Dynamic logic circuits using transistors having differing threshold voltages and delayed low threshold voltage leakage protection

[0002] This application is related to U.S. patent application Ser. No. 10/307,035 entitled “Hold Time Latch With Decreased Precharge Node Voltage Leakage”, and filed on the same date as the present application.

[0003] Not Applicable.

[0004] The present embodiments relate to electronic circuit devices and are more particularly directed to dynamic logic circuits using transistors having differing threshold voltages and delayed low threshold voltage leakage protection.

[0005] Electronic circuits have become prevalent in numerous applications, including uses for devices in personal, business, and other environments. Demands of the marketplace affect many aspects of circuit design, including factors such as device power consumption and speed. Various electronic circuits now implement what is known in the art as dynamic, or domino, logic. Dynamic logic circuitry operates in two phases, a precharge phase during which one or more precharge nodes are precharged to a first voltage, and an evaluate phase during which the data is read based on the voltage at each precharge node, where the precharge voltage therefore may be read as a first logic state if undisturbed or where that precharge voltage may be first discharged and then read as a second logic that is complimentary to the first logic state. Further, dynamic logic often passes signals through two or more stages, so that an evaluation by a first stage that discharges the first stage precharge node results in an output to the input of a second stage, and where that signal also thereby triggers a discharge of a precharge node in the second stage. In this manner, a number of stages may be triggered as the evaluate phases of the stages overlap and the output of each stage propagates through to the next successive stage.

[0006] The art has recognized that current may leak from the dynamic logic precharge node during its evaluate phase in the instance when the discharge path to that precharge node is not enabled during that evaluate phase. Indeed, it is known in the art that the discharge path of the dynamic logic stage may include one or more so-called low threshold voltage (“LVT”) transistors, where such transistors are sometimes used in a discharge path in order to increase the speed of that path when the path is enabled. In such a case, each LVT transistor has a relatively lower threshold transistor as compared to other transistors in the same circuit, and such other transistors are therefore referred to as high threshold voltage (“HVT”) transistors. In one previous approach, at least one HVT transistor is also included in the discharge path. Indeed, for numerous additional details relating to such an approach and related aspects, the reader is invited to review U.S. Pat. No. 5,831,451, entitled “Dynamic Logic Circuits Using Transistors Having Differing Threshold Voltages,” issued Nov. 3, 1998, having the same inventor and assignee of the subject application, and U.S. Pat. No. 5,821,778, entitled “Using cascode transistors having low threshold voltages,” issued Oct. 13, 1998, having the same inventor and assignee of the subject application, where both of these two patents are hereby incorporated herein by reference. In any event, the use of an LVT transistor (or more than one such transistor) correspondingly provides increased leakage when it is not enabled and, as such, when implemented in the discharge path of a dynamic logic stage, that transistor increases leakage when that discharge path is not enabled. Increased leakage is undesirable because it increases device power consumption and, indeed, if severe, may jeopardize the proper operation of the circuit.

[0007] By way of further background, in dual rail dynamic logic circuits an approach has been implemented in an effort to alleviate the above-described leakage that occurs from the precharge node during the evaluate phase. Specifically, dual rail dynamic logic circuits include two precharge nodes, where during the evaluate phase one or the other of the precharge nodes is, by definition of the dual-rail system, assured to discharge, while at the same time therefore the other of the two precharge nodes remains charged. Thus, the charged precharge node can leak, particularly if there are one or more LVT transistors in a discharge path connected to the charged precharge node. Also in such a circuit, however, often a cross-coupled pair of LVT p-channel transistors is connected such that the source/drain path of each such transistor is connected to a respective precharge node while the gate of each respective transistor is connected to the opposite respective precharge node. As a result, when one precharge node is discharged low, that same node provides an enabling voltage to the opposing cross-coupled p-channel transistor, and that transistor, when enabled, conducts a source voltage to the opposing precharge node, thereby maintaining it at a sufficient level despite any leakage through its respective discharge path. Moreover, note in this case that the cross-coupled transistors are LVT transistors; as such, the LVT transistor provides a less-resistive connection between V_DDand the precharge node as compared to an HVT transistor. This is desirable, particularly at low operating voltages, when the relative drive currents of HVT transistors may diminish considerably as compared to their LVT counterparts. Also, the use of an LVT p-channel transistor in this manner is desirable because using an HVT transistor to provide such a coupling may prove insufficient to hold the precharge node at a sufficient voltage against the leakage of one or more LVT transistors in the discharge path. Still further, without sufficiently holding the precharge voltage high under these circumstances, insufficient noise margin may result and dynamic nodes may switch when they are not supposed to, due to noise or leakage currents, or both.

