Semiconductor device with a voltage regulator

This application is based on and claims the benefit of priority from the prior Japanese Patent Application No. 2001-052588, filed on Feb. 27, 2001, the entire contents of which are incorporated herein by reference.

1. Field of the Invention

This invention relates generally to a semiconductor device having a voltage regulator for output of a regulated voltage that is variable in response to an output voltage of an internal voltage generation circuit.

2. Description of the Related Art

Prior known electrically rewritable non-volatile semiconductor memory devices include electrically erasable programmable read-only memory (EEPROM) chips, also known as “Flash” memories in the semiconductor device art. Currently available Flash EEPROM chips come with an array of rows and columns of nonvolatile memory cells, each of which is typically formed of a metal insulator semiconductor field effect transistor (MISFET) of the so-called “stack gate” structure with a floating gate and a control gate being stacked above a semiconductor chip substrate. Each stack-gate MISFET memory cell stores therein a digital binary data bit of a logic “0” or “1” in accordance with the charge accumulation state of its floating gate. One example is that a storage data bit is assumed to be a logic “0” in the event that the floating gate storing thereon electrons is high in threshold voltage while letting the bit be a logic “1” when the floating gate releases the electrons and thus is low in threshold voltage.

Flash memories include EEPROMs of the NOR type. In the case of such NOR-EEPROMs, a memory cell array is arranged so that a respective one of rows of memory cells is associated with a corresponding one of parallel bit lines in a manner such that drains of these cells are connected together or “common-coupled” to the bit line whereas each column of memory cells is common-coupled at control gates to a corresponding one of parallel word lines that cross over the bitlines. A data write operation is performed after having erased all the cells of the memory array at a time, known as “all-at-a-time” or “all-at-once” erase among those skilled in the art. The all-at-once erase is achievable in a way as follows. Firstly all the wordlines of the memory cell array are applied a voltage of the negative polarity having a specific potential level—typically, −7 volts (V), or more or less. Then, apply a positive voltage of about +10V to a common source, causing electrons presently residing on floating gates to release toward the substrate side by Fowler-Nordheim (F-N) tunneling effects. Whereby, all the memory cells are thus set in the erase state of data “1.”

Data writing on a per-cell basis is done by applying a write voltage of about 10V to a word line being presently selected from among the wordlines and then giving to the selected bitline either one of an on-chip power supply voltage Vdd and a source or “ground” voltage Vss in a way depending on whether the data being written is a logic “0” or “1.” This voltage application results in that in a cell to which logic “0” data is given, “hot” electrons are injected into the floating gate thereof causing its threshold voltage to shift or “offset” in a positive direction. In the case of logic “1” data, no appreciable threshold voltage changes occur.

Data read is done by giving a read voltage to a selected wordline and then detecting the presence or absence of a cell current flow.

In the above operations, the data write is typically combined with a verify-read operation for confirmation or verification of the resulting write state. More specifically, after having written data under application of the write voltage, the verify-read operation is done. Repeated execution of such write voltage application and its following verify-read session forces the threshold voltage of a written memory cell to finally fall within a prespecified distribution range. Similarly in the case of data erase, recurrent execution of erase voltage application and its following erase-verify operation enables the threshold voltage of an erased cell to finally fall within a specified distribution range.

For adequate control of the above-noted operations in the write and erase modes, a need is felt to supply a stabilized and regulated voltage with its potential optimized on a per-mode basis. A presently available approach to generating such a regulated voltage is to employ a voltage regulator with an ability to generate a plurality of types of regulated voltages based on an output voltage of potential rise-up or “booster” circuitry for use as an internal power supply. A scheme using this approach is disclosed, for example, in J. F. Dickson, “On-Chip High-Voltage Generation in NMOS Integrated Circuits Using an Improved Voltage Multiplier Technique,” IEEE Journal of Solid-State Circuits, Vol. SC-11, No. 3, June 1976 at pp. 374-378. Another technique is taught from A. Umezawa et al., “A 5V-Only Operation 0.6 μm Flash EEPROM with Row Decoder Scheme in Triple-Well Structure,” IEEE J. Solid-State Circuits, Vol. 27, No. 11, November 1992, pp. 1540-1546.

See FIG. 18 . This diagram shows an arrangement of one prior art voltage regulator adaptable for use with Flash EEPROM chips. This regulator shown herein includes a driver unit 1 . The driver 1 has a serial connection of MOS transistors: a PMOS transistor QP 2 for potential pull-up drive, and an NMOS transistor QN 2 for pull-down drive. The pull-up PMOS transistor QP 2 has a source, to which a potentially increased or “boosted” voltage Vpp as output from a potential booster circuit (not shown) is supplied. The pull-down NMOS transistor QN 2 has its source which is coupled to ground. These transistors QP 2 , QN 2 are coupled together at a connection node N 0 , which is for use as an output terminal of a regulated voltage Vreg.

An operational amplifier OP 1 is provided for controlling a gate of the pull-down NMOS transistor QN 2 . It is an ensemble of operational amplifier OP 2 and NMOS transistor QN 1 plus PMOS transistor QP 1 that controls a gate of the pull-up PMOS transistor QP 2 . The NMOS transistor QN 1 is operable under control of an output of the op-amp OP 2 . The PMOS transistor QP 1 functions as a current source load of NMOS transistor QN 1 . PMOS transistors QP 1 , QP 2 make up a current mirror circuit.

