Charge-pump circuit and control method thereof

The present invention relates to a charge-pump circuit outputting voltage fluctuation within a step or power source voltage Vdd and a control method thereof, particularly a control method of a charge-pump circuit capable of normal charging pump operation removing influence of a parasitic diode following to a charge transfer device.

The charge-pump circuit developed by Dicson generates higher voltage than power source voltage Vdd of an LSI chip by voltage fluctuation of each pumping packet connecting plural stages of the pumping packet in series. For example, it is used for generating voltage for program/erase of flash memories.

However, the conventional charge-pump circuit carries out voltage fluctuation with a step of the power source voltage Vdd, and a circuit capable of carrying out voltage fluctuation with lower voltage step than the Vdd was not yet proposed. So, the inventor has already proposed a charge-pump circuit capable of carrying out voltage fluctuation with lower voltage step than the Vdd and improving efficiency η of the circuit (U.S. patent application No. 09/732,944 filed on Dec. 8, 2000).

The outline thereof will be described below. FIG. 10 to FIG. 12 are circuit diagrams showing a structure of an operation of −0.5 Vdd voltage fluctuation charge-pump circuit. The charge-pump circuit generates voltage of voltage fluctuation of −0.5 Vdd to earth voltage (0 V).

In FIG. 10, diodes D 1 and D 2 are connected in series as charge transfer devices. To a cathode of the diode D 1 , earth voltage (0 V) is supplied. The diodes D 1 and D 2 generally consist of MOS transistors for charge transfer in order to integrate into an LSI.

Switches S 1 , S 2 and S 3 connect two capacitors 1 and 2 to a connecting point between the diodes D 1 and D 2 switching in parallel or in series. These switches S 1 , S 2 and S 3 can consist of MOS transistors. Thus, on and off of the switches S 1 , S 2 and S 3 corresponds to on and off of the MOS transistors. A clock driver 3 supplies clock CLK to the capacitor 2 . Output voltage output from the diode 2 is applied to a load 4 .

An outline of control method of the charge-pump circuit will be described below. Power source voltage of the clock driver is assumed 5 V. Although forming the diodes D 1 and D 2 and the switches S 1 , S 2 and S 3 actually occurs voltage drop, the voltage drop is assumed 0 V omitting the voltage drop here.

When input clock of the clock driver 3 is high level (CLK=High), assuming that S 1 is off, S 2 is on and S 3 is off, two capacitors 1 and 2 are connected in series and each node voltage is: VL 1 ≈0V, VA=VB=2.5V, VC=5V. VL 1 is voltage of a connecting node between the diode D 1 and the capacitor 1 (a pumping node), VA is voltage of a connecting node between the capacitor C 1 and the switch S 2 , VB is voltage of a connecting node between the switch S 2 and the capacitor 2 , and VC is voltage of a connecting node between output of the clock driver 3 and the capacitor 2 .

That is, if capacitance values which capacitors 1 and 2 have are equal, capacitors 1 and 2 are respectively charged to voltage of Vdd/ 2 by distributing equally electric charge to the capacitors 1 and 2 (see FIG. 10 ).

Next, when S 2 is off and S 1 and S 3 are on in the state of CLK=High, two capacitors 1 and 2 are switched to parallel connection. Thus, each node voltage becomes: VL 1 ≈2.5V, VA=5V, VB=2.5V, VC=5V (See FIG. 11 ).

Next, when the input clock CLK is transferred to low level (CLK=Low) in the state of the parallel connection, each node voltage becomes: VL 1 ≈2.5V, VA=0V, VB=−2.5V, VC=5V by effect of the capacitor coupling because the capacitors 1 and 2 are connected to the pumping node (See FIG. 12 ).

Thus, by repeating switching the capacitors 1 and 2 alternately to series and parallel according to the input clock CLK, output voltage of −2.5 V (=(−½)·Vdd) is supplied to the load 4 from the diode D 2 .

When the diodes D 1 and D 2 consist of MOS transistors for charge transfer where a source and a gate are connected, there are a problem that needless current flows transitionally by that the diode D 1 is biased to forward direction when voltage of the pumping node VL 1 becomes 2.5V. Then, in order to avoid the problem, gate voltage of the MOS transistor for charge transfer may be controlled separating from source voltage.

