Semiconductor integrated circuit device and microcomputer

Information

  • Patent Grant
  • 6819158
  • Patent Number
    6,819,158
  • Date Filed
    Thursday, May 29, 2003
    21 years ago
  • Date Issued
    Tuesday, November 16, 2004
    20 years ago
Abstract
A semiconductor integrated circuit comprises a logic circuit which is formed of p-channel MIS transistors and n-channel MIS transistors, a first oscillation circuit of variable oscillation frequency which is formed of p-channel MIS transistors and n-channel MIS transistors, a control circuit which produces a control signal for controlling the threshold voltage of the p-channel and n-channel MIS transistors, and a second oscillation circuit which produces multiple reference clock signals of different frequencies depending on the operation mode. The control circuit receives a reference clock signal and controls the first oscillation circuit with the control signal so that the oscillation frequency of the first oscillation circuit is correspondent to the frequency of the reference clock signal.
Description




BACKGROUND OF THE INVENTION




The present invention relates to a semiconductor integrated circuit, and particularly to a semiconductor integrated circuit device and microcomputer that are intended for fast and low-voltage operation, and to a microcomputer system based on this microcomputer.




In order for a MOS transistor to operate at a low voltage around 1 V, it must have a lower threshold voltage for the enhancement of driving ability and operation speed. However, when the threshold voltage is set too low, the MOS transistor cannot turn off completely due to its subthreshold characteristics (tailing characteristics), causing a subthreshold leakage current to flow, resulting in an increased power consumption, as described in publication “1993 Symposium on VLSI Circuits Digest of Technical Papers”, pp. 45-46 (May 1993).




As the sub-micron MOS transistor pattern design advances, the variation among devices of the fundamental characteristics including the threshold voltage attributable to the unequality of manufacturing process increases, as described in publication “1994 Symposium on VLSI Circuits Digest of Technical Papers”, pp. 13-14 (June 1994).





FIG. 15

shows the variation of threshold voltage in connection with the gate length Lg of a MOS transistor. The variation of threshold voltage due to the gate length variation increases as the gate length Lg becomes shorter.




Assuming the lower limit of threshold voltage to be 0.2 V for making the subthreshold leakage current below a certain value and the above-mentioned process-causing threshold variation to be ±0.15 V, the actual lower limit of threshold voltage as the sum of these values becomes 0.35 V.




On this account, conventional semiconductor integrated circuit devices cannot have their threshold voltage set much lower. Particularly, MOS transistors with lower power voltages operate in a state of incomplete saturation, and the operation speed of MOS transistor circuits falls sharply in response to a slight rise of the threshold voltage. Therefore, it is difficult for the conventional design methodology based on the worst-case consideration to attain the intended performance of semiconductor integrated circuit devices.




SUMMARY OF THE INVENTION




The present invention is intended to overcome the foregoing prior art deficiency.




An object of the invention is to provide a semiconductor integrated circuit device formed of MOS transistors capable of properly arbitrating the conflicting factors of the increased power consumption caused by the subthreshold leakage current and the higher operation speed of MOS transistors.




Another object of the invention is to provide a semiconductor integrated circuit device capable of controlling the threshold voltage properly so that the power consumption and the operation speed are well balanced.




Still another object of the invention is to provide a semiconductor integrated circuit device capable of controlling the threshold voltage easily based on the external clock frequency.




Still another object of the invention is to provide a semiconductor integrated circuit device which is operative at multiple clock frequencies and capable of controlling the threshold voltage in correspondence to a selected clock frequency.




Still another object of the invention is to provide a semiconductor integrated circuit device capable of controlling the threshold voltage optimally for each MOS transistor even in case it is unequal among individual transistors.




Still another object of the invention is to provide a microcomputer which is controlled to operate at an optimally balanced power consumption and operation speed, and a microcomputer system based on the microcomputer.




In order to achieve the above objectives, the inventive semiconductor integrated circuit device comprises MOS transistors that constitute a main circuit and “delay-monitoring” MOS transistors that are provided in correspondence to the main-circuit MOS transistors, with these MOS transistors being operated under the corresponding substrate biases.




The operation speed or delay characteristics of the circuit is detected based on the comparison of a signal produced by the delay-monitoring MOS transistors with a clock signal or other reference timing signal. The substrate bias level is controlled in response to the result of detection so that the main-circuit MOS transistors have an appropriate threshold voltage. The substrate bias level control is a sort of monitor control based on the delay-characteristics detecting transistors and the reference signal.




The difference of the delay characteristics of the circuit from the reference can be detected from a frequency error or phase error of the signal produced by the delay-monitoring MOS transistors with respect to the reference signal.




As a preferable arrangement, an oscillation circuit is designed such that the delay-monitoring MOS transistors determine the output frequency of the oscillation circuit. In this case, the delay characteristics is detected based on frequency error information that takes a continuous signal form derived from the oscillation signal.




According to a preferred embodiment of this invention, a main circuit that is a logic circuit is connected with a substrate-bias dependent oscillation circuit which shares the substrate bias with the main circuit and an operation-mode dependent oscillation circuit which switches the output frequency depending on the operation mode, and a substrate bias control circuit controls the substrate bias of the main circuit by making both oscillation outputs synchronous.




A semiconductor integrated circuit device according to a preferred embodiment of this invention comprises a logic circuit which implements a certain logical operation, a control circuit which controls the threshold voltage of transistors that form the logic circuit, and a variable-frequency oscillation circuit, wherein the logic circuit includes MOS transistors formed on a semiconductor substrate, the oscillation circuit has its output delivered to the control circuit which also receives a reference clock of a certain frequency and adapted to produce a first control signal for controlling the oscillation circuit so that the oscillation output has a frequency correspondent to the frequency of the reference clock, and the MOS transistors of the logic circuit have their threshold voltage controlled by a second control signal that corresponds to the first control signal.




A semiconductor integrated circuit device according to a preferred embodiment of this invention comprises a logic circuit including MIS transistors formed on a semiconductor substrate, a control circuit which controls the threshold voltage of the MIS transistors of the logic circuit, and an oscillation circuit including MIS transistors formed on the semiconductor substrate and adapted to vary the output frequency, wherein the control circuit receives the output signal of the oscillation circuit and a clock signal of a certain frequency and compares the oscillation output frequency with the clock signal frequency to produce a first control signal, the oscillation circuit has its threshold voltage controlled by the first control signal so that the oscillation output has a frequency correspondent to the clock signal frequency, and the MIS transistors of the logic circuit have their threshold voltage controlled by a second control signal that corresponds to the first control signal.




A semiconductor integrated circuit device according to a preferred embodiment of this invention comprises a logic circuit including p-channel MIS transistors and n-channel MIS transistors, a first oscillation circuit of variable frequency including p-channel MIS transistors and n-channel MIS transistors, a control circuit which produces a control signal for controlling the threshold voltages of these p-channel MIS transistors and n-channel MIS transistors, and a second oscillation circuit which produces reference clocks of different frequencies depending on the operation mode, wherein the control circuit receives one of the reference clocks and controls the first oscillation circuit so that the oscillation output has a frequency that corresponds to the frequency of the reference clock.




