|Publication number||USRE35313 E|
|Application number||US 07/875,088|
|Publication date||Aug 13, 1996|
|Filing date||Apr 28, 1992|
|Priority date||Apr 17, 1981|
|Publication number||07875088, 875088, US RE35313 E, US RE35313E, US-E-RE35313, USRE35313 E, USRE35313E|
|Inventors||Ryoichi Hori, Kiyoo Itoh, Hitoshi Tanaka|
|Original Assignee||Hitachi, Ltd., Hitachi Micro Computer Engineering, Ltd.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (19), Non-Patent Citations (6), Referenced by (3), Classifications (21), Legal Events (3)|
|External Links: USPTO, USPTO Assignment, Espacenet|
.[.This application is a continuation of application Ser. No. 562,969, filed December 19, 1983, now abandoned..].
.Iadd.This is a reissue of application Ser. No. 140,628, filed Jan. 4, 1988, now U.S. Pat. No. 4,916,389, which is a Continuation of application Ser. No. 562,969, filed Dec. 19, 1983, now abandoned, which is a Continuation-In-Part of application Ser. No. 368,162, filed Apr. 14, 1982, now U.S. Pat. No. 4,482,958..Iaddend.
The present invention relates to a voltage converter which lowers an external supply voltage within a semiconductor integrated circuit chip to drive circuits on the chip having small geometries.
Reduction in the geometries of devices such as bipolar or MOS transistors has been accompanied by a lowering in the breakdown voltages of the devices, which has made it necessary to lower the operating voltage of small geometry devices with an integrated circuit. From the viewpoint of users, however, a single voltage source of for example 5 V which is easy to use is desirable. As an expedient for meeting such different requests of IC manufacturers and the users, it is considered to be necessary to lower the external supply voltage VCC within a chip and to operate the small geometry devices with the lowered voltage VL.
FIG. 1 shows an example of such an expedient, in which the circuit A' of the whole chip 10 including, e.g., an input/output interface circuit is operated with the internal supply voltage VL lowered by a voltage converter 13.
FIG. 2 shows an integrated circuit disclosed in Japanese Patent Application No. 56-57143 (Japanese Laid-open Patent Application No. 57-172761). The small geometry devices are employed for a circuit A determining the substantial density of integration of the chip 10, and are operated with the voltage VL obtained by lowering the external supply voltage VCC by means of a voltage converter 13. On the other hand, devices of comparatively large geometries are employed for a driver circuit B including, e.g., an input/output interface which does not greatly contribute to the density of integration which are operated by applying VCC thereto. Thus, a large-scale integrated circuit (hereinbelow, termed "LSI") which operates with VCC when viewed from outside the chip becomes possible.
However, when such an integrated circuit is furnished with the voltage converter, an inconvenience is involved in an aging test which is performed after the final fabrication step of the integrated circuit.
The terminology "aging test" as used herein identifies a test performed after the final fabrication step of the integrated circuit during which voltages higher than in an ordinary operation are intentionally applied to the respective transistors in the circuit to test the integrated circuit for break down due to an inferior gate oxide film.
The aforementioned voltage converter in Japanese Patent Application No. 56-57143 functions to feed the predetermined voltage. Therefore, the circuit fed with the supply voltage by the voltage converter cannot be subjected to the aging test.
In order to solve this problem, an invention disclosed in U.S. Pat. No. 4,482,985 has previously been made, but it has had difficulty in the performance for actual integrated circuits. As illustrated in FIGS. 2 to 6 in that patent, according to the cited invention, an internal voltage increases up to an aging point rectilinearly or with one step of change as an external supply voltage increases. Accordingly, the internal voltage changes greatly with the change of the external supply voltage. This has led to the disadvantage that the breakdown voltage margins of small geometry devices in an ordinary operation become small.
An object of the present invention is to further advance the invention disclosed in U.S. Pat. No. 4,482,985 referred to above, and to provide a voltage converter which can widen the margins of the breakdown voltages of small geometry devices in an ordinary operation and which affords sufficient voltages in an aging test.
The present invention consists in that the output voltage of a voltage converter is set at a voltage suitable for the operations of small geometry devices against the change of an external supply voltage when a semiconductor integrated circuit is in its ordinary operation region, and at an aging voltage when the ordinary operation region is exceeded.
To this end, according to the voltage converter of the present invention, when the external supply voltage has been changed from the lower limit value of the ordinary operation range thereof to the aging operation point thereof, the output voltage of the voltage converter changes up to the aging voltage without exhibiting a constant changing rate versus the change of external supply voltage.
FIGS. 1 and 2 show semiconductor integrated circuits each having a voltage converter.
FIGS. 3 and 5 show basic circuits each of which constitutes a device embodying the present invention.
FIGS. 4 and 6 show the characteristics of the circuits in FIGS. 3 and 5, respectively.
FIGS. 7, 9 and 11 show devices embodying the present invention.
FIGS. 8, 10 and 12 show the characteristics of the circuits in FIGS. 7, 9 and 11, respectively.
FIGS. 13(A) to (C) and 14(A) to (C) show prior art voltage regulators and FIGS. 13B and 14B show the formation of the circuit of FIG. 3 in practicable forms and characteristics of such practicable forms.
FIG. 15 shows the characteristic in FIG. 4 more specifically.
FIG. 16 shows another practicable example of the circuit in FIG. 3.
FIG. 17 shows the characteristic in FIG. 8 concretely.
FIG. 18 shows a circuit for producing the characteristics in FIG. 17.
FIG. 19 shows the characteristic in FIG. 8 concretely.
FIG. 20 shows a circuit for producing the characteristics in FIG. 19.
FIG. 21 shows the characteristic in FIG. 10 concretely.
FIG. 22 shows a circuit for producing the characteristic in FIG. 21.
FIG. 23 shows a characteristic in another embodiment of the present invention.
FIG. 24 shows a circuit for producing the characteristic in FIG. 23.
FIG. 25 shows the characteristic in FIG. 12 concretely.
FIG. 26 shows a circuit for producing the characteristic in FIG. 25.
FIG. 27 shows a practicable example of the circuit in FIG. 26.
FIG. 28 shows the actual characteristics of the circuit in FIG. 27.
FIG. 29(A) shows a gate signal generator for use in an embodiment of the present invention.