The regulator of FIG. 18 has its output node N 0 , at which a voltage division circuit 2 is provided. This voltage divider circuit 2 includes a serial combination of resistors R 1 to R 3 and a switch element formed of an NMOS transistor QN 3 . Connected to a connection node N 2 of the resistors R 2 -R 3 is an NMOS transistor QN 4 which operates under control of a write-use control signal PROG for coupling the node N 2 to ground. NMOS transistor QN 3 has its gate which is to be controlled by a verify-read control signal VRFY. When NMOS transistor QN 3 is driven to turn on, a connection node N 3 of resistor R 3 and NMOS transistor QN 3 is grounded.

One of the op-amps OP 1 -OP 2 —here, opamp OP 2 —is given a reference voltage Vref at its non-inverting input terminal, with a voltage potential at node N 1 being fed back to an inverting input terminal thereof. The other opamp OP 1 is such that the reference voltage Vref is given to its inverting input terminal while a potential at node N 1 is fed back to its non-inverting input terminal.

This regulator experiences application of feedback control in a way such that the node N 1 of voltage divider 2 becomes equal in potential to the reference voltage Vref being supplied to the opamps OP 1 - 2 whereby it outputs a potentially regulated voltage Vreg with potential “trackability” to a change in boosted output voltage Vpp. More specifically, while the potentially divided output voltage that is obtainable at the node N 1 of voltage divider 2 stays lower than the reference voltage Vref, the opamp OP 1 derives its output voltage of high level causing NMOS transistor QN 1 to turn on whereas an output of opamp OP 1 is at low level letting NMOS transistor QN 2 turn off. This results in the current mirror circuit of PMOS transistors QP 1 - 2 permitting a pull-up current Iup to flow in pull-up PMOS transistor QP 2 , thereby forcing the regulated voltage Vreg obtained at output node N 0 to increase in potential.

When the regulated voltage Vreg potentially rises up causing the voltage at voltage divider node N 1 to become higher than the reference voltage Vref, the opamp OP 2 's output drops down at low level, letting NMOS transistor QN 1 turn off. Simultaneously the opamp OP 1 's output becomes high level, causing NMOS transistor QN 2 to turn on. This precludes flow of the pull-up current Iup while alternatively allowing a pull-down current Idn to flow in pull-down NMOS transistor QN 2 . This results in a likewise decrease in the regulated voltage Vreg. With the above-stated feedback control operation, there is obtainable the regulated voltage Vref that is stabilized at a preselected potential level lower than the boosted output voltage Vpp.

The stabilization level of the regulated voltage Vreg is variable at different levels depending upon whether each of the write control signal PROG and verify-read control signal VRFY is presently set at either “High” (“H”) level or “Low” (L) level. This can be said because the voltage divider 2 must differ in potential division ratio in light of the fact that when the write control signal PROG stays at “H” and verify-read control signal VRFY is at “L,” node N 2 is grounded and, alternatively, when the former is at “L” and the latter is at “H,” node N 3 is grounded.

Practically, a write operation of an EEPROM chip using the FIG. 18 circuitry is as follows. When the write control signal PROG potentially goes high to reach “H” level, the boosted output voltage Vpp begins to rise up in potential. This potential increase permits a pull-up current Iup to flow in pull-up PMOS transistor QP 2 . Thus the regulated voltage Vreg at regulated voltage output node N 0 behaves to potentially increase with an increase in the boosted voltage Vpp—say, the former “tracks” the latter. The required write voltage is thus obtained, which is well stabilized at a level of Vreg=7V, for example. After elapse of a “fixed” length of time, the write control signal PROG goes low to “L” level and the verify control signal VRFY goes high at “H” level. This forces node N 3 to be grounded. Whereby, the regulated voltage Vreg is such that a lower level becomes a point of stabilization. Thus a verify-read voltage of for example about Vreg=5V is obtained.

With the voltage regulator shown in FIG. 18, in cases where the opamps OP 1 - 2 have “ideal” characteristics with no appreciable input offsets, the resulting pull-up and pull-down currents Iup, Idn—these are derived from the driver 1 's PMOS transistor QP 2 and NMOS transistor QN 2 respectively—vary with a change in potentially divided output voltage VN 1 of the divider circuit node N 1 in a pattern as demonstrated in FIG. 19 A. The current-versus-voltage characteristic of FIG. 19A indicates the state that a penetration current Ion occurring due to simultaneous turn-on of these transistors QP 2 , QN 2 stays less. In contrast, when opamps OP 1 - 2 are relatively large in input offset, a large penetration current Ion can flow therein as shown in a characteristic diagram of FIG. 19 B.

This penetration current would result in an increase in power consumption or dissipation of integrated circuits concerned. Another problem faced with the prior art is that the flow of larger penetration current in booster circuitry badly behaves to limit the performance of the booster, i.e. voltage boost ability or “boostability,” in view of the fact that most booster circuits for generation of the boosted voltage Vpp are less in current supplying abilities. The limitation to boostability in turn leads to an increase in time as taken to create the boosted voltage Vpp and regulated voltage Vreg with potential trackability thereto. In the worst case, any desired regulated voltages will no longer be obtainable in any way.

