Row Decoder for a Flash-Eeprom Memory Device With the Possibility of Selective Erasing of a Sub-Group of Rows of a Sector

The present invention relates to a row decoder for a flash-EEPROM memory device with the possibility of selective erasing of a subgroup of rows of a sector.

As is known, flash memory devices comprise an array of cells arranged in rows and columns, wherein word lines connect the gate terminals of the cells arranged on a single row, and bit lines connect the drain terminals of the cells arranged on a single column. Individual memory array rows are thus addressed by a row decoder receiving an input coded address.

In flash memory devices, the memory array is divided into different sectors comprising cell blocks; in these devices the rows decoders permit reading and programming of individual cells of a sector, and only erasing of all the cells in the sector.

Such memory arrays are currently erased by applying a negative voltage to the gate terminals of the cells, for example, -10V, taking the source terminal to a positive voltage, for example, +5V, and leaving the drain terminal floating.

The method of erasing by sectors requires special measures when allocating the memory, since each time a datum stored in a sector is modified, it is necessary to erase and rewrite completely all the sector, and each sector has a separator device (which infer alia permits independent erasing from the other sectors) which has a large size (of approximately one hundred times the height of an array row). Consequently the design usually requires a compromise between the requirement to section the memory blocks as far as possible in order to erase only the components which need to, and not to increase excessively the total dimensions of the memory, owing to the presence of a large number of separators. However, this compromise does not always make it possible to obtain the best results.

An object of the present invention is thus to provide a row decoder allowing selective erasing of one or a plurality of rows of a sector of a memory array of a flash memory device and reducing the above-described compromises.

In one embodiment of the invention, the row decoder includes a decoding stage having at least one decoding circuit, a first line set to a first reference potential, a second line set to a second reference potential switching between said first reference potential and an erase potential lower than the first reference potential wherein said decoding circuit includes a first path and a second path arranged in parallel to one another and connected respectively to said first line and said second line, and a first selection circuit selectively enabling said first and second paths.

For the comprehension of the present invention, a preferred embodiment is now described, purely by way of non-limiting example, with reference to the attached drawings.

FIG. 1 shows the architecture of a flash memory device including a row decoder according to the present invention.

FIGS. 2-6 show simplified electric circuit diagrams relating to the row decoder of the present invention.

In FIG. 1, the numeral 1 indicates as a whole a memory device of the flash type which uses the negative gate erasing technique. The memory device 1 includes a memory array 2 having a plurality of cells 4 arranged in rows and columns; a supply stage 6 generating supply voltages V.sub.CC, V.sub.PP, V.sub.ERP, V.sub.ERN, V.sub.GND ; a logic stage 8 generating address signals A<0-n>, a program enabling signal EHN and an erase enabling signal EEH; a pre-decoding stage 10 receiving at the input the address signals A<0-n> and the program and erase enabling signals EHN and EEH and generating at the output pre-decoding signals Lx<0-x>, Ly<0-y>, Lz<0-z> and PFN<0-p>, in which x, y, z and p are pre-determined numbers; and a final decoding stage 12 receiving at the input the pre-decoding signals Lx, Ly, Lz and PFN, and generating at the output bias signals R<0-r>, r being the number of rows, each output bias signal being supplied to a respective row of the memory array 2.

Pre-decoding stage 10 includes a plurality of pre-decoding circuits 14 having the same structure, one for each pre-decoding signal Lx, Ly and Lz, and a plurality of pre-decoding circuits 15 having the same structure, one for each predecoding signal PFN; the circuits 14 and 15 receive the program and erase enabling signals EHN and EEH, but different address signals A. Similarly, the final decoding stage 12 includes a plurality of final decoding circuits 16 having the same structure, but receiving different combinations of the pre-decoding signals Lx, Ly, Lz and PFN.

FIG. 2 shows the block diagram of a pre-decoding circuit 14 receiving at the input an address signal, for example A<0>, and the program and erase enabling signals EHN and EEH, and generating at the output a pre-decoding signal, for example Lx<0>.