A semiconductor integrated circuit device according to a preferred embodiment of this invention comprises a logic circuit having at least a first and second circuit blocks, a first and second oscillation circuits of variable frequencies, a first control circuit which controls the threshold voltage of MIS transistors of the first circuit block and first oscillation circuit, a second control circuit which controls the threshold voltage of MIS transistors of the second circuit block and second oscillation circuit, a clock supply circuit which supplies a clock signal of a certain frequency to the first and second control circuits, wherein the first control circuit controls the threshold voltage of the MIS transistors of the first circuit block and first oscillation circuit so that the oscillation output frequency is equal to the frequency of the clock signal and the second control circuit controls the threshold voltage of the MIS transistors of the second circuit block and second oscillation circuit so that the oscillation output frequency is equal to the frequency of the clock signal.




A microcomputer according to a preferred embodiment of this invention comprises a load monitoring means for detecting the volume of processing (load) of the logic circuit formed of the above-mentioned semiconductor integrated circuit device, wherein the load monitoring means operates to vary the frequency of the clock signal depending on the detected volume of processing.




According to the preferred embodiments of this invention, the substrate-bias dependent oscillation circuit provided in the circuit device operates at a known frequency selected for the operation mode of the logic circuit as the main circuit, whereby the threshold voltage of MOS transistors of the oscillation circuit can be controlled optimally depending on the operation mode.




Because of the substrate bias of the substrate-bias dependent oscillation circuit common to that of the main circuit, the threshold voltage of MOS transistors used in the main circuit can be controlled optimally in each operation mode. Consequently, it becomes possible to minimize the power consumption caused by the subthreshold leakage current. The oscillation circuit operates in synchronism with a known frequency selected for each operation mode, allowing easy estimation of the delay of MOS transistors used in the main circuit and facilitating the main circuit design.




According to the preferred embodiments of this invention, the threshold voltage of transistors can be controlled by a simple circuit arrangement based on the reference clock, whereby the high operation speed and the smaller power consumption can be balanced.




According to the preferred embodiments of this invention, the oscillation circuit has its output frequency controlled to the reference clock frequency based on the threshold voltage control, whereby the logic circuit of the semiconductor integrated circuit device can be controlled efficiently.




According to the preferred embodiments of this invention, the logic circuit of the circuit device has its threshold voltage varied to meet each operation mode, whereby it has an optimal operation speed depending on the operation mode.




According to the preferred embodiment of this invention, the logic circuit is controlled by being divided into multiple blocks, whereby the threshold voltage control can be implemented, with the unequality of characteristics among transistors attributable to the manufacturing process being compensated.




According to the preferred embodiment of this invention, it is possible to build a microcomputer in which the internal logic circuit has its operation speed controlled through the monitoring of the volume of processing.











BRIEF DESCRIPTION OF THE DRAWINGS





FIG. 1

is a block diagram of the semiconductor integrated circuit based on an embodiment of this invention;





FIG. 2

is a block diagram of the semiconductor integrated circuit based on another embodiment of this invention;





FIG. 3

is a block diagram of the semiconductor integrated circuit based on still another embodiment of this invention;





FIG. 4

is a block diagram showing more specific circuit arrangement of the embodiment shown in FIG.


3


:





FIGS. 5A and 5B

are schematic diagrams each showing the detailed circuit arrangement of the frequency/phase comparator PFD1 and low-pass filter LPF1 used in the circuit shown in

FIG. 4

;





FIG. 6

is a timing chart showing the relation of the operation frequency and the substrate bias of the embodiment shown in

FIG. 4

;





FIGS. 7A and 7B

are graphs showing the relation of the substrate bias and the threshold voltage of an NMOS transistor and PMOS transistor, respectively;





FIG. 8

is a graph showing the relation of the operation frequency and the substrate bias of a MOS transistor;





FIG. 9

is a graph showing the relation of the gate voltage and the drain current of a MOS transistor;





FIG. 10

is a graph showing the relation of the operation frequency and the power consumption based on an embodiment of this invention;





FIG. 11

is a cross-sectional diagram showing an example of the device structure of the semiconductor integrated circuit based on an embodiment of this invention;





FIG. 12

is a block diagram of the semiconductor integrated circuit based on still another embodiment of this invention;





FIG. 13

is a graph showing the relation of the gate voltage and the drain current of the inventive MOS transistor;





FIG. 14

is a block diagram of the semiconductor integrated circuit based on still another embodiment of this invention;





FIG. 15

is a graph showing the relation of the gate length and the threshold voltage of the inventive MOS transistor;





FIG. 16

is a block diagram of the microcomputer based on an embodiment of this invention; and





FIG. 17

is a block diagram of the microcomputer system based on an embodiment of this invention.











DESCRIPTION OF THE PREFERRED EMBODIMENTS




The embodiments this invention will be explained in detail with reference to the drawings.





FIG. 1

shows the first embodiment of this invention, in which the semiconductor integrated circuit device includes a main circuit LOG


0


such as a logic circuit, a variable-frequency oscillation circuit OSC


0


, and a control circuit CNT


0


.




The oscillation circuit OSC


0


receives a control signal B


1


from the control circuit CNT


0


and has its oscillation frequency dependent on the value of the control signal B


1


. The control circuit CNT


0


receives an external clock signal CLK


0


and the oscillation output S


0


of the oscillation circuit OSC


0


, and produces control signals B


0


and B


1


that are proportional to the difference of frequency between the external clock signal CLK


0


and oscillation output S


0


. The control signal B


0


is fed to the main circuit LOG


0


, while the control signal B


1


is fed to the oscillation circuit OSC


0


.




The control circuit CNT


0


and variable-frequency oscillation circuit OSC


0


in unison form a stable negative feedback system so that the frequency of the oscillation output S


0


of the oscillation circuit OSC


0


is correspondent to the frequency of the external clock signal CLK


0


. For example, these frequencies are equal, and the frequency of S


0


is locked to the frequency of CLK


0


, as will be explained in more detail later.




The oscillation circuit OSC


0


is formed of n-channel MOSFETs and p-channel MOSFEs formed on a semiconductor substrate, with the substrate bias being varied by the control voltage provided by the control circuit CNT


0


. The variation of the substrate bias varies the threshold voltage, and thus varies the oscillation frequency.




The main circuit LOG


0


receives the control signal B


0


from the control circuit CNT


0


, and its MOSFETs have their substrate bias and thus their threshold voltage controlled by the signal B


0


. This circuit arrangement enables to control the threshold voltage of the main circuit and accordingly vary its power consumption and operation speed in response to the external clock frequency.




More specifically, the oscillation circuit OSC


0


receives the control signal B


1


from the control circuit CNT


0


and oscillates at a certain frequency. The oscillation circuit OSC


0


is controlled by the control signal B


1


so that the frequency of oscillation output S


0


corresponds to the frequency of external clock signal CLK


0


. The control signal B


1


controls the substrate bias of MOS transistors of the oscillation circuit OSC


0


thereby to control their threshold voltage, and the MOS transistors operate at such an operation speed as to oscillate at the frequency that corresponds to the frequency of external clock signal CLK


0


.




Accordingly, by controlling the substrate bias of the main circuit LOG


0


with the signal B


0


which corresponds to the control signal B


1


for the substrate bias of the oscillation circuit OSC


0


, it is possible to control the threshold voltage of the MOS transistors of the main circuit LOG


0


so that it operates at the operation speed which corresponds to the external clock signal CLK


0


.