A semiconductor device in accordance with one aspect of this invention has a voltage regulator for outputting a regulated voltage potentially varying in response to an output voltage of an internal voltage generation circuit. The voltage regulator includes: a driver including a pull-up transistor and a pull-down transistor serially connected between an output node of the internal voltage generation circuit and a reference potential terminal, for outputting a regulated voltage at its regulated voltage output node corresponding to the connection node of the pull-up and pull-down transistors; a voltage divider circuit configured to potentially subdivide the regulated voltage output at the regulated voltage output node; a first operational amplifier configured to control current drivability of the pull-down transistor in accordance with a difference between a first reference voltage and a divided output voltage of said voltage divider circuit; and a second operational amplifier configured to control current drivability of the pull-up transistor in accordance with a difference between a second reference voltage and the divided output voltage of the voltage divider circuit in such a way as to vary in a reverse direction to the current drivability of the pull-down transistor.

FIG. 1 is a diagram showing a configuration of an EEPROM device in accordance with an embodiment of this invention.

FIG. 2 is a diagram showing a configuration of a memory cell array of the EEPROM.

FIG. 3 is a diagram showing a structure of a memory cell of the EEPROM.

FIG. 4 is a diagram showing a voltage relationship in each operation mode of the EEPROM.

FIG. 5 is a graph showing a data distribution status of the EEPROM.

FIG. 6 is a diagram showing a control signal relation at part of an internal voltage generation circuit and a regulator of FIG. 1 .

FIG. 7 is a diagram showing waveforms of several major voltage signals in a write cycle of the same.

FIG. 8 is a diagram showing a practically reduced configuration of the internal voltage generation circuit of the same EEPROM.

FIG. 9 is a diagram showing a practical arrangement of the regulator of the EEPROM.

FIG. 10 is a diagram showing a configuration of an operational amplifier as used in the regulator of FIG. 9 .

FIG. 11 is a graph showing a current-versus-voltage characteristic of the regulator of FIG. 9 .

FIG. 12 is a diagram showing a configuration of a reference voltage generation circuit of FIG. 9 .

FIG. 13 is a diagram showing another available configuration of the reference voltage generation circuit of FIG. 9 .

FIG. 14 is a diagram showing an exemplary configuration of the reference voltage generation circuit of FIG. 12 .

FIG. 15 is a diagram showing another configuration example of the reference voltage generation circuit of FIG. 12 .

FIGS. 16A to 16 D are diagrams each showing a configuration example of a second reference voltage generator unit of FIG. 13 .

FIG. 17 is a diagram showing a configuration of a voltage regulator in accordance with another embodiment of the invention.

FIG. 18 is a diagram showing a configuration of one typical prior known regulator.

FIGS. 19A-19B are graphs each showing a current-voltage characteristic of the prior art regulator.

Referring now to FIG. 1, there is shown an overall configuration of a “Flash” EEPROM chip of the NOR type, which incorporates the principles of this invention. This Flash NOR-EEPROM includes an array 11 of rows and columns of memory cells. As shown in FIG. 2, this memory cell array 11 is constituted from a plurality of nonvolatile memory cells MC as organized into a matrix form. As shown in FIG. 3, each memory cell MC is formed of a metal oxide semiconductor field effect transistor (MOSFET) of the stacked gate structure type, with a floating gate 33 and a control gate 34 being stacked above a silicon substrate 30 . Drains 35 of a column of memory cells MC are connected in common to a corresponding one of parallel bit lines BL whereas control gates 34 of a row of memory cells MC are commonly coupled to one of parallel word lines WL crossing over the bitlines BL. Their sources 36 are coupled to a common source line SL.

As shown in FIG. 1, the bitlines BL of the memory cell array 11 are connected through a column gate circuit 12 to a sense amplifier circuit 14 for sensing data being read and a data latch circuit 15 for holding to-be-written data therein. The sense amp circuit 14 and data latch circuit 15 are connected via a data line 24 to an input/output (I/O) buffer 16 .

A row-decoder/word-line driver 13 is provided for performing selection and driving of the wordlines of memory cell array 11 . Any external address will be once held in an address buffer 17 and then supplied to a column decoder 18 and the row-decoder/wordline driver 13 .

An internal voltage generation circuit 21 is the one that is operable to generate a variety of kinds of control voltages required, including but not limited to a write voltage to be supplied to a presently selected control gate line of the memory cell array 11 during writing and an erase voltage being supplied during data erasing to a well region in which the memory cell array 11 is formed. Practically the internal voltage generator circuit 21 is a voltage rise-up circuit for potentially increasing or “boosting” the power supply voltage used to thereby produce a potentially boosted voltage Vpp for use during data writing, which will then be passed to a voltage regulator 23 . Also input to this voltage regulator 23 is a reference voltage as supplied from a reference voltage generation circuit 22 . The voltage regulator 23 is operatively responsive to receipt of a control signal incoming from a controller 20 for generating and issuing a regulated voltage, which is variable in potential with a change in the boosted voltage Vpp in such a way as to keep track of the voltage Vpp while having a different level in accordance with an operation mode. This regulated voltage is supplied to the row-decoder/wordline driver 13 .