At timing connecting the capacitors 1 and 2 in series, gate voltage of the transistor for charge transfer suitable for the diode D 1 is made on by setting low level (see FIG. 10 ), at timing connecting the capacitors 1 and 2 in parallel, gate voltage of the transistor for charge transfer suitable for the diode D 1 is made off by setting high level (see FIG. 11 ).

However, in the above-mentioned control method of the charge-pump circuit, voltage of the pumping node VL 1 repeats change such as 0V→2.5V→−2.5V. Because of that, even if the MOS transistors for charge transfer are any of P-channel and N-channel, a problem occurs that a parasitic diode formed incidentally to the MOS transistors is biased to forward direction and voltage fluctuation is not carried out normally.

FIGS. 13A and 13B are views showing a problem in the case that the diode D 1 is made by P-channel MOS transistor as a charge transfer device. In this case, a source S and a substrate B are earthed to improve efficiency of the charge-pump circuit depressing back gate bias effect of the MOS transistor.

As shown in FIG. 13A, there is not any problem in case that voltage of the pumping node VL 1 is −2.5V. However, as shown in FIG. 13B, when a parasitic diode formed between a drain and the substrate in the case that voltage VL 1 is 2.5V of the pumping node is biased to forward, forward direction current of the diode flows between the drain and the substrate, power efficiency becomes bad, and charge-pump operation is not carried out normally.

FIGS. 14A and 14B are views showing a problem in the case that the diode D 1 is made by N-channel MOS transistor as a charge transfer device. In this case, a drain D (pumping node) and a substrate B are connected to depress back gate bias effect of the MOS transistor.

As shown in FIG. 14A, there is not any problem in case that voltage of the pumping node VL 1 is −2.5V. However, as shown in FIG. 14B, a parasitic diode formed between the substrate and a source is biased to forward in the case that voltage VL 1 is 2.5V. Then, forward direction current of the diode flows between the drain and the substrate, power efficiency becomes bad, and charge-pump operation is not carried out normally.

An object of the invention is to prevent that the parasitic diode is biased to forward direction and needless current flows, and to make normal operation of the charge-pump circuit possible.

A charge-pump circuit of the invention comprises, at least first and second MOS transistors for charge transfer connected in series, first and second capacitors, clock supplying means supplying clock to one end of the second capacitors, first switching means for connecting said first and second capacitors to a connecting point of the first and second MOS transistors for charge transfer in series, and second switching means for connecting said first and second capacitors to the connecting point of the first and second MOS transistors for charge transfer in parallel, wherein said clock supplying means changes the state of said clock when said first and second switching means turn off.

By such the structure, a timing that the clock is supplied to the capacitor changes to high level from low level (or to low level from high level) is adjusted in the state that both of the first and second switch means are off. In this state, the first and second capacitors are separated from the connecting point (pumping node) of the first and second MOS transistors for charge transfer.

Thus, it is prevented that the parasitic diodes following the first and second MOS transistors for charge transfer are biased to forward direction because change of potential of the pumping node is depressed.

FIG. 1 is a circuit diagram showing a charge-pump circuit and a control method thereof according to an embodiment of the invention.

FIG. 2 is a circuit diagram showing a charge-pump circuit and a control method thereof according to an embodiment of the invention.

FIG. 3 is a circuit diagram showing a charge-pump circuit and a control method thereof according to an embodiment of the invention.

FIG. 4 is a circuit diagram showing a charge-pump circuit and a control method thereof according to an embodiment of the invention.

FIG. 5 is a circuit diagram showing a charge-pump circuit and a control method thereof according to an embodiment of the invention.

FIG. 6 is a circuit diagram showing a charge-pump circuit and a control method thereof according to an embodiment of the invention.

FIG. 7 is a timing chart showing a charge-pump circuit and a control method thereof according to an embodiment of the invention.

FIGS. 8A and 8B are views showing a case that a charge transfer device is made of a P-channel MOS transistor.

FIGS. 9A and 9B are views showing a case that a charge transfer device is made of an N-channel MOS transistor.

FIG. 10 is a circuit diagram showing a construction and

FIG. 11 is a circuit diagram showing a construction and an operation of the conventional charge-pump circuit.

FIG. 12 is a circuit diagram showing a construction and an operation of the conventional charge-pump circuit.

FIGS. 13A and 13B are views showing a problem of the case that a charge transfer device is made of a P-channel MOS transistor.

FIGS. 14A and 14B a views showing a problem of the case that a charge transfer device is made of an N-channel MOS transistor.

