Boost device for nonvolatile memories with an integrated stand-by charge pump

1. Field of the Invention

The present invention relates to a boost device for nonvolatile memories with and integrated stand-by charge pump.

2. Description of the Related Art

As is known, when a memory device is not used, most of its internal circuits are switched off so as to reduce consumption (waiting or stand-by phase). However it is necessary to keep some circuits operating to allow a faster return of the memory device from stand-by to the main operating condition (user-mode). In particular, during stand-by, the stand-by charge pump, which feeds the memory cell of the device with a voltage close to the reading voltage, is kept on.

On this matter, FIG. 1 shows a stand-by charge pump 1 , of the diode type with two opposed phases, included in a memory device 2 .

In greater detail, the stand-by charge pump 1 comprises a plurality of stages 3 connected in series between a supply line 4 , set at a supply voltage V cc , and an output terminal 5 connected to an output capacitor 6 , on which an output voltage V out is present. Each stage 3 comprises a boost capacitor 7 and a switch 8 , implemented by a diode, which is switched on or off alternately with the adjacent one. The charge transfer from one stage to the other takes place under the control of phase signals CK and CKn which are complementary to each other. The value of the output voltage V out is given by the equation:

V out =( N+ 1)*( V cc −V th )

where V th is the threshold voltage of each diode and N is the number of stages 3 of the stand-by charge pump 1 .

As previously stated, the stand-by charge pump comes into operation only when the memory device is in stand-by. Consequently, during stand-by there is a considerable consumption of energy since the stand-by charge pump needs a certain time to supply the memory device with a voltage close to the reading voltage.

Moreover, the problem of area occupation is serious because the stand-by charge pump is distinct from the main charge pump, which is used during programming and reading of the memory device, and each charge pump occupies a not negligible area, since it comprises high capacities.

An embodiment of the present invention is to overcome the limitations and drawbacks indicated above with reference to the prior art.

According to the present invention a boost device for nonvolatile memories with an integrated stand-by charge pump is provided by a boost device for nonvolatile memories, comprising a charge pump circuit having an input and a first output, the charge pump circuit comprising at least one input stage and a first output stage cascade-connected between the input and the first output of the charge pump circuit, characterized in that the charge pump circuit comprises a second output stage connectable in parallel to the first output stage.

For a better understanding of the invention, an embodiment is now described, purely as a non-limitative example, with reference to the enclosed drawings, wherein:

FIG. 1 shows a circuit diagram of a boost device according to the prior art;

FIG. 2 shows a circuit diagram of a boost device according to the invention;

FIG. 3 shows the plot of electrical quantities measured in the boost device of FIG. 2;

FIG. 4 shows a circuit diagram of a detail of the boost device of FIG. 2;

FIG. 5 shows the plot of further electrical quantities measured in the boost device of FIG. 2; and

FIG. 6 presents a flow-chart of the operation of the detail of FIG. 4 .

FIG. 2 shows a boost device 10 incorporated in a memory 100 , of flash type, and comprises a four-phase type, charge pump circuit 11 , a phase generator stage 12 and a clock generator stage 13 .

In detail, the charge pump circuit 11 has an input 11 a, receiving a supply voltage V DD , a main output 11 b 1 and a stand-by output 11 b 2 . The charge pump circuit 11 comprises an input stage 15 , an intermediate stage 16 and a main output stage 17 , cascade-connected to one another. The charge pump circuit 11 further comprises a stand-by output stage 18 connected in parallel to the main output stage 17 .

The input stage 15 , the intermediate stage 16 , the main output stage 17 and the stand-by output stage 18 each have an input node 15 a - 18 a and an output node 15 b - 18 b. In greater detail, the input node 15 a of the input stage 15 is connected to the input terminal 11 a of the charge pump circuit 11 while the output node 17 b of the main output stage 17 and the output node 18 b of the stand-by output stage 18 are connected respectively to the main output 11 b 1 and to the stand-by output 11 b 2 of the charge pump circuit 11 . Furthermore, the output node 16 b of the intermediate stage 16 is connected to the input nodes 17 a, 18 a of the main output stage 17 and of the stand-by output stage through a circuit node 19 .