The controller 20 is responsive to receipt of a command as taken or “accommodated” into a command register 19 for generating and issuing control signals as used to control several operations, typically including a data write operation and its following verify-read operation along with a data erase operation and its following verify-read operation.

FIG. 4 shows a voltage relation in each operation mode of the EEPROM chip. Data erasing is such that all cells of the memory cell array 11 are erased at a time—say, all-at-once erasure. During such data erasing, all wordlines WL of the memory cell array 11 are applied a specified voltage of −7 volts (V). Simultaneously a voltage of +10V is applied to source lines SL, causing each bitline BL to be set in an electrically floating state. With this voltage application scheme, each memory cell is such that electrons residing at its floating gate are released toward the substrate by Fowler-Nordheim (F-N) tunnel effects, resulting in the cell being set in an erase state low in threshold voltage. FIG. 5 shows current versus voltage characteristics for demonstration of a typical threshold voltage distribution pattern of a memory cell storing therein data of a logic “0” and that of a memory cell of logic “1” data, wherein the data “1” low in threshold voltage is the erase state.

Data write is achievable by applying a write voltage of +9V to a presently selected wordline WL while at the same time applying a specific voltage to its associative bitline BL. This voltage is set at either Vdd=5V or Vss=0V in a way pursuant to the logic level of a data bit stored—i.e. logic “0” or “1.” This results in creation of “hot” electrons at a memory cell that is given logic data “0.” The hot electrons thus created are then injected into this cell's floating gate, causing it be set in the logic “0” data state that is high in threshold voltage as shown in FIG. 5 .

Data read is to be done by giving a 5-V voltage for use as a read voltage Vr shown in FIG. 5 to a selected wordline to thereby detect whether the memory cell of interest is presently turned on or off. In FIG. 5, Vpv is a threshold voltage used for determination or “judgment” of write-verify read, and Vev is a threshold voltage used for judgment of erase-verify read. During data write, control is made for driving into a prespecified threshold voltage distribution through repeated alternate execution of pulsate write voltage application and verify-read operation. Similarly, during erasing, control is done for driving into a prespecified threshold voltage distribution through repeated alternate execution of read pulse voltage application and verify-read operation.

In the data write event, it is the voltage regulator 23 that generates a write voltage and a verify-read voltage to be used thereafter, based on a potentially boosted output voltage Vpp of the internal voltage generator circuit 21 . Accordingly, as shown in FIG. 6, a combination of the internal voltage generator circuit 21 and voltage regulator 23 receives a variety of control signals—including boost enable signal VPPE, regulator enable signal REGE, write control signal PROG, and verify control signal VRFY—which are generated and supplied from the controller 20 to thereby produce a regulated voltage Vreg as shown in FIG. 7 .

As better seen in FIG. 7, when the boost enable signal VPPE and regulator enable signal REGE plus write control signal PROG potentially go high at a time point t 0 , the internal voltage generator 21 generates a boosted output voltage Vpp that rises up in potential from a level of Vdd=5V along with a control voltage Vreg which is potentially variable with a change in voltage Vpp—that is, the former potentially “tracks” the latter. This regulated voltage Vreg is for use as the write voltage to be given to a selected wordline. The boosted output voltage Vpp stays at approximately 10V. The regulated voltage Vreg is at about 9V during writing.

Thereafter when the write control signal PROG potentially drops down to “Low” or “L” level while the verify control signal VRFY goes high to “High (H)” level at a time point t 1 , the regulated voltage Vreg being output from the regulator 23 behaves to decrease in potential. This becomes the verify read-use voltage of about 7V to be given to the selected wordline. Further, when the verify control signal VRFY reaches “L” level at a time point t 2 , the output voltage of regulator 23 returns to 5V.

A practically implemented configuration of the internal voltage generator circuit 21 is shown in FIG. 8 . As shown herein, this is a booster circuit that is operable to boost the power supply voltage Vdd to generate a potentially raised or “boosted” voltage Vpp. The booster circuit 21 has its main or “core” part 81 , which is a charge pump circuit made up from a serial combination of three N-channel metal oxide semiconductor (NMOS) transistors QN 11 to QN 13 , each of which has a threshold voltage of almost zero volts, and capacitors C 1 -C 2 associated therewith. The capacitors C 1 - 2 have terminals which are complementarily driven by a ring oscillator 82 as coupled thereto. The charge pump circuit 81 has its boosted voltage output node 80 , which is connected to a voltage limiter circuit 83 . This circuit 83 is for detection of the exact potential level of a boosted output voltage Vpp being derived from node 80 on a real time basis to thereby perform turn-on/off control of ring oscillator 82 .

The voltage limiter circuit 83 is configured to have, between the boosted output node 80 and ground terminal Vss, a serial combination of a switching P-channel MOS (PMOS) transistor QP 13 and resisters R 11 , R 12 for level detection. Voltage limiter 83 also has an operation amplifier or “opamp” OP 11 , which is the one that determines or “judges” a potential level at a connection node of resistors R 11 -R 12 to thereby perform turn-on/off control of the ring oscillator 82 . Circuit part consisting of the PMOS transistors QP 11 -QP 12 and NMOS transistors QN 11 - 12 makes up a switch control circuit which performs turn-on/off control of switching PMOS transistor QP 13 under control of the boost enable signal VPPE stated previously.