An embodiment of the invention will be described below with reference to the drawings. FIG. 1 to FIG. 6 are circuit diagrams showing constructions and operations of charge-pump circuits outputting voltage fluctuation of −0.5 Vdd. These charge-pump circuits generate voltage fluctuation of −0.5 Vdd to earth voltage (0 V).

P-channel MOS transistors M 1 and M 2 are connected in series. The MOS transistors M 1 and M 2 have constructions where a substrate and a source are connected in order to prevent back gate effect. In the MOS transistors M 1 and M 2 , though it is not limited, for example, a gate and a source are connected so as to construct a kind of diode.

Switches S 1 , S 2 and S 3 connect two capacitors 1 and 2 to a connecting point (pumping node) between MOS transistors M 1 and M 2 switching in parallel or in series. That is, when the switch S 2 (first switching means) in on, the MOS transistors M 1 and M 2 are connected in series, when the switches S 1 and S 3 (second switching means) are on, the MOS transistors M 1 and M 2 are connected in parallel.

As described later, the switch S 2 and the switches S 1 and S 3 are controlled so as to repeat on and off alternately in outline. Even these switches S 1 , S 2 and S 3 consist of MOS transistors. Thus, on and off of the switches S 1 , S 2 and S 3 corresponds to on and off of the MOS transistors.

A clock driver 3 supplies clock CLK to the capacitor 2 . The clock driver 3 , though it is not especially limited, consists of a CMOS inverter to which power source voltage Vdd is supplied. Output voltage output from a diode D 2 is applied to a load 4 .

A control method of the charge-pump circuit having the above-mentioned construction will be described referring FIG. 1 to FIG. 7 . FIG. 7 is a timing chart for describing control method of a charge-pump circuit.

Although it is not limited, it is assumed that power source voltage Vdd of the clock driver 3 is 5 V and capacitor values of the capacitors 1 and 2 are equal. Voltage fall caused by the MOS transistors M 1 and M 2 and the switches S 1 , S 2 and S 3 is described as 0 V.

(1) First Control Step

At time t 1 , the switches S 1 and S 2 are off, and all of the switches S 1 , S 2 and S 3 become off state. Input clock CLK of the clock driver 3 is low level (CLK=Low). At this state, each node voltage is: VL 1 ≈−2.5V, VA=0V, VB=−2.5V, VC=0V. VL 1 is voltage of a connecting node between the diode D 1 and the capacitor 1 (a pumping node), VA is voltage of a connecting node between the capacitor C 1 and the switch S 2 , VB is voltage of a connecting node between the switch S 2 and the capacitor 2 , and VC is voltage of a connecting node between output of the clock driver 3 and the capacitor 2 (see FIG. 1 and FIG. 7 ).

(2) Second Control Step

Next, at time t 2 in the state that all of the switches S 1 , S 2 and S 3 are off, the clock CLK is changed to high level from low level. Then, VC changes to 5V and VB changes to 2.5V by effect of capacitor coupling. Voltage of pumping node VL 1 does not change because all of the switches S 1 , S 2 and S 3 are off (see FIG. 2 and FIG. 7 ).

( 3 ) Third Control Step

After that, at time t 3 in the state that input clock of the clock driver 3 maintains high level (CLK=High), S 2 is changed to on. Thus, two capacitors 1 and 2 are connected in series to the pumping node.

Thus, the capacitors 1 and 2 are charged to voltage of Vdd/2, and each node voltage is: VL 1 ≈0V, VA=VB=2.5V, VC=5V. That is, average output current Iout flows through the MOS transistor M 1 and further flows from output of the clock driver 3 (see FIG. 3 and FIG. 7 ).

(4) Fourth Control Step

Next, at time t 4 in the state the clock CLK is high, the switch S 2 is off. Thus, all of the switches S 1 , S 2 and S 3 become again off. Voltage of each node is maintained in the state (see FIG. 4 and FIG. 7 ).

(5) Fifth Control Step

Next, at time t 5 in the state that all of the switches S 1 , S 2 and S 3 are off, input clock CLK is changed to low level (CLK=Low). Then, by effect of capacitor coupling, each node voltage is: VL 1 ≈0V, VA=2.5V, VB=−2.5V, VC=0V (see FIG. 5 and FIG. 7 ).

