Asynchronous circuit

An asynchronous circuit, also known as a clockless or self-timed circuit, is a type of sequential digital logic. In digital electronics, sequential logic is a system where the output depends on both current inputs and past inputs. Most modern digital devices use synchronous circuits to manage this process. These synchronous systems rely on a global clock signal to coordinate all components. In contrast, an asynchronous circuit does not use a central clock or signal generator. Instead, it relies on a handshaking circuit to manage the flow of information. This method uses data transfer protocols to indicate when a set of instructions is complete.

To understand how these circuits work, it is helpful to compare them to synchronous systems. In a synchronous circuit, an electronic oscillator creates a repetitive series of pulses called a clock signal. This signal is sent to every component in the integrated circuit. Components like flip-flops only change their state when they receive the edge of a clock pulse. This ensures all signals change at regular intervals. However, signals take time to move through logic gates, a delay known as propagation delay. In large circuits, the clock signal might arrive at different parts at different times. This problem is called clock skew.

Asynchronous circuits avoid the clock skew problem by operating without a global beat. The state of the circuit changes as soon as the inputs change. These circuits use a handshaking mechanism to communicate between functional blocks.

One part of the circuit sends data and then signals that it has finished. The receiving part then responds to complete the protocol. Because they do not have to wait for a specific clock pulse to begin processing, asynchronous circuits can operate much faster. Their theoretical speed limit is defined only by the propagation delays of the logic gates and other elements.

There are different design styles within asynchronous logic. One style is the bundled delay model. This model uses conventional data processing elements and indicates completion using a locally generated delay model. Another style is delay-insensitive design. In this approach, the circuit can accommodate any amount of delay through its elements. While delay-insensitive circuits are often larger, they are considered "correct by design" because they are insensitive to layout and parametric variations.

Researchers also use Petri nets to model these circuits. A specific type called Signal Transition Graphs (STGs) was proposed in 1985 by Leonid Rosenblum, Alex Yakovlev, and Tam-Anh Chu. These models help engineers analyze and build asynchronous control circuits.

Asynchronous technology has a long history of development. The original theory was created by David E. Muller in the mid-1950s. This work was later published in the book "Switching Theory" by Raymond Miller. While synchronous circuits were used in early asynchronous systems like the ORDVAC in the 1950s, asynchronous circuits did not see real-life commercial products until the 1990s. Since then, new logical operations have been developed to handle the complexity of these systems. In 1984, Vadim O. Vasyukevich introduced "venjunction" and "sequention." These operations allow the logic to account for the history of an element, not just its current value.

Asynchronous circuits offer several significant advantages in performance and efficiency. They provide better average-case performance because they are not limited by the worst-case completion time of a component. They also feature inherent elasticity, which allows them to handle variable input and output rates gracefully. One major benefit is lower power consumption. In 2005, Epson reported that their flexible 8-bit asynchronous microprocessor used 70% less power than a synchronous design. Additionally, these circuits produce less electromagnetic interference (EMI). While synchronous circuits create EMI near their clock frequency, asynchronous circuits spread this interference more evenly across the spectrum.

Despite these benefits, designing asynchronous circuits is quite difficult. They are susceptible to a problem called a race condition. This happens if two inputs arrive at a gate at nearly the same time. Slight differences in propagation delays can cause the circuit to enter the wrong state. Because of this complexity, asynchronous circuits can sometimes require double the area or resources of a synchronous design. They also require specialized training, as most engineers are used to synchronous systems. Currently, asynchronous circuits are most often used in critical parts of larger systems, such as high-speed signal processing circuits.