An operation of this voltage limiter 83 is as follows. When the boost enable signal VPPE is at “L” level, the NMOS transistor QN 11 turns on whereas NMOS transistor QN 12 turns off. The PMOS transistor QP 12 turns on. At this time the switching PMOS transistor QP 13 is kept off with its gate connected to the boosted voltage output node 80 . When the boost enable signal VPPE goes high at “H” level, NMOS transistor QN 11 turns off, NMOS transistor QN 12 turns on, and PMOS transistor QP 12 turns off. Whereby, switching PMOS transistor QP 13 is grounded at its gate, making voltage limiter 83 active.

While the boosted voltage Vpp of the boosted voltage output node 80 stays low in potential level, an output of the opamp OP 11 is at “H” level. This causes the oscillator 82 to perform an oscillation operation, thereby rendering the charge pump circuit 81 operative. When the boosted output voltage Vpp reaches a prespecified level, the output of opamp OP 11 is set at “L” level, causing oscillator 82 to turn off. The voltage boost operation is thus terminated. With the voltage boost operation control stated above, a constant output voltage Vpp is obtainable at the boosted voltage output node 80 in any events.

See next FIG. 9 . This diagram shows a circuit configuration of the voltage regulator 23 as used in this embodiment. As in the prior art regulator shown in FIG. 18, the regulator circuit shown herein is principally configured from a driver circuit 1 and a resistor-based potential dividing circuit 2 operatively associated with a couple of operational amplifiers OP 1 , OP 2 . The driver 1 has a PMOS transistor QP 2 for pull-up drive and an NMOS transistor QN 2 for pull-down drive. The resistive voltage divider circuit 2 is operable to receive an output control voltage of driver 1 and then potentially divide this voltage at an appropriate potential division ratio. The opamps OP 1 -OP 2 are for control of driver 1 through feedback of an output of voltage divider 2 .

The PMOS transistor QP 2 and NMOS transistor QN 2 have drains common-coupled together at a circuit node N 0 , which is for use as an output terminal of the regulated voltage Vreg stated supra. PMOS transistor QP 2 has a source connected to the boosted voltage output node 80 . NMOS transistor QN 2 has its source coupled to the ground terminal Vss.

The opamp OP 1 is responsible for control of the current drivability of NMOS transistor QN 2 . The opamp OP 2 is for control of the current drivability of PMOS transistor QP 2 . One of these opamps, i.e. opamp OP 1 , derives an output voltage, which is used to directly control the gate of pull-down NMOS transistor QN 2 . The remaining opamp OP 2 produces its output as used to control the gate of a source-grounded NMOS transistor QN 1 . A PMOS transistor QP 1 is such that its gate and drain are connected to the drain of NMOS transistor QN 1 with its source coupled to the boosted voltage output node 80 . The PMOS transistor QP 1 is for use as a current source load. This PMOS transistor QP 1 and the pull-up PMOS transistor QP 2 are such that their gates are common-coupled together, thus making up a current mirror circuit. With such an arrangement, controlling the current drivability of NMOS transistor QN 1 enables control of a current of the pull-down PMOS transistor QP 2 .

Provided between the driver 1 's output node N 0 and the ground terminal Vss is a serial combination of resistors R 1 to R 3 and a switching NMOS transistor QN 3 for constitution of the voltage divider circuit 2 . Resistors R 2 -R 3 have a common connection node N 1 , to which an NMOS transistor QN 4 is coupled. This transistor QN 4 is for ground connection of a node N 2 under control of the write control signal PROG. NMOS transistor QN 4 has its gate which is controlled by the verify-read control signal VRFY, causing a connection node N 3 of resistor R 3 and NMOS transistor QN 3 to be grounded selectively.

A difference of the voltage regulator 23 of the illustrative embodiment from the arrangement shown in FIG. 18 is that the opamps OP 1 and OP 2 are different in reference voltage from each other. More specifically, a reference voltage Vref 1 being input to an inverting (−) input terminal of one opamp OP 1 is different in potential from a reference voltage Vref 2 as input to a non-inverting (+) input terminal of the other opamp OP 2 in a manner such that Vref 1 is set at a carefully chosen value which is slightly higher than Vref 2 . Regulator 23 is similar to that of FIG. 18 in that a potentially divided output voltage obtainable at the connection node N 1 of resistors R 1 -R 2 in voltage divider circuit 2 is fed back to the non-inverting input terminal of opamp OP 1 and the inverting input terminal of opamp OP 2 , respectively.

Another feature unique to the regulator 23 of FIG. 9 is that the required two reference voltages Vref 1 , Vref 2 are generated by a single reference voltage generator circuit 22 having its integrated core circuitry with two separate reference voltage output nodes as shown in FIG. 12, rather than by mere use of separate or “discrete” reference voltage generator circuits. This point will be set forth later in greater detail.