[0008] By way of still further background, in single rail dynamic logic circuits, it is known in the art to include an HVT p-channel transistor as a feedback, or sometimes referred to as a “keeper” transistor, connected such that the transistor gate is connected to the circuit output, the transistor source is connected to a voltage source, and the transistor drain is connected to the precharge node. In this manner, when the dynamic logic circuit is in its evaluate phase but its discharge path is not enabled, then the high precharge voltage is inverted and applied to the gate of the feedback transistor. In response, that transistor conducts and thereby couples the source voltage to the precharge node. In theory, therefore, this coupling of voltage offsets any leakage that may occur through the discharge path of the circuit. Moreover, the single feedback p-channel transistor cannot simply be made to be an LVT transistor because then, in instances when the discharge path is fully enabled and includes an HVT transistor, that HVT transistor included in the discharge path may provide a limit that keeps the discharge path from satisfactorily pulling down the precharge node, since that precharge node would be connected to a source voltage through a less-resistive LVT p-channel transistor. In other words, the HVT discharge path transistor may not provide sufficient drive current to overcome the LVT feedback p-channel transistor.

[0009] While the preceding approaches implemented with respect to dynamic logic circuits have proven satisfactory in many applications, the leakage in the discharge path of the circuit may be unacceptable, particularly when the leakage in that path is increased due to the use of one or more LVT n-channel transistors. Further, while an HVT p-channel keeper transistor may provide some current to compensate for the leakage of an LVT n-channel transistor in a single rail dynamic logic circuit, it has been recognized in connection with the preferred embodiments that such an approach alone may not be satisfactory for various reasons. Specifically, in order for the HVT p-channel keeper transistor to serve its purpose in this regard, the current it provides when enabled (i.e., the “on-current” of the p-channel transistor) must be sufficiently larger than the leakage current of the LVT n-channel transistor (i.e., the “off-current” of the n-channel transistor). However, this goal is becoming increasingly more difficult because there are various factors that are tending to decrease the ratio of the on-current of the HVT p-channel transistor relative to the off-current of the LVT n-channel transistor. For example, as the system source voltage, V_DD, decreases in circuit design as is the current trend, the above-described current ratio decreases. As another factor, transistor threshold voltage is a function of temperature, that is, high temperatures decrease threshold voltage, and this too decreases the subject current ratio. As still another factor, there is a trend to build many low threshold voltage transistors in a manner to provide more drive current, once more decreasing the above-discussed ratio. Finally, n-channel transistors are stronger in current provision than p-channel transistors, still again decreasing the above-discussed ratio. With these factors, note therefore that the HVT p-channel transistor attempts to source sufficient current to compensate for the leakage of the LVT n-channel transistor, but in that attempt an increasingly larger amount of voltage is dropped across the HVT p-channel transistor, particularly in a relative sense as its strength is reduced relative to that of the n-channel transistor. The increase in voltage drop may be detrimental for various reasons, including calling into question the reliability of data validity of the dynamic logic circuit. However, because the feedback transistor is an HVT transistor, it is likely unable to provide sufficient current to offset the larger amount of current that may be lost through leakage of any LVT transistor in the circuit discharge path.

[0010] Given the preceding, the preferred embodiments seek to improve upon the limitations and drawbacks of the prior art, as discussed in detail below.

[0011] In the preferred embodiment, there is a dynamic logic circuit. The dynamic logic circuit comprises a precharge node to be precharged to a precharge voltage during a precharge phase and a conditional discharge path connected to the precharge node. The conditional discharge path is operable, during an evaluate phase, to conditionally couple the precharge node to a voltage different than the precharge voltage. The dynamic logic circuit also comprises an output for providing a signal in response to a state at the precharge node. Lastly, the dynamic logic circuit comprises voltage maintaining circuitry, coupled to the output, for coupling the precharge voltage to the precharge node during a portion of an instance of the evaluate phase when the conditional discharge path is not enabled during the instance of the evaluate phase.

[0012] Other aspects are also disclosed and claimed.

[0013] FIG. 1 illustrates a schematic of a prior art dual rail dynamic logic circuit with cross-coupled LVT p-channel transistors connected to the precharge nodes of the circuit.

[0014] FIG. 2 illustrates a schematic of a prior art single rail dynamic logic circuit with an HVT p-channel feedback transistor connected to provide voltage to the precharge node when the precharge node is not discharged by the circuit discharge path.

[0015] FIG. 3 illustrates a schematic of a preferred embodiment dynamic logic circuit including a delay circuit and an LVT p-channel transistor supply circuit, wherein the delay circuit enables the supply circuit after the dynamic logic circuit has commenced its evaluate phase and in the instance when the precharge node has not been discharged by the circuit discharge path.

[0016] FIG. 4 illustrates a timing diagram of the operation of the preferred embodiment of FIG. 3 where the conditional data path from the precharge node to ground is enabled during the evaluate phase of operation.

[0017] FIG. 5 illustrates a timing diagram of the operation of the preferred embodiment of FIG. 3 where the conditional data path from the precharge node to ground is disabled during the evaluate phase of operation.