(6) Sixth Control Step

Next, at time t 6 in the state that input clock CLK is maintained low level, S 1 and S 3 are turned on. Thus, the capacitors 1 and 2 are connected in parallel to the pumping node. Therefore, each node voltage is: VL 1 ≈−2.5V, VA=0V, VB=−2.5V, VC=0V (see FIG. 6 and FIG. 7 ).

After that, returning to the above-mentioned first control step, the first to sixth steps are repeated.

According to the above-mentioned control method, since voltage of the pumping node VL 1 is depressed to 0V in maximum differing from the conventional example, it is prevented that charge-pump operation is not normally carried out because parasitic diode is biased to forward direction and needless current flows.

FIGS. 8A and 8B are views showing a case that the charge transfer device is made of a P-channel MOS transistor. In this case, though a source and a substrate are earthed in order to depress back gate bias effect, there is not any problem because the parasitic diode is not biased to forward direction in the any cases that the pumping node is 0V and −2.5V.

FIGS. 9A and 9B are views showing a case that the charge transfer device is made of an N-channel MOS transistor. In this case, a drain (pumping node) and a substrate are connected in order to depress back gate bias effect. There is not any problem because the parasitic diode is not biased to forward direction in the any cases that the pumping node is 0V and −2.5V.

The charge-pump circuit of the invention, in short wards, the clock CLK from the clock driver 3 is changed at the state that all of the switches S 1 , S 2 and S 3 are off (state that the capacitors 1 and 2 are separated from the pumping node) first. Second, after changing the clock CLK to high level, the switch S 2 is made on and the capacitors 1 and 2 are connected the pumping node in series. Third, after changing the clock CLK to low level, switches S 1 and S 2 are made on and the capacitors 1 and 2 are connected to the pumping node in parallel. According to the rule, it is avoided that the parasitic diode following to the MOS transistor is biased to forward direction in the case that the charge transfer device of the charge-pump circuit is made of a MOS transistor.

The construction that the MOS transistors M 1 and M 2 for charge transfer are diode-connected in the embodiment generates voltage loss suitable for threshold voltage of the MOS transistors M 1 and M 2 . The invention is not limited to this and is applied for a charge-pump circuit in which the MOS transistors M 1 and M 2 for charge transfer are made on and off alternately in response to the clock CLK and fluctuated voltage (for example, 2 Vdd in absolute value) is supplied to gates thereof when the MCS transistors M 1 and M 2 for charge transfer are turned on.

In this case, these gate voltages are controlled so that M 1 is on and M 2 is off during the capacitors 1 and 2 are connected in series and M 1 is off and M 2 is on during the capacitors 1 and 2 are connected in parallel.

Thus, threshold voltage loss of the MOS transistors M 1 and M 2 are removed, and a charge-pump circuit with high efficiency and large output current is realized because on resistances of the MOS transistors M 1 and M 2 are decreased.

Although the MOS transistors M 1 and M 2 for charge transfer are made of a P-channel MOS transistors in the embodiment, it is not limited to this, is may be made of an N-channel MOS transistor.

Although an applied example for one stage charge-pump circuit outputting boosted voltage of −0.5 Vdd is shown, it is applicable to a two stages charge-pump circuit outputting fluctuated voltage of −1.5 Vdd by increasing number of stage of the charge-pump in the invention. Generally, the invention is applicable to a multi-stages charge-pump circuit assembling the charge-pump circuit of the embodiment as a core. Such the multi-stages charge-pump circuit outputs voltage of −0.5 Vdd at the first stage, and at the second stage or more, the circuit is a general Dicson type charge-pump circuit.

Although the charge-pump circuit of the embodiment is a type carrying out voltage fluctuation of voltage step of −0.5 Vdd changing to two capacitors 1 and 2 to series and parallel, voltage fluctuation of further smaller voltage step can be carried out by changing two or more capacitors to series and parallel. The invention is applicable to such the charge-pump circuit.

Although the charge-pump circuit outputting fluctuated voltage of minus is described in the embodiment, the invention is applicable similarly to a charge-pump circuit having a step of +0.5 Vdd.

According to the charge-pump circuit and the control method thereof of the invention, since it is prevented that the parasitic diode is biased to forward direct on in the charge-pump circuit carrying out voltage fluctuation with a step less than power source voltage by repeating connecting the capacitor to the pumping node in series and in parallel, such the charge-pump operation is carried out normally and the invention has effect improving electric power efficiency.