Moore’s Law makes integrated analog sexy

As the shrinking phenomenon predicted by Gordon Moore in 1965 approaches its atomic endpoint, engineers are tasked with finding innovative solutions for creating better and faster designs, independent of merely shrinking transistor size. The consistent pattern of shrinking transistors has led designers to incorporate additional functionality in less space, by effectively increasing the overall number of transistors to enable the creation of faster and more powerful microprocessors, denser memory devices, and more capable system-on-a-chip IC’s (integrated circuits). However, the components cannot maintain their nature semiconductor nature once they’ve reach atomic proportions; thus, semiconductors are turning to integrated analog in an effort to maximize performance before Moore’s Law reaches its limits.

The rise in wireless applications such as smartphones, tablets, and laptops makes analog integration especially useful. Wireless systems such as 4G cellular phone systems, wireless sensor networking systems and broadband wireless networking systems, all require analog ICs in their transceiver chips for signal processing. Thereby, some of the latest techniques in postponing the inevitable transistor halt, or maximizing transistor performance, focus on integrating analog functions along-side digital on a common die; this may include basic comparators to ADCs, DACs, sensors, mixers, analog muxes and more.

Integrating analog into what is typically an all-digital IC provides several benefits: Reducing the amount of uncertainty inherent in a strictly analog design, board space reduction, reduced power dissipation, higher performance, and massive cost savings. The benefits are two fold; they grant digital circuits newfound flexibility while adding a robustness unseen in discrete analog circuits. For example, the integrated components allow for a level of customization that was unprecedented with traditional discrete level analog circuit. The ease-of-use and pre-verified aspects of integration shorten design time by shortening time spent selecting individual discrete components and checking their parameters. Ultimately, it’s a numbers game that culminates in reduced time to market.

Systems that use generic microcontrollers and outsource analog functions to outside devices are very flexible; their parts can be swapped with equivalent outside components if better performing ICs become available. However, the substituted microcontroller must perform identically to the previous MCU as well as be able to communicate and control the slave analog devices.

Contemporary analog integration has proliferated beyond mixed signal ASIC’s and standard cell IC’s, which have existed for many years, to application-specific standard products or ASSP.  But while it’s safe to say that the majority of microcontrollers and microprocessors include some form of analog integration, analog and digital integration is not without its challenges.

The primary issue is one of juxtaposition: digital functions thrive on miniaturization and the higher transistor density/lower power that ensues; however, analog properties worsen when held to the same size standards as the digital IC. If process node reaches 28nm, then the analog properties will not perform as efficiently. As result, designers need to consider the economic implications of when it makes most sense to stay digital or bring in some analog.

A second challenge exists in the form of limited-specialty. It is often difficult to find a partner which has leveraged similar levels of expertise in both analog design, which is considered a specialized art, and digital IC design. Circuit designers must have access to the best-in-class components from companies such Maxim Integrated.

To a large extent, the resurgence of analog technology is crucial for the increasing range of circuit complexity and the value placed on the performance. The multitudes of benefits exist, but they are not without their challenges and tradeoffs.

Via Embedded.com