A practically reduced configuration of a respective one of the opamps OP 1 -OP 2 is shown in FIG. 10 . As shown herein, each opamp OP 1 , OP 2 may be a differential amplifier of the complementary MOS (CMOS) current mirror type. This CMOS current-mirror differential amplifier includes a current mirror load made up from a pair of PMOS transistors QP 21 , QP 22 and a driver unit formed of another pair of NMOS transistors QN 21 -QN 22 . These driver transistors QN 21 - 22 have common-coupled sources which are grounded via a current source I 0 . PMOS transistors QP 21 - 22 have common sources which are connected to a power supply terminal through an activation-use PMOS transistor QP 20 . This PMOS transistor QP 20 has its gate, to which the above-noted regulator enable signal REGE enters through an inverter INV 11 . This inverter INV 11 derives its output, which will also be supplied to a gate of NMOS transistor QN 23 as provided to reset an output terminal.

The voltage regulator 23 of this embodiment is controllable by the two different reference voltages Vref 1 , Vref 2 discussed previously. Appropriate setting of these two reference voltages Vref 1 - 2 permits a pull-up current Iup flowing in the pull-up PMOS transistor QP 2 of regulator 23 and a pull-down current Idn flowing in the pull-down NMOS transistor QN 2 thereof to exhibit specific characteristics as shown in FIG. 11 . Unlike the current-voltage characteristics of FIGS. 19A-19B, the pull-up current Iup behaves to become zero at a point at which a potentially divided output voltage VN 1 at the node N 1 of voltage divider circuit 2 becomes equal to the reference voltage Vref 2 whereas the pull-down current Idn becomes zero at a point whereat the divided output voltage VN 1 is equal to the reference voltage Vref 1 , which is higher in potential than reference voltage Vref 2 .

A concrete explanation will be given of an operation of the regulator 23 in accordance with the timing of FIG. 7 . At a time point t 0 , the boost enable signal VPPE and regulator enable signal REGE potentially go high at “H” level. This potential increment permits a voltage boost operation and regulator operation to get started. Assume that at this time, the write control signal PROG is also set at “H” simultaneously. This causes the NMOS transistor QN 4 to turn on, resulting in the node N 2 being grounded in the voltage divider circuit 2 . At this time the potential division ratio of voltage divider 2 is given as R 2 /(R 1 +R 2 ).

While the regulated voltage Vreg stays lower in potential, the opamp OP 2 derives its output of high level, whereby the NMOS transistor QN 1 is driven to turn on so that the pull-up current Iup rushes to flow owing to the PMOS transistor QP 2 . The other opamp OP 1 provides its output which stays at low level, causing NMOS transistor QN 2 to be kept nonconductive, i.e. off. Thus the regulated voltage Vreg potentially increases with an increase in potential of the boosted output voltage Vpp.

Upon potential increase of the regulated voltage Vref up to a predetermined level as represented by Vreg=Vref 2 ·(R 1 +R 2 )/R 2 , the output of opamp OP 2 drops down at low level. This causes NMOS transistor QN 1 to turn off, thereby letting PMOS transistor QP 2 turn off, resulting in the pull-up current Iup becoming zero. Thus the regulated voltage Vreg stops its potential rise-up. The exact value of control voltage Vreg is determinable by the Vref 2 , R 1 , R 2 values from the above Equation. For instance, a potentially stabilized write-use voltage of Vreg=9V is obtained as shown in FIG. 7 . During this session, the output of opamp OP 1 stays at low level; thus, any appreciable pull-down current Idn does not flow.

At instant t 1 of FIG. 7, when the write control signal PROG drops down in potential to reach “L” level while the verify-read control signal VRFY goes high at “H” level, the node N 3 of voltage divider circuit 2 is grounded causing the potential division ratio of voltage divider 2 change to measure (R 2 +R 3 )/(R 1 +R 2 +R 3 ). This forces the potentially divided output voltage being fed back to opamps OP 1 - 2 to increase in potential, which in turn permits the output of opamp OP 1 to change or “transit” to the high level side, thereby driving NMOS transistor QN 2 to turn on. This results in flow of pull-down current Idn, causing the regulated voltage Vreg at output terminal N 0 to drop down accordingly.

When the regulated voltage Vreg decreases at a level defined as Vref=Vref 1 ·(R 1 +R 2 +R 3 )/(R 2 +R 3 ), the output of opamp OP 1 is set at low level. This forces NMOS transistor QN 2 to turn off causing the pull-down current Idn to become zero; thus, the regulated voltage Vreg's level reduction stops. The value of such regulated voltage Vreg at this time is determinable by the Vref 1 , R 1 -R 3 values. One example is that a potentially stabilized verify-read voltage of Vref=7V is obtained as shown in FIG. 7 .

In the way stated above, with the voltage regulator 23 of this embodiment, the use of two different reference voltages Vref 1 , Vref 2 precludes both the pull-up PMOS transistor QP 2 and the pull-down NMOS transistor QN 2 from attempting to turn on at a time. Consequently, even in cases where slight input offset voltage(s) is/are present at the opamps OP 1 -OP 2 , it becomes possible to prevent without failure the occurrence of any unwanted flow of large penetration currents as shown in FIG. 19B, while still retaining the possibility of establishment of the “ideal” state shown in FIG. 19 A. To be brief, it is possible to virtually perfectly eliminate any possible penetration currents at driver 2 , which in turn makes it possible to shorten a time as taken for the boosted output voltage Vpp to fully rise up at the required potential level—say, potential riseup or boost time period. It is thus possible to lessen or minimize a write time period of the Flash EEPROM device embodying the invention. Such write time reduction leads to a decrease in electrical power consumption thereof.