FIG. 29(B) shows a time chart of the circuit in FIG. 29(A).
FIG. 30 shows a protection circuit which connects the circuit of FIG. 29(A) with the circuit of FIG. 16, 18, 20, 22, 24, or 26.
FIG. 31 shows a practicable circuit of an inverter for use in the circuit of FIG. 29(A).
FIG. 32 shows a practicable circuit of an oscillator for use in the circuit of FIG. 29(A).
FIG. 33 shows an example of a buffer circuit for the output of the circuit shown in FIG. 16, 18, 20, 22, 24 or 26.
FIG. 34 shows the characteristics of the circuit in FIG. 33.
FIGS. 35, 36 and 37 show other examples of buffer circuits, respectively.
FIG. 38 shows a time chart of the circuit in FIG. 37.
FIG. 39 shows a practicable example of the circuit in FIG. 3.
FIG. 40 shows an example of a buffer circuit.
FIG. 41 shows the characteristics of the circuit in FIG. 40.
Voltage converter circuit forms for affording various output characteristics versus an external supply voltage VCC, as well as practicable examples thereof, will be first described, followed by practicable embodiments on a method of feeding power to the voltage converter and on a buffer circuit for the voltage converter well suited to drive a large load.
FIGS. 3 and 5 show basic circuits which are used for forming voltage converter embodiments of the present invention for providing a voltage VL to circuits such as shown in FIGS. 1 and 2.
In the circuit of FIG. 3, a resistance R3 in FIG. 23 of U.S. Pat. No. 4,482,985 is replaced by a variable impedance arrangement described below, and a transistor Q is employed in order to enhance a current driving ability for a load to which an output voltage VL is applied. Here, the control terminal voltage VG of the transistor Q has a characteristic which changes versus the change of an external supply voltage VCC and which is the output voltage of a reference voltage generator REF. More specifically, as illustrated in FIG. 4, in a case where the external supply voltage VCC is gradually increased from 0 (zero) V, the voltage VG rises abruptly when a certain voltage VP has been reached, so that the transistor Q turns "on". For VCC not smaller than VP, Q continues to turn "on". Therefore, the effective impedance of the whole basic circuit BL decreases, and the ratio thereof with the effective impedance R changes, so that the voltage VL becomes a straight line of different slope for VCC not smaller than VP as shown in FIG. 4. Here in FIG. 4, the example is illustrated in which VG rises abruptly from O V to a certain voltage for VCC not smaller than VP. However, it is also allowed to adopt a characteristic in which, in case of changing VCC from O V, VG rises gradually from O V and becomes, at the point VP, a voltage level to turn "on" the transistor Q. Regarding the example in which VG rises abruptly at and above the certain voltage VCC, the reference voltage generator can be realized by the cascade connection of devices having rectification characteristics as taught in U.S. Pat. No. 4,482,985. Regarding the example in which VG rises gradually, the reference voltage generator can be realized by a simple resistance divider circuit. In FIG. 4, the coefficient of VL relative to VCC can be changed at will by the designs of the resistance and the transistor Q.
FIG. 5 shows another example which employs the same basic circuit BL as in FIG. 3. Whereas the example of FIG. 3 derives VL from the VCC side, this example derives VL from the ground side. When the characteristic of the output voltage VG from the reference voltage generator is set in advance so that the transistor Q may turn "on" at VCC not smaller than VP, VL is determined by the effective impedance of the whole basic circuit BL and the effective impedance R, and hence, VL becomes as shown in FIG. 6.
While FIGS. 3 and 5 have exemplified the transistors as being MOS transistors, bipolar transistors may be used if desired. Particularly in a case where whole chips are constructed of MOS transistors in the examples of FIGS. 1 and 2, it is usually easier to design them when the circuits of FIGS. 3 and 5 are constructed of MOS transistors. In a case where the whole chips are of bipolar transistors, it is more favorable to use bipolar transistors. It is sometimes the case, however, that the chip includes both MOS transistors and bipolar transistors. It is to be understood that, in this case, the MOS transistor or/and the bipolar transistor can be used for the circuit of FIG. 3 or FIG. 5 in accordance with an intended application. In addition, although the examples of FIGS. 4 and 6 have been mentioned as the characteristics of the circuit REF, these examples are not especially restrictive, but the characteristic of the circuit REF may be set according to the purpose of the design of VL.
Now, a voltage converter based on the circuit of FIG. 3 will be described. FIGS. 7 and 8 illustrate an example in which the basic circuits BL numbering k are connected in parallel with the effective impedance R of the circuit of FIG. 3 (formed by the resistor and the basic circuit BL0). Each of the basic circuits BL corresponds in structure to the basic circuit BL shown in FIG. 3, but are respectively set to turn on their transistors Q at different levels of the supply voltage VCC. For example, the circuits REF in the respective basic circuits BL are set so that BL0 may first turn "on" at VP0, BL1 may subsequently turn "on" at VP1, and BLk may lastly turn "on" at VPk as shown in FIG. 8. The transistors in the respective circuits BL are designed so that the coefficients of the changes of the respective voltages VL versus the voltage VCC may be varied. As VCC increases more, impedances are successively added in parallel with the impedance R of the resistor and the basic circuit BL0, so that the entire characteristic of VL becomes concave for VCC not smaller than VP0.
The coefficients of the changes are varied for the following reason. For example, in a case where the aging operation points are VP2, VP3, . . . and VPk and where the aging voltages of circuits to be fed with the supply voltages by the voltage converter are VL2, VL3, . . . and VLk, the transistion is smoothed when the first aging operation point shifts to the next one.
