ShareCG: Power, accuracy and noise aspects in CMOS mixed-signal



Prev TOC Next

1.2. Problem definition

The performance required points out towards higher integration and processing capability at lower and lower power consumption. Digital signal processing has become the major solution in all applications due to versatility, processing capabilities and last but not least lower power consumption. That is why the design effort in the last years has been directed towards low-power in digital. For this reasons, the CMOS process is tuned towards digital performance with negative consequences on analog functionality.

What is then the situation in analog? Although digital dominates in size, analog dominates the interface side and the high speed side. A/D and D/A conversion, filters, amplifiers, voltage references, power-up and power-down converters are possible analog applications. Analog remains the bottleneck in the design trajectory due to the efforts to cope with digital requests. If we are considering the analog part up to few years ago, the evolution has been going on at a slower pace. In the 90ís new technologies are evolved that are merging the high performance needed for analog with the density of CMOS. The design philosophy has to change: you have not to design the best analog circuit per se but the one that fits the best when used together with the DSP section [13]. As often it is told, DSP is the new name of analog: in this situation analog is not an art by itself but has to be in contact with DSP architectural definition. Many technologies are possible from CMOS analog-enhanced to BiCMOS, GaAs etc. CMOS can quite often make the job and guarantee a single chip solution. That is why it should be the first choice whenever possible.

In spite of these concerns related to performance degradation of analog circuits, there has not been a major focus on a design methodology of analog circuits which addresses power. When power is on the discussion floor, all aspects of the circuit, important from specifications point of view, are contributing to it. The approach which is presented here takes another viewpoint, in which possible aspects of a system design are investigated with the goal of reducing power consumption from the analog side.

We state that low power in analog means to fit the design within the specifications with the minimum possible power consumption by using the most efficient architecture.

Claims of low power should be doubled by specifications and the proof that the choice of the architecture is the best. The large variability of analog circuits makes almost impossible a thorough analysis. Only the inventiveness of the analog designer which is, most of the time, experience based or heuristic can generate new solutions. Mixed-signal design should make the best use of different signal representations like: sampled-data, time-discrete or time-continuous to find the best partitioning of the system in terms of performance and power. The ultimate goal is to shift from one domain to another e.g. digital or analog, sampled-data or continuous time and to find the best architecture to fit within the requirements.

© 2017 Internet Business Systems, Inc.
25 North 14th Steet, Suite 710, San Jose, CA 95112
+1 (408) 882-6554 — Contact Us
ShareCG™ is a trademark of Internet Business Systems, Inc.

Report a Bug      Report Abuse      Make a Suggestion      About      Privacy Policy      Contact Us      User Agreement      Advertise