Preferably in this embodiment an undefined output range in which both the pull-up PMOS transistor QP 2 and the pull-down NMOS transistor QN 2 are turned off simultaneously is made smaller or minimized for the purpose of circuit operation stabilization. To this end, the two reference voltages Vref 1 , Vref 2 are specifically designed to have a small difference of about 10 millivolts (mV), by way of example.

Accordingly with this embodiment, the two reference voltages Vref 1 , Vref 2 are such that the stability of a potential difference is more important thereto than the stability of the absolute values thereof. From this point of view, as shown in FIG. 12, the single reference voltage generator circuit 22 is specifically employed to generate these two reference voltages Vref 1 -Vref 2 . With the “single-circuit/two-voltage creation” feature, even where the reference voltage generator circuit 22 suffers from certain variation or deviation occurring due to manufacturing processes or the like, it is possible to allow the two reference voltages Vref 1 - 2 to potentially increase or decrease together. This in turn makes it possible to force a difference between Vref 1 and Vref 2 to stay at a prefixed level in any events.

Typical examples of the reference voltage generator circuit 22 are band-gap reference (BGR) circuits as shown in FIGS. 14 and 15. Each BGR circuit shown herein includes three current flow paths 141 , 142 , 143 . The first current flow path 14 has a serial connection of a PMOS transistor QP 31 for use as a current source and a diode D 1 along with a resistor R 21 as connected in parallel to the diode D 1 . The second current flow path 142 has a serial combination of a current-source PMOS transistor QP 32 and a resistor R 22 plus a diode D 2 and also has a resistor R 23 as parallel-connected to the resistor R 22 and diode D 2 . The diodes D 1 and D 2 are designed to have a predetermined area ratio.

The current-source PMOS transistors QP 31 -QP 32 have gates which are common-coupled together. An operational amplifier OP 21 is provided, which has its output terminal connected to gates of these current-source PMOS transistors QP 31 - 32 and two input terminals that are connected to nodes N 11 , N 12 of the first and second current flow paths 141 - 142 respectively. Making the current-source PMOS transistors QP 31 - 32 equal in size to each other permits the same current to flow in the first current flow path 141 and second current flow path 142 . The opamp OP 21 controls the degree of electrical conduction of current-source PMOS transistors QP 31 - 32 in such a way that the node N 11 of the first current flow path 141 and the node N 12 of second current flow path 142 are constantly held at the same potential.

The third current flow path 143 is an output circuit which is formed of a current-source PMOS transistor QP 33 that makes up a current mirror circuit together with the PMOS transistors QP 31 - 32 and a resistor for use as a load. In FIG. 14 the load resistor constitutes a voltage divider circuit 144 with a serial combination of resistors R 24 , R 25 . The resistor R 24 is coupled to the PMOS transistor QP 33 at a connection node, which acts as an output node of the first reference voltage Vref 1 . A connection node of resistors R 24 and R 25 is an output node of the second reference voltage Vref 2 .

The circuitry of FIG. 15 is similar to that of FIG. 14 with its output circuit configuration being slightly modified. More specifically, the load resistor makes up the voltage divider circuit 144 with a serial combination of resistors R 24 -R 26 . A connection node of two resistors R 24 - 25 serves as the output node of first reference voltage Vref 1 whereas a node of resistors R 25 - 26 is the output node of second reference voltage Vref 2 .

In the reference voltage generator circuit 22 of FIG. 14, the first reference voltage Vref 1 without any appreciable temperature dependabilities is generated by optimal setup of the area ratio of the diodes D 1 -D 2 and the values of resistors R 21 , R 22 , R 23 and (R 24 +R 25 ). This reference voltage Vref 1 is then subjected to potential division to thereby generate the second reference voltage Vref 2 . Similarly with the reference voltage generator circuit 22 of FIG. 15, the first reference voltage Vref 1 with no temperature dependabilities is produced by optimally setting the area ratio of diodes D 1 - 2 and the values of resistors R 21 , R 22 , R 23 and (R 24 +R 25 +R 26 ). This reference voltage Vref 1 is potentially divided to thereby generate the second reference voltage Vref 2 .

In this way, the two reference voltages Vref 1 , Vref 2 are generated by a single BGR circuit and its associative voltage divider circuit for potential division of an output of the former. With such an arrangement, the two reference voltages Vref 1 - 2 are expected to vary together in the same direction in spite of any possible fabrication process variation, thus enabling a difference therebetween to stay substantially at a fixed level constantly.