The present circuit is a circuit which is practical in terms of the operating stability of the ordinary operation and an effective aging for the system of FIG. 2. By way of example, the VCC operation point in the ordinary operation is set at a point at which VL changes versus VCC as slightly as possible, that is, the coefficient of change is the smallest, in order to achieve a stable operation. In fact, if desired, the coefficient of change of VL versus VCC can be set to be zero in the range between VP1 and VP2 for the ordinary operation so that a constant voltage VL is held in this entire range. Alternatively, a small positive slope can be used in this, as shown in FIG. 8. On the other hand, the VCC operation point in the aging test is set at a point at which the coefficient of change is great, in order to approximately equalize the stress voltage conditions of transistors of large geometries receiving VCC to stress voltage conditions of transistor of small geometries receiving VL as described in, U.S. Pat. No. 4,482,985. Specifically, large geometry devices such as those found in the interface circuit B of FIG. 2 are operated during aging tests at a higher potential than small geometry devices in circuit A at the reduced potential produced by voltage converter 13. More concretely, in case of using only BL0 and BL1 in the circuit of FIG. 7, the coefficient of change in FIG. 8 may be made small between the lower limit voltage VP0 (e.g., 2-3 V) and the upper limit voltage VP1 (e.g., 6 V), to set the ordinary operation point (e.g., 5 V) concerning VCC for the ordinary operation range in this section, while the coefficient of change may be made great between VP1 and VP2 (e.g., VP2 being 7-9 V), to set the aging operation point (e.g., VCC =8 V) in this section. The ordinary operation range is solely determined by ratings, and it is usually set at 5±0.5 V. It is to be understood that, for some purposes of designs, the operation voltage points and the aging voltage points can be set at any desired VCC points by employing the basic circuits BL2, BL3 . . . etc. When more circuits BL are used, the VL characteristic can also be made smoother versus VCC, so that the operation of the internal circuit can be stabilized more. Further, since the VCC voltage is high in the aging test, it is effective to construct the voltage converter itself using high breakdown voltage transistors. To this end, the voltage converter may be constructed of transistors of large geometry in the system of FIG. 2 by way of example.
FIGS. 9 and 10 show an example of using the FIG. 4 arrangement with additional basic circuits BL being connected in parallel on the ground side. In this arrangement, by setting the respective circuits BL to have different turn-on times, the characteristic of the whole VL can be made convex relative to VCC, as shown in FIG. 10. This characteristic is effective for protecting the circuit A' from any overvoltage VL in the system of FIG. 1 by way of example. This achieves the advantage that, in case of measuring the VCC voltage margin of the whole chip, a sufficiently high voltage VCC can be applied without destroying small geometry devices.
In some uses, it is also possible that the circuits of FIGS. 7 and 9 can coexist. By way of example, the ordinary operation point is set at a point at which the coefficient of change is small, and the aging operation point is set at a point at which the coefficient of change is great. These are realized by BL0 and BL1 in the circuit of FIG. 7. Further, in order to make the coefficient of change small again at and above the VCC point of the aging condition to the end of preventing the permanent breakdown of devices, the basic circuits BL other than BL0 are connected so as to operate in parallel with the latter as in the circuit form of FIG. 9. This makes it possible to design a circuit in which the devices are difficult to break down at and above the VCC point of the aging operation.
Thus, even when the supply voltage has erroneously been made abnormally high, by way of example, the breakdown of the devices can be prevented.
FIGS. 11 and 12 show an example in which a basic circuit BL' is connected in parallel with the circuit of FIG. 3, whereby the changing rate of VL is made negative at and above VP ' which is a certain value of VCC. More specifically, when VCC is increased, the transistor Q first turns "on" while the output voltage VG of the reference voltage generator 1 in the basic circuit BL is not lower than VP, so that the gradient of VL versus VCC decreases. A reference voltage generator 2 is designed so that a transistor Q' in the basic circuit BL' may subsequently turn "on" at the certain VCC value, namely, VP '. In addition, the conductance of Q' is designed to be sufficiently higher than that of Q. Then, the VL characteristic after the conduction of the transistor Q' is governed by the characteristic of BL', so that VL comes to have the negative gradient as shown in FIG. 12.
The merit of the present circuit is that, when the aforementioned point at which VL lowers is set at or below the breakdown voltages of small geometry devices, these small geometry devices are perfectly protected from breakdown even when the voltage VCC has been sufficiently raised. For example, a measure in which the output voltage VL lowers when a voltage higher than the external supply voltage VCC at the aging point has been applied is especially effective because any voltage exceeding the aging point is not applied to the devices.
It is to be understood that an external instantaneous voltage fluctuation can also be coped with.
Obviously, the circuit of FIG. 5 can afford any desired VL characteristic by connecting the basic circuit BL' in parallel as in the example of FIG. 3.
While, in the above, the conceptual examples of the voltage converters have been described, practicable circuit examples based on these concepts will be stated below.
FIG. 13A shows an example of the circuit of FIG. 3 which employs a bipolar transistor. A voltage regulator circuit CVR is, for example, a cascade connection of Zener diodes or ordinary diodes the terminal voltage of which becomes substantially constant. FIG. 13(A) indicates a well-known voltage regulator which has the characteristic shown by (A) in FIG. 13(C). This voltage regulator is described in detail in "Denpa-Kagaku (Science of Electric Wave)", February 1982, p. 111 or "Transistor Circuit Analysis", by Joyce and Clarke, Addison-Wesley Publishing Company, Inc., p. 207. Since, however, VL is a fixed voltage in this condition, a resistance r can be connected in series with the CVR as shown in FIG. 13(B) in accordance with the present invention to slope the curve as desired. Thus, VL comes to have a slope relative to VCC as shown by the characteristic (B) shown in FIG. 13(C).
FIG. 14 shows another embodiment. FIG. 14(A) indicates a well-known voltage regulator which employs an emitter follower and which has the characteristic shown by (A) in FIG. 14(C). Since VL is also a fixed voltage, a resistance r is used in FIG. 14(B) in order to provide a desired slope. Thus, a characteristic as shown as characteristic (B) in FIG. 14(C) is provided.
These examples of FIGS. 13 and 14 are especially suited to the system as shown in FIG. 1. In FIG. 1, usually a great current flows through the circuit associated with the input/output interface. Therefore, a high current driving ability is required of the voltage converter correspondingly. Obviously, the voltage converter constructed of the bipolar transistor is suited to this end.
Next, there will be explained practicable examples in which voltage converters are constructed of MOS transistors on the basis of the circuits of FIGS. 3, 7, 9 and 11.
FIG. 15 shows a concrete example of the characteristic of FIG. 4 in which VL is endowed with a slope m for VCC of and above a certain specified voltage V0. Since the change of VL decreases for the voltage not smaller than V0, the breakdown of small geometry devices is less likely to occur to that extent.
VL =VCC is held for VCC smaller than V0, for the following reason. In general, MOSTs have their operating speeds degraded by lowering in the threshold voltages thereof as the operating voltages lower. To the end of preventing this drawback, it is desirable to set the highest possible voltage on a lower voltage side such as VCC smaller than V0. That is, VL should desirably be equal to VCC.