The pre-decoding circuit 14 includes a first supply line 20 set to a voltage V.sub.PC which, during the reading, is equivalent to a supply voltage V.sub.CC of the memory device (i.e., 3-5 V), during programming is equivalent to a program voltage V.sub.PP, greater than V.sub.CC (for example, 12 V), and during erasing is equivalent to a first erase voltage V.sub.ERP smaller than V.sub.CC (for example, 2 V); a second supply line 21 set to a voltage V.sub.NEG which, during reading and programming, is equivalent to a ground voltage V.sub.GND (typically 0 V) and during erasing is equivalent to a second erase voltage V.sub.ERN smaller than the ground voltage V.sub.GND (for example, -8 V); a first drive circuit 22 which has first and second input terminals 24, 26 which receive address signal A<0> and program enabling signal EHN respectively, and an output terminal 28 supplying a first drive signal DSH; a second drive circuit 32 which has first and second input terminals 34, 36 receiving address signal A<0> and erase enabling signal EEH respectively, and an output terminal 38 supplying a second drive signal DSD; and an output circuit 40 of the push-pull type which comprises a pull-up PMOS transistor 42 and a pull-down NMOS transistor 44, which have gate terminals receiving respectively the first and second drive signal DSH, DSD, drain terminals connected to one another (output node 46, to which the pre-decoding signal Lx<0> is supplied), and source terminals connected respectively to the first and the second supply line 20, 21.

The pre-decoding circuits 15 have the same structure as the pre-decoding circuits 14, except for the fact that the output node 46 of circuits 15 (which supplies signals P<i>) is connected to an EXNOR gate to output signals PFN<i>, as explained in greater detail hereinafter with reference to FIG. 6.

FIG. 3 shows the electric diagram of the first drive circuit 22. As shown, first drive circuit 22 includes a first and a second path 50, 52 separated from one another and disposed in parallel, and a selection circuit 54 for selectively enabling the two paths 50, 52.

The first path 50 is used for reading, i.e., when the voltage V.sub.PC of the first supply line 20 is equivalent to voltage V.sub.CC, to supply voltage V.sub.CC or voltage V.sub.GND to the gate terminal of PMOS transistor 42 of output circuit 40, as specified by the address signal A<0>.

The first path 50 includes substantially a CMOS switch 58 and an inverter 60 interposed between first input terminal 24 and CMOS switch 58. In particular, CMOS switch 58 includes a PMOS transistor 64 and an NMOS transistor 66 having source terminals connected to one another and to the output of inverter 60, drain terminals connected to one another and to the output terminal 28, and gate terminals respectively connected to an output terminal of the selection circuit 54 and the second input terminal 26.

The second path 52 is used when voltage V.sub.PC is different from V.sub.CC and specifically during programming, when voltage V.sub.PC of the first supply line 20 is equivalent to voltage V.sub.PP, and during erasing, when voltage V.sub.PC is equivalent to voltage V.sub.FRP, to supply voltage V.sub.PC or voltage V.sub.GND to the gate terminal of PMOS transistor 42 of the output circuit 40, as specified by the address signal A<0>.

The second path 52 comprises substantially a voltage shifter 70 and a NOR gate 72. In particular, voltage shifter 70 includes two NMOS transistors 74, 76 and three PMOS transistors 78, 80, 82.

In detail, NMOS transistor 74 has a source terminal connected to ground (defining a third supply line 71 set to the voltage V.sub.GND), a drain terminal connected to the drain terminal of PMOS transistor 78, and a gate terminal connected to the source terminal of NMOS transistor 66; the NMOS transistor 76 has a source terminal connected to ground 71, a drain terminal connected to the drain terminal of PMOS transistor 80 and to the output terminal 28, and a gate terminal connected to an output terminal of NOR gate 72; PMOS transistor 78 has a source terminal connected to the drain terminal of PMOS transistor 82 and a gate terminal connected to the drain terminal of PMOS transistor 80; PMOS transistor 80 has a source terminal connected to the drain terminal of PMOS transistor 82 and a gate terminal connected to the drain terminal of PMOS transistor 78; and PMOS transistor 82 has a source terminal connected to the first supply line 20 and a gate terminal connected to the second input terminal 26.

Finally NOR gate 72 has a first input terminal connected to the source terminal of NMOS transistor 66 and a second input terminal connected directly to the second output terminal 26.

The selection circuit 54 comprises a voltage shifter 84 formed by two NMOS transistors 90, 92, three PMOS transistors 94, 96, 98 and an inverter 86.

