition, the encoding technique may change the logic state of the flipped bit throughout all possible combinations of the logic state of the flipped bit to find a specific encoded user data with a specific weight within a predetermined interval (e.g., an interval of [50%, 62.5%] when k=4). When there are k flipped bits (e.g., k=4), there are a total of 2^k (e.g., 2^4=16) combinations, such as 1111, 1110, 1101, 1100, ..., 0001, and 0000.
[0049] When the logic state of the flip bit corresponds to a logic state of 0, the adaptive read may invert the logic state of the corresponding portion of the user data and evaluate the weight. As illustrated, the user data 406 does not include any inversions, and therefore all flip bits are set to 1. The inversion of the data may occur subsequently. For example, when the flip bit is 1110 as shown in the encoded user data 408, the logic state of the fourth portion (e.g., bits a13 to a16) is inverted from 0110 to 1001. Subsequently, the encoding technique may determine that the encoded user data pattern 408 has a weight of 25% (e.g., 4 bits with a logic state of 1 among the 16 bits in the user data), which does not meet the predetermined condition of the weight within the interval of [50%, 62.5%). The encoding technique may restore the logic state of the fourth portion back to 0110 and change the content of the flip bit to the next combination (e.g., 1101 as shown in the encoded user data 410). The encoding technology may invert the logic state of the third portion (e.g., bits a9 to a12) from 0100 to 1011, as shown in the encoded user data 410, and determine that the encoded user data pattern 410 has a weight of 38% (e.g., 6 bits having a logic state of 1 out of 16 bits in the user data), which also does not satisfy the predetermined condition of the weight within the interval of [50%, 62.5%).
[0050] The adaptive read may continue to change the content of the flip bit, flip the logic value of the bit of the corresponding portion of the user data according to the flip bit, and then evaluate the weight of the encoded user data until the encoded user data satisfies the predetermined condition (e.g., the interval of [50%, 62.5%]). For example, the encoded user data 412 has a weight of 38% and does not satisfy the predetermined condition of the [50%, 62.5%] weight interval. The encoded user data pattern 414 has a flip bit content of 1011, and the second portion of the user data (e.g., bits a5 to a8) is flipped from 0000 to 1111. The weight of the encoded user data 414 is 50% (e.g., 8 bits having a logic state of 1 among 16 bits in the user data), which satisfies the predetermined condition of having a weight between [50%, 62.5%].
[0051] The decoding technique may stop changing the content of the flip bit based on determining that the encoded user data pattern 414 satisfies a predetermined condition, and the decoded user data pattern 414 may be stored in the memory cell. The flip bit content (e.g., 1011) may then be used to decode the encoded user data when the encoded user data is read from the memory cell. For example, the logic state of bits a5 to a8 of the encoded user data 414 (e.g., 1111) may be reversed back to its original logic state (e.g., 0000) based on the value of the flip bit b2 when the encoded user data 414 is read (e.g., the logic state 0 of b2 indicates that bits a5 to a8 have been reversed). By storing the encoded bits in a desired weight range, the techniques described herein may read data stored in the memory device 100 more quickly.
[0052] Figure 5 4 is a block diagram illustrating an embodiment of two ramping shapes that may be used to read data stored in the memory device 100. In the depicted embodiment, a first ramping shape (e.g., a staircase shape) 450 is depicted, where the X-axis represents time and the Y-axis represents voltage. In the depicted embodiment, the ramping shape 450 includes four steps with fixed durations and numbers. A read command 452 may be received, and the control circuit 122 may then initiate a first ramping step 454 to read data, such as the data 400. The first ramping step 454 may initiate a series of switching events in the memory cell 102 that may store data having a logic 1, for example.
[0053] As previously mentioned, the memory cells 102 storing data may exhibit a bell curve distribution based on voltage. That is, the first ramp step 454 does not cause all memory cells 102 storing logic 1 data to exhibit switching behavior. Therefore, additional voltage ramps may be applied. Furthermore, the memory cells 102 storing logic 0 data will not exhibit a switching event until a voltage higher than the voltage used to read all memory cells 102 storing logic 1 data is applied.
