The invention relates to dynamic circuits.
Dynamic circuits are not robust under noisy condition. For example, a typical dynamic domino circuit uses a half or full latch to hold its dynamic nodes. However, under noisy conditions, these nodes can couple to noise and lose their data. Thus, a need exists for a dynamic domino circuit that is robust under noisy condition.
As another example, in many kinds of dynamic adders, inversion of carry that is required at the end of the carry chain is either accomplished by using a static inverter (extra gate delay) followed by static circuits, or having a separate chain of logic to generate carry bar (area penalty). For a really fast implementation, four chains of logic (g, ˜g, p, ˜p) are needed, thereby consuming a lot of area. Thus, a need exists for a dynamic adder that does not use a static inverter that causes extra gate delay A further need exists for a dynamic adder that does not consume a lot of area.
Furthermore, in most dynamic adders, a latch is required at the output to preserve the generated sum during circuit precharge. A need exists to improve the speed of the adders by cutting the latch delay while not requiring complex clocking.
The invention provides a dynamic domino circuit that is robust under noisy condition. The invention provides a dynamic adder that does not use a static inverter that causes extra gate delay. The invention also provides a dynamic adder that does not consume a lot of area. The invention also improves speed of the adders by cutting the latch delay while not requiring complex clocking.
Preferably, a dynamic circuit includes a logic portion and three dynamic output portions, each of which having a dynamic node for holding data. A first and a second transistors have their gates coupled to the first dynamic node. The first transistor has its drain coupled to the second dynamic node, and the second transistor has its drain coupled to the third dynamic node. A third and a fourth transistors have their gates coupled to the second dynamic node. The third transistor has its drain coupled to the first dynamic node, and the fourth transistor has its drain coupled to the third dynamic node. A fifth and a sixth transistors having their gates coupled to the third dynamic node. The fifth transistor has its drain coupled to the first dynamic node, and the sixth transistor has its drain coupled to the second dynamic node.
Reference is made in detail to the preferred embodiments of the invention. While the invention is described in conjunction with the preferred embodiments, the invention is not intended to be limited by these preferred embodiments. On the contrary, the invention is intended to cover alternatives, modifications and equivalents, which may be included within the spirit and scope of the invention as defined by the appended claims. Furthermore, in the following detailed description of the invention, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, as is obvious to one ordinarily skilled in the art, the invention may be practiced without these specific details. In other instances, well-known methods, procedures, components, and circuits have not been described in detail so that aspects of the invention will not be obscured.
Referring now to
Referring still to
Transistors 132 and 142 are coupled to dynamic node 124. Transistors 122 and 143 are coupled to dynamic node 134. Transistor 123 and 133 are coupled to dynamic node 144. Assume dynamic nodes 124, 134 and 144 are mutually exclusive. That is to say, one and only one dynamic node among nodes 124, 134 and 144 will be at logic zero when the clock is high. As such, when dynamic node 124 holds a logic zero, dynamic node 134 holds a logic one and dynamic node 144 holds a logic one. Similarly, when dynamic node 134 holds a logic zero, dynamic node 124 holds a logic one and dynamic node 144 holds a logic one. Also similarly, when dynamic node 144 holds a logic zero, dynamic node 124 holds a logic one and dynamic node 134 holds a one.
This exclusive property can be used to ensure that the dynamic nodes 124, 134 and 144 will recover fully from noise. Specifically, as an example, a logic zero being held on dynamic node 124 turns on transistor 132 to provide a current to ensure that a logic one is held in dynamic node 134. As such, if a noise spike occurs at dynamic node 134 to change the logic one held therein into a logic zero, the current supplied by transistor 132 recovers the logic one at dynamic node 134. Also, a logic zero being held on dynamic node 124 turns on transistor 142 to provide a current to ensure that a logic one is held in dynamic node 144. As such, if a noise spike occurs at dynamic node 144 to change the logic one held therein into a logic zero, the current supplied by transistor 142 recovers the logic one at dynamic node 144. As understood herein, the idea of noise recoverability applies not only to circuit 100, but to any class of dynamic circuit that uses exclusive signals.
Continuing with
Referring now to
In one embodiment of the invention, the terms k, p, g can be defined as:
As an example, referring still to
˜p0=g0+k0,
˜g0=p0+k0, and
˜k0=g0+p0,
Similar equations apply for inversion at all other levels of carry chain 205.
Additionally, the terms k, p, g can be defined in groups as follows:
gk(i:0)=k(i)+p(i)&gk(i−1:0)
gp(i:0)=p(i)&gp(i−1:0)
gg(i:0)=g(i)+p(i)&gg(i−1:0)
For example, when i=1, gk, gg, gp corresponds to gk1 0, gg1 0, gp1:0, as referenced in
Also, for i>j>k,
gk(i:k)=gk(i:j)+gp(i:j)+gk(j−1:k)
gp(i:k)=gp(i:j)+gp(j−1:0)
gg(i:k)=gg(i:j)+gp(i:j)+gg(j−1:k)
Group level inversions can be accomplished as a function of true terms as follows:
˜gk(i:k)=gp(i:k)+gg(i:k)
˜gp(i:k)=gg(i:k)+gk(i:k)
˜gg(i:k)=gp(i:k)+gk(i:k)
In this embodiment, the terms k, p, g are mutually exclusive and the terms gk, gp, gg are mutually exclusive. By using exclusivity nature of the terms k/p/g and gk/gp/gg, inversion of signals can be expressed in terms of monotonic signals.
By using true terms to generate complements, a common problem in dynamic signal inversion is circumvented. In a regular dynamic circuit, inversion is frequently implemented by demorganization all the way to the beginning of carry chain 205. In so doing, four terms (generate (g), generate bar (˜g), propagate (p), and propagate bar (˜p)) are required per dynamic circuit. On the other hand, by using only three terms rather than the conventional four terms, more area can be saved.
Referring now to
Referring now to
This application is a Continuation-in-Part of U.S. patent application Ser. No. 09/970,250, entitled “Dynamic Circuits Using Exclusive States” by Pak, et al., filed on Oct. 4, 2001.
Number | Name | Date | Kind |
---|---|---|---|
4962471 | Cornelissen | Oct 1990 | A |
5208490 | Yetter | May 1993 | A |
5600583 | Bosshart et al. | Feb 1997 | A |
5719803 | Naffziger | Feb 1998 | A |
6144228 | Matson et al. | Nov 2000 | A |
6175852 | Dhong et al. | Jan 2001 | B1 |
6329857 | Fletcher | Dec 2001 | B1 |
6496846 | Bradley | Dec 2002 | B1 |
Number | Date | Country | |
---|---|---|---|
Parent | 09562063 | May 2000 | US |
Child | 09970250 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 09970250 | Oct 2001 | US |
Child | 09996865 | US |