The control circuit CNT


0


consists of a frequency/phase comparator (not shown in detail) which detects the difference of frequency between the external clock signal CLK


0


and the oscillation output S


0


, and a control signal generation circuit which produces in response to the output of the frequency/phase comparator a control signal B


0


to be fed to the main circuit LOG


0


and a control signal B


1


to be fed to the gates of MOSFETs of the oscillation circuit OSC


0


. The frequency/phase comparator, which has a delay detecting function or operation speed detecting function, delivers frequency error information in terms of the delay or operation speed of the main circuit LOG


0


and oscillation circuit OSC


0


. In this invention, the frequency and phase have the same significance in representing the delay or operation speed, although these items should be treated separately in concept since phase control accompanies frequency control as will be appreciated in the following explanation.




The control signals B


0


and B


1


have certain reference levels when the oscillation output S


0


corresponds in frequency to the external clock signal CLK


0


, or vary to increase the substrate bias level if the frequency of S


0


is higher than CLK


0


or decrease the substrate bias level if the frequency of S


0


is lower than CLK


0


.




It is known that a larger substrate bias raises the threshold voltage level of the related MOSFETs and a smaller substrate bias lowers their threshold voltage level. Accordingly, if the control signal B


1


increases the substrate bias, the MOSFETs of oscillation circuit OSC


0


have their conductance decreased accordingly, causing the time constant of OSC


0


, which is determined from the MOSFETs and parasitic capacitance, to increase, resulting in a lower oscillation frequency. Whereas, a decreased substrate bias causes the threshold voltage of MOSFETs to fall and their conductance to increase, resulting in a higher oscillation frequency of the oscillation circuit OSC


0


. In this manner, the oscillation frequency is controlled through the control of threshold voltage of MOSFETs.




In the embodiment of

FIG. 1

, the control signal generation circuit is designed to produce two control signals B


0


and B


1


, which facilitates the circuit design and the achievement of intended response characteristics. In case the main circuit LOG


0


and oscillation circuit OSC


0


need to be supplied with control signals for their individual substrate biasing or in case it is intended to reduce the electrical coupling between the main circuit LOG


0


and oscillation circuit OSC


0


through their substrate, the control signal generation circuit may be two separate circuits for producing the control signals B


0


and B


1


. In this case, the two control signal generation circuits receive the output of the frequency/phase comparator to produce the control signals B


0


and B


1


having an equal signal level. The two control signal generation circuits may have different circuit arrangements.




For example, in case the main circuit LOG


0


has two states of active mode and sleep mode, the control signal B


0


at sleep mode specifies a higher substrate bias in response to a certain sleep-mode command signal irrespective of the detected level of the frequency/phase comparator. With the control signal B


1


being made responsive to the output of the frequency/phase comparator, the oscillation circuit OSC


0


is controlled by the signal. Consequently, the power consumption can be reduced during sleep mode of the main circuit LOG


0


, and the control signal B


0


to the main circuit LOG


0


can be restored quickly when the circuit LOG


0


has regained active mode.




Generally, when a high oscillation frequency is selected for the high-speed operation by setting a lower threshold voltage, the power consumption increases due to the subthreshold leakage current, whereas when a lower oscillation frequency is selected by setting a higher threshold voltage, the power consumption decreases in exchange for a lower operation speed. When the main circuit LOG


0


is a logic circuit that bases the operation on an external clock signal CLK


0


, the circuit operates at the speed corresponding to the external clock CLK


0


through the control of the threshold voltage of MOS transistors of the main circuit LOG


0


. The threshold voltage is controlled so that the power consumption is smaller when the operation speed is lower.




Although in this embodiment the frequency of oscillation output S


0


is controlled through the control of substrate bias of the oscillation circuit OSC


0


, another control scheme may be adopted by making the operation speed of MOS transistors variable in correspondence to the frequency of external clock signal CLK


0


. In this case, the oscillation circuit OSC


0


functions as a means of monitoring the threshold voltage of the main circuit LOG


0


for achieving the operation speed that corresponds to the frequency of external clock signal CLK


0


, and therefore it is necessary to convert the signal B


1


for controlling the oscillation frequency of oscillation circuit OSC


0


into the signal B


0


for controlling the threshold voltage of the main circuit LOG


0


.




Although the foregoing embodiment is designed to vary the substrate bias in order to control the threshold voltage of transistors used in the main circuit LOG


0


, the transistor threshold voltage may be controlled in a different manner instead of the substrate bias control.




In contrast to the conventional variable-frequency oscillation circuit, in which in general the operation speed of the inverter circuit which constitutes a ring oscillator is made variable through the additional provision of transistors for controlling the supply current to the inverter circuit, the circuit arrangement for varying the oscillation frequency through the variation of substrate bias as in this embodiment eliminates the need of these current control transistors and supplies the power voltage to the inverter circuit without a loss, raises the upper limit of oscillation frequency, and expands the range of variable frequency.




The circuit arrangement of this embodiment also enables the oscillation frequency of the oscillation circuit to be locked to the frequency of an external clock signal, whereby it becomes possible to build a phase-locked loop (PLL) circuit which can follow a high-frequency external clock signal.





FIG. 2

shows another embodiment of this invention. The principal portion of this embodiment is identical to the first embodiment, and detailed explanation thereof is omitted. This embodiment differs from the first embodiment shown in

FIG. 1

in the use of a common control signal B


3


in place of the control signals B


0


and B


1


for controlling the threshold voltage and oscillation frequency. Accordingly, the oscillation frequency of the oscillation circuit OSC


0


is controlled through the variation of the substrate bias by the control signal B


3


provided by the control circuit CNT


0


, and the substrate bias of the main circuit LOG


0


is also controlled by the same control signal B


3


.




This embodiment using the common control signal B


3


of the control circuit CNT


0


simplifies the circuit arrangement and yet accomplishes the same effectiveness as the first embodiment. The embodiment enables to build an oscillation circuit having a wide range of frequency and a PLL circuit which can follow a high-frequency external clock signal. It also enables to control the operation speed and power consumption of the main circuit in response to the external clock frequency.





FIG. 3

shows still another embodiment of this invention, in which the substrate bias of MOS transistors is used for the threshold voltage control.




The semiconductor integrated circuit device includes a main circuit LOG


1


and a substrate-bias dependent oscillation circuit OSC


1


which varies the oscillation frequency in response to the substrate bias. A substrate bias control circuit CNT


1


uses the output S


1


of the oscillation circuit OSC


1


and the oscillation output CLK


1


of an operation-mode dependent oscillation circuit VCLK


1


to produce substrate voltages BP


1


and BN


1


for the main circuit LOG


1


.




The oscillation circuit OSC


1


of this embodiment is a ring oscillator made up of series-connected inverters formed of p-channel MOSFETs and n-channel MOSFETs on a semiconductor substrate. The main circuit LOG


1


is a CMOS logic circuit formed of p-channel MOSFETs and n-channel MOSFETs. The substrate bias voltages BP


1


and BN


1


are applied to the n-substrate (or n-well electrode) of PMOS transistors and to the p-substrate (or p-well electrode) of NMOS transistors, respectively, of the main circuit LOG


1


.




The operation-mode dependent oscillation circuit VCLK


1


which operates based on the system clock SCLK


1


as the reference clock to produce an oscillation output CLK


1


having a frequency that depends on an internal or external operation mode signal MODE


1


. The main circuit LOG


1


is a logic circuit which operates based on the oscillation output CLK


1


as the clock in this embodiment.