[0054] Thus, the first ramp shape 450 may result in three additional ramp steps 456, 458, 460 being applied to the memory cell 102 in order to read logic 1 data (e.g., set cell data) and logic 0 data (e.g., reset cell data). In the depicted embodiment, each of the ramp steps 454, 456, 458, and 460 includes the same time t. That is, each ramp step 454, 456, 458, and 460 is performed for a duration of t. After the last ramp step 460 is used to apply the read voltage, the control circuit 122 may then use the data output 462 to decode into return data, such as to provide to a device connected to the memory device 100.
[0055] In some cases, the stored data can be read more efficiently. For example, the adaptive reading technique can use a variable number of ramp steps and / or a variable ramp timing, as shown in the ramp shape 464. More specifically, the ramp shape 464 includes ramp steps 466, 468, and 470, each of which can have different times t1, t2, tN from each other. In practice, the techniques described herein can adaptively determine the number of steps to use, as well as the time to apply the voltage for each step, as further described below.
[0056] In certain embodiments, the techniques described herein may derive a distribution of a set unit and divide the set distribution into a plurality of σ regions (eg, −5σ to +5σ) in units of slices (eg, 1σ slices), such as Figure 6 More specifically, the figure illustrates the above relative Figure 4 The example of the data that has been encoded is set to have a distribution 500 with weights between, for example, [50%, 62.5%]. The set distribution is shown as having a statistical mean That is, 0σ (0σ), where the X-axis represents the read voltage threshold (Vth) value and the Y-axis represents the number of set memory cells 102 that exhibit a switching event at a given Vth. That is, a cell can switch at a given Vth, and therefore is shown to store a logical 1.
[0057] For example, due to the above Figure 4 The described encoding scheme, therefore, approximately half (eg, 50%) of the memory cells storing logic 1 may exhibit a switching event at the average voltage. Under , approximately half of the memory cells 102 may exhibit switching events. The distribution 500 is further shown to have standard deviations -5σ, -4σ, -3σ, -2σ, -1σ, +1σ, +2σ, +3σ, +4σ, +5σ, where σ is one standard deviation from the mean. Thus, +2σ is two standard deviations to the right of the mean, while -2σ is two standard deviations to the left of the mean, and so on. The standard deviation formula may be used Where X is the Vth value of the memory cell 102 for the data point in the X-axis and n is the total number of data points to find 1σ or the first standard deviation and / or any other σ.
[0058] As illustrated, region 502 is inside 1σ, region 504 is inside 2σ, region 506 is inside 3σ, region 508 is inside 4σ, and region 510 is inside 5σ. Points outside region 510 may then be included inside 6σ. For a 64-bit codeword, set distribution 500 may be derived via 256 (i.e., 64 x 4) and data points (e.g., n=256), with each data point (e.g., X) representing the Vth of an individual set memory cell 102 for storing a bit included in the codeword. Techniques described herein may adaptively define a ramp shape such as ramp shape 464 by using a statistical distribution such as set distribution 500 to determine certain characteristics of the ramp shape (e.g., number of steps, voltages for a step, length of a step), as further described below.
[0059] Figure 7 64-bit codeword. The process 600 may be implemented, for example, by the control circuit 122. In summary, the process 600 may be based on dividing a set distribution (e.g., the distribution 500) into an approach region or range and a final step region or range, for example, by using the regions 502, 504, 506, 508, 510. The approach region may be used to apply one or more ramping steps, while the final step region may apply a single and final step, resulting in an adaptive ramping shape (e.g., the ramping step 564).