FIG. 16 shows an embodiment of a practicable circuit DCV therefor, which corresponds to a practicable example of the circuit of FIG. 3.
The features of the present circuit are that the output voltage VL is determined by the ratio of the conductances of MOS transistors Q0 and Ql, and that the conductance of the MOS transistor Ql is controlled by VL.
With the present circuit, letting the gate voltage VG of QO be VCC +Vth(O) (where Vth(O) denotes the threshold voltage of the MOST QO), the control starting voltage VO and the slope m are expressed as follows: ##EQU1## Here, β(O) and β(1) denote the channel conductances of QO and Ql, Vth(i) (i=1-n) and Vth(l) denote the threshold voltages of the MOS transistors Qi (i=1-n) and Ql, and n denotes the number of stages of Qi.
Accordingly, VO and m can be varied at will by n, Vth(i), Vth(l) and β(1)/β(O). It has been stated before that VL =VCC is desirable for VCC smaller than VO. In this regard, for VCC smaller than VO, VL is determined by VO because Ql is "off". Therefore, the voltage VG of QO must be a high voltage of at least VCC +Vth(O).
In order to simplify the computation and to facilitate the description, the circuit of FIG. 16 is somewhat varied from an actual circuit. As a practical circuit, as shown in FIG. 27 to be referred to later, a transistor of similar connection (QS(1.6) in FIG. 27) needs to be further connected between the n-th one of the transistors connected in cascade and the ground. That is, a kind of diode connection is made toward the ground. With this measure, when VCC has been varied from the high voltage side to the low voltage side, the nodes of the transistors connected in cascade are prevented from floating states to leave charges behind. For the sake of the convenience of the description, the transistor of this measure shall be omitted in the ensuing embodiments.
FIG. 17 shows a characteristic in which, when the external supply voltage VCC changes between the lower limit value VO and upper limit value VO ' of the ordinary operation range, the slope m of the output voltage VL is small, and a slope m' which corresponds to the external supply voltage greater than VO ' is made steeper than m.
FIG. 18 shows an example of a circuit for producing the characteristic of FIG. 17.
These correspond to a practicable form of the example of FIGS. 7 and 8.
The feature of the present circuit is that, between the terminals 1 and 2 of the circuit DCV shown in FIG. 16, a circuit DCV2 similar to DCV1 is added, whereby the conductance of a load for DCV1 is increased at and above VO ' so as to increase the slope of VL.
With the present circuit, the second control starting voltage VO ' is expressed by: ##EQU2## In addition, the slope m' is determined by the ratio between the sum of the conductances of the MOS transistors QO and Q'l and the conductance of the MOS transistor Ql. Here, V'th(i) (i=1-n') and V'th(l) denote the threshold voltages of the MOS transistors Q'i (i=1-n') and Q'l, respectively.
Accordingly, V'O and m' can be varied at will by n, n', β(1), β'(1), Vth(i), Vth(l), V'th(i) and V'th(l). Here, β'(1) denotes the channel conductance of the MOS transistor Q'l.
This circuit has the ordinary operation range between the lower limit value VO and the upper limit value VO ', and is effective when the aging point has a value larger than VO '. That is, since the slope m is small in the ordinary operation region, margins for the breakdown voltages of small geometry devices are wide, and power consumption does not increase. Here, the slope m' for the external supply voltage higher than the ordinary operation region is set for establishing a characteristic which passes an aging voltage (set value).
In an example illustrated in FIG. 19, a characteristic in which the slope of VL becomes m" greater than m' when the external supply voltage VCC has reached VO " is further added to the characteristic shown in FIG. 17.
FIG. 20 shows an example of a practicable circuit therefor. These correspond to a concrete form of the example of FIGS. 7 and 8. The feature of the present circuit is that circuits DCV2 and DCV3 similar to the circuit DCV1 are added between the terminals 1 and 2 of the circuit shown in FIG. 16, whereby the conductance of the load for DCV1 is successively increased so as to increase the slope of VL in two stages at the two points VO ' and VO ".
With the present circuit, the second and third control starting voltages VO ' and VO " are respectively expressed by: ##EQU3## Here, V"th(i) (i=1-n") and V"th(l) denote the threshold voltages of the MOS transistors Q"i (i=1-n") and Q"l, respectively. Besides, the slope m' is determined by the ratio between the sum of the conductances of the MOS transistors QO and Q'l and the conductance of the MOS transistor Ql, and the slope m" by the ratio between the sum of the conductances of the MOS transistors QO, Q'l and Q"l and the conductance of the MOS transistor Ql.
Accordingly, VO ' and m' can be varied at will by n, n', β(O), β(l), β'(l), Vth(i), Vth(l), V'th(i) and V'th(l), while V"O and m" by n, n', n", β(O), β(l), β'(l), β"(l), Vth(i), Vth(l), V'th(i), V'th(l), V"th(i) and V"th(l). Here, β"(l) denotes the channel conductance of Q"l.
This circuit is effective when the ordinary operation range extends between the lower limit value VO and the upper limit value VO ', and aging tests are carried out in the two sections of the external supply voltage VCC ≧VO " and VO '<VCC <VO ". The aging tests in the two sections consist of the two operations: aging for a short time, and aging for a long time. The former serves to detect a defect occurring, for example, when an instantaneous high stress has been externally applied, while the latter serves to detect a defect ascribable to a long-time stress.
FIG. 21 shows an example wherein, when the external supply voltage VCC is greater than VO ', the slope m' of the voltage VL is set at m>m' under which the output voltage VL follows up the external supply voltage VCC.
FIG. 22 shows an embodiment of a practicable circuit therefor. These correspond to a concrete form of the example of FIGS. 9 and 10. The feature of the present circuit is that a circuit DCV2 similar to DCV1 is added between the terminal 2 and ground of the circuit shown in FIG. 16, whereby the conductance of a load for the transistor QO is increased at VO ' so as to decrease the slope of VL.
With the present circuit, the second control starting voltage VO ' is expressed by: ##EQU4## In addition, the slope m' is expressed by the ratio between the conductance of QO and the sum of the conductances of Ql and Q'l.