The substrate bias voltages BP


1


and BN


1


are also supplied to the substrate-bias dependent oscillation circuit OSC


1


. The substrate-bias dependent oscillation circuit OSC


1


and the substrate bias control circuit CNT


1


which receives the output S


1


of the oscillation circuit OSC


1


as a feedback signal form a stable negative feedback system so that the output S


1


of the oscillation circuit OSC


1


has its frequency locked to the frequency of oscillation output CLK


1


of the operation-mode dependent oscillation circuit VCLK


1


.




The operation mode mentioned here is set in advance based on the intended operation speed and power consumption, such as high-speed operation mode, low-speed operation mode and sleep mode, of the main circuit LOG


1


in this embodiment.




Since both substrate bias voltages of the substrate-bias dependent oscillation circuit OSC


1


and main circuit LOG


1


are controlled by the outputs BP


1


and BN


1


of the substrate bias control circuit CNT


1


, it is possible to control the substrate bias of the main circuit LOG


1


by the oscillation output CLK


1


of the operation-mode dependent oscillation circuit VCLK


1


. The MOS transistors have their threshold voltage varied by the substrate bias voltage, and accordingly it is possible to switch the threshold voltage of MOS transistors of the main circuit LOG


1


appropriately to meet the operation mode by switching the operation mode signal MODE


1


.




Since the main circuit LOG


1


has its operation speed and power consumption varied in response to the threshold voltage as mentioned above, it is possible to control the operation speed and power consumption of the main circuit LOG


1


appropriately to meet the operation mode by switching the frequency of the clock signal CLK


1


supplied to the control circuit CNT


1


by use of the operation mode signal MODE


1


.




Although in this embodiment and following embodiments, the oscillation frequency of the operation-mode dependent oscillation circuit VCLK


1


is varied-with the signal indicative of the operation mode thereby to produce the oscillation output CLK


1


, an alternative arrangement is to feed an external clock signal to the control circuit CNT


1


directly, as shown in FIG.


1


and FIG.


2


. In this case, it becomes possible to control the operation speed and power consumption of the internal main circuit LOG


1


through the prior determination of the frequency of the external clock (system clock), and the circuit arrangement can be simplified owing to the elimination of the operation-mode dependent oscillation circuit VCLK


1


.




Although in this embodiment the substrate bias voltages of both of n-type MOSFETs and p-type MOSFETs are controlled with the BP


1


and BN


1


voltages, an alternative arrangement is to use one of BP


1


and BN


1


to vary the substrate bias of MOSFETs of only one type.




More specific circuit arrangements of the foregoing embodiment will be explained in detail in connection with

FIG. 4

,

FIGS. 5A and 5B

, and FIG.


6


.





FIG. 4

shows in detail the internal circuit arrangement of each circuit block shown in FIG.


3


. The substrate-bias dependent oscillation circuit OSC


1


is a ring oscillator made up of 5-stage CMOS inverter circuits (each p-channel or n-channel MOSFET has a gate length of 0.25 μm and a gate width of 5 μm). The MOS transistors of the ring oscillator have their substrate electrodes (or well electrodes) used for the oscillation frequency control line.




The main circuit LOG


1


is a CMOS logic circuit (gate length: 0.25 μm; gate width: 5 μm) of 2-input NAND gate having input signals A and B and an output signal C as an example of this embodiment. The substrate bias control circuit CNT


1


consists of a substrate bias generation circuit BGEN


1


, a low-pass filter LPF


1


and a frequency/phase comparator PFD


1


.

FIG. 5A

shows the specific circuit arrangement of the frequency/phase comparator PFD


1


, and

FIG. 5B

shows the specific circuit arrangement of the low-pass filter LPF


1


.




The frequency/phase comparator PFD


1


, which has a well-known circuit arrangement, receives a reference frequency signal REF


1


and a signal S


1


of a certain frequency and produces an output signal S


2


indicative of the phase difference of the two input signals. The low-pass filter LPF


1


is a well-known lag-lead filter formed of resistors and a capacitor. Further explanation of these circuit blocks are omitted.




The substrate bias generation circuit BGEN


1


is a DC-to-DC converter for converting the output voltage S


3


of the low-pass filter LPF


1


into an optimal voltage level for the substrate bias. The control signal BP


1


delivered by the substrate bias generation circuit BGEN


1


is higher than the power voltage Vdd, while the control signal BN


1


is lower than the ground voltage Vss.




The control signals BP


1


and BN


1


and the output voltage S


3


of the low-pass filter LPF


1


have the following relationship. If the frequency of output S


1


of the oscillation circuit OSC


1


is lower than the frequency of the clock signal CLK


1


as the reference signal REF


1


, the low-pass filter LPF


1


shown in

FIG. 5B

responds to the output S


2


of the frequency/phase comparator PFD


1


shown in

FIG. 5A

to produce a voltage (detected signal) S3 having a relatively lower voltage level close to the ground voltage Vss. In response to the lower detected signal S


3


, the control signal BP


1


has a relatively lower voltage level close to the power voltage Vdd, and the control signal BN


1


has a relatively lower voltage level close to the ground voltage Vss.




Otherwise, if the frequency of the oscillation output S


1


is higher than the frequency of the clock signal CLK


1


, the low-pass filter LPF


1


delivers a detected signal S


3


having a relatively higher voltage level close to the power voltage Vdd. In response to the higher detected signal S


3


, the control signal BP


1


has a relatively higher voltage level above the power voltage Vdd and the control signal BN


1


has a deeply low voltage level below the ground voltage Vss.




The control signals BP


1


and BN


1


have voltage levels that are beyond the operational voltage range Vdd-Vss of the circuit and have a complementary transition of voltage levels in response to the detected signal S


3


. Through the control of conductance of MOSFETs based on the control of substrate bias as described above, the oscillation circuit OSC


1


and main circuit LOG


1


operate as intended.




The substrate bias generation circuit BGEN


1


which produces these control signals BP


1


and BN


1


can be configured in a relatively simple circuit arrangement including a substrate bias oscillation circuit, voltage level control circuit, positive charge pump circuit, and negative charge pump circuit, for example. In this case, the voltage level control circuit controls the level of the oscillation signal fed to the positive charge pump circuit and negative charge pump circuit in accordance with the output S


3


of the low-pass filter LPF


1


.




The voltage level control circuit, which receives the output S


3


of the low-pass filter LPF


1


, is configured as a voltage follower circuit formed of a MOS operational amplifier, for example. The output of the voltage follower circuit is considered to be an internal voltage source, and it is used at least for the power source of the oscillation signal output circuit of the substrate bias oscillation circuit. Based on this circuit arrangement, the oscillation signal output circuit has its output signal level determined by that power voltage, and it is controlled to have a voltage level which corresponds to the voltage level of the output S


3


of the low-pass filter LPF


1


.