[0060] In the depicted example, the process 550 may begin at block 552 and then initialize (block 554) one or more variables. For example, a Last variable indicating whether the process is the final step region may be initialized to 0, while a V ramp variable indicating the voltage to be applied to the step or ramp may also be initialized to zero. A V step variable indicating the amount by which the voltage for each step will be increased may also be initialized. For example, the V step variable may be initialized to start at a start voltage (V start ) value. In the depicted embodiment, the process 550 may then increase (block 556) the V ramp to itself plus the value of the V step variable. The process 550 may also (block 556) transmit a read voltage having the value of the V ramp to the memory cell 102 storing the data to be read. For example, the bit line 104 and the word line 106 may be used to transmit the V ramp voltage.
[0061] The set memory cell 102 may then exhibit a switching event representing a cell storing a logical 1. The process 550 may then count the number of memory cells 102 that have been read (block 558). The process 550 may then determine (decision 560) whether the variable Last is set to 1. If the variable Last is set to 1 (decision 560), the process 500 may then stop (block 562). If the variable Last is not set to 1 (decision 560), the process 500 may then determine (decision 564) whether the number of set cells exhibiting a switching event is less than or equal to 6. The number 6 may have been selected because it falls outside the 3σ region 504 for a 64-bit codeword. Other numbers may be used based on the size of the codeword, such as 8 for a 128-bit codeword, 12 for a 256-bit codeword, and so on. The process 550 may then set (block 566) the V step value to 3σ, and then iterate to block 556. If the number of set cells exhibiting switching events is not less than or equal to 6 (decision 564), process 550 may then determine (decision 568) whether the number of set cells exhibiting switching events is greater than or equal to 7 and less than or equal to 41. The numbers 7 and 41 may be selected because values in this range are outside of 2σ. Other values may be used based on the codeword length.
[0062] If the number of set cells that exhibit switching events is greater than or equal to 7 and less than or equal to 41 (decision 568), the process 550 may then set (block 570) the V step value to 2σ and iterate to block 556. If the number of set cells that exhibit switching events is not greater than or equal to 7 and less than or equal to 41 (decision 568), the process 550 may then determine (decision 572) whether the number of set cells that exhibit switching events is greater than or equal to 42 and less than or equal to 128. The numbers 42 and 128 may be selected because values in this range are outside of 1σ.
[0063] If the number of set units that exhibit switching events is greater than or equal to 42 and less than or equal to 128 (decision 572), the process 550 may then set (block 574) the V step value to 6σ, set the Last variable to 1, and iterate to block 556. If the number of set units that exhibit switching events is not greater than or equal to 42 and less than or equal to 128 (decision 572), the process 550 may then determine (decision 576) whether the number of set units that exhibit switching events is greater than or equal to 129 and less than or equal to 215. The numbers 129 and 215 may be selected because values in this range are within 1σ.
[0064] If the number of set units that present switching events is greater than or equal to 129 and less than or equal to 215 (decision 576), the process 550 may then set (block 578) the V step value to 5σ, set the Last variable to 1, and iterate to block 556. If the number of set units that present switching events is not greater than or equal to 129 and less than or equal to 215 (decision 576), the process 550 may then determine (decision 580) whether the number of set units that present switching events is greater than or equal to 216 and less than or equal to 256. The numbers 216 and 256 may be selected because values in this range are within 6σ.
[0065] If the number of set cells exhibiting a switching event is greater than or equal to 216 and less than or equal to 256 (decision 580), the process 550 may then set (block 584) the Vstep value to 1σ, set the Last variable to 1, and iterate to block 556. In certain embodiments, the process 550 may utilize a lookup table (LUT) or a portion of a LUT as described below with respect to Table 1. The LUT may be used to more quickly determine the voltage to be applied to read the set memory cell 102 and determine whether additional ramp steps are to be used.
