Many circuit designs have been tried in the development of analog-to-digital converters (ADC). The field has been narrowed to four basic topologies: flash, successive-approximation (SAR), pipeline, and sigma-delta converters. Each has its strong and weak points.

The optimum device for a particular system depends on the application. More specifically, the optimum device depends on what needs to be done with the data. Applications requiring waveform reconstruction have different requirements from those applications performing calculations or decisions based on a snapshot reading.

The speed-resolution comparison between the four most popular converter types is shown in Figure 1. As new circuit techniques are developed, the performance boundaries between them have become somewhat blurred.

Figure 1: ADC type comparison
(Click on image to enlarge)

Beyond the speed-resolution distinction, the next performance point is time-of-reading. The flash converter is a very high-speed device, and time synchronization usually is not an issue. In contrast, the SAR converter uses a start-convert signal. With this capability, two or more devices can be synchronized to an external event. The data appears several clock cycles following the convert command.

Both the pipeline and sigma-delta topology converters are continuous-conversion devices. This makes it nearly impossible to synchronize multiple devices for simultaneous reading, or to obtain a snapshot reading at a defined moment. There will be a time lapse between an event at the analog input and the appearance of that event in the digital data stream. That time delay is termed data latency.

Data latency should not to be confused with the conversion time associated with the SAR converter, where there are no other conversion operations in progress. Sigma-delta converters can be compared to performing a running average. The pipeline converter, however, can be compared to an assembly line. After a station contributes to the final result, the signal moves on, making way for that station to process the next sample.

One result of this difference is that sigma-delta converters effectively transpose noise energy to frequency bands above the frequencies of interest. Pipeline-style converters accomplish high data rates with moderate-to-high resolution.

Another significant change brought about by the new techniques is the disappearance of the stand-alone sample-and-hold (S/H) amplifier. To obtain a valid conversion results, the analog input must be stable to within less than one least significant bit (LSB) during the conversion time. Several advancements in converter performance including increased resolution, shorter conversion times, and reduced full-scale voltage, all point to the need for higher-performance S/H amplifiers. Advanced circuit topology has allowed the S/H amplifier to be built into the converter.

Each of these converter types will be investigated in detail in future articles.

About the author

William P. (Bill) Klein is a Senior Applications Engineer with the High Performance Analog group at Texas Instruments. Bill joined TI through its acquisition of Burr-Brown in August 2000. His experience as an analog circuit designer covers over 40 years in fields ranging from mineral exploration to medical nuclear imaging. One current role Bill has is hosting the Analog e-LAB Web Cast, presenting real world solutions to real world problems in analog circuit design. In addition to a BSEE from Arizona State University and registration as a Professional Engineer in the State of Arizona, he has authored numerous magazine articles, application notes and conference papers.

Previous installments of this series:

• "SIGNAL CHAIN BASICS Series (Part 3): Analog and the digital world", www.planetanalog.com/features/showArticle.jhtml;?articleID=204400376, click here
• "SIGNAL CHAIN BASICS Series (Part 2): Op Amp--Basic operations", www.planetanalog.com/features/showArticle.jhtml;?articleID=203101699, click here
• "SIGNAL CHAIN BASICS: Operational Amplifier--The Basic Building Block", www.planetanalog.com/features/showArticle.jhtml;?articleID=202801320, click here