The positive charge pump circuit is made up of a positive charge pump capacitor which consists of a MOS capacitor having its one terminal coupled to the output terminal of the oscillation signal output circuit, a first rectifying element which consists of a p-channel MOSFET in diode configuration having its source connected to the power terminal Vdd and its gate and drain connected to another terminal of the positive charge pump capacitor, and a second rectifying element which consists of a p-channel MOSFET in diode configuration having its source connected to the other terminal of the capacitor and its gate and drain connected to the substrate gates of the p-channel MOSFETs of the main circuit LOG


1


and oscillation circuit OSC


1


. The positive charge pump circuit provides periodically a positive charge to the substrate gate of the p-channel MOSFETs in response to the oscillation signal from the substrate bias oscillation circuit. This positive charge is smoothed by a parasitic capacitance such as a junction capacitance existing in the substrate gate, and it provides the substrate gate with a bias voltage as the control signal BP


1


.




The negative charge pump circuit is made up of an negative charge pump capacitor which consists of a MOS capacitor having its one terminal coupled to the output terminal of the oscillation signal output circuit, a third rectifying element which consists of an n-channel MOSFET in diode configuration having its source connected to the ground terminal Vss and its gate and drain connected to another terminal of the negative charge pump capacitor, and a fourth rectifying element which consists of an n-channel MOSFET in diode configuration having its source connected to substrate gates of the n-channel MOSFETs of the main circuit LOG


1


and oscillation circuit OSC


1


and its gate and drain connected to the other terminal of the negative charge pump capacitor.




In case another voltage source such as an external power source having an output voltage level higher than the power voltage Vdd is available, the circuit for producing the control signal BP


1


in the substrate bias generation circuit BGEN


1


may be arranged as follows. A level conversion circuit or level shift circuit which is supplied with power by the above-mentioned another voltage source is provided, by which the control signal BP


1


is produced from the output of the low-pass filter LFP


1


through the conversion or level shift. Similarly, in case another voltage source such as an external power source having an output voltage level lower than the ground voltage Vss is available, the control signal BN


1


may be produced by use of a similar level conversion circuit or level shift circuit.




In the case of using these level conversion circuit or level shift circuit, it becomes possible to reduce the number of relatively large circuit components such as the charge pump capacitors, and also reduce the transitional current of charge pump operation which can create noises in the power wiring system.




In

FIG. 4

, the operation-mode dependent oscillation circuit VCLK


1


is made up of a frequency multiplier PLL


1


and operation mode control circuit MCU


1


. A fixed-frequency oscillator CK


1


is used for the reference clock source of the operation-mode dependent oscillation circuit VCLK


1


.




The frequency multiplier PLL


1


has its multiplying factor varied by the operation mode control circuit MCU


1


in accordance with a mode control signal MODE associated with the operation of the main circuit LOG


1


. The output frequency of the fixed-frequency oscillator CK


1


is multiplied by a multiplying factor that depends on the operation mode of the main circuit LOG


1


, and a resulting oscillation output CLK


1


with a rectangular waveform and a frequency f0 (ranging from 0 to 100 MHz) is released from the frequency multiplier PLL


1


.




The oscillation output S


1


of the CMOS ring oscillator OSC


1


is fed together with the oscillation output CLK


1


of the frequency multiplier PLL


1


to the frequency/phase comparator PFD


1


, which has its output S


2


fed through the low-pass filter LPF


1


to the substrate bias generation circuit BGEN


1


, by which the substrate bias voltages BP


1


and BN


1


are produced.




The substrate bias voltages BP


1


and BN


1


are applied commonly to the CMOS ring oscillator OSC


1


and main circuit LOG


1


. The substrate bias voltage BN


1


is applied commonly to the substrate electrodes of n-type MOSFETs (transistors with gate-pointing arrows at the channel section), while the substrate bias voltage BP


1


is applied commonly to the substrate electrodes of p-type MOSFETs (transistors with gate-originating arrows at the channel section), as shown in FIG.


4


. These transistors have their operating point set so as to operate in a low voltage range of about 1 V based on a power voltage Vdd of about 1 V and a ground voltage Vss.





FIGS. 7A and 7B

show the relationship of the substrate bias voltage and threshold voltage of the MOS transistors.

FIG. 7A

is for the NMOS transistors, and

FIG. 7B

is for the PMOS transistors.




The MOS transistors have their threshold voltage varied depending on the substrate bias as shown in

FIGS. 7A and 7B

and have a reduced current drive ability for a smaller value of threshold voltage in general, and accordingly the CMOS ring oscillator OSC


1


of

FIG. 4

operates as a voltage-controlled oscillator (VCO) in response to the substrate bias signals BP


1


and BN


1


. On this account, the whole circuit of

FIG. 4

has a structure of phase-locked loop (PLL) and operates so that the oscillation frequency and phase of the CMOS ring oscillator OSC


1


coincide with those of the oscillation output CLK


1


of the frequency multiplier PLL


1


.




The oscillation circuit (VCO) and PLL circuit of this embodiment enable to build an oscillation circuit having a wide frequency range and a PLL circuit capable of following an external high-frequency clock signal, and also enable the control of operation speed and power consumption of the main circuit depending on the external clock frequency. The PLL circuit is described in publication “IEEE JOURNAL OF SOLID-STATE CIRCUITS”, Vol.22, No. 2 (April 1987), pp. 255-261, for example.




Next, the variation of the substrate bias voltages BP


1


and BN


1


and threshold voltage of the MOS transistors of the main circuit LOG


1


in response to the switching of operation mode of the main circuit LOG


1


will be explained in connection with the timing chart of FIG.


6


.




When the operation mode is switched, the operation mode control circuit MCU


1


switches the frequency multiplying factor of the frequency multiplier PLL


1


, and its oscillation output CLK


1


varies. In this example, the frequency f0 varies with time from 75 MHz to 50 MHz. Consequently, the substrate bias voltage BP


1


rises from 1.3 V to about 1.6 V, and the substrate bias voltage BN


1


falls from −0.3 V to about −0.6 V. Consequently, the threshold voltage of MOS transistors rises (refer to FIGS.


7


A and


7


B), causing the oscillation frequency f1 of the substrate-bias dependent oscillation circuit OSC


1


to fall gradually to become equal to the frequency f0 (50 MHz) of oscillation output CLK


1


of the frequency multiplier PLL


1


. At the same time, the output of the circuit OSC


1


becomes in-phase with the oscillation output CLK


1


of the frequency multiplier PLL


1


.




If the frequency f0 of the oscillation output CLK


1


of the frequency multiplier PLL


1


rises from 50 MHz to 100 MHz in response to the switching of the operation mode signal MODE, the substrate bias voltage BP


1


falls from 1.6 V to about 1.0 V and the substrate bias voltage BN1 rises from −0.6 V to about 0 V. Consequently, the threshold voltage of MOS transistors falls (refer to FIGS.


7


A and


7


B), causing the oscillation frequency f1 of the substrate-bias dependent oscillation circuit OSC


1


to rise gradually to become equal to the frequency f0 (100 MHz) of oscillation output CLK


1


of the frequency multiplier PLL


1


.





FIG. 8

shows the relationship of the frequency f0 of oscillation output CLK


1


and the substrate bias voltages BP


1


and BN


1


of the frequency multiplier PLL


1


. In this manner, the substrate bias of the MOS transistors of the substrate-bias dependent oscillation circuit OSC


1


can be varied depending on the oscillation output CLK


1


of the frequency multiplier PLL


1


, and the threshold voltage of the MOS transistors can be varied accordingly.