[0018] By way of background to the preferred embodiments, FIG. 1 is now discussed to provide additional context and detail with respect to the prior art introduced earlier in the Background Of The Invention section of this document. Specifically, FIG. 1 illustrates a schematic of a prior art dual rail dynamic logic circuit system designated generally at 10, where dynamic logic is also sometimes referred to as a “domino” logic. In general, system 10 may be considered about a vertical imaginary line down the middle of the Figure, where symmetric items about both sides of that line contribute to provide dual rail output signals, that is, complementary signals, during the data output period for system 10, where as described below that period is the evaluate phase of system 10.

[0019] Looking toward the left of FIG. 1, system 10 includes a precharge transistor 12_PT, a precharge node 12_PN, an output inverter 12_INV, and a low threshold voltage (“LVT”) p-channel transistor keeper transistor 12_KT, where the LVT transistor is designated with an asterisk to indicate its LVT attribute and this convention is used throughout the remainder of this document for other transistors as well. Precharge node 12_PNis connected to the input of output inverter 12_INV, and the output of inverter 12_INVprovides an output signal OUT₁. Precharge node 12_PNis also connected to the drain of precharge transistor 12_PTwhich is an HVT p-channel transistor, where the source of precharge transistor 12_PTis connected to a system voltage source, V_DD, and the gate of precharge transistor 12_PTis connected to a clock signal, CLK1. Further, precharge node 12_PNis connected to the drain of keeper transistor 12_KT, which is an LVT p-channel transistor, where the source of keeper transistor 12_KTis connected to V_DDand the gate of keeper transistor 12_KTis cross-coupled to precharge node 14_PN. Precharge node 12_PNalso is coupled to a discharge node 10_DNthrough the source/drain paths of different transistor combinations. First, precharge node 12_PNis coupled to discharge node 10_DNthrough the source/drain path of an LVT n-channel transistor 12_N1having its gate connected to a signal B and the source/drain path of an LVT n-channel transistor 12_N2having its gate connected to a signal A. Second, precharge node 12_PNis coupled to discharge node 10_DNthrough the source/drain path of an LVT n-channel transistor 14_N1having its gate connected to a signal BX and the source/drain path of an LVT n-channel transistor 14_N2having its gate connected to a signal AX. Note with respect to signals BX and AX that the “X” designation is to indicate a complementary signal, that is, AX is the complement of A and BX is the complement of B, where this convention of using an “X” in this manner is used with other signals also in this document. Lastly, discharge node 10_DNis connected via the source/drain path of an HVT n-channel discharge transistor 10_DTto ground, where discharge transistor 10_DThas its gate connected to CLK1. Given the preceding, two alternative paths are created from precharge node 12_PNto ground, and each such path may be referred to as a conditional discharge path because, as shown below, it permits a discharge along the entire path if the input to each device along that path is enabling such that the respective device is enabled.

[0020] Looking toward the right of FIG. 1, system 10 includes a precharge transistor 14_PT, a precharge node 14_PN, an output inverter 14_INV, and an LVT p-channel keeper transistor 14_KT. Precharge node 14_PNis connected to the input of output inverter 14_INV, and the output of inverter 14_INVprovides an output signal OUTX₁, that is, the dual rail of output OUT₁. Precharge node 14_PNis also connected to the drain of precharge transistor 14_PT, which is an HVT p-channel transistor, where the source of precharge transistor 14_PTis connected to V_DDand the gate of precharge transistor 14_PTis connected to CLK1. Further, precharge node 14_PNis connected to the drain of keeper transistor 14_KT, which is an LVT p-channel transistor, where the source of keeper transistor 14_KTis connected to V_DDand the gate of keeper transistor 14_KTis cross-coupled to precharge node 12_PN. Precharge node 14_PNalso is coupled to discharge node 10_DNthrough the source/drain paths of various transistor combinations. First, precharge node 14_PNis coupled to discharge node 10_DNthrough the source/drain path of an LVT n-channel transistor 14_N3having its gate connected to signal B and the source/drain path of LVT n-channel transistor 14_N2having its gate connected to signal AX. Second, precharge node 14_PNis connected to discharge node 10_DNthrough the source/drain path of an LVT n-channel transistor 12_N3having its gate connected to BX and the source/drain path of LVT n-channel transistor 12_N2. Thus, two alternative paths are created from precharge node 14_PNto ground, and each such path also may be referred to as a conditional discharge path for reasons discussed above.

