Abstract:
The present invention proposes a Field Programmable Gate Array (FPGA) device (100) comprising a plurality of configurable electrical connections (115 1 -115 2 ,115 1 -115 3 ), a plurality of controlled switches (205), each one adapted to activating/de-activating at least one respective electrical connection in response to a switch control signal and a control unit (125) including an arrangement of a plurality of control cells (200). Each control cells controls at least one of said controlled switches by the respective switch control signal, each control cell including a volatile storage element (210,215) adapted to storing in a volatile way a control logic value corresponding to a preselected status of the at least one controlled switch, and providing to the controlled switch said switch control signal corresponding to the stored logic value. Each control cell further includes a non-volatile storage element (P1;F1) coupled to the volatile storage element, the non-volatile storage element being adapted to storing in a non-volatile way the control logic value.
Abstract:
The invention relates to an architecture for distributing supply voltages to a plurality of memory modules (Mod1, ..., ModN) supplied through a plurality of charge pump circuits (Pump1, ..., PumpM). Advantageously according to the invention, the architecture for distributing supply voltages comprises a sorting block (11) bidirectionally-connected to the plurality of memory modules (Mod1, ..., ModN) from which it receives a plurality of power requests and it is capable of providing a sorting signal (ORD) of said power requests on the basis of a priority scale in order to drive the plurality of charge pump circuits (Pump1, , PumpM) and distribute convenient supply voltages (Vhigh1, ..., VhighN; Vneg1, ..., VnegN) to the plurality of memory modules (Mod1, ..., ModN). Moreover, this architecture is software-configurable.
Abstract:
The invention relates to a programming method for a multilevel memory cell able to store a plurality of bits in a plurality of levels (N), which method comprises the phases of:
initially programming (I) a cell threshold value (VthDATI) to a first set of levels [O;(m-1)] being (m) a submultiple of the plurality (N) of levels of the multilevel cell; reprogramming without erasing (II) another set of levels [m;(2m-1)] containing the same number of levels (m) as the first set; reiterating (N R - 1 times) the reprogramming without erasing phase (III,IV, ...) until the levels (N) of the multilevel cell are exhausted.
The invention makes also reference to a multilevel memory device of the type comprising a plurality. of multilevel memory cells organised into sectors, the sectors being themselves split into a plurality of data units (UD) wherein a data updating operation is performed in parallel, the data units (UD) being programmed by means of the programming method according to the invention.
Abstract:
The present invention relates to a sensing circuit (10, 100) for a memory cell (11) inserted between a first (Vdd) and a second voltage reference (GND) and connected, in correspondence with a first inner circuit node (XBL), to the memory cell (11) of the type comprising:
a first bias current (Ip) generator (12) inserted between the first voltage reference (Vdd) and the first inner circuit node (XBL); a comparator (17) having a first input terminal (-) connected to a comparison circuit node (Xrif), connected in turn to the first voltage reference (Vdd) by means of at least a second reference current (Iref) generator (15), as well as a second input terminal (+) connected to a circuit node (Xmat) connected in turn to the first inner circuit node (XBL), an output terminal (OUT) of the comparator (17) corresponding to an output terminal of the sensing circuit (10).
Advantageously according to the invention, the sensing circuit comprises also:
a cascode-configured bias circuit (13) inserted between the inner circuit node (XBL) and the matching circuit node (Xmat) and connected to a third voltage reference (Vref); and a current/voltage conversion stage (16) connected to the matching circuit node (Xmat) and to the comparison circuit node (Xrif), as well as to the second voltage reference (GND).
Abstract:
Described herein is a method for storing a datum in a first and a second memory cells of a nonvolatile memory. The storage method envisages programming the first and second memory cells in a differential way, by setting a first threshold voltage in the first memory cell and a second threshold voltage different from the first threshold voltage in the second memory cell, the difference between the threshold voltages of the two memory cells representing a datum stored in the memory cells themselves.
Abstract:
A content addressable memory cell (105) for a non-volatile Content Addressable Memory (100), including non-volatile storage means (S1,S2,S) for storing a content digit, a selection input (WL i ;WL i ,BLP j ) for selecting the memory cell, a search input for receiving a search digit (BLR j ,BLL j ), and a comparison circuit arrangement for comparing the search digit to the content digit and for driving a match output (ML i ) of the memory cell so as to signal a match between the content digit and the search digit. The non-volatile storage means include at least one Phase-Change Memory element (S1,S2,S) for storing in a non-volatile way the respective content digit.
Abstract:
The invention relates to a programming method of a multilevel memory cell able to store a plurality of bits in a plurality of levels (N), the method comprising at least a step of writing a logic value in the multilevel memory cell by setting one of the programming levels (LA) thereof, these levels being included in the plurality of levels (N), with respect to a reference level (LR) according to the symbol to be written and to a previous programming level. The writing step is repeated until a highest possible value (Lmax) for the levels (LS, LR) is reached. The invention relates also to a multilevel memory device comprising a plurality of multilevel memory cells organised into sectors, split into a plurality of data units (UD) whereon a programming operation is performed in parallel according to the method of the invention.