Accordingly, VO ' and m' can be varied at will by n, n', β(O), β(l), β'(l), Vth(i), Vth(l), V'th(i) and V'th(l).
This circuit is applicable to devices of lower breakdown voltages. Usually, when the breakdown voltages of devices are low, the output voltage VL of the ordinary operation region (VO <VCC <VO ') may be suppressed to a low magnitude. In some cases, however, the magnitude of VL cannot be lowered because the operating speeds of a circuit employing small geometry devices and a circuit employing large geometry devices are matched. In such cases, the slope ma of the output voltage VL is decreased in order for the aging operation point to be passed. Thus, the magnitude of the output voltage VL can be raised near to the withstand voltage limit of the devices within the range of the ordinary operation region, and the operating speed of the circuit employing the small geometry devices can be matched with that of the circuit employing the large geometry devices.
In an example shown in FIG. 23, a characteristic in which the slope of VL becomes m" smaller than m' when the external supply voltage CCC has reached VO " is further added to the characteristic illustrated in FIG. 17.
FIG. 24 shows an embodiment of a practicable circuit therefor. This corresponds to an example in which the examples of FIGS. 7 and 9 coexist. The feature of the present circuit is that the embodiments of FIGS. 18 and 21 are combined thereby to increase and decrease the slope of VL at the two points VO ' and VO " respectively.
With the present circuit, the second and third control starting voltages VO ' and VO " are respectively expressed by: ##EQU5## In addition, the slope m' is expressed by the ratio between the sum of the conductances of QO and Ql ' and the conductance of Ql, while m" is expressed by the ratio between the sum of the conductances of QO and Ql ' and the sum of the conductances of Ql and Ql ".
Accordingly, VO ' and m' can be varied at will by n, n', β(O), β(l), β'(l), Vth(i), V'th(i) and V'th(l), while VO " and m" can be varied by n, n', n", β(O), β(l), β'(l), β"(l), Vth(i), Vth(l), V'th(i), V'th(l), V"th(i) and V"th(l).
This circuit protects small geometry devices from permanent breakdown in such a way that, even when VCC has become higher than the withstand voltage limit VO " of the devices due to some fault of the external power source, it does not exceed a breakdown voltage VB. That is, the slope m" of VL for VCC not smaller than VO " is made gentler than the slope m' in the aging, whereby even when the external supply voltage VCC has become VO " or above, the output voltage VL is prevented from exceeding the breakdown voltage (usually, higher than the withstand voltage limit) of the devices. This makes it possible to prevent the device breakdown even when the supply voltage has been raised abnormally by way of example.
FIG. 25 shows an example in which the slope m' is made negative when the external supply voltage VCC has exceeded VO '.
FIG. 26 shows an embodiment of a practical circuit therefor. These correspond to a concrete form of the example of FIGS. 11 and 12. The feature of the present circuit is that the drain of Q1 ' in DCV2 is connected to the terminal 1 of the circuit shown in FIG. 16, the drain of Ql ' to the terminal 2, and the source of Ql ' to the ground, whereby the conductance of Ql ' is controlled by VCC, and besides, it is made greater than the conductance of QO so as to establish m'<O.
With the present circuit, the second control starting voltage VO ' and the slope m' are expressed by the following on the assumption of β'(l)≧β(O): ##EQU6##
Accordingly, VO ' and m' can be varied at will by n', V'th(i), V'th(l) and β'(l)/(O).
FIGS. 27 and 28 show a practicable example of the present circuit and examples of the characteristics thereof. All the threshold voltages of transistors are 1 (one) V, and VG =VCC +Vth(O) is held. In addition, numerals in parentheses indicate values obtained by dividing the channel widths by the channel lengths of the transistors. FIG. 28 illustrates VL with a parameter being the corresponding value Wl /Ll of Ql '. By way of example, the voltage in the ordinary operation is set at 5 V, and the aging voltage at 8 V.
This circuit consists in that the slope of the voltage at and above VO " in the characteristic shown in FIG. 23 is made negative, thereby to intensify the aspect of the device protection of the circuit in FIG. 24.
With this circuit, the breakdown due to the external application of a high voltage is perfectly prevented, and the power consumption in the integrated circuit does not exceed an allowable value. Thus, even when the instantaneous high voltage has been externally applied, the prevention of the breakdown of the devices is ensured.
Thus far, the voltage converters and their characteristics have been described. Next, the method of feeding the voltage converter with power will be described.
In the above, the gate voltage of QO has been presumed to be VCC +Vth. This has intended to simplify the computation and to clearly elucidate the characteristics of the circuits. Essentially, however, this voltage need not be limited to VCC +Vth, but may be chosen at will for the convenience of design.
FIG. 29(A) shows a practicable circuit which boosts the gate voltage VG to above the supply voltage VCC within the chip as stated with reference to FIG. 15.
When a pulse φi of amplitude VCC from an oscillator OSC included within the chip rises from 0 (zero) V to VCC, a node 4' having been previously charged to VCC -Vth by Q1 ' is boosted to 2 VCC -Vth.
In consequence, a node 4 becomes a voltage 2 (VCC -Vth) lowered by Vth by means of Q2 '. Subsequently, when φi becomes 0 V and a node 2 rises to VCC, the node 4 is further boosted into 3 VCC -2 Vth. Accordingly, a node 5 becomes a voltage 3 (VCC -Vth) lowered by Vth by means of Q2. Each of Q2 ' and Q2 is a kind of diode, so that when such cycles are continued a large number of times, VG becomes a D.C. voltage of 3 (VCC -Vth). VG of higher voltage is produced by connecting the circuits CP1, CP2 in a larger number of stages. The reason why the two stages are comprised here, is as follows. Assuming VCC to lower to 2.5 V and Vth to be 1 (one) V, one stage affords VG =2 (VCC -Vth), and hence, VG =3 V holds. Under this condition, however, the source voltage VL of QO in FIG. 15 becomes 2 V lower than VCC. In contrast, when the two stages are disposed, VG =4.5 V holds because of VG =3 (VCC -Vth). Accordingly, VL can be equalized to VCC, so that VL =VCC can be established below VO as in FIG. 15. Conversely, however, as VCC becomes a higher voltage, it is more of a concern that VG may become an excess voltage which can break down the associated transistors. Therefore, some circuit for limiting VG is required on the high voltage side of VCC.