[0066] area Range of the switched-on master setting unit Read voltage step Final step near ≤6 3σ 0 near [7:41] 2σ 0 Final Step [42:128] 6σ 1 Final Step [129:215] 5σ 1 Final Step [216:256] 4σ 1 Final Step >256 1σ 1
[0067] Table 1 - Read voltages to be used based on the number of set cells turned on
[0068] As mentioned earlier, a first starting voltage may be applied, which may then cause one or more set cells to turn on. Table 1 may then be queried, for example by looking at the value range of the second column, to determine a new voltage to be applied (e.g., storing the voltage of the third column), and also to determine whether more ramping steps will be provided (e.g., via the fourth column). By subsequently applying a voltage from the third column, a new set of cells may be read, and the total number of cells that have been read may then be used to determine whether more ramping steps will be used to read the remaining set cells. Once all set cells have been read (e.g., after applying the final step voltage), all remaining unread memory cells 102 may be derived as reset memory cells.
[0069] It should be understood that the above decisions 564, 568, 572, 576, 580 and / or the values in Table 1 are merely examples. Other values may be used. For example, given a codeword having N bits, N may be divided into regions 502, 504, 506, 508, 510, e.g., assuming a normal distribution of encoded data (and / or balanced data, e.g., data having approximately equal logical 1 counts and logical 0 counts). In other embodiments, the manufacturer may perform a statistical analysis on the read behavior of the memory cells of one or more memory devices 100, and then derive a distribution (e.g., distribution 500) for the memory device 100 and store the resulting analysis as a LUT (e.g., Table 1).
[0070] In one embodiment, the boundaries between the ranges (6, 41, 128, and 215) have been selected considering a Gaussian distribution of 256 set cells for certain mean values. For example, if the mean value is 4.4V and σ=0.85mV. The values may correspond to mean-2σ, mean-1σ, mean, and mean+1σ of the set distribution, respectively, as summarized in the following table.
[0071] Vth count Mean - 2σ 4.230 6 Mean - 1σ 4.315 41 average value 4.400 128 Mean + 1σ 4.485 215
[0072] Table 2 - Example only table to illustrate an average value of 4.4V and a Vth value of σ = 0.85mV for 256 set cells.
[0073] It may be helpful to illustrate a series of read voltage steps that may be provided by using the process 550 described above. Figure 8 , the figure includes a graph 600 illustrating an embodiment of a total curve 602 that can be used to read set memory cells 102. The graph 600 includes a read voltage on the X-axis and a number of set memory cells on the Y-axis. The Y-axis is also shown to include a σ range of 1 to 6 for the total count curve 602, and the X-axis is also shown to include a σ step range of 1 to 6 for the read voltage. In the depicted embodiment, a dividing line 604 divides the graph 600 into a near range or region 606 and a final step range or region 608. In certain embodiments, the dividing line 604 can be equal to -2σ for the distribution 610 (e.g., equivalent to the distribution 500). In the first step 612, a first read voltage can be applied. For example, a V start read voltage can be applied to the memory cells 102 to cause one or more set memory cells 102 to exhibit a switching event representing a cell storing a logical 1.
[0074] The number of switching units 102 may then be counted, and if the number falls within the proximity range 606 (e.g., as listed in Table 1 and Figure 76 (described in ), then a second step 614 may then be taken. In the depicted embodiment, the second step 614 may be applied at a 3σ voltage, and a new total number of set memory cells 102 that have exhibited a switching event may be counted. The total count is shown to be within the proximity region 606, and therefore the third step 616 may apply a voltage, such as a 2σ voltage. The 2σ voltage may result in the total count of switched set memory cells having a value that falls outside the proximity region 606 and inside the final step region 608. Thus, a single final step 618 may be applied, for example, at a 5σ voltage. The final step 618 may result in all set memory cells 102 being found, and the remaining memory cells 102 may be assumed to be reset cells. By adaptively determining the number of total read steps and the voltage to be applied for each read step, the techniques described herein may more efficiently read data stored in the memory cells 102.
[0075] As previously mentioned, the techniques described herein may also enable adaptive times between ramping steps. That is, rather than using fixed times for the durations of 454, 456, 458, 460, the steps may have varying durations, such as steps 466, 468, 470. In one embodiment, the duration may depend on the number of memory cells 102 sensed in parallel, such as from the same word line 106. A larger number of sensed memory cells 106 may introduce reciprocal interference and thus may result in a longer duration of the step, such as a longer sensing time. After sensing the number of activated memory cells 102 (e.g., set cells) for the first step, the control circuit 122 may then adaptively decide the duration for the second step, count the second number of activated memory cells 102, and so on. The adaptive read technique may also include fixed and / or variable delays suitable for handling various data scenarios (e.g., best case, typical case, worst case). Thus, a more efficient and faster converging self-referenced read operation may be provided, with a lower raw bit error rate (RBER).
