Clock signal

A clock signal is an electronic logic signal used to synchronize digital circuits. This signal consists of voltage or current that oscillates between a high and a low state. It maintains a constant frequency to act as a metronome for the system. In a synchronous logic circuit, the clock signal is applied to all storage devices, such as flip-flops and latches. This ensures that all components change their state at the exact same time. By doing this, the signal prevents race conditions, where signals arrive out of order.

An electronic oscillator called a clock generator produces this signal. Most common clock signals take the form of a square wave. This wave typically has a 50% duty cycle, meaning the high and low states last for equal amounts of time. Circuits may become active during the rising edge, which is the transition from low to high. Others may trigger on the falling edge, the transition from high to low. In double data rate systems, the circuit uses both the rising and falling edges to increase speed. Engineers also use clock gating to save power. This technique combines the clock with a controlling signal to enable or disable specific parts of a circuit when they are not in use.

Historically, the way clock signals are distributed has changed significantly. In the 1970s, many metal oxide semiconductor (MOS) integrated circuits used a two-phase clock. This system distributes signals on two separate wires with non-overlapping pulses. This method allows the use of gated latches instead of edge-triggered flip-flops. Because a gated latch uses only four gates compared to six in a flip-flop, this can reduce the total gate count. Early microprocessors like the Motorola 6800 and the Intel 8080 required these two-phase signals. To support them, companies produced specific hardware like the Intel 8224 clock generator and the Motorola MC6875. Eventually, newer chips began to include the clock generation circuitry directly on the microprocessor chip itself.

Different architectures have required different numbers of clock phases. While two-phase logic was common in the 1970s, some very early integrated circuits used four-phase logic. This required four separate, non-overlapping clock signals. Examples of this include the National Semiconductor IMP-16 and the Texas Instruments TMS9900. However, four-phase clocks are rarely seen in modern CMOS processors. Most modern microprocessors and microcontrollers have moved to a single-phase clock. This means all clock signals are effectively transmitted on a single wire. The MOS Technology 6502 is an interesting example, as it uses two-phase logic internally but includes an on-chip generator to accept a single-phase input.

Performance and speed are often managed through specialized clocking techniques. Many microcomputers use a clock multiplier to boost performance. This device multiplies a lower frequency external clock into a much higher rate for the microprocessor. This allows the CPU to operate at a higher frequency than the rest of the computer. Such a setup provides performance gains when the CPU does not have to wait for external factors like memory. Additionally, many devices use dynamic frequency change. Instead of a fixed frequency, the clock generator can vary the time between edges. This is used in techniques like spread-spectrum clock generation or dynamic frequency scaling. Some devices using static logic can even be slowed or paused indefinitely.

Distributing the clock signal across a large chip is a massive engineering challenge. In a large microprocessor, the power required to drive the clock signal can exceed 30% of the chip's total power. To reach every part of the chip with low skew, engineers use a metal grid or a clock distribution network. This network is sometimes shaped like a tree, known as an H-tree. Because the clock signal provides the temporal reference for all data, the waveforms must be extremely clean and sharp. As technology scales down, long interconnect lines become more resistive. This increased resistance makes efficient clock distribution even more critical for maintaining synchronous performance.

Finally, different types of circuits may require different signal shapes. While digital circuits use square waves, some sensitive mixed-signal circuits use sine waves. Precision analog-to-digital converters might use sine waves to avoid high-frequency harmonics. These harmonics can create noise that interferes with analog circuitry. Some engineers also use differential signals, which can have twice the slew rate and half the timing uncertainty of single-ended signals. These signals also radiate less strongly than a single line. Researchers continue to look for new ways to improve these systems, such as resonant clocking or on-chip optical interconnects.