The input stage 15 comprises a charge storage capacitor 15 c; a charge transfer transistor 15 d; a boost capacitor 15 e; a precharge transistor 15 f.

Specifically, the charge storage capacitor 15 c has a first terminal connected to the output node 15 b of the input stage 15 and a second terminal receiving a phase signal B supplied by the phase generator stage 12 . The charge transfer transistor 15 d is connected between the input node 15 a and the output node 15 b of the input stage 15 and has a gate terminal connected to a precharge node 15 g. The boost capacitor 15 e has a first terminal connected to the precharge node 15 g and a second terminal receiving a phase signal A supplied by the phase generator stage 12 . The precharge transistor 15 f is diode-connected and is connected between the input node 15 a of the input stage 15 and the precharge node 15 g.

The intermediate stage 16 comprises a charge storage capacitor 16 c; a charge transfer transistor 16 d; a boost capacitor 16 e; a precharge transistor 16 f; a precharge node 16 g.

The intermediate stage 16 , as indicated, is structurally identical to the input stage 15 and differs from the latter only in that the charge storage capacitor 16 c receives a phase signal D, the boost capacitor 16 e receives a phase signal C and the precharge transistor 16 f has a gate terminal connected to the output node 16 b.

In turn the main output stage 17 comprises a charge transfer transistor 17 d ; a boost capacitor 17 c ; a precharge transistor 17 f ; a precharge node 17 g.

The main output stage 17 , as indicated, is structurally identical to the input stage 15 and differs from the latter only in that the precharge transistor 17 f has a gate terminal connected to the main output 11 b 1 of the charge pump circuit 11 and the charge storage capacitor is not present.

The stand-by output stage 18 comprises a charge storage capacitor 18 c; a charge transfer transistor 18 d; a boost capacitor 18 e; a precharge transistor 18 f; a precharge node 18 g.

The stand-by output stage 18 , as indicated, is structurally identical to the input stage 15 and differs from the latter only in that the precharge transistor 18 f has a gate terminal connected to the stand-by output 11 b 2 of the charge pump circuit 11 , and the boost capacitor 18 e and the charge storage capacitor 18 c receive respectively a phase signal STBYA and a phase signal STBYB. In detail, the phase signals STBYA and STBYB are equal to the phase signals A and B during the reading and stand-by conditions of the memory 100 and differ from these during programming, as better described below.

In turn, the phase generator stage 12 is a logic circuit of a generally known type, having a first input 12 a receiving a clock signal Ckint 1 , generated by the clock generator stage 13 , and four outputs supplying the phase signals A, B, C, D. The phase signals A, B, C, D are logic signals switching between V DD and 1V, as shown in FIG. 3, with two different frequency values depending on the value assumed by the clock signal Ckint 1 as described below.

The clock generator stage 13 has a first signal input 13 a receiving a first timing signal CK 1 , having a first frequency value; a second signal input 13 b receiving a second timing signal CK 2 , having a second frequency value, lower than the first value; a third input signal 13 c receiving a first pump enable signal PXPH which enables the charge pump circuit 11 during the main operation condition (reading or programming) of the memory 100 ; a fourth signal input 13 d receiving a stand-by signal PXST; a fifth signal input 13 e receiving a second pump enable signal PULSE which enables the charge pump circuit 11 during stand-by operation of the memory 100 ; an output 13 f supplying the clock signal Ckint 1 .

As shown in FIG. 4, the clock generator stage 13 comprises a clock selection circuit 20 and a clock processing logic circuit 21 .

In greater detail, the clock selection circuit 20 has a first and a second signal input connected respectively to the first and the second signal input 13 a, 13 b of the clock generator stage 13 , a selection input 20 a receiving a frequency selection signal SEL generated by the clock processing logic circuit 21 and an output terminal 20 b generating an output signal CK 3 that is equal to the first or the second timing signal CK 1 , CK 2 depending on the logic value of the frequency selection signal SEL.