[0021] The operation of system 10 is now explored, first by introducing three general aspects. First, as indicated above, each transistor that is an LVT transistor in system 10 is shown in FIG. 1 by including an asterisk next to the transistor schematic. With respect to each such LVT transistor, it has a relatively low threshold voltage relative to the other like-conductivity types of transistors in the circuit, that is, LVT n-channel transistors have lower threshold voltages than HVT n-channel transistors in system 10, and LVT p-channel transistors have lower threshold voltages than HVT p-channel transistors in system 10. A greater discussion of such “LVT” transistors is provided in the U.S. Pat. Nos. 5,831,451 and 5,821,778, incorporated earlier in the Background Of The Invention section of this document. Second, in general, system 10 operates in two phases of operation, where those phases are caused by the state of CLK1. A precharge phase of operation occurs when CLK1 is low, and during this phase precharge nodes 12_PNand 14_PNare precharged high and, thus, outputs OUT₁and OUTX₁are both low. The precharge phase is followed by an evaluate phase wherein CLK1 is high; in this latter phase, outputs OUT₁and OUTX₁are valid, and as a dual-rail system, one of those two outputs is high while the other is low. Third, during the evaluate phase of operation, the choice of which output is high and which is low is provided by the states of the inputs A, B, AX, and BX, where those inputs therefore provide a logic equation that dictates the output values for system 10. Each of these aspects is further explored below.

[0022] When CLK1 is low, it enables both p-channel precharge transistors 12_PTand 14_PT, and at the same time it disables n-channel discharge transistor 10_DT, thereby placing system 10 in its precharge phase of operation. Thus, each of p-channel precharge transistors 12_PTand 14_PTcouples V_DDto one of precharge nodes 12_PNand 14_PN, respectively, to which its drain is connected. Thus, during the precharge phase, precharge nodes 12_PNand 14_PNremain precharged without any conduction through any of the conditional discharge paths. Also during this phase, the high voltage on each precharge node 12_PNand 14_PNis inverted by the respective inverter 12_INVand 14_INV, respectively, to provide low signals at outputs OUT₁and OUTX₁, respectively. Further, the high signals at precharge nodes 12_PNand 14_PNdisable both p-channel keeper transistors 12_KTand 14_KT.

[0023] Following the precharge phase of operation of system 10, CLK1 transitions high, in which case it disables both p-channel precharge transistors 12_PTand 14_PT, and at the same time it enables n-channel discharge transistor 10_DT, thereby placing system 10 in its evaluate phase of operation. During the evaluate phase, inputs are A, AX, B, and BX are valid, where they may be provided from some other circuit, being either a static or dynamic circuit. By way of example, therefore, assume that both A and B are high during the evaluate phase. As such, both n-channel transistors 12_N1and 12_N2are enabled, as is n-channel discharge transistor 10_DTdue to the high CLK1; thus, a discharge path is enabled from precharge node 12_PNto ground via these three transistors. As a result, the voltage from precharge node 12_PNdrops to ground, and this falling transition causes a rising transition at output OUT₁from inverter 12_INV. At the same time, by definition as complements, AX and BX are low, and thus no conditional path is enabled from precharge node 14_PNto ground. As a result, the voltage from precharge node 14_PNremains high, and this high voltage causes a low output OUTX₁from inverter 14_INV. Indeed, given the other signal possibilities and connectivity of system 10, then one skilled in the art will appreciate that outputs OUT₁and OUTX₁realize the logical equations shown in the following Equations 1 and 2:

[0024] OUT₁=(B AND A) OR (BX AND AX)  Equation 1

[0025] OUTX₁=(B and AX) OR (BX AND A)  Equation 2

[0026] In addition to the preceding, a noteworthy observation is made with respect to the effect of the cross-coupled LVT p-channel keeper transistor 12_KTand 14_KTduring the evaluate phase. From the preceding discussion, one skilled in the art will appreciate that during any evaluate phase, one of the two precharge nodes 12_PNand 14_PNdischarges while the other remains charged. For the precharge node that remains charged, however, such a precharge node is connected to two different conditional discharge paths, and each of those paths includes two LVT n-channel transistors. For example, precharge node 12_PN, when charged, is still connected via transistors 12_N1and 12_N2to discharge node 10_DN, or alternatively it is connected via transistors 14_N1and 14_N2to discharge node 10_DN. While these conditional discharge paths are not enabled at this time, the LVT nature of those transistors causes them to leak current, and as a result there is a draw on the voltage at precharge node 12_PN. However, also at this time, recall that the opposite precharge node 14_PNis low, and recall further that this low voltage is cross-connected to the gate of p-channel keeper transistor 12_KT. As a result, keeper transistor 12_KTis enabled, thereby coupling V_DDto precharge node 14_PN. As a result, by virtue of keeper transistor 12_KTbeing an LVT transistor, it provides an LVT current source to precharge node 12_PN, thereby offsetting the LVT current leakage that will occur from precharge node 12_PNto ground along the disabled conditional discharge paths. Having shown how p-channel keeper transistor 12_KToperates in this respect when precharge node 12_PNis high and precharge node 14_PNis therefore low, one skilled in the art will appreciate the complementary operation of p-channel keeper transistor 14_KTwhen precharge node 14_PNis high (and precharge node 12_PNis therefore low). Thus, such an approach provides improved operation for a dual rail circuit; however, as shown below, a single p-channel keeper transistor per precharge node is not workable in a single rail dynamic logic circuit, where later the preferred embodiments are instead shown to provide a solution in this regard.

