We use time to measure our perception of the world and to mark and coordinate events. Ideally, everything would run smoothly and on schedule. Unfortunately, in the real world, this is not always true. Timing issues can cause a disruption of the entire system.

Electronic systems hold to the same principle. Early systems had simple clocks and oscillators for timing. A main clock would provide timing to a CPU which would coordinate the system’s activities. As the CPUs, memories, and I/O peripherals became more complex, so did the timing. Systems no longer run based on a single clock frequency. Memory products, such as DRAM or communication interfaces, have their own timing requirements which can differ from the CPU. Additionally, there can be requirements for Real-Time-Clock functions. In most systems, there exists a need to coordinate or synchronize the operations of the entire system. This may be generating a high or lower-frequency version of a system clock or distributing a “clean” jitter-free version of the clock throughout the system.

As the complexity of timing requirements grew, semiconductor manufacturers created dedicated products to address these needs. Oscillators, timers, multivibrators, and simple buffers were sufficient to meet the early requirements but quickly grew with the demands for higher performance. Designers of larger systems began to implement timing networks within their systems. Jitter performance and skew variations became critical as clock frequencies increased and signal levels decreased. Noise and clock edge variations could introduce errors if they were not addressed. Conversely, functionality, such as spread-spectrum clocking that dynamically moves clock edges, was introduced to reduce EMI issues.

The higher performances also introduced the need for improved electrical interfaces. The use of PECL, LVDS, HSTL, CML, and other interfaces was utilized to meet performance needs. Designers now faced the challenge of distributing timing to devices that may require different interfaces. This created a need for buffers that could translate between different I/O types. Suppliers introduced dedicated devices to solve the problem, as well as programmable devices whose inputs and outputs could be configured to the various I/O standards and frequencies configured on derivatives of the input.