In turn, the clock processing logic circuit 21 has a first input terminal connected to the output terminal 20 b of the selection circuit 20 , a second, a third and a fourth input terminal connected respectively to the third, fourth and fifth signal input 13 c, 13 d, 13 e of the clock generator stage 13 , a first output terminal connected to the selection input 20 a of the clock selection circuit 20 and a second output terminal connected to the output 13 f of the clock generator stage 13 .

With reference to FIGS. 3 and 5, the operation of the boost device will now be described.

When the stand-by signal PXST is low, the memory 100 is in the main operating condition.

In this condition, if the first pump enable signal PXPH is high, the charge pump circuit 11 is enabled and the clock signal Ckint 1 is equal to the first timing signal CK 1 .

If only one voltage value is required during the main operating condition (reading of the memory 100 ), the two outputs 11 b 1 and 11 b 2 of the charge pump 11 are short-connected (for example by closing a switch 30 ) and the main output stage 17 is in parallel with the stand-by output stage 18 . In these conditions, the charge pump circuit 11 , receiving the main timing signal CK 1 , acts as a normal four-step charge pump determining the sequence of the charging and charge transfer steps in a known way (for example as described in U.S. Pat. No. 6,075,402 in the name of the same applicant) and depending on the value of the phase signals A, B, C, D (the phase signals STBYA and STBYB having the same behavior respectively as the phase signals A and B).

Advantageously, the connection between the main output stage 17 and the stand-by stage 18 gives a greater driving capability of the memory cells connected to the charge pump circuit 11 , thanks to the additional presence of the charge storage capacitor 18 c of the stand-by output stage 18 .

If instead, during the main operating condition, two different voltage values are required (programming of the memory 100 ), the stand-by output stage 18 and the main output stage 17 are separated (switch 30 off and outputs 11 b 1 , 11 b 2 isolated) and are regulated by two different regulators. In this case, the phase signals STBYA and STBYB have different timing signals with respect to the phase signals A and B, though being generated from the same timing signal (that is the first timing signal CK 1 ).

Vice versa, when the stand-by signal PXST is high, the memory 100 is in stand-by operating condition.

In this condition, if the second pump enable signal PULSE is also high, the charge pump circuit 11 is enabled and the clock signal Ckint 1 is equal to the second timing signal CK 2 .

In this case too, the two outputs 11 b 1 and 11 b 2 of the charge pump 11 are connected (since the switch 30 is closed) and the main output stage 17 is in parallel with the stand-by output stage 18 . In these conditions, the charge pump circuit 11 , receiving the second timing signal CK 2 , acts as a normal stand-by charge pump driven by the phase signals A, B, C, D generated with a longer period (for example 1 μs) with respect to the main operating condition (wherein the phase signals have a period of, for example, 50 ns to 100 ns). In this case too, the phase signals STBYA and STBYB have the same plot as the phase signals A and B.

Advantageously, when stand-by operation is enabled, the first two stages 15 and 16 of the charge pump 11 are already precharged at an intermediate voltage V X (present on the circuit node 19 ) close to the required voltage.

In this way a considerable current saving is obtained and therefore lower consumption, since the charge pump 11 requires a shorter time to supply the memory 100 with the required voltage.

According to a further aspect of the present invention, the clock generator stage 13 having the structure described above makes it possible to avoid the occurrence of undesirable charge sharing phenomena generated by spurious impulses.

It is supposed indeed that during the transition from a main operating condition to a stand-by operating condition and vice versa, the first timing signal CK 1 and the second timing signal CK 2 are in phase opposition. By performing a simple multiplexing operation on the two timing signals CK 1 and CK 2 , a resulting clock signal Ckint 0 would be obtained with a logic state transition which would give rise to undesired “pumping actions” (FIG. 5, time intervals Δt 1 , Δt 2 ).

Considering that the passage from main operating condition to stand-by operating condition can be forced by the user by regulating the memory enable signal CE and that the specifications of the memories allow “fast” transitions between the two above-mentioned operating conditions, even with periods of less than 100 ns, there would be a high number of undesired pumping actions per unit of time. This would cause an uncontrolled increase of the voltages in the last stages of the pumps, causing the exit from the SOA (Safe Operating Area) of the MOS transistors that compose them.