In particular, NMOS transistor 90 has a source terminal connected to ground, a drain terminal connected to the drain terminal of PMOS transistor 94 (node 93) and to the gate terminal of PMOS transistor 64 of CMOS switch 58, and a gate terminal connected to the second input terminal 26; NMOS transistor 92 has a source terminal connected to ground, a drain terminal connected to the drain terminal of PMOS transistor 96, and a gate terminal connected to the output terminal of inverter 86; PMOS transistor 94 has a source terminal connected to the drain terminal of PMOS transistor 98 and a gate terminal connected to the drain terminal of PMOS transistor 96; PMOS transistor 96 has a source terminal connected to the drain terminal of PMOS transistor 98 and a gate terminal connected to the drain terminal of PMOS transistor 94; and PMOS transistor 98 has a source terminal connected to the first supply line 20 and a gate terminal connected to ground 71. Finally the inverter 86 has an input terminal connected to the second input terminal 26.

FIG. 4 shows the electric diagram of the second drive circuit 32, which, similarly to first drive circuit 22, includes a first path 100 and a second path 102, separated from one another and arranged in parallel, and a selection circuit 104 for selectively enabling the two paths 100, 102.

The first path 100 is used for reading and programming, i.e., when voltage V.sub.NEG of the second supply line 21 is equivalent to ground voltage V.sub.GND, to supply voltage V.sub.CC or voltage V.sub.GND to the gate terminal of NMOS transistor 44 of output circuit 40, as specified by address signal A<0>.

The first path 100 includes substantially a CMOS switch 108 and an inverter 110 interposed between the first input terminal 34 and CMOS switch 108. In particular, CMOS switch 108 includes a PMOS transistor 114 and an NMOS transistor 116 which have source terminals connected to one another and to the output terminal of the inverter 110, drain terminals connected to one another and to output terminal 38 and gate terminals connected respectively to the second input terminal 36 and to an output terminal of selection circuit 104, as described in greater detail hereinafter.

The second path 102 is used during erasing, i.e., when voltage V.sub.NEG of the second supply line 21 is equivalent to second erase voltage V.sub.ERN, so as to supply second erase voltage V.sub.ERN or voltage V.sub.CC to the gate terminal of NMOS transistor 44 of output circuit 40, as specified by the address signal A<0>.

The second path 102 includes substantially a voltage shifter 120, a NAND gate 122 and an inverter 123.

In particular, voltage shifter 120 includes three NMOS transistors 124, 126, 128 and two PMOS transistors 130, 132. In detail, NMOS transistor 124 has a source terminal connected to second supply line 21, a drain terminal connected to the source terminals of NMOS transistors 128, 126, and a gate terminal connected to the second input terminal 36; NMOS transistor 126 has a drain terminal connected to the drain terminal of PMOS transistor 130 and to the gate terminal of NMOS transistor 128, and a gate terminal connected to the drain terminal of NMOS transistor 128; NMOS transistor 128 has a drain terminal connected to the drain terminal of PMOS transistor 132 and to the gate terminal of NMOS transistor 126, and a gate terminal connected to the drain terminal of NMOS transistor 126; PMOS transistor 130 has a source terminal connected to a fourth supply line 134 set to voltage V.sub.CC, and a gate terminal connected to an output terminal of inverter 123, which in turn has an input terminal connected to the first input terminal 34; and PMOS transistor 132 has a source terminal connected to the fourth supply line 134, and a gate terminal connected to an output terminal of NAND gate 122. Finally NAND gate 122 has a first input terminal connected to the output terminal of inverter 123, and a second input terminal connected directly to the second input terminal 36.

Selection circuit 104 includes a voltage shifter 136 having an inverter 138 and three NMOS transistors 140, 142, 144 and two PMOS transistors 146, 148, similar to and connected like NMOS transistors 124, 126 and 128 and PMOS transistors 130, 132, with the only differences that NMOS transistor 140 has a gate terminal connected to the fourth supply line 134, PMOS transistor 146 has a gate terminal connected to the second input terminal 36, and PMOS transistor 148 has a gate terminal connected to an output terminal of inverter 138. The latter has its input terminal connected to second input terminal 36.

The pre-decoding circuit 14 functions as follows.

As already stated, during reading voltage V.sub.PC is equal to voltage V.sub.CC, second supply line 21 is set to V.sub.GND, program enabling signal EHN is high (set to V.sub.CC) and the erase enabling signal EEH is low (set to V.sub.GND).

