Abstract:
The invention relates to an analog-to-digital conversion method and relevant device, in high-density multilevel non-volatile memory devices. The method applies to multilevel memory cells comprising a floating gate transistor with drain and source terminals; the cell to be read is subjected to a reading operation by applying predetermined bias voltage values to its drain and source terminals, while to its drain terminal is applied a predetermined current value (Iref), and by measuring the value of its gate voltage (Vg). The method of the invention comprises a first conversion phase the most significant bits (MSB) contained in the memory cell, followed by a second conversion phase of the least significant bits (LSB). The first step is completed within a time gap (T1-T0) which corresponds to the rise transient of the gate voltage signal (Vg), while the second step is started at the end of the transient.
Abstract:
A semiconductor memory, particularly of the electrically programmable and erasable type such as a flash memory, comprises at least one two-dimensional array (SCT) of memory cells (MC) with a plurality of rows (row0-row511) and a plurality of columns (COL) of memory cells. The columns of the two-dimensional array are grouped in a plurality of packets (CP0-CP1), and the memory cells belonging to the columns of each packet are formed in a respective semiconductor region (4) with a first type of conductivity, this region (4) being distinct from the semiconductor regions (4) with the first type of conductivity in which the memory cells belonging to the columns of the remaining packets are formed. The semiconductor regions with the first type of conductivity divide the set of memory cells belonging to each row into a plurality of subsets of memory cells that constitute elemental memory units which can be modified individually. It is thus possible to produce memory units of very small dimensions (for example, bytes, words, or long words) which can be erased individually, without excessive overhead in terms of area.
Abstract:
This invention relates to a non-volatile memory device (20) with configurable row redundancy, comprising:
a non-volatile memory (11) comprising of at least one matrix (11') of memory cells and at least one matrix (11") of redundant memory cells, both organised into rows and columns; row and column decoding circuits (12,13); read and modify circuits for reading and modifying data stored in the memory cells; and at least one associative memory matrix (14), also organised into rows and columns, able to store the addresses of faulty rows, and control circuits for controlling the associative memory matrix.
The memory device (20) of this invention further comprises:
at least one circuit for recognising and comparing selected row addresses (ADr) with faulty row addresses (ADrr) contained in the associative memory matrix (14), such as to produce de-selection of the faulty row and selection of the corresponding redundant cell row in the event of a valid recognition; and at least one configuration register (17), also comprising a matrix of non-volatile memory cells, and associated control circuits.
Abstract:
The sensing circuits (30, 31, 32) comparing the current flowing in the cell with a plurality of reference currents are not identical to each other but differently amplify the compared currents. In particular, the sensing circuit (32) associated with the lowest reference current (I R3 ) amplifies (33b) the cell current more than the other sensing circuits (30, 31) and to the respective reference current (33c). The current dynamics is thereby increased and it is possible to keep the reading voltage low, since the inherent characteristic of the lowest reference current (I R3 ) may be very close to or directly superimposed on that of the immediately preceding memory cell current distribution (I M3 ), retaining the possibility of discriminating between the different logic levels.
Abstract:
The reading method comprises the steps of: supplying simultaneously two memory cells (F1, F2), both storing a respective unknown charge condition; generating two electrical quantities (Va, Vb), each correlated to a respective charge condition; comparing the two electrical quantities (Va, Vb) with each other; and generating a two-bit signal (01, 02) on the basis of the result of the comparison. The reading circuit comprises a two-input comparator (58) comprising two branches in parallel, each branch being connected to a respective memory cell by a current/voltage converter (41). Both the two-input comparator (58) and the current/voltage converter (41) comprise low threshold transistors (49, 50, 65-68).
Abstract:
The invention relates to a row decoding circuit for an electronic memory cell device, particularly in low supply voltage applications, being of the type adapted to boost, through at least one boost capacitor (Cboost), a read voltage to be applied to a memory column containing a memory cell to be read. The circuit is powered between a first supply voltage reference (Vpcx) and a second ground potential reference (GND), and comprises a hierarchic structure (13) of cascade connected inverters (15,16) and a circuit means of progressively raising the read voltage level dynamically. First means (Cboost0,D1) are provided for raising the read voltage level to a value equal to the supply voltage (Vpcx) plus a threshold voltage (Vtp), and second means (Cboost1,D2) are provided for raising the read voltage level to a value equal to the supply voltage (Vpcx) plus twice said threshold voltage (Vtp).
Abstract:
The invention relates to a method and a circuit for regulating a pulse synchronization signal (ATD) for the memory cell read phase in semiconductor integrated electronic memory devices. The pulse signal (ATD) is generated upon detection of a change in logic state of at least one of a plurality of address input terminals of the memory cells, so as to also generate an equalization signal (SAEQ) to a sense amplifier. The SAEQ pulse is blocked (STOP) upon the row voltage reaching a predetermined sufficient value to provide reliable reading. Advantageously, the pulse blocking is produced by a logic signal (STOP) activated upon a predetermined voltage value being exceeded during the overboost phase of the addressed memory row.