According to the second aspect of the present invention, this problem is overcome by making the clock generator stage 13 work with an operation described below with reference to FIGS. 4, 5 and 6 .

Initially (FIG. 5, instant to; FIG. 6, block 200 ) the stand-by signal PXST switches high (memory 100 in stand-by condition) and controls the clock processing logic circuit 21 so that it generates a first logic level for the frequency selection signal SEL (FIG. 6, block 201 ) such as to select the second timing signal CK 2 ; consequently the output signal CK 3 of the clock selection circuit 20 is the same as the second timing signal CK 2 .

Moreover, the first and the second pump enable signals PXPH, PULSE and the clock signal Ckint 1 are low.

Subsequently, the second pump enable signal PULSE switches to high (instant t 1 , block 202 of FIG. 6 ). In these conditions, as soon as the clock processing logic circuit 21 detects the switching (here to low) of the second timing signal CK 2 (block 203 , FIG. 6 ), it controls the switching of the clock signal Ckint 1 (block 204 , FIG. 6 ). In the example in FIG. 5, the clock signal Ckint 1 switches to high.

The clock signal Ckint 1 then follows the transitions of the second timing signal CK 2 until the second pump enable signal PULSE switches to low (instant t 2 , block 205 of FIG. 6 ); then the clock processing circuit 21 “freezes” the present status of the clock signal Ckint 1 (block 206 of FIG. 6 ). In the example in FIG. 5, the clock signal Ckint 1 is maintained low. Then, the present status of the clock signal Ckint 1 is stored in a status register (not shown in FIG. 4) inside the circuit (FIG. 6, block 207 ). Thereby, spurious transitions (and therefore undesired pumping actions), affecting the known clock signal Ckint 0 in the interval Δt 1 , are avoided.

In the instant t 2 , also the stand-by signal PXST switches to low (memory 100 in the main operating condition, block 200 in FIG. 6 ), so that the clock processing circuit 21 generates a second logic level for the frequency selection signal SEL so as to select the first timing signal CK 1 ; consequently the output signal CK 3 of the clock selection circuit 20 is equal to the first timing signal CK 1 (FIG. 6, block 201 ).

Subsequently (instant t 3 ), the first pump enable signal PXPH switches over (here to high, FIG. 6, block 202 ). Since the clock processing circuit 21 also detects the switching of the first timing signal CK 1 , here to high (block 203 in FIG. 6 ), it controls the switching of the clock signal Ckint 1 (block 204 in FIG. 6 ). In the example in FIG. 5, the clock signal Ckint 1 switches to high.

The clock signal Ckint 1 then follows the transitions of the first timing signal CK 1 until the first pump enable signal PXPH switches to low (instant t 4 , block 205 ); then the clock processing circuit 21 maintains the present (low) status of the clock signal Ckint 1 (FIG. 6, block 206 ) and stores it in the status register (FIG. 6, block 207 ).

Subsequently (instant t 5 ), the first pump enable signal PXPH switches over again to high level and the clock processing circuit 21 , detecting also the switching of the first timing signal CK 1 (here to high), controls switching of the clock signal Ckint 1 (FIG. 6, block 204 ). In FIG. 5, the clock signal Ckint 1 switches to high.

The clock signal Ckint 1 then follows the transitions of the first timing signal CK 1 until the first pump enable signal PXPH switches to low (instant t 6 , block 205 of FIG. 6 ), controlling the clock processing circuit 21 so as to “freeze” the present (high) status of the clock signal Ckint 1 (block 206 in FIG. 6) and to store it in the status register (FIG. 6, block 207 ). Thereby, the spurious transition affecting the known clock signal Ckint 0 in the interval Δt 2 is avoided.

Thereby, a clock signal Ckint 1 was obtained free of spurious transitions in the passage from stand-by operation to main operating condition and vice versa.

The advantages of the described boost device 10 are clear from the above. A further advantage is that the boost device 10 occupies a small area, so that the structure of the stand-by charge pump does not have to be made separate from the structure of the main charge pump, and only one phase generator is required for both charge pumps.

Lastly it is clear that modifications and variations may be made to the described boost device, without leaving the scope of the present invention.