In this condition, in first drive circuit 22 (FIG. 3), NMOS transistor 90 is switched on, node 93 is low and CMOS switch 58 is enabled. In addition, PMOS transistor 82 is switched off, and thus level shifter 70 is disabled.

Thereby, first drive signal DSH is generated by the first path 50 and is inverted with respect to address signal A<0>.

In addition, in second drive circuit 32 (FIG. 4), because of the low logic level of erase enabling signal EEH, PMOS transistor 146 is on, node 145 is high and CMOS switch 108 is enabled. On the other hand, NMOS transistor 124 is off and thus level shifter 120 is disabled.

Thereby, second drive signal DSD is generated via the first path 100 and is inverted with respect to address signal A<0>.

Consequently, during reading, both NMOS and PMOS transistors 42, 44 (FIG. 2) are driven by a drive signal which is inverted with respect to A<0>, and thus form an inverter of conventional type, wherein output terminal 46 of pre-decoding circuit 14 is connected selectively to first and second supply line 20, 21 by NMOS and PMOS transistors 42, 44, thus assuming logic levels which are set by voltages V.sub.CC.sup.- V.sub.GND.

As already stated, during programming first supply line 20 assumes value V.sub.PP, second supply line 21 is set to V.sub.GND, and program and erase enabling signals EHN and EEH are at a low logic level (V.sub.GND).

In this condition, in first drive circuit 22 (FIG. 3) NMOS transistor 90 is off, NMOS transistor 92 is on, as is PMOS transistor 94, node 93 is high (set to V.sub.PP) and CMOS switch 58 is disabled. In addition, PMOS transistor 82 is on, and thus level shifter 70 is enabled.

Thereby, first drive signal DSH is generated via the second path 52 and has a low logic level (V.sub.GND) or a high logic level (V.sub.PP) according to the logic level of the address signal A<0>. In practice, first drive signal DSH is inverted with respect to address signal A<0>, and has a high translated logic level.

In addition, because of the low logic level of erase enabling signal EEH, second drive circuit 32 functions in the same manner as described above for reading, with CMOS switch 108 enabled and level shifter 120 disabled, thus second drive signal DSD is generated via first path 100.

Output terminal 46 of pre-decoding circuit 14 (FIG. 2) thus assumes logic levels which are set by voltages V.sub.PP.sup.- V.sub.GND.

As already stated, during erasing, first supply line 20 is set to V.sub.ERP, second supply line 21 is set to V.sub.ERN, and program enabling signal EHN is at a low logic level (V.sub.GND) and erase enabling signal EEH is at a high logic level (V.sub.CC).

In this condition, first drive circuit 22 (FIG. 3) operates as described above for the programming step, with CMOS switch 58 disabled and level shifter 70 enabled, thus first drive signal DSH is generated via the second path 52, and has a low logic level (V.sub.GND) or a high logic level (V.sub.ERP) according to the logic level of address signal A<0>.

In addition, the second drive circuit 32 (FIG. 4), owing to the high logic level of erase enabling signal EEH, PMOS transistor 146 is off and PMOS transistor 148 is on; consequently NMOS transistor 142 is on and node 145 has a low logic level (equal to V.sub.ERN); CMOS switch 108 is thus disabled. In addition, NMOS transistor 124 is on and thus the level shifter 120 is enabled.

Thereby, second drive signal DSD is generated via the second path 102 and has a low logic level (V.sub.ERN) or a high logic level (V.sub.CC) according to the address signal A<0>. In practice, second drive signal DSD is inverted with respect to address signal A<0> and has low translated logic levels.

Consequently during erasing, the output terminal 46 of pre-decoding circuit 14 assumes logic levels defined by voltages V.sub.ERP -V.sub.ERN. This biasing is particularly useful for limiting the stress on final inverters of the final decoding circuit 16, as explained in greater detail hereinafter. As an alternative, during erasing also the first supply line is set to V.sub.CC, in which case first drive circuit 22 operates as described for reading, generating first drive signal DSH via CMOS switch 58.

FIG. 5 shows schematically a portion of final decoding circuit 16, for 25 driving two rows of the memory array, for example rows 0 and 1, and thus receiving as input pre-decoding signals Lx<0>, Ly<0>, Lz<0>, PFN<0>, PFN<1> and erase enabling signal EEH, and generating bias signals R<0>, R<1> as outputs.