Although the frequency f0 of oscillation output CLK


1


is switched among 50 MHz, 75 MHz and 100 MHz in this embodiment, arbitrary oscillation frequencies that meet the control range of threshold voltage can be chosen obviously. It is possible for this embodiment to set the oscillation frequency f1 of the CMOS ring oscillator OSC


1


to a preset value, enabling the estimation of the signal propagation delay time of MOS transistors of the main circuit LOG


1


in each operation mode.





FIG. 9

shows the gate voltage dependency of MOS transistors in their subthreshold region of the drain current. Generally, the lower the threshold voltage of MOS transistors, the greater is their driving ability, enabling higher-speed operation. However, the comparison of points A and


6


in

FIG. 9

reveals that the on-off ratio of MOS transistors decreases, the subthreshold current increases, and the power consumption increases.




According to the inventive circuit arrangement, it is possible in case in need of high-speed operation to lower the threshold voltage of MOS transistors of the main circuit LOG


1


by choosing from outside an operation mode for a higher oscillation frequency of the output CLK


1


of frequency multiplier PLL


1


. Although in this case the subthreshold current and power consumption of main circuit LOG


1


increase, it can operate at a higher speed. It is also possible in case in need of low-speed operation to raise the threshold voltage of MOS transistors of the main circuit LOG


1


by choosing from outside an operation mode for a lower oscillation frequency of the output CLK


1


of frequency multiplier PLL


1


, thereby reducing the subthreshold current and power consumption of the main circuit LOG


1


.





FIG. 10

shows these affairs. Shown by curve P


2


is the power consumption of the main circuit LOG


1


when the operation mode signal MODE is varied to vary the frequency f0 of oscillation output CLK


1


of the frequency multiplier PLL


1


. The main circuit LOG


1


operates at the same frequency f0 of the oscillation output CLK


1


of PLL


1


in this example. Shown by line P


0


is the power consumption of LOG


1


of the case without power a loss in subthreshold current T0, and shown by line P


1


is the case with the subthreshold current T0.




The power consumption of the main circuit LOG


1


has a linear relationship with its operation frequency f0 as shown. The subthreshold current does not depend on the operation frequency f0, and line P


1


is parallel to line P


0


. In the case of the line P


2


relevant to this invention, the threshold voltage which relates with the subthreshold current varies with the operation frequency f0, and the power consumption approaches the P


0


without power loss in subthreshold current as the operation frequency f0 falls in response to the operation mode signal. Consequently, the main circuit LOG


1


can be operated at a minimal power consumption caused by a minimal subthreshold current needed for the operation at the operation frequency f0. This effect is particularly significant for the low-voltage operation at around 1 V in which case the subthreshold current is a matter and for the higher-density circuit integration.




Since the threshold voltage of the main circuit LOG


1


is controlled automatically to take the preset value depending on the operation mode of the main circuit LOG


1


the inventive circuit arrangement can cope with unequal characteristics among MOS transistors of the main circuit LOG


1


and external varying factors such as the temperature and power voltage.




Although shown in

FIG. 4

is a 2-input NAND gate as an example of the main circuit LOG


1


of

FIG. 3

, it can be an inverter or other logic gates such as a NOR gate, or can be a logic circuit made up of these logic gates. The main circuit LOG


1


of CMOS transistors can alternatively be formed of NMOS transistors and/or PMOS transistors, or even can include bipolar transistors. The CMOS ring oscillator shown in

FIG. 4

can also be any other oscillation circuit that varies the output frequency in response to the substrate bias.





FIG. 11

shows the cross section of the CMOS structure for fabricating the integrated circuit shown in FIG.


4


.




N well


116


and p well


118


are formed on part of a p-type silicon substrate


117


. Formed on the surface of the n well


116


is a PMOS transistor consisting of a p+ type source, drain diffusion regions


104


and


105


, a gate electrode


110


and a gate oxide film


113


, and formed on the surface of the p well


118


is an NMOS transistor consisting of an n-+ type source, drain diffusion regions


108


and


109


, a gate electrode


108


and a gate oxide film


115


. Device isolation films


100


,


101


and


102


are formed between the PMOS transistor and NMOS transistor. Power lines for supply the substrate bias voltages BP


1


and BN


1


to the PMOS transistor and NMOS transistor are connected to the respective well regions, although these lines are not shown in the figure.




An n-type substrate may be used instead of the p-type substrate of this example. Instead of the twin well structure using the n well


116


and p well


118


in the example of

FIG. 11

, a single well structure, with one of the n well and p well being substituted by the substrate, may be used.




MOS transistors of the triple well structure as described in publication “ISSCC Digest of Technical Papers” (February 1989), pp. 248-249, or MOS transistors of the SOI structure as described in publication “1992 IEDM Technical Digest”, pp. 35-38 may be formed.




Although in the foregoing embodiments, the substrate bias voltage for NMOS transistors is 0 V or lower and that of PMOS transistors is the power voltage (e.g., 1.0 V) of the main circuit or higher, it may be applied so as to provide a forward bias for the pn junction between the diffusion layer of MOS transistors and the substrate.




Particularly, in case the forward bias voltage does not exceed the diffusion potential (around 0.6 V), the leakage current between the diffusion layer and substrate is small enough to prevent an increased power consumption, enabling the application of a forward bias. In this case, the substrate bias factor of the threshold voltage (the variation of threshold voltage against the substrate bias) is large in the substrate bias region, and therefore the threshold voltage of MOS transistors can be controlled efficiently. Another advantage is the elimination of the need of a special substrate biasing circuit such as a negative voltage source owing to the setting of the substrate bias voltage within the range of power voltage.




Although in the foregoing embodiments, the threshold voltage of MOS transistors of the main circuit is controlled through the control of substrate bias, an alternative scheme is to configure a main circuit with MOS transistors having a threshold voltage control terminal (e.g., SOIMOSFETs having an electrode that is insulated from the silicon substrate) and apply a substrate bias control voltage to the terminal.





FIG. 12

shows an embodiment of the configuration of a main circuit LOG


1


that is divided into circuit blocks LOG


10


, LOG


20


and LOG


30


. The circuit blocks LOG


10


-LOG


30


have their substrate bias voltages BP


10


-BP


30


and BN


10


-BN


30


controlled by the oscillation output CLK


10


of a single operation-mode dependent oscillation circuit VCLK


10


. The oscillation output CLK


10


of the circuit VCLK


10


is fed commonly to the control circuits CNT


10


-CNT


30


that are provided along with oscillation circuits OSC


10


-OSC


30


for the respective circuit blocks LOG


10


-LOG


30


. The operation-mode dependent oscillation circuit VCLK


10


and each of the control circuits CNT


10


-CNT


30


and oscillation circuits OSC


10


-OSC


30


are arranged as shown in FIG.


4


.




In this embodiment, the substrate bias of the circuit blocks LOG


10


-LOG


30


are controlled independently by the oscillation output CLK


10


of the operation-mode dependent oscillation circuit VCLK


10


. Accordingly, it is possible to compensate the unequality of threshold voltage and substrate bias characteristics of MOS transistors among the circuit blocks LOG


10


-LOG


30


. In case MOS transistors of the main circuit LOG


10


have a threshold voltage different from that of other circuits due to the manufacturing process, MOS transistors of the oscillation circuit OSC


10


conceivably have the same variation of threshold voltage, and the substrate bias for these MOS transistors can be controlled appropriately by the clock signal CLK


10


. Specifically, in case the threshold voltage differs by 0.15 V among the circuit blocks LOG


10


-LOG


30


, the difference can be reduced to around 0.05 V according to this embodiment.