[0027] By way of additional background to the preferred embodiments, FIG. 2 is now discussed also to elaborate with respect to the prior art introduced earlier in the Background Of The Invention section of this document. Specifically, FIG. 2 illustrates a schematic of a prior art single rail dynamic logic circuit system designated generally at 20. System 20 includes a precharge transistor 20_PT, a precharge node 20_PN, an output inverter 20_INV, an HVT p-channel feedback transistor 20_FP, and a conditional discharge path including both a logic circuit 20_Land a discharge transistor 20_DT. Looking in more detail to these devices, precharge transistor 20_PTtypically is an HVT p-channel transistor and is typically smaller than the n-channel transistors shown in FIG. 2. Precharge transistor 20_PThas a source connected to a source of the system voltage level (e.g., V_DD) and a drain connected to precharge node 20_PN. A clock signal, CLK2, is connected to the gate of precharge transistor 20_PT, and it is also connected to the gate of discharge transistor 20_DT. Discharge transistor 20_DTis an HVT n-channel transistor having its source connected to a low reference voltage (e.g., ground) and its drain connected to a node 20_N. Between precharge node 20_PNand node 20_Nis connected a logic circuit 20_L. The particular logic of logic circuit 20_Lmay implement any type of logic equation. For example, often logic circuit 20_Lincludes numerous n-channel transistors connected in various fashions to implement a logic equation, and the logic function of that equation is determined based on the particular transistor configuration as well as the respective input signals connected to the gates of those transistors. For purposes of the present discussion, the particular logic equation and inputs are not important and, thus, the word INPUT(S) is generally shown with it understood that such signals may come from any of various other circuits, being static, dynamic, or both. Moreover, preferably the n-channel transistor transistor(s) within logic circuit 20_Lare LVT n-channel transistors and, thus, have a low threshold voltage relative to the n-channel discharge transistor 20_DT, where again a greater discussion of such “LVT” transistors is provided in the U.S. Pat. Nos. 5,831,451 and 5,821,778, incorporated earlier in the Background Of The Invention section of this document. Once more to designate the use of LVT transistors in logic circuit 20_L, it is labeled with an asterisk.

[0028] The operation of system 20, as a dynamic logic circuit, also includes both a precharge and an evaluate phase. Specifically, when CLK2 is low, it enables p-channel precharge transistor 20_PT, and at the same time it disables n-channel discharge transistor 20_DT, thereby placing system 20 in its precharge phase of operation. Thus, p-channel precharge transistor 20_PTcouples V_DDto precharge node 20_PN. Thus, precharge node 20_PNremains precharged during the precharge phase without any conduction through the conditional discharge path. Also during this precharge phase, the high voltage on precharge node 20_PNis inverted by inverter 20_INVto provide a low signal at output OUT₂. Further, the low signal at output OUT₂is connected to the gate of, and thereby enables, p-channel feedback transistor 20_FP, and this HVT transistor also couples V_DDto precharge node 20_PN. Alternatively, when CLK2 transitions high, it disables p-channel precharge transistor 20_PT, and at the same time it enables n-channel discharge transistor 20_DT, thereby placing system 20 in its evaluate phase of operation. During the evaluate phase, the INPUT(S) are valid, where they may be provided from some other circuit, being either a static or dynamic circuit. Thus, based on the states of the INPUT(S), if a logic equation of logic circuit 20_Lis satisfied, then a conductive path is created through that circuit, whereas if no such logic equation is satisfied, then no conductive path is created through that circuit. In the first instance, that is, where the conditional discharge path is enabled, then precharge node 20_PNis discharged through logic circuit 20_Land the already-enabled discharge transistor 20_DT, thereby causing the voltage at precharge node 20_PNto drop to ground and the output OUT₂to rise. In the second instance, that is, where the conditional discharge path is not enabled, then the voltage at precharge node 20_PNis not initially discharged; however, also at this time, the one or more LVT transistors in logic circuit 20_Lwill leak, thereby pulling down the precharge voltage that is desired to remain undisturbed at precharge node 20_PN. At the same time, note further that the precharge voltage is inverted, thereby providing a low voltage at output OUT₂, which enables p-channel feedback transistor 20_FP. However, p-channel feedback transistor 20_FPis an HVT transistor, and therefore there is the result that this HVT transistor attempts to offset the leakage provided by the LVT transistor(s) in logic circuit 20_L. As described earlier, this mis-match in threshold voltage strength may not prove suitable in various circumstances.