FIG. 30 shows an example in which VG ≃3 (VCC -Vth) is held as a high voltage on the low voltage side of VCC, and besides, VCC +2 Vth is held on the high voltage side of VCC in order to protect the associated transistors. Here, any of the circuits thus far described, for example, the whole circuit in FIG. 16, 18, 20, 22, 24 or 26, is indicated by LM1 as the load of VG. A protection circuit CL1 is such that, when VG is going to exceed VCC +2 Vth, current flows through Q1 and Q2, so VG results in being fixed to VCC +2 Vth. With the present circuit, VCC at which CL1 operates ranges from 3 (VCC -Vth)=VCC +2 Vth to VCC =5/2 Vth.
FIG. 31 shows a practicable circuit of the inverter 1 or 2 in FIG. 29(A). An output pulse φO is impressed on the circuit CP1 or CP2.
While the oscillator OSC can be constructed as a circuit built in the chip, FIG. 32 shows an example utilizing a back bias generator which is built in the chip in order to apply a back bias voltage VBB to a silicon substrate. The advantage of this example is that the oscillator need not be designed anew, which is effective for reducing the area of the chip. In general, when VL is applied to respective transistors with VBB being 0 (zero) V, the threshold voltages Vth of the respective transistors are not normal values. Therefore, an excess current flows, or stress conditions on the transistors become severe, so the transistors can break down. In contrast, when this circuit is used, VBB is generated upon closure of a power source, and VL is generated substantially simultaneously, so that the operations of respective transistors are normally executed.
Next, practicable embodiments of buffer circuits will be described. As the load of the voltage converter, there is sometimes disposed a load of large capacity or of great load fluctuation. In this case, such a heavy load needs to be driven through a buffer circuit of high driving ability. In order to accomplish this, the ordinary method is to drive the load through a single transistor of high driving ability, namely, a transistor having a large width-to-length ratio W/L as shown in FIG. 33. With this method, however, the performance degrades because a voltage drop of Vth arises on the low voltage side of VCC as shown in FIG. 34. FIG. 35 shows a practicable example of the buffer circuit which has a high driving ability without the Vth drop. When a voltage VPP is made greater than VL =Vth and a resistance RP is made much higher than the equivalent "on" resistance of a transistor Q1, the gate voltage of a transistor Q2 becomes VL +Vth. Accordingly, the source voltage VL1 of Q2 equalizes to VL. When the W/L of Q2 is made great, the desired buffer circuit is provided. Here, VL becomes VCC on the low voltage side of VCC, so that VPP must be at least VCC -Vth. As a circuit therefor, the circuit shown in FIG. 29(A) is usable. Regarding connection, the node 5 of the circuit in FIG. 29(A) may be connected to the drain of Q1 in a regulator in FIG. 35. Here, in order that the effective output impedance as viewed from the node 5 may be made sufficiently higher than the equivalent "on" resistance of Q1 of the circuit in FIG. 35, the value of the W/L of Q2 or the value of CB in FIG. 29(A) or the oscillation frequency of OSC may be properly adjusted by way of example.
As to some loads, it is necessary to apply VL to the drain of a transistor constituting a part of the load and to apply VL +Vth to the gate thereof, so as to prevent the Vth drop and to achieve a high speed operation. FIG. 36 shows an embodiment therefor. The circuit LM1 is, for example, the circuit in FIG. 16, and the voltage VL1 equalizes to VL as stated before. In addition, the gate voltage of Q4 is VL +2 Vth. Therefore, VL2 becomes VL +Vth. Here, transistors Q6 and Q7 serve to prevent unnecessary charges from remaining in VL1 at the transient fluctuation of VCC. Q6 and Q7 are connected into LM1 as shown in figure so as to operate at VCC of at least VO and at VCC of at least VO -Vth. Here, the ratio V/L of Q6, Q7 is selected to be sufficiently smaller than that of Q2, minimize the influence of the addition of Q6, Q7 on VL. It has been previously stated that Q7 operates in the region not greater than VO. Since Q2 and Q4 are in the operating states of unsaturated regions (VGS -Vth ≧VDS, VGS : gate-source voltage, VDS : drain-source voltage) in the region not greater than VO, surplus charges are discharged to VCC through Q2, Q4, and hence, Q7 is unnecessary in principle. However, when VCC is near VO, the "on" resistances of Q2, Q4 increase unnecessarily, and it is sometimes impossible to expect the effects of these transistors. Accordingly, Q7 is added, whereby stable values of VL1 can be obtained in a wide range from the region (VO -Vth) where VCC is not greater than VO, to the region where VCC is greater than VO and where the converter is normally operating.
The function of Q5 is that, when VL1 is going to fluctuate negatively relative to VL2, current flows to Q5 so as to keep the difference of VL2 and VL1 constant. In addition, in the present embodiment, the example of VL and VL +Vth has been stated. However, when the pairs of Q1, Q2 or the pairs of Q3, Q4 are connected in cascade, a voltage whose difference from VL1 becomes an integral multiple of Vth can be generated.
A circuit shown in FIG. 37 is another buffer circuit which is connected to the output stage of the circuit of FIG. 35 or 36 in order to further enhance the driving ability of the buffer circuit of FIG. 35 or 36. By connecting such a buffer circuit of higher driving ability, a large load capacity can be driven. First, VL1 becomes VL1 +2 Vth and VL1 +Vth at respective nodes 4 and 2. Eventually, however, it is brought into VDP being the level of VL1 at a node 5 by Q4. Problematic here is the load driving ability of Q4 which serves to charge a large capacitance CD in the load LC1 at high speed. In order to enhance the ability, the node 2 being the gate of Q4 needs to be boosted in a time zone for charging the load. Transistors therefor are Q6 -Q11, and capacitors are C1 and C2. A node 6 discharged by Q13 owing to the "on" state of φ2 is charged by Q12 and Q4 when the next φ1 is in the "on" state. At this time, the node 2 being at VL1 +Vth and a node 3 being at VL1 are boosted by the "on" state of φ1. Consequently, the conductances of Q10, Q11 increase, so that the boosted voltage of the node 2 is discharged to the level of VL1 +Vth by Q10, Q11. Here, when the boosting time is made longer than the charging time of CD based on Q4, Q12, the capacitor CD is charged rapidly. The transistor Q6 cuts off the nodes 3 and 1 when the node 3 is boosted by φ1. When φ2 is "on", Q7 -Q9 turn "off" subject to the condition of VL1 ≦3Vth, so that Q11 has its gate rendered below Vth to turn "off". Accordingly, no current flows through Q3, Q10 and Q11, so that the power consumption can be rendered low. In addition, in order to reduce to power consumption in the case of VL1 >3 Vth, the "on" resistance of Q6 may be increased to lower current. The voltage of the node 3 at this time becomes a stable value of approximately 3 Vth. Thus, the boosting characteristic of the node 3 is also stabilized, with the result that the operation of the whole circuit can be stabilized.
