publications
2024
- TechRxivA Unified Analysis of Continuous-Time A/D Converters, Dec 2024
As a generalization of the continuous-time pipeline (CTP), we present a theoretical model that can be used to analyze any conventional continuous-time (CT) analog-to-digital converter (ADC). In addition to providing a unified approach for deriving reconstruction filter coefficients, the general model enables new ADC architectures that do not fit in the framework of neither the cascaded continuous-time sigma-delta modulators (CT-Σ∆Ms), nor the CTP ADCs. As an example exploiting this design flexibility, we present a 7th order Leapfrog ADC. The aggressive 7th order filter enables a low oversampling ratio (OSR), while stability is ensured by local digital feedback with single-bit quantizers. The modular structure makes it well-suited for a programmatic design methodology, and the layout of the ADC is generated using a custom-made Python tool for analog layout compilation. Implemented in a 130nm CMOS technology, the prototype achieves a dynamic range of 82 dB with an OSR of 11, while consuming 2.5 mW from a 1.5 V supply. The resulting Schreier figure-of-merit (FOM) is 168 dB.
- arXivThe Continuous-Time RC-Chain ADCHampus Malmberg, and Fredrik Feyling, Oct 2024
An amplifier-less continuous-time analog-to-digital converter consisting of only passives, comparators, and inverters is presented. Beyond simplicity, the architecture displays significant robustness properties with respect to component variations and comparator input offsets. We give an analytical design procedure demonstrating how to parameterize the architecture to a range of signal-to-noise and bandwidth requirements and validate the procedure’s accuracy with behavioral transient simulations.
- TCAS-IA Control-Bounded Quadrature Leapfrog ADCHampus Malmberg, Fredrik Feyling, and José M. de la RosaIEEE Transactions on Circuits and Systems - I: Regular Papers, Feb 2024
In this paper, the design flexibility of the controlbounded analog-to-digital converter principle is demonstrated. A band-pass analog-to-digital converter is considered as an application and case study. We show how a low-pass controlbounded analog-to-digital converter can be translated into a band-pass version where the guaranteed stability, converter bandwidth, and signal-to-noise ratio are preserved while the center frequency for conversion can be positioned freely. The proposed converter is validated with behavioral simulations on several filter orders, center frequencies, and oversampling ratios. Additionally, we consider an op-amp circuit realization where the effects of first-order op-amp non-idealities are shown. Finally, robustness against component variations is demonstrated by Monte Carlo simulations.
2023
- VLSIDesign and Analysis of the Leapfrog Control-Bounded A/D ConverterIEEE Transactions on Very Large Scale Integration (VLSI) Systems, Oct 2023
This article presents analytical tools for high-level design of the leapfrog (LF) control-bounded analog-to-digital converter (CBADC). We derive closed-form design equations for parameterizing the analog system for a target signal-to-noise ratio (SNR) and bandwidth. Furthermore, we show how the parameterization can be modified to compensate for finite amplifier gain-bandwidth product (GBWP) and to control the signal swing at different nodes of the system. Behavioral circuit simulations are used to compare the LF CBADC to relevant continuous-time sigma–delta modulators (CT- Σ∆Ms) in terms of nominal performance and sensitivity to component variations, clock jitter, and finite GBWP. Simulations show that the nominal performance of the LF is similar to that of a CT- Σ∆M of the same loop-filter order and with the same number of quantization levels. The simple, modular structure, analytical stability guarantee, and single-bit quantizers make the LF an interesting alternative to conventional CT- Σ∆Ms.
- MWSCASQuadrature Control-Bounded ADCsHampus Malmberg, Fredrik Feyling, and José M. de la Rosa, Aug 2023
In this paper, the design flexibility of the control-bounded analog-to-digital converter principle is demonstrated by considering band-pass analog-to-digital conversion. We show how a low-pass control-bounded analog-to-digital converter can be translated into a band-pass version where the guaranteed stability, converter bandwidth, and signal