[0029] FIG. 3 illustrates a schematic of an inventive single rail dynamic logic circuit designated generally at 30. System 30 in some respects resembles system 20 of FIG. 2, where one skilled in the art will readily appreciate certain commonality. However, system 30 includes additional aspects, thereby providing in combination an overall improved system, as will be apparent following the remaining discussion. Looking now in more detail to those items in system 30 that are comparable in some respects to those in system 20, system 30 includes a precharge transistor 30_PT, a precharge node 30_PN, an output inverter 30_INV, and a conditional discharge path including both a logic circuit 30_Land a discharge transistor 30_DT. Precharge transistor 30_PTis preferably an HVT p-channel transistor and is typically smaller than the n-channel transistors shown in FIG. 3. Precharge transistor 30_PThas a source connected to a source of the system voltage level (e.g., V_DD) and a drain connected to precharge node 30_PN. A clock signal, CLK3, is connected to the gate of precharge transistor 30_PT, and it is also connected to the gate of discharge transistor 30_DT. Discharge transistor 30_DTis preferably an n-channel transistor having its source connected to a low reference voltage (e.g., ground) and its drain connected to a node 30_N1. Between precharge node 30_PNand node 30_N1is connected a logic circuit 30_L. The particular logic of logic circuit 30_Lmay implement any type of logic equation, and preferably logic circuit 30_Lincludes one or more LVT n-channel transistors connected in various fashions to implement a logic equation, where the logic function of that equation is determined based on the particular transistor configuration as well as the respective input signals connected to those transistors (e.g., to the gates or one of the source/drain nodes). Again, the particular logic equation and inputs are not important and, thus, the word INPUT(S) is generally shown with it understood that such signals may come from any of various other circuits, being static, dynamic, or both.

[0030] Looking now to various differences between the inventive system 30 and prior art system 20, CLK3 is also connected as an input to a delay circuit 30_DEL, which for reasons detailed below provides an output delayed clock signal, CLKD, that preferably represents an inverted form of CLK3 and has transitions that are delayed with respect to CLK3 by a period of time that is shorter than the evaluation period of system 30, yet longer than the time that system 30 takes to transition when its conditional discharge path is enabled. In the illustrated example, this delay is achieved by making it equal to three signal inversions, that is, it is accomplished by connecting CLK3 into a series connection of three inverters 30_DINV1, 30_DINV2, 30_DINV3, so that the output of inverter 30_DINV3provides the CLKD signal. The CLKD signal is connected to the gate of an LVT p-channel keeper transistor 30_KT1that has its source connected to V_DDand its drain connected to a node 30_N2. Node 30_N2is also connected to the source of an LVT p-channel keeper transistor 30_KT2that has its drain connected to precharge node 30_PNand its gate connected to output OUT₃, that is, to the output of inverter 30_INV. Also, consistent with the earlier conventions, both LVT p-channel keeper transistors 30_KT1and 30_KT2are labeled with asterisks to indicate the preference that each such transistor is an LVT device. Lastly, note for sake of simplification that FIG. 3 does not illustrate an additional HVT feedback transistor such as p-channel feedback transistor 20_FPshown in FIG. 2; however, such a device also could be connected in the same manner as shown in FIG. 2, thereby providing an additional coupling of V_DDto precharge node 30_PNwhen output OUT₃is low and, indeed, even in the instances when p-channel keeper transistors 30_KT1and 30_KT2are disabled.

[0031] Before detailing the operation of system 30, additional observations are noteworthy with respect to the use of a low threshold voltage transistor versus a high threshold voltage transistor as those devices are also included in the preferred embodiment. In this regard, note that these terms are used to designate the relative values of one transistor's threshold voltage versus another transistor's threshold voltage and are not necessarily limiting those terms to certain absolute values. Particularly, it is known in the art that a transistor's threshold voltage may be determined by measuring its current-voltage characteristics to determine a gate-to-source voltage where the transistor conducts a small amount of current. To demonstrate the relative nature of threshold voltages, consider the following. For a first transistor with a first threshold voltage, it will provide a first amount of drive current at a given drain-to-source voltage (with its gate connected to its drain). For a second transistor with a second threshold voltage lower than that of the first transistor, the second transistor will provide a second amount of drive current at the given drain-to-source voltage (with its gate connected to its drain), and which drive current is larger than the first amount of drive current provided by the first transistor at that same drain-to-source voltage. Thus, within this document and with respect to the preferred embodiments, a transistor of the first type is referred to as an HVT transistor (i.e., a transistor with a relatively high V_T) while a transistor of the second type is referred to as an LVT transistor (i.e., a transistor with a relatively low V_T). Further, note that an HVT transistor preferably has a V_Ton the order of that for transistors which are now used throughout a circuit where all transistors share the same V_T, and where that V_Twas determined to be high enough to avoid unacceptably large leakage currents. Note also that although an LVT transistor provides the benefit of a higher drive current as opposed to an HVT transistor, in contrast it gives rise to the detriment that it provides a greater leakage current than an HVT transistor. Indeed, the typical leaking by an LVT transistor may be greater than that of an HVT transistor by two or three orders of magnitude or more. Lastly, since LVT and HVT have been shown as relative terms, note in FIG. 3 that each of p-channel transistors 30_KT1and 30_KT2, as LVT transistors, preferably has a lower V_Tas compared to at least one other p-channel transistor in system 30, such as in comparison to p-channel precharge transistor 30_PTor in comparison to the p-channel transistor which will be understood by one skilled in the art to be included in inverter 30_INV.

