Time-Encoding Analog-to-Digital Converters: Bridging the Analog Gap to Advanced Digital CMOS-Part 1: Basic Principles

Abstract

I. MOTIVATION FOR TIME ENCODINGThe scaling of CMOS technology deep into the nanometer range has created challenges for the design of high-performance analog integrated circuits. The shrinking supply voltage and the presence of mismatch and noise restrain the dynamic range, causing analog circuits to be large in area and have a high power consumption in spite of the process scaling. Analog circuits based on time encoding [1], [2] and hybrid analog/digital signal processing [3] have been developed to overcome these issues. Realizing analog circuit functionality with highly digital circuits results in more scalable design solutions that can achieve excellent performance. This article reviews the basic principles of time encoding, in particular applied to analog-to-digital converters (ADCs) based on Voltage-Controlled Oscillators (VCOs), one of the most successful time-encoding techniques to date. Although VCO-based ADCs have been around for a long time [4], [5] they really received a significant boost in interest after Straayer and Perrotts highly cited 2008 paper [6]. Since then, many other advancements from different research groups worldwide have been published, in application domains such as sensor readout [7]-[10], telecom [11]-[13], wideband wireless [14]-[18], Internet of Things [19], automotive [20], [21], biomedical [22]-[24], to name a few. Note that the integrating properties of VCOs have also been used to implement continuous-time filters, time to digital converters and other analog signal processing blocks [25]-[27].This overview article is divided in two parts. This first part will introduce the basic principles of time encoding with emphasis on VCO-based ADCs, and will compare them to traditional analog circuits. The follow-up article (part 2) will describe and compare different time-encoding circuit architectures for analog-to-digital converters, and will discuss the corresponding design challenges of the building blocks. Together, they will demonstrate how time-encoding circuits can overcome many of the problems encountered when designing high-performance analog circuits that fully take advantage of advanced digital CMOS technologies, rather than suffering from it, hence bridging the analog gap to advanced CMOS. The main idea behind time encoding consists of representing an analog signal with a modulated square wave, where the signal information is encoded in the transitions instead of the instantaneous amplitude. Such a signal is easy to handle with digital circuits and is robust against noise and distortion. Representing the signal is then done through pulse modulations, known since long ago. Pulse Width Modulation (PWM) is one of the best examples, and used extensively in, for instance, power electronics. The typical block diagram of a time-encoding ADC is shown in Fig. 1. The analog input signal is applied to a pulse modulator that encodes the information in the pulse width, frequency or position of a signal (or set of signals) with two levels only. This two-level signal, which is still analog, ...

Time-Encoding Analog-to-Digital Converters: Bridging the Analog Gap to Advanced Digital CMOS?Part 2: Architectures and Circuits

“…1 (whether bottom or top driven) is obtained, which is catastrophically nonlinear. A standard technique to improve this is the use of a pseudo-differential set-up (see discussion in Part 1 [3]). However, even then the large-signal linearity is still far from acceptable for most applications.…”

“…The phase detector shown in the figure uses only 1 VCO output phase, but it is also possible to use multiple VCO output phases for improved performance [3]. The underlying mechanism of this architecture can readily be understood by reasoning on the established phase-domain model of a VCO as an integrator (see Part 1 [3]). In this way the depicted architecture of Fig.…”

“…Circuit noise is an important limitation in any analog circuit, including VCO-based ADCs. Altough many designers think about VCO noise in terms of oscillation phase noise, we advocate to reason about this in terms of input-referred circuit noise (see also Part 1 [3]), similar to amplifier noise. The white noise translates in a pretty conventional noise versus power trade-off.…”

“…Analog circuits based on time encoding [1], [2], where the signal information is encoded in the waveform transitions instead of its amplitude, have been developed to overcome these issues. While Part 1 of this overview article [3] has presented the basic principles of time encoding, this Part 2 follow-up article will describe and compare the main time-encoding architectures for analog-todigital converters and will discuss the corresponding design challenges of the circuit blocks. Focus will be on structures that avoid as much as possible the use of traditional analog blocks like opamps or comparators, but instead use digital circuitry, ring oscillators, flipflops, counters, etc.…”

“…Additionally, similar to conventional continuous-time Delta-Sigma modulators, VCO-based ADCs exhibit an intrinsic anti-aliasing filtering property. Combined with their very high sensitivity [3], [4], this allows to eliminate other traditional analog blocks such as pre-amplifiers and anti-aliasing filters. We will conclude with presenting the current state of the art in VCO-based analog-to-digital converters as published in literature.…”