[0076] Fig. 9Included are timing diagrams illustrating embodiments of fixed latency data transfers for certain systems that may interface with the memory device 100 and may use fixed latency communications. In the depicted embodiment, a first timing diagram 650 illustrates worst (and also best) case fixed latency times for steps with fixed durations (e.g., steps 454, 456, 458, and 460). A second timing diagram 652 illustrates fixed latency times for steps with adaptive (e.g., varying) durations (e.g., steps 466, 468, or 470). In the depicted embodiment, all steps in the first timing diagram 650 have approximately the same duration 654. The fixed duration step 656 is shown as including a sense time 658, an error correction (ecc time) 660, and a data transfer time 662. In use, the memory device 100 may issue a read command on the command bus at time 662. The control circuit 122 may then respond to the read command by sensing the data stored in the memory cell, such as by sending a voltage to activate the set memory cell 102 storing, for example, a logical 1. The sensing operation may transmit a stepped voltage during a sensing time 658, apply error correction to the sensed data during an error correction time 660, and then transmit the data during a data transmission time 662. A data value 664 (e.g., 32 data bits) may then be placed on the data bus at time 666.
[0077] The timing diagram 652 when using adaptive steps 668, 670 is shown as also using the same fixed delay time, e.g., time 654 for communication. Thus, while the best time step 668 and the worst time step 670 are shown as having, for example, different sensing times 672, 674, different error correction times 676, 678, and / or different transmission times 680, 682, the adaptive steps 668, 670 both wait for the end of the delay time 654 before transmitting data 684 (e.g., 32 data bits). That is, the read command is initiated at time 686, which is illustrated as the same time as time 662, and data 684 is provided to both steps 668, 670 at the same time 666, regardless of the timing for the adaptive steps used (e.g., steps 668 and / or 670). Therefore, the techniques described herein may utilize a data not valid (DNV) flag and / or a ready buffer (R / B) signal to reduce data read time for certain read steps.
[0078] Fig.10 Timing diagrams of embodiments including typical case read steps and worst case read steps. More specifically, timing diagram 700 illustrates a typical read step 702 and timing diagram 704 illustrates a worst case read step 706. Also shown is Fig. 9The fixed delay time 654 of the typical case read step 706 is used for comparison. In the illustrated embodiment, a read command is issued at time 708. The control circuit 122 may then issue a typical case read step 702 in response to the read command. The typical case read step 702 may include a sensing time 710, an error correction time 712, and a data transfer time 714. The typical case read step 702 may complete its data transfer time at or before the end of the fixed delay time 654. However, the worst case read step 706 may include a longer sensing time 716, an error correction time 718, and / or a data transfer time 720 than the typical case read step 702. Therefore, the typical read step 702 may be ready for data communication before the worst case read step 706.
[0079] To improve data read efficiency, the techniques described herein may use a data valid (DV) flag 722 and / or a data not valid (DNV) flag 724. For example, at the end of a certain time period (e.g., at the fixed latency period 654, or 10, 20, 30, 50 nanoseconds before the end of the fixed latency period 654), a DV flag 722 may be transmitted by the control circuit 122 on the DNV bus to indicate that the data 726 (e.g., 32 bits) now placed on the data bus is valid. If the time period (e.g., time 654) expires and the data is not ready to be processed, the control circuit 122 may transmit a DNV flag 724 on the DNV bus to indicate that the data 728 is not ready to be processed. Thus, read steps of different durations may more efficiently transfer data when the data is ready to be processed.