[0032] The operation of system 30 is now described, and the following demonstrates that the inclusion of delay circuit 30_DELand p-channel keeper transistors 30_KT1and 30_KT2provides additional protection against voltage being lost at precharge node 30_PN, where such protection is provided in that in certain instances these LVT transistors provide additional drive current to that node to offset the current leakage by any one or more LVT transistors in logic circuit 30_L, thereby operating to maintain the precharge voltage at precharge node 30_PN. In addition, however, the inclusion of these devices is such that in other instances proper data switching at precharge node 30_PNis not disturbed. These attributes and benefits are further explored below in connection with the two possible instances of operation of the conditional data path in response to the INPUT(S), wherein in a first instance the INPUT(S) are such that the conditional data path is enabled as described in connection with FIG. 4, and wherein in a second instance the INPUT(S) are such that the conditional data path is not enabled as described in connection with FIG. 5.

[0033] FIG. 4 illustrates a timing diagram of various signals in connection with system 30 and for an example where the conditional data path from precharge node 30_PNto ground is enabled during the evaluate phase of operation, while the figure also illustrates for completeness the preceding precharge phase of operation. Looking, then, to the signals, at a time t₀, CLK3 transitions low, thereby enabling p-channel precharge transistor 30_PTand disabling n-channel discharge transistor 30_DT. With precharge p-channel transistor 30_PTenabled, precharge node 30_PNis coupled to V_DD; thus, following to FIG. 4 labels the CLK3 signal as corresponding to the precharge phase of operation. Also during the precharge phase, the V_DDcoupled to precharge node 30_PNis inverted by inverter 30_INVso that output OUT₃is low, and this low is connected to the gate of p-channel keeper transistor 30_KT2and thereby enables that transistor. Also at the beginning of the precharge phase, CLKD is low, and this low is connected to the gate of p-channel keeper transistor 30_KT2and thereby enables that transistor. Thus, the combination of p-channel keeper transistors 30_KT1and 30_KT2is enabled (or “ON” as shown in FIG. 4), thereby further coupling V_DDto precharge node 30_PNvia the series-connected source/drain paths of those two transistors.

[0034] The falling transition of CLK at t₀is input to delay circuit 30_DEL, and thereby passes through the three inverters 30_DINV1, 30_DINV2, and 30_DINV3; as a result after a combined delay corresponding to these three signal inversions, at time t₁CLKD rises to a high value. This high value is connected to the gate of p-channel keeper transistor 30_KT1and thereby disables that transistor. Thus, starting at t₁, the combination of p-channel keeper transistors 30_KT1and 30_KT2is disabled (or “OFF” as shown in FIG. 4), thereby no longer coupling V_DDto precharge node 30_PNvia the series-connected source/drain paths of those two transistors. However, p-channel precharge transistor 30_PTremains enabled, thereby continuing to couple V_DDto precharge node 30_PNfor the duration of the precharge phase.

[0035] At a time t₂, CLK3 transitions high, thereby disabling p-channel precharge transistor 30_PTand enabling n-channel discharge transistor 30_DT. With this transition, system 20 is deemed to enter its evaluate phase of operation, with such a label appropriately shown in FIG. 4 following t₂. Also during the evaluate phase, the INPUT(S) to logic circuit 30_Lare valid. Further, recall that the example of FIG. 4 is intended to illustrate the instance wherein the conditional discharge path from precharge node 30_PNis enabled. Thus, it is assumed that the INPUT(S) to circuit 30_Lare such that at least one conductive path is formed through logic circuit 30_L, as illustrated in FIG. 4 by indicating that logic circuit 30_Lturns “ON” at a time t₃, where the difference between t₂and t₃is intended to depict the delay of enabling discharge transistor 30_DTas well as the relevant transistor(s) in logic circuit 30_L. Further, as a result of the then-enabled conditional discharge path, the precharge voltage at precharge node 30_PNis discharged to ground through that path. Thus, following a delay after t₃for this discharge to occur and for the operation of inverter 30_INV, then at a time t₄output OUT₃begins to rise.