Here, since the sources and gates of Q7 and Q10 are connected in common, the conditions of biasing the gates are quite equal. Accordingly, when ##EQU7## is held in advance, the boosting characteristics of the nodes 2, 3 can be quite equal, so the circuit design can be facilitated advantageously. That is, one merit of the present embodiment consists in that the boosting characteristic of the node 2 can be automatically controlled with the boosting characteristic of the node 3. In this way, the D.C. path from the node 2 to VSS in the case of performing no boosting can be relieved, and it becomes possible to lower the power consumption.
Here, Q5 has the function of discharging the surplus charges of the node 2 when Q10 is "off".
As regards the embodiment of FIG. 37, various modifications can be considered. While the drain of Q6 in FIG. 37 is connected to VL1 in order to stabilize the boosting characteristics of the nodes 2, 3 to the utmost, it can also be connected to VCC so as to relieve a burden on VL1. Likewise, while Q10 subject to the same operating condition as that of Q7 is disposed in order to stabilize the boosting characteristics of the nodes 2, 3, it may well be removed into an arrangement in which the nodes 2 and 9 are directly connected, with the source of Q7 and the node 9/disconnected. Since, in this case, the relationship of Q9 and Q11 is in the aforementioned relationship of Q7 and Q10, the boosting characteristics can be similarly designed, and the occupying area of the circuit can be effectively reduced. Further, the 3-stage connection arrangement of Q7, Q8 and Q9 is employed here. This is a consideration for efficiently forming the circuit in a small area by utilizing a capacitance Chd 2 (for example, the capacitance between the gate of a MOST and an inversion layer formed between the source and drain thereof, known from ISSCC 72 Dig. of Tech. Papers, p. 14, etc.) for the reduction of the power consumption described above. That is, in order to use the inversion layer capacitance, the gate voltage to be applied needs to be higher by at least Vth than the source and drain. Accordingly, in case of forming C2 by the use of a MOST of low Vth or an ordinary capacitor, it is also possible to reduce the connection number of Q7 -Q9 to two or one.
The buffer circuit as shown in FIG. 37 is indispensable especially to the LSI systems as shown in FIGS. 1 and 2. In general, the voltage converter for generating VL in FIG. 1 or 2 is desired to have an especially high ability of supplying current because the circuit current in the circuit A, A' or B flows toward the ground. Accordingly, when the whole circuit including the circuit of FIG. 37 thus far described is regarded as the voltage converter of FIG. 1 or 2, it is applicable to general LSIs.
With the embodiments stated above, when the actual circuit of FIG. 18 which is diode-connected as shown in FIG. 27 is operated at VCC of or above VO as shown in FIG. 17, current flows through Q1 '-QS ' (FIG. 27) to increase the power consumption. This increase of the power consumption poses a problem in case of intending to back up the LIS power source, namely, the externally applied supply voltage with a battery. More specifically, in an apparatus wherein the ordinary external power source is backed up by a battery when turned "off"; when the power consumption of the LSI itself is high, the period of time for which the power source is backed up is limited because the current capacity of the battery is small. Therefore, with a measure wherein VCC to be applied by the battery is set at below VO during the time interval during which the battery is operated for backup, no current flows through Q1 '-QS ', and hence, the period of time for which the power source can be backed up can be extended to that extent. Alternatively, the number of stages of Q1 '-QS ' (FIG. 27) can be determined so as to establish VO which is greater than VCC being the battery supply voltage in the case of the backup.
The supply voltage VCC in the ordinary operation can be selected at VCC <VO besides at VCC >VO. Since this permits no current to flow through Q1 '-QS ' under the ordinary VCC condition, the power consumption can be lowered. Another merit is that design is facilitated because the circuit can be designed while avoiding a region where the relation of VCC and VL becomes a polygonal line. More specifically, when the polygonal region is used, an imbalance of characteristics concerning VCC arises between a circuit directly employing VCC and a part of a certain circuit employing VL by way of example, so that the operation sometimes becomes unstable. When VCC <VO holds, this drawback can be eliminated.
In the above, the practicable embodiments have been described in which the voltage converters are constructed of MOS transistors. These are examples which chiefly employ MOS transistors of positive threshold voltages Vth, namely, of the enhancement mode. Needless to say, however, it is also possible to employ a MOS transistor of negative Vth, namely, of the depletion mode as disclosed in FIG. 16 of Japanese Patent Application No. 56-168,698. For example, in the embodiment of FIG. 16, in order to establish VL =VCC in the region of VCC ≦VO as illustrated in the characteristic of FIG. 15, the gate voltage of QO needs to be VG ≧VCC +Vth(O), and it has been stated that the circuit of FIG. 29(A) may be used as the VG generator therefor. In this regard, the circuit can be further simplified by employing the MOS transistor of the depletion mode. FIG. 39 shows such a practicable embodiment. It differs from the circuit of FIG. 16 in that QO is replaced with the depletion mode MOS transistor QO ', the gate of which is connected to the terminal 2. With this measure, since the V'th(O) of QO ' is negative, QO ' is in the "on" state at all times, and the desired characteristic illustrated in FIG. 15 can be realized without employing the VG generator as shown in FIG. 29(A). With the present embodiment, not only the circuit arrangement can be simplified as stated above, but also the merit of attaining a stable characteristic is achieved because current I(QO ') to flow through QO ' becomes a constant current determined by β'(O) (channel conductance) and V'th(O) (threshold voltage) as ##EQU8## Although the present embodiment has exemplified FIG. 16, it is applicable as it is by substituting QO ' for QO in any other embodiment and connecting its gate to the terminal 2 as in the present embodiment.