[0080] Fig.11 A timing diagram of an embodiment of a typical case read step and a worst case read step that can use ready busy (R / B) signaling. In the depicted embodiment, timing diagram 750 includes a typical case read step 702 and timing diagram 752 includes a worst case read step 706. In the depicted embodiment, a read command can arrive via a command bus at time 754. Therefore, a ready busy signal 756 for the typical case read step 702 is depicted as having a logic low value at time 754 to indicate that the data is not ready to read. Similarly, a ready busy signal 758 for the worst case read step 706 is also depicted as having a logic low value at time 754 to indicate that the data is not ready to read.
[0081] The typical case read step 702 is illustrated as ending at time 760. Thus, the ready busy signal 756 may now be set to a logic high at time 760, indicating that the data 762 placed on the data bus is now ready for processing. Similarly, the ready busy signal 758 may be set to a logic high at time 764, indicating that the data 766 placed on the data bus is now ready for processing. When the read step results in valid data being placed on the data bus, the control circuit 122 and / or an external system coupled to the memory device 100 may then use the ready busy signals, such as signals 756, 758, to read and process data, such as data 762, 764. By using the ready busy signals, the techniques described herein may optimize the timing of the reads.
[0082] As mentioned previously, partitions 202 may be used, for example, to more efficiently read and write data in parallel. However, one side effect of using partitions 202 may include introducing reciprocal interference when sensing data. The techniques described herein may adaptively adjust the read time for a step based on the number of cells activated during a read operation. Turning now to Fig.12 , is a block diagram illustrating an embodiment of a partition architecture 800 that may cause some "interference" based on parallel reading or writing.
[0083] In the depicted embodiment, a plurality of partitions 202 are shown, such as included in a memory device 100. Each partition 202 may include a plurality of tiles 802 disposed on various layers or stacks. The bit count of each tile 802 may vary, for example, depending on the desired application. That is, the memory device 100 may be configured to have a certain bit count per tile. Similarly, the tiles 802 may be manufactured in various tile sizes. The figure illustrates a tile 804 of 1 bit per tile (bit / tile), a tile 806 of 4 bits / tile, a tile 808 of 8 bits / tile, a tile 810 of 16 bits / tile, and a tile 812 of 32 bits / tile. It should be noted that other bits per tile and / or tile sizes may be used.
[0084] As the bit per tile count increases and / or as the tile size increases, "disturbance" may be introduced. That is, when sensing memory cells 102 in parallel in the same word line 106, the voltage introduced on the word line 106 may "disturb" nearby memory cells 102 and may require more overall read time than a non-parallel read. For example, a memory cell 102 may switch and inject current into the word line 106, and the next memory cell 102 may switch at a later time than expected due to the changed RC value. The greater the number of memory cells 102 that can be sensed in parallel from the same word line 106, the more sensing times (e.g., sensing times 672, 674, 710, 716).
[0085] The techniques described herein may use adaptive adjustment of the duration for each read step (e.g., steps 466, 468, 470), for example, based on a count of the number of steps that have been activated. In some embodiments, the duration for a read step may be derived more quickly by using a lookup table (LUT), which may include a first column with a count of set cells that have been activated and a second column that may include the time to apply a voltage for the next read step in the read ladder. An example table is shown below.
[0086]
[0087]
[0088] Table 3 - Timing adjustments based on the number of set units switched on
[0089] Table 3 illustrates some example values that may be used for a 128-bit codeword. In use, the control circuit 122 may use a counter-based circuit to count the number of set memory cells 102 activated after the first step (e.g., step 466), and then use Table 3 above to determine the time for the second step (e.g., step 468), for example, via a comparator circuit. The time for the third step, fourth step, fifth step, sixth step, and so on may then be determined based on the count of set memory cells that have been activated in the previous step. In practice, adaptive timing for the read steps may be provided via a lookup table. With respect to Fig.13 Describes an example procedure that uses a lookup table.