[0036] Completing the illustration of FIG. 4, recall at t₂CLK3 transitioned high, and this high signal transition is connected as an input to delay circuit 30_DEL. Thus, following the three inversion delay of that circuit as provided by inverters 30_DINV1, 30_DINV2, and 30_DINV3, at a time t₅CLKD transitions low. While this low signal is connected to the gate of p-channel keeper transistor 30_KT1, at that same time however the high from output OUT₃is connected to the gate of n-channel keeper transistor 30_KT2. Thus, continuing after time t₄, the combination of n-channel keeper transistors 30_KT1and 30_KT2remains off and thereby do not impact the operation of system 20 during its evaluate phase.

[0037] FIG. 5 illustrates a timing diagram of various signals in connection with system 30 and for an example where the conditional data path from precharge node 30_PNto ground is disabled during the evaluate phase of operation. At the outset, the signals in FIG. 5 are the same as shown in FIG. 4 from times t₀through time t₂. Thus, the reader is assumed familiar with the previous discussion or is referred thereto for additional details, where such details describe the precharge phase from t₀to t₂and the commencement of the evaluate phase at t₂. Following t₂in FIG. 5, again the INPUT(S) to logic circuit 30_Lare valid, and for the example of FIG. 5 it is assumed that the INPUT(S) to circuit 30_Lare such that no conductive path is formed through logic circuit 30_L, as illustrated in FIG. 5 by indicating that logic circuit 30_Lremains “OFF” as of, and following, t₂. At this point, therefore, although discharge transistor 30_DTis enabled by the high CLK3, there is no enabled current path from precharge node 30_PNto ground. However, the disabled transistors in logic circuit 30_Lwill leak, and indeed because these transistors are preferably LVT transistors, their leakage will exceed that of their HVT counterparts. Thus, immediately following t₂, such leakage may tend to pull down the voltage at precharge node 30_PN, although as described below this leakage is thereafter alleviated as of a time t_3.1.

[0038] At time t_{3 1}(indicated as such to distinguish it from the reference t₃in FIG. 4), CLKD transitions low. Specifically, recall that at t₂CLK3 transitioned high. Thus, following the three-inverter delay of delay circuit 30_DEL, at t_3.1is shown the responsive low transition of CLKD. Further, also as of t_3.1, OUT₃was already low because the conditional discharge path to precharge node 30_PNwas not enabled. As a result, as of t_{3 1}, a low signal is connected by CLKD to the gate of p-channel keeper transistor 30_KT1and a low signal is connected by output OUT₃to the gate of p-channel keeper transistor 30_KT2. Thus, as shown in the last signal of FIG. 5, as of t_3.1the combination of p-channel keeper transistors 30_KT1and 30_KT2is enabled (i.e., turned on). The enabled combination of p-channel keeper transistors 30_KT1and 30_KT2provides a conductive path, along the series-connected source/drains of those two transistors, between V_DDand precharge node 30_PN. Additionally, because p-channel keepers transistor 30_KT1and 30_KT2are both preferably LVT devices, then as of t_3.1an LVT source, formed by these two transistors, provides current to precharge node 30_PNto offset any leakage that may be occurring in the conditional discharge path due to the LVT transistor(s) in logic circuit 30_Lof that path. Moreover, note that this offset continues for the remainder of the evaluate phase of system 30.

[0039] From the preceding, one skilled in the art should appreciate that system 30 provides various benefits over the prior art. First, in the instance where a single rail dynamic logic circuit does not change state (e.g., provides a low output) due to its conditional discharge path(s) remaining disabled during its evaluate phase, then LVT transistor leakage that might otherwise occur in the discharge path is offset in part or in whole due to the inclusion of LVT transistor current sourcing to the circuit precharge node. Second, in the instance where the single rail dynamic logic circuit does change state (e.g., provides a high output) due to its conditional discharge path(s) being enabled during its evaluate phase, then the LVT transistor current source is disabled and, thus, does not disturb the ability of the circuit to properly discharge during the evaluate phase. Given these benefits, the conditional discharge path may employ LVT n-channel devices to benefit from the speed of those devices when enabled during the evaluate phase, yet any additional leakage provided by those devices when not enabled during the evaluate phase can be offset by the additional current source. In the preferred embodiments, the additional current source implements LVT p-channel devices, and the above functionality is provided by using a delay circuit triggered by the circuit CLK signal, where preferably that delay is three inversions following the CLK signal. However, in an alternative embodiment, other delay configurations may be devised. For example, five inverters could be used. In still another alternative embodiment, yet another structure may be implemented, where a net benefit is realized so long as the LVT p-channel transistor current source is enabled prior to the end of the evaluate phase, but also not too early in the evaluate phase so as to interfere with proper data switching of the circuit in the instances where a conditional discharge path is enabled. Given these timing goals, the LVT p-channel transistor source can be enabled, thereby guarding against leakage and noise. Moreover, this functionality is provided without any additional clocks. Lastly, note that while the present embodiments have been described in detail, various substitutions, modifications or alterations could be made to the descriptions set forth above without departing from the inventive scope which is defined by the following claims.