FIG. 40 shows an embodiment in which a buffer circuit is constructed using a single depletion-mode MOS transistor, while FIG. 41 shows the characteristic thereof. Although the present embodiment is the same in the circuit arrangement as the foregoing embodiment of FIG. 33, it differs in that the MOS transistor is changed from the enhancement mode into the depletion mode. As shown in FIG. 41, the output VL' of the present buffer circuit bends from a point P at which the difference of VCC and VL equalizes to the absolute value |VthD | of the threshold voltage VthD of the MOS transistor, and it thereafter becomes a voltage which is higher than VL by |VthD |. Accordingly, VL may be set lower than a desired value by |VthD |. The present embodiment has a simple circuit arrangement, and can meritoriously eliminate the problem, as in the characteristic of the embodiment of FIG. 33 illustrated in FIG. 34, that only the output lower than VCC by Vth can be produced in the range of VCC ≦VO.
As set forth above, the present invention can provide, in an integrated circuit having small geometry devices, an integrated circuit which has a wide operating margin even against the fluctuations of an external supply voltage in an ordinary operation and which can apply a sufficient aging voltage.
It is to be understood that the above-described arrangements are simply illustrative of the application of the principles of this invention. Numerous other arrangements may be readily devised by those skilled in the art which embody the principles of the invention and fall within its spirit and scope.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US2697201 *||Sep 27, 1949||Dec 14, 1954||Westinghouse Electric Corp||Adjustable nonlinear resistance|
|US3386030 *||Oct 21, 1964||May 28, 1968||Collins Radio Co||Voltage regulator|
|US3523195 *||Jun 7, 1967||Aug 4, 1970||Bendix Corp||Function generator network utilizing a transistor including a multiple tap emitter follower|
|US3553487 *||Oct 12, 1967||Jan 5, 1971||Honeywell Inc||Circuit for generating discontinuous functions|
|US3588675 *||Mar 26, 1969||Jun 28, 1971||Meidensha Electric Mfg Co Ltd||Voltage regulator circuit effective over predetermined input range|
|US3742338 *||Mar 8, 1972||Jun 26, 1973||Matsushita Electronics Corp||Dc voltage regulator circuit|
|US3895239 *||Dec 26, 1973||Jul 15, 1975||Motorola Inc||MOS power-on reset circuit|
|US4239980 *||Aug 4, 1978||Dec 16, 1980||Hitachi, Ltd.||Integrated circuit having an operation voltage supplying depletion type MISFET of high breakdown voltage structure|
|US4368524 *||Jul 18, 1980||Jan 11, 1983||Fujitsu Limited||Semiconductor device|
|US4385243 *||May 19, 1980||May 24, 1983||Fujitsu Limited||Automatic reset circuit|
|US4409677 *||Dec 17, 1979||Oct 11, 1983||Fujitsu Limited||Semiconductor integrated circuit device|
|US4433252 *||Jan 18, 1982||Feb 21, 1984||International Business Machines Corporation||Input signal responsive pulse generating and biasing circuit for integrated circuits|
|US4446381 *||Apr 22, 1982||May 1, 1984||Zilog, Inc.||Circuit and technique for initializing the state of bistable elements in an integrated electronic circuit|
|US4455628 *||Nov 4, 1982||Jun 19, 1984||Mitsubishi Denki Kabushiki Kaisha||Substrate bias generating circuit|
|US4482985 *||Apr 14, 1982||Nov 13, 1984||Hitachi, Ltd.||Semiconductor integrated circuit|
|US4495622 *||May 13, 1982||Jan 22, 1985||Thomson-Csf||System for selecting high-reliability integrated circuits|
|US4585955 *||Nov 30, 1983||Apr 29, 1986||Tokyo Shibaura Denki Kabushiki Kaisha||Internally regulated power voltage circuit for MIS semiconductor integrated circuit|
|JPS5547414A *||Title not available|
|JPS59191935A *||Title not available|
|1||*||Buchsbaum, W., Encyclopedia of Integrated Circuits, Prentice Hall, 1987, pp. 65 67, 113, 114.|
|2||Buchsbaum, W., Encyclopedia of Integrated Circuits, Prentice-Hall, 1987, pp. 65-67, 113, 114.|
|3||Itoh, K. et al., "An Experimental 1Mb DRAM With On-Chip Voltage Limiter", 1984 IEEE International Solid-State Circuits Conf., pp. 282-283.|
|4||*||Itoh, K. et al., An Experimental 1Mb DRAM With On Chip Voltage Limiter , 1984 IEEE International Solid State Circuits Conf., pp. 282 283.|
|5||Love, R. et al., "Width-to-Length Ratio Design Program for Interacting Static FET Circuits", IBM Technical Diclosure Bulletin, vol. 16, No. 11, Apr. 1974, pp. 3671-3673.|
|6||*||Love, R. et al., Width to Length Ratio Design Program for Interacting Static FET Circuits , IBM Technical Diclosure Bulletin, vol. 16, No. 11, Apr. 1974, pp. 3671 3673.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US6351420||Jun 16, 2000||Feb 26, 2002||Advanced Micro Devices, Inc.||Voltage boost level clamping circuit for a flash memory|
|WO2001057874A2 *||Feb 5, 2001||Aug 9, 2001||Advanced Micro Devices, Inc.||Voltage boost level clamping circuit for a flash memory|
|WO2001057874A3 *||Feb 5, 2001||Apr 18, 2002||Advanced Micro Devices Inc||Voltage boost level clamping circuit for a flash memory|
|U.S. Classification||714/724, 327/538, 323/269, 323/270, 323/303, 324/750.3, 324/762.02|
|International Classification||G05F1/46, G01R31/30, H01L27/02, G11C5/14|
|Cooperative Classification||G01R31/30, G11C5/147, G11C5/145, G05F1/465, H01L27/0218|
|European Classification||G05F1/46B3, G11C5/14R, G01R31/30, G11C5/14P, H01L27/02B3B|
|Sep 26, 1997||FPAY||Fee payment|
Year of fee payment: 8
|Sep 26, 2001||FPAY||Fee payment|
Year of fee payment: 12
|Dec 3, 2003||AS||Assignment|
Owner name: RENESAS TECHNOLOGY CORPORATION, JAPAN
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:HITACHI, LTD.;REEL/FRAME:014754/0218
Effective date: 20031203