In view of the fact that the two reference voltages Vref 1 - 2 may be designed to be varied together in accordance with a variation or deviation in circuit characteristics in the way stated above, the reference voltage generator circuit 22 may alternatively be modified in configuration as shown in FIG. 13 . This reference voltage generator circuit 22 is made up from a couple of circuit modules: a first or “pre stage” reference voltage generator circuit 221 and a second or “post stage” reference voltage generator circuit 222 . The pre-stage voltage generator 221 is for generation of an original or “root” reference voltage Vref. The post-stage reference voltage generator 222 is responsive to receipt of an output of the pre-stage circuit for potentially dividing it to thereby generate the two reference voltages Vref 1 , Vref 2 required.

The pre-stage reference voltage generator circuit 221 may be a BGR circuit. As shown in FIGS. 16A to 16 D, the post-stage reference voltage generator circuit 222 is configurable from a voltage follower 161 employing an operational amplifier OP 31 for receipt of the reference voltage Vref as output from the prestage reference voltage generator 221 and a resistive voltage divider circuit 162 that is provided at the output of the voltage follower 161 . With such an arrangement, a potential difference between two reference voltages Vref 1 , Vref 2 is stably held or stabilized constantly even in cases where the original reference voltage Vref per se varies little in potential.

In the circuits of FIGS. 16A-16B, the voltage follower 161 is formed of an operational amplifier OP 131 having an inverting input terminal, a non-inverting input terminal for use as an input terminal of the reference voltage Vref, and an output terminal being fed back to the inverting input. In FIG. 16A an output of voltage follower 161 is used as the reference voltage Vref 1 whereas a connection node of resistors R 31 - 32 to which this voltage is supplied for use as a terminal from which the reference voltage Vref 2 is obtained. Obviously the first reference voltage Vref 1 has its potential value that is equal to that of the original reference voltage Vref whereas the second reference voltage Vref 2 is slightly lower in value than it. In the circuitry of FIG. 16B, its voltage divider circuit 162 is formed of a serial combination of three resistors R 31 -R 33 for generating two reference voltages Vref 1 , Vref 2 lower in potential than the reference voltage Vref.

In the circuits of FIGS. 16C-16D, the voltage follower 161 is arranged including, in addition to the operational amplifier OP 31 , a current-source PMOS transistor QP 41 having its gate as controlled by an output of opamp OP 31 . In this case the reference voltage Vref is input to the inverting input terminal of opamp OP 31 . In the circuitry of FIG. 16C the current-source PMOS transistor QP 41 's drain serves as an output terminal of the first reference voltage Vref 1 , which is fed back to the non-inverting input of opamp OP 31 while being connected to the resistive voltage divider circuit 162 . Accordingly the first reference voltage Vref 1 is made equal in potential to the original reference voltage Vref whereas the second reference voltage Vref 2 is slightly lower than it. In the circuitry of FIG. 16D the current-source PMOS transistor QP 41 's drain is for use as the output terminal of the reference voltage Vref 1 , with the connection node of resistors R 41 -R 42 being coupled and fed back to the non-inverting input of opamp OP 31 . Thus the second reference voltage Vref 2 is equal to the original reference voltage Vref whereas the first reference voltage Vref 1 is little higher than it.

Turning to FIG. 17, there is depicted a configuration of a voltage regulator 23 in accordance with another embodiment of this invention. Parts or components corresponding to those of the FIG. 9 embodiment are designated by the same reference characters, and any detailed explanations thereof will be eliminated herein for brevity purposes. The FIG. 17 circuitry is similar to that of FIG. 9 with the two operational amplifiers OP 1 , OP 2 being modified to receive the same or “common” reference voltage Vref while alternatively letting them be different from each other in feedback voltage.

More specifically, a voltage divider circuit 2 of FIG. 17 is formed of a serial combination of four resistors R 1 to R 4 . A first potentially divided output voltage DVout 1 as obtainable at a connection node N 2 of serial-coupled resistors R 2 , R 3 is fed back to the non-inverting input of the opamp OP 1 that is used for control of a pull-down current Idn. In addition, let a second divided output voltage DVout 2 obtainable at a connection node N 1 of serial resistors R 1 -R 2 be fed back to the inverting input of the opamp OP 2 used for control of a pull-up current Iup.

With such an arrangement, the potentially divided output voltage DVout 1 for feedback to the opamp OP 1 is lower in potential than the divided output voltage DVout 2 for feedback to opamp OP 2 . This in turn guarantees achievement of substantially equivalent operability to the previous embodiment with its two opamps OP 1 , OP 2 being given mutually different reference voltages Vref 1 , Vref 2 and with the same potentially divided output voltage fed back thereto. Consequently as in the previous embodiment, it becomes possible for the FIG. 17 embodiment to perform the regulated voltage control operation with any penetration current at driver 1 being greatly suppressed to virtually zero even where the opamps OP 1 - 2 exhibit some input offsets.

It has been stated that the voltage regulator incorporating the principles of this invention is specifically arranged to give slightly different reference voltages to two operational amplifiers for performing current drivability control of pull-up and pull-down transistors, or alternatively permit feedback thereto of potentially divided output voltages with a slight difference therebetween. Thus it is possible to successfully minimize or virtually perfectly eliminate any possible penetration currents flowing in the voltage regulator.

While the invention has been described with reference to specific embodiments, the description is illustrative of the invention and is not to be construed as limiting the invention. Various modifications and applications may occur to those skilled in the art without departing from the true spirit and scope of the invention as defined by the appended claims.