Since the unequality of circuit characteristics attributable to the manufacturing process is dependent on the location on the semiconductor chip of the semiconductor integrated circuit, it is desirable to arrange each main circuit LOG and associated oscillation circuit OSC close to each other. Similarly, in dividing a main circuit into circuit blocks, it is desirable to partition a semiconductor chip symmetrically into two, four, or the like so that a circuit block is configured with transistors that are close to each other.




The allowable range of threshold voltage has its lower limit determined from the subthreshold leakage current and its upper limit determined from the required operation speed, as mentioned previously. On this account, it is obliged conventionally to set a higher threshold voltage against high-speed circuit operation when the threshold voltage differ greatly among transistors, whereas the scheme of this embodiment allows the setting of threshold voltage to its lower limit, enabling the high-speed circuit operation.




The effectiveness of these embodiments is particularly significant for the low-voltage operation at a power voltage of around 1 V.





FIG. 13

shows this affair, showing the variation of drain current when the threshold voltage varies against the gate voltage applied to the gate terminal. For a relatively high power voltage of VDD


1


=


2


.


0


V, there is little variation of drain current caused by the variation of threshold voltage as shown by the distance of points A


1


and B


1


, whereas for a relatively low power voltage of VDD2=1.0 V, the drain current is varied greatly by the variation of threshold voltage as shown by the distance of points A


2


and B


2


. The distance of these points further expands when the power voltage is 1 V or lower.




As a variant embodiment derived from the embodiment of

FIG. 12

, in which the circuit blocks LOG


10


-LOG


30


are supplied with a common oscillation output CLK


10


from one operation-mode dependent oscillation circuit VCLK


10


, two operation-mode dependent oscillation circuits VCLK


20


and VCLK


30


may be provided as shown in FIG.


14


. The two oscillation circuits VCLK


20


and VCLK


30


are supplied with a common system clock SCLK


20


and mode signal MODE


20


.




The circuit blocks LOG


10


-LOG


30


have individual oscillation circuits OSC


10


-OSC


30


and control circuits CNT


10


-CNT


30


, of which the control circuits CNT


10


and CNT


20


are commonly supplied with the oscillation output CLK


20


of the oscillation circuit VCLK


20


and the control circuit CNT


30


is supplied with the oscillation output CLK


30


, of the oscillation circuit VCLK


30


.




The operation-mode dependent oscillation circuits VCLK


20


and VCLK


30


, control circuits CNT


10


-CNT


30


, and oscillation circuits OSC


10


-OSC


30


are arranged as shown in FIG.


4


. The scheme of controlling multiple circuit blocks LOG


10


and LOG


20


with a single operation-mode dependent oscillation circuit VCLK


20


is identical to the embodiment of

FIG. 12

, and the same effectiveness is attained.




Based on this circuit arrangement, circuit blocks, e.g., LOG


10


and LOG


30


, can be controlled separately with different operation-mode dependent oscillation circuits, e.g., VCLK


20


and VCLK


30


, so that individual circuit blocks can operate under suitable substrate bias control. For example, based on the setting of optimal threshold voltages for individual circuit blocks, a circuit block can operate at a higher speed in exchange for a larger subthreshold leakage current and another circuit block can operate at a smaller subthreshold leakage current in exchange for a lower operation speed.




In regard to the unequality of threshold voltage among transistors as mentioned previously in connection with the embodiment of

FIG. 12

, it is desirable for this embodiment to divide the main circuit into circuit blocks on a semiconductor chip based on the operation speed so that transistors located close to each other are allotted to a same circuit block.




Although in

FIG. 14

, the operation mode signal MODE


20


is fed commonly to the oscillation circuits VCLK


20


and VCLK


30


, these oscillation circuits may be supplied with different operation mode signals so as to produce different output frequencies. This modified circuit arrangement enables a certain circuit block including a main circuit for a specific logical operation to operate at a higher or lower speed. For example, in case a circuit block LOG


30


is a microprocessor having a floating-point arithmetic unit, only this circuit block can be controlled to operate at a higher speed during the floating-point arithmetic process.




Although in this embodiment, the circuit blocks LOG


10


and LOG


20


are controlled commonly with the oscillation output CLK


20


of the operation-mode dependent oscillation circuit VCLK


20


, these circuit blocks may be controlled by oscillation outputs of different operation-mode dependent oscillation circuits.




Although in the embodiments of FIG.


12


and

FIG. 14

, the main circuit and its associates are divided into three circuit blocks, they may be divided into a larger number of circuit blocks, and the above-mentioned effectiveness becomes more pronounced as the circuit is divided into much smaller circuit blocks.




The circuit blocks LOG


10


-LOG


30


may be formed on a single LSI chip, or may be formed on multiple LSI chips dividedly. In the latter case, substrate bias control circuits (CNT


10


, etc.) and corresponding oscillation circuits (OSC


10


, etc.) are formed on individual semiconductor chips and an operation-mode dependent oscillation circuit is formed commonly for these semiconductor chips, for example. It is desirable to form oscillation circuits (OSC


10


, etc.), which function to monitor the threshold voltage of main circuits, on the same semiconductor chips of the main circuits so that their threshold voltages are controlled accurately.





FIG. 16

shows the arrangement of a microcomputer based on an embodiment of this invention. The microcomputer


500


, which is formed on a single semiconductor substrate based on the CMOS technology or the like, receives a clock signal


502


from a fixed-frequency crystal oscillator


501


and an operation mode signal


503


. The operation mode signal


503


which consists of one bit or more is produced by a load detector


505


within the microcomputer


500


. The load detector


505


detects the volume of processing of the microcomputer


500


and produces the operation mode signal


503


which controls the operation speed of the microcomputer


500


to meet the volume of processing.




The load detector


505


may be formed of a semiconductor circuit included in the microcomputer


500


, or may be accomplished in the form of a program executed by the microcomputer


500


. Alternatively, it may be an external device adapted to monitor the load of the microcomputer


500


.




In case the microcomputer


500


runs at a small load, the load detector


505


delivers an operation mode signal of a small value, by which the oscillation circuit VCLK


1


is controlled to produce a lower operation clock signal


504


which determines the operation speed and power consumption of the microcomputer


500


. If the load of the microcomputer


500


increases, the load detector


505


produces an operation mode signal


503


of a large value, and the microcomputer


500


runs at an increased operation clock frequency to meet the load condition.




In this manner, the microcomputer


500


can run at an optimal operation clock frequency that meets the load condition. The operation mode signal


503


may be an analog signal instead of a multi-bit digital signal.





FIG. 17

shows a microcomputer system based on an embodiment of this invention. The microcomputer system


600


which includes the microcomputer shown in

FIG. 16

is connected with an input device


601


such as a keyboard or mouse device through a system bus


602


which is controlled by the microcomputer


500


.




A load detector


505


monitors the active time of the input device


601


and produces an operation mode signal


503


. During a period of the frequent operation of the input device


601


, the operation mode signal


503


specifies such an operation mode that the microcomputer


500


runs at a higher clock frequency, or during the infrequent operation of the input device


601


, the microcomputer


500


runs in operation mode of a lower clock frequency.




