Considering that our extensive blog series on the analog-to-digital conversion (ADC) process was spread out over a relatively long period of time, we felt that a short recap might be in order. Here's links to the ten posts from this series:

The goal of our series on ADCs was two-fold: first to present a broad overview of the ADC process itself, with its basic theory and practical implications, and then to describe how the devices actually doing the analog-to-digital conversion are characterized and how these characterizations can be interpreted. Our intention was to be informative while keeping things relatively light, so that the reader could learn a few key concepts (or maybe refresh ones that had been forgotten) and be able to view them within a meaningful context.

A recap of Part 1: Introduction

Part 1 of this series consisted of a short introduction to the recent and radical transformations that affect almost every aspect of technology and of our daily lives, namely when everything migrated from analog to digital. The conversion of analog information to a digital format requires the understanding of a few mathematical concepts and the use of dedicated hardware. These two different (but very inter-dependent) subjects were the basis of the blog posts that followed.

A recap of Part 2: The conversion process

Part 2 of this series provided a theoretical introduction to the analog-to-digital conversion process itself and described how the conversion is actually performed as two individual and successive stages. The first stage captures samples (i.e. snapshots) of the analog signal, while the second stage assigns digital values to the sampled signal. This second stage, called quantization, is the most demanding one and is the one that ultimately determines how accurately the analog signal is represented by the series of generated digital values.

A recap of Part 3: Signal sampling

Part 3 of this series demonstrated how a signal is sampled, It emphasized how important it is to sample the signal rapidly enough. It was shown, in the most familiar and intuitive way possible, that a signal must be sampled at least twice as fast as its highest frequency component in order to be accurately represented in a digital format. This is one of the fundamental principles necessary for the accurate digital representation of analog signals (known as the Nyquist-Shannon Sampling Theorem) and it must always be taken into account. Sampling a signal at less that half its highest frequency will create incorrect (aliased) representations of the signal's frequency components that will always be folded back at a location below half of the sampling frequency.

A recap of Part 4: Signal bandwidth, parts a/b

The two posts for Part 4 of this series concentrated on the relationship between the representation of a signal as it varies in time (its time-domain representation) and its equivalent form when represented as a sum of individual frequency components (its frequency-domain representation). Using the frequency-domain representation of a signal helps us establish where its highest frequency components are located, so that its actual "bandwidth" can be determined. Different categories of signals were presented (periodic, quasi-periodic, and transient) along with their typical bandwidth characteristics. Knowing these characteristics is of great help when trying to determine the most appropriate signal conditioning and sampling strategy in preparation for the analog-to-digital conversion process.

A recap of Part 5: Signal conditioning

Part 5 of this series explained why analog front-ends are usually required to condition the analog input signal so that it can be optimally converted into its digital equivalent. The post also described the analog building blocks typically found in analog front-ends, like amplifiers, filters (low-pass, high-pass, band-pass and notch), mixers, and drivers. Examples illustrated how these blocks could be assembled into configurations that optimized the conditioning and isolation of the analog signal prior to digital conversion.

A recap of Part 6: ADC performances, parts a/b/c

The three posts for Part 6 of this series explained how the concepts from the previous blogs could help us understand the performance characteristics and limitations of actual ADC devices. The first item covered was how an ADC's quantization resolution, the corresponding noise floor in the frequency domain, and the ADC's resulting signal-to-noise (SNR) figure are all interrelated. Following this, we explained how the non-linearity of an ADC's quantizer is linked to the converter's resulting harmonic distortion. Harmonic distortion is the fundamental component necessary to determine an ADC's most important performance characteristics: total harmonic distortion (THD), spurious-free dynamic range (SFDR), total harmonic distortion plus noise (THD+N), signal to noise and distortion (SINAD), and effective number of bits (ENOB).

A recap of Part 7: Precautions

Part 7 of this series was not related to the ADC process itself, but rather discussed two important considerations when using ADCs. The first consideration was the importance of using differential drivers to protect the integrity of the analog signal if the source was a relatively long distance from the ADC. The second consideration was the importance of using clock signals with very low phase noise to control the sampling of the ADC. The post included some mathematical equations to determine a clock's exact phase noise requirements when having to work with a specific ADC resolution, sampling rate, and signal frequency.


It's my hope that by reading (and sticking with) this lengthy series that you have gained a better understanding and deeper appreciation of the analog-to-digital conversion process. Digital technologies are commonplace, but that only means designers are expected to come up with even better, more efficient, and more accurate ways to process information from the real, analog world. Fine-tuning and optimizing your ADC processes and components means a better end-result. At Nutaq, our goal is to offer products and services that let you work better and faster. Starting out with the required knowledge is the first step to make it happen.