Abstract:
A non volatile memory device based on the realization of two distinct orders of main wordlines and bitlines, distinguishing a first order as main read wordlines (MWLR) and main read bitlines (MBLR) and a second order as main program wordlines (MWLP) and main program bitlines (MBLP), to each pair of main lines of which are associated a certain number of local lines, for example four lines, in each sector of subdivision of the array of memory cells, according to the criteria that are normally used in the so called hierarchical decoding structures of a memory whose array of elementary memory cells is subdivided in banks and/or sectors. Distinct decoders are associated to the two distinct orders of main wordlines and of main bitlines, the relative inputs of which are two address buses for reading and programming/erasing operations, respectively. A first address bus (RADD) coming from the address input pads of the device, through commonly used circuits, realizes a path for the read addresses to be input to main read wordlines decoders and to main read bitlines decoders. The input of the decoders of the main program wordlines and of the main program bitlines is, on the contrary, provided by an internal bus (PADD), generated by controllers of program and erase operations that control all the verification and load operations of data to be programmed.
Abstract:
The method for timing reading of a memory cell envisages supplying the memory cell (10) with a constant current (I) by means of a first capacitive element (23), integrating said current (I) in a time interval (Δt), and controlling the duration of the time interval (Δt) in such a way as to compensate for any deviations in the current (I) from a nominal value. In particular, a reference current (I R ) is supplied to a reference cell (101) by means of a second capacitive element (122); next, a first voltage (Var) present on the second capacitive element (122) is measured; finally, the memory cell (10) is deactivated when the first voltage (Var) is equal to a second voltage (Vref), which is constant.
Abstract:
The method for reading a memory cell includes supplying the cell (30) with a first charge quantity (ΔQb) through a capacitive integration element (39) and reintegrating the first charge quantity (ΔQb) through a plurality of second charge quantities (Qa) supplied alternately and in succession to the capacitive integration element (39). In a first embodiment, the second charge quantities (Qa) are initially stored in a plurality of capacitive charge-regeneration elements (37, 38) connected alternately and in succession to the capacitive integration element (39); the second charge quantities (Qa) are then shared between the capacitive integration element (39) and the capacitive charge-regeneration elements (37, 38).
Abstract:
The multilevel memory device (1) has a memory section containing cells (2a) that can be programmed with a predetermined number of levels greater than two, i.e. a multilevel array (2), and a memory section containing cells that can be programmed with two levels, i.e. a bilevel array (3). The multilevel array (2) is used for storing high density data, for which speed and reliability of reading are not essential, for example for storing the operation code of the system including the memory device. On the other hand, the bilevel array (3) is used for storing data for which high speed and reliability of reading is essential, such as the BIOS of personal computers, and the data to be stored in a cache memory. The circuitry parts dedicated to programming, writing of test instructions, and all the functions necessary for the operation of the memory device (1), are common to both arrays.
Abstract:
The memory device (21) has hierarchical sector decoding (24, 25). A plurality of groups of supply lines (28-32) is provided, one for each sector row, extending parallel to the sector rows. A plurality of switching stages (35) are each connected between a respective sector (15) and a respective group of supply lines (28-32); the switching stages (35) connected to sectors (15) arranged on a same column are controlled by same control signals (S0, S1) supplied on control lines (40) extending parallel to the columns of sectors. For biasing the sectors, modification voltages (NW, SB, V NEG ) are sent to at least one selected group of biasing lines (28-32), and control signals (SO, S1) are sent to the switching stages connected to a selected sector column.
Abstract:
The present invention relates to a semiconductor memory device with an improved address signal generator. The memory device comprises an array of memory elements (10), first decoding circuit means (8,15) for decoding a first set of address signals (7,14) for the selection of said memory elements, and second circuit means (4) for the generation internally to the memory of a sequence of values for said address signals (3,11). The second circuit means (4) generates said sequence so that successive values in the sequence differ for the logic state of only one of said address signals (3,11).