Generally, the infrequent operation of the input device


601


signifies a small load of the microcomputer


500


. Based on this manner of load detection for the microcomputer


500


, the power consumption can be reduced without reducing the practical operation speed of the microcomputer system.




Although the microcomputer system of

FIG. 17

uses the input device


601


such as a keyboard or mouse device for the estimation of the load of microcomputer, the load condition may be detected in a different manner. For example, the load of the microcomputer may be estimated by measuring the time length of user's CPU access. It is also possible for the system to allow the computer user to set the operation mode through the input device


601


.




According to the foregoing embodiments of this invention, the threshold voltage of MOS transistors of the main circuit LOG


1


can be controlled by use of the operation mode signal MODE


1


. For example, for the higher-speed operation of the main circuit, an operation mode in which the operation-mode dependent oscillation circuit VCLK


1


produces a higher-frequency output CLK


1


is specified from outside so that the threshold voltage of MOS transistors of the main circuit LOG


1


is lowered. In this case, the main circuit LOG


1


operates fast in exchange for an increased power consumption due to an increased subthreshold current. For the lower-speed operation of the main circuit, an operation mode in which-the operation-mode dependent oscillation circuit VCLK


1


produces a lower-frequency output CLK


1


is specified from outside so that the threshold voltage of MOS transistors of the main circuit LOG


1


is raised. In this case, the main circuit LOG


1


operates at a reduced power consumption owing to a reduced subthreshold current.



Claims
  • 1. A microcomputer comprising:a logic circuit; a first circuit which receives an operation mode signal and supplies a clock signal having a frequency according to the operation mode signal a second circuit which outputs a monitoring signal having a frequency; and a control circuit which receives the clock signal and the monitoring signal and generates a control signal to the logic circuit and the second circuit to control an operation speed of the logic circuit and a frequency of the monitoring signal, wherein the control circuit controls the frequency of the monitoring signal to synchronize the clock signal with the monitoring signal, wherein the operation mode signal is set according to loads of the microcomputer.
  • 2. The microcomputer according to claim 1, wherein an amount of subthreshold leakage current of the logic circuit is a first value when the frequency of the clock signal is a first frequency, and the amount of subthreshold leakage current of the logic circuit is a second value lower than the first value when the frequency of the clock signal is a second frequency lower than the first frequency.
  • 3. The microcomputer according to claim 1, wherein the control signal controls threshold voltages of MIS transistors of the logic circuit and the second circuit so as to lock the frequency of the monitoring signal to the frequency of the clock signal.
  • 4. The microcomputer according to claim 1, further comprising:a load detector which detects the loads of the microcomputer and produces the operation mode signal.
  • 5. The microcomputer according to claim 1, wherein a program executed by the microcomputer produces the operation mode signal.
  • 6. The microcomputer according to claim 1, wherein an external device of the microcomputer produces the operation mode signal.
  • 7. The microcomputer according to claim 1, wherein the second circuit is formed on a substrate which the logic circuit is formed thereon.
  • 8. A method of controlling power consumption of a microcomputer comprising:providing a microcomputer having a logic circuit, a first circuit and a second circuit, setting an operation mode signal according to loads of the microcomputer; supplying a clock signal generated by the first circuit, which has a frequency according to the operation mode signal, to the logic circuit; outputting a monitoring signal from the second circuit having a frequency; comparing the clock signal and the monitoring signal; and generating a control signal and the second circuit according to a result of comparing in order to control an operation speed of the logic circuit and a frequency of the monitoring signal, wherein the control signal controls the second circuit so as to synchronize the clock signal with the monitoring signal.
  • 9. The method of controlling power consumption according to claim 8, wherein an amount of subthreshold leakage current of the logic circuit is a first value when the frequency of the clock signal is a first frequency, and the amount of subthreshold leakage current of the logic circuit is a second value lower than the first value when the frequency of the clock signal is a second frequency lower than the first frequency.
  • 10. The method of controlling power consumption according to claim 8, wherein the control signal controls threshold voltages of MIS transistors of the logic circuit and the second circuit so as to lock the frequency of the clock signal to the frequency of the monitoring signal.
  • 11. The method of controlling power consumption according to claim 8, wherein the microcomputer detects the loads of the microcomputer and produces the operation mode signal.
  • 12. The method of controlling power consumption according to claim 8, wherein a program executed by the microcomputer produces the operation mode signal.
  • 13. The method of controlling power consumption according to claim 8, wherein an external device of the microcomputer produces the operation mode signal.
  • 14. The method of controlling power consumption according to claim 8, wherein the second circuit is formed on a substrate which the logic circuit is formed thereon.
Priority Claims (1)
Number Date Country Kind
7-071136 Mar 1995 JP
Parent Case Info

This is a continuation application of Ser. No. 10/241,505, filed Sep. 12, 2002; now U.S. Pat. No. 6,597,220, which is a continuation application of Ser. No. 09/994,645, filed Nov. 28, 2001, now U.S. Pat. No. 6,472,916; which is a continuation application of Ser. No. 09/688,234, filed Oct. 16, 2000, now U.S. Pat. No. 6,388,483; which is a continuation application of Ser. No. 09/415,220, filed Oct. 12, 1999, now U.S. Pat. No. 6,166,577; which is a divisional application of U.S. Ser. No. 08/622,389, filed Mar. 27, 1996, now U.S. Pat. No. 6,608,509.

US Referenced Citations (16)
Number Name Date Kind
4899071 Morales Feb 1990 A
4922141 Lofgren et al. May 1990 A
5039893 Tomisawa Aug 1991 A
5146121 Searles et al. Sep 1992 A
5184027 Masuda et al. Feb 1993 A
5216302 Tanizawa Jun 1993 A
5231319 Crafts et al. Jul 1993 A
5252867 Sorrells et al. Oct 1993 A
5420531 Wetlaufer May 1995 A
5420544 Ishibashi May 1995 A
5461338 Hirayama et al. Oct 1995 A
5475344 Maneatis et al. Dec 1995 A
5557231 Yamaguichi et al. Sep 1996 A
5568072 Schoellkopf Oct 1996 A
5703522 Arimoto et al. Dec 1997 A
6081146 Shiochi et al. Jun 2000 A
Foreign Referenced Citations (8)
Number Date Country
62-272619 Nov 1987 JP
63-69315 Mar 1988 JP
1-293559 Nov 1989 JP
4-247653 Sep 1992 JP
5-108194 Apr 1993 JP
5-152935 Jun 1993 JP
5-235714 Sep 1993 JP
6-089574 Mar 1994 JP
Non-Patent Literature Citations (5)
Entry
1993 Symposium on VLSI Circuits Digest of Technical Papers, May 1993, pp. 45-46.
1994 Symposium on VLSI Technology Digest of Technical Papers, Jun. 1994, pp. 13-14.
IEEE Journal of Solid-State Circuits, vol. 22, No. 2, Apr. 1987, pp. 255-261.
ISSCC Digest of Technical Papers, Feb. 1989, pp. 248-249.
1992 IEDM Technical Digest, pp. 35-38.
Continuations (4)
Number Date Country
Parent 10/241505 Sep 2002 US
Child 10/446797 US
Parent 09/994645 Nov 2001 US
Child 10/241505 US
Parent 09/688234 Oct 2000 US
Child 09/994645 US
Parent 09/415220 Oct 1999 US
Child 09/688234 US