As shown in FIG. 5, an NAND gate 150 has three input terminals receiving respectively pre-decoding signals Lx<0>, Ly<0>, Lz<0> and an output terminal supplying a signal EGN and connected to a first input terminal of a NAND gate 152; the latter has a second input terminal receiving erase enabling signal EEH and an output terminal connected to the input terminal of an inverter 162. Output 163 of inverter 162 is connected to a plurality of drive branches 153.0, 153.1, . . . each for a respective row, enabled by respective pre-decoding signals PFN<i> (and in the example in FIG. 5 by signals PFN<0> and PFN<1>).

In particular, each drive branch includes a NOR gate 156.0, 156.1, . . . having a first input connected to the output of inverter 162, and a second input connected to an output of a respective NOR gate 154.0, 154. 1, . . . The NOR gates 154.0, 154.1, . . . , in turn respectively have a first input terminal connected to the output of NAND gate 150 and a second input terminal receiving a respective signal PFN<i>.

Pre-decoding signals PFN<i> are generated by respective EXNOR gates 172, as shown in FIG. 6. In detail, EXNOR gate 172 has a first input terminal receiving erase enabling signal EEH and a second input terminal receiving a respective signal P<i> present on node 46 of pre-decoding circuits 15 similarly structured to the pre-decoding circuit 14 of FIG. 2, such as to implement in practice the function: ##EQU1##

With reference again to FIG. 5, the outputs of NOR gates 156.0, 156. 1, . . . have output terminals connected to input terminals of inverters 164.0, 164.1, respectively, . . . , which supply at their own outputs the bias signals R<0>, R<1>, . . . to a respective row.

In final decoding circuit in FIG. 5, all logic gates are supplied via the first and the second supply line 20, 21, and thus between V.sub.PC, and V.sub.NEG.

Final decoding circuit 16 operates as follows.

During reading and programming, erase enabling signal EEH is low and node 163 is low, irrespective of whether or not the group of rows biased by final decoding circuit 16 has or has not been selected by pre-decoding signals Lx<0>, Ly<0>, Lz<0>. Consequently, bias signals R<0>, R<1>, . . . assume a high or low logic value according to the logic value of pre-decoding signals PFN<0> and PFN<1>, . . . ; thereby the address of a single row or a plurality of rows within the group can be obtained.

During erasing, erase enabling signal EEH is high and node 163 is low only when the group of rows associated with pre-decoding signals Lx<0>, Ly<0>, L,z<0> is selected by a high logic value of pre-decoding signals.

In this situation, bias signals R<0>, R<l>, . . . assume a high logic value (equal to V.sub.ERP) when the respective pre-decoding signal PFN<0>, PFN<1>, . . . has a low logic value, and vice versa they have a low logic value (equal to V.sub.ERN) when predecoding signals PFN<0>, PFN<1>, . . . are high.

The groups of rows which are not selected are biased to voltage V.sub.ERP irrespective of the value of pre-decoding signals PFN<i>.

Consequently, in the erasing step it is possible to address also a single row, making the respective pre-decoding signal PFN assume a high logic level. However the possibility still exists of erasing the entire sector, simply by making all pre-decoding signals LIx, Ly, Lz and PFN assume a high logic level.

Thereby, memory device 1 has greater operative flexibility, and memory allocation control is more efficiently than in known memory devices, with the disadvantage of a slight increase of area occupied by the decoding circuits. However a larger area increase is obtained for the pre-decoding circuits 14, I5, which nevertheless, since arranged outside the memory array, are less critical in this respect.

The use of four separate paths allows very short reading access times. In fact the direct path through CMOS switches (used during reading) provides very fast propagation times, since feedback structures are not used for most of the uses of the memory. On the other hand, the paths with the shifters, allowing the voltage translation necessary for programming and erasing, are slower, but in fact are used less frequently than direct paths.

In addition, in row decoders of known flash memory devices, NMOS transistors 42, 44 define an output inverter which, during switching, has consumption caused by crowbar current; on the other hand, in the present decoder driving of these NMOS transistors 42, 44 is separate and permits elimination of this consumption.

The described decoder can advantageously be used to replace flexible memory components (floppy discs) for computers, since it can reproduce their internal organization into sectors with the advantage of eliminating mechanical drive units.

Finally, it is apparent that changes and variants can be made to the row decoder described and illustrated here, without departing from the scope of the present invention.