[0090] Fig.13 8 is a flow chart of a process 850 that may be used to adaptively determine a time interval for a read step, such as adaptive steps 466, 468, 470, etc. The process 850 may be implemented by the control circuit 122. In the depicted embodiment, the process 850 may apply a starting voltage (e.g., a starting voltage step) and then count the number of set memory cells 102 that may have been activated (block 852). The set memory cells 102 may be used, for example, relative to the time interval of the adaptive step 466, 468, 470. Figure 4 The encoding process described above stores data. The process 850 may then use a lookup table (LUT) 854, such as Table 3 described above, to determine the value of the duration to be used during the next step in the ramp (e.g., a staircase ramp) (block 856). That is, the process 850 may determine the duration for applying the voltage in the next step of the staircase based on the current count of the activated set cells. The process 850 may then apply the duration to read (block 858) the next step, such as by applying a voltage at a certain value and for a desired duration. In some embodiments, a value relative to the value of the set cell may be used. Figure 7 The process 550 depicted determines the applied voltage value.
[0091] The process 850 may then iterate back to block 852 and count the number of set memory cells 102 that may have been activated after completion of the current read step and continue using the LUT 854 if there are more steps to be applied. In some embodiments, a LUT 854 may be used relative to the Figure 7 The process 550 depicted determines the number of steps that will be used to read the data. By iteratively applying voltages for the desired duration, counting the resulting activated set memory cells, and determining the timing duration for the next read step via the LUT 854, the techniques described herein can adaptively determine the timing for the next step and thus more efficiently read data from the memory array 112. The timing for the next step, such as the timing stored in the second column of Table 3, can be derived based on statistical analysis. For example, a manufacturer of the memory device 100 can test one or more of the memory devices 100 to empirically determine, for example, more efficient time values for reading at various codeword sizes.
[0092] It should be noted that the lookup tables described above may be created based on memory technology, usage models (e.g., use in servers storing data in long-term, short-term, etc.), applications (e.g., portable device memory, server memory, desktop computer memory, and the like), and may be updated during the life of the memory device. It should be noted that the lookup tables described above (e.g., Tables 1 and 3) may be created based on memory technology, usage models (e.g., use in servers storing data in long-term, short-term, etc.), applications (e.g., portable device memory, server memory, desktop computer memory, and the like), and may be updated during the life of the memory device. Other step durations in the lookup tables (e.g., Tables 1 and 3) may also include non-monotonic trends. Similarly, parallel access within one tile may be used, and longer resistance-capacitance (RC) recovery times may be considered when accessing tiles in parallel. For example, it may be expected that an increased number of snap cells (with a sudden current increase and corresponding voltage drop at each snap) may occur.
[0093] Although various modifications and alternative forms may occur in the present disclosure, specific embodiments have been shown by way of example in the accompanying drawings and described in detail herein. However, it should be understood that the present disclosure is not intended to be limited to the specific forms disclosed. In fact, the present disclosure is intended to cover all modifications, equivalents and alternatives that fall within the spirit and scope of the present disclosure as defined by the appended claims.
[0094] The technology presented and claimed herein is by reference and application to substantial objects and specific examples of a practical nature that improve the art in a demonstrable manner and are therefore not abstract, intangible, or purely theoretical. In addition, if any claim appended at the end of this specification contains one or more elements that are expressed as "means for [performing] [the function] ..." or "steps for [performing] [the function] ...", then it is intended that these elements be interpreted under 35 U.S.C. 112(f). However, for any claim containing elements specified in any other manner, it is expected that such elements will not be interpreted under 35 U.S.C. 112(f).
Claims
1. A memory device, comprising: a memory array comprising a plurality of memory cells; a control circuit operatively coupled to the memory array, the control circuit being configured to: applying a first voltage to the memory array in response to receiving a read request; counting a total number of the plurality of memory cells that have switched to a valid read state based on the first voltage; as well as applying a second voltage to the memory array based on the total number of the plurality of memory cells that have switched to the valid read state based on the first voltage, Wherein the control circuit is configured to apply the first voltage for a first duration and to apply the second voltage for a second duration, and wherein the first duration is different from the second duration.
2. The memory device of claim 1 , wherein a first set of memory cells in the plurality of memory cells are configured to be read as storing a first logic value based on the first voltage, and wherein a second set of memory cells included in the plurality of memory cells are configured to be read as storing the first logic value based on the second voltage.
3. The memory device of claim 2, wherein data is returned based on bits stored in the first set of memory cells and the second set of memory cells, and wherein the bits comprise encoded data having a weight range between 40% and 70%.
4. The memory device of claim 1, wherein the control circuit is configured to apply the second voltage by determining whether the total number is within an approach value range or within a final step value range. 5 . The memory device of claim 4 , wherein the control circuit is configured to use a lookup table (LUT) to determine whether the total number is within the approximate value range or within the final step value range.
6. The memory device of claim 4, wherein the control circuit is configured to apply one or more voltage steps if the total number is within the approach value range, and to apply a single voltage step if the total number is within the final step value range.
7. The memory device of claim 4, wherein the approach value range, the final step value range, or a combination thereof is divided into one or more statistical regions.
8. The memory device of claim 7, wherein the one or more statistical regions comprise 1σ, 2σ, 3σ, 4σ, 5σ, 6σ, or a combination thereof, wherein σ is one standard deviation from a statistical mean of the one or more statistical regions.
9. The memory device of claim 1, wherein the first voltage and the second voltage are included in a voltage step shape.
10. A method comprising: applying a first voltage to a memory array of the memory device at a control circuit included in the memory device in response to receiving a read request; counting a total number of a plurality of memory cells that have switched to a valid read state based on the first voltage; as well as applying a second voltage to the memory array based on the total number of the plurality of memory cells that have switched to the valid read state based on the first voltage, Wherein applying the first voltage comprises applying the first voltage for a first duration and wherein applying the second voltage comprises applying the second voltage for a second duration, and wherein the first duration is different from the second duration.
11. The method of claim 10, wherein applying the second voltage comprises determining whether the total number is within an approach value range or a final step value range, and wherein the approach value range, the final step value range, or a combination thereof is divided into one or more statistical regions.
12. The method of claim 11, comprising performing one or more voltage steps if the total number is within the approach value range, and performing a single voltage step if the total number is within the final step value range.
13. The method of claim 10, comprising encoding data as encoded data and storing the encoded data in the memory array prior to receiving the read request, and wherein returning the data comprises decoding the encoded data.
14. A memory device comprising: a memory array comprising a plurality of memory cells; a control circuit operatively coupled to the memory array, the control circuit being configured to: applying a first voltage to the memory array in response to receiving a read request; counting a total number of the plurality of memory cells that have switched to a valid read state based on the first voltage; Using a lookup table LUT to determine whether the total number falls within a range of approximate values, a range of final step values, or a combination thereof; as well as One or more voltage steps are applied if the total number is within the approach value range and a single voltage step is applied if the total number is within the final step value range to return data.
15. The memory device of claim 14, wherein a first set of memory cells in the plurality of memory cells are configured to be read as storing a first logic value based on the first voltage, and wherein a second set of memory cells in the plurality of memory cells are configured to be read as storing the first logic value based on a second voltage.
16. The memory device of claim 14, wherein the approach value range, the final step value range, or a combination thereof, is divided into one or more statistical regions, the one or more statistical regions comprising 1σ, 2σ, 3σ, 4σ, 5σ, 6σ, or a combination thereof, where σ is one standard deviation from a statistical mean of the one or more statistical regions.
17. The memory device of claim 16, wherein the approximate value range is divided into a first statistical region including the 2σ, the 3σ, or a combination thereof, and wherein the final step value range is divided into a second statistical region including 1σ, 4σ, 5σ, 6σ, or a combination thereof.
18. The memory device of claim 14, wherein the control circuit is configured to encode data into encoded data having a weight between 40% and 70% and store the encoded data in the memory array before receiving the read request, and wherein the control circuit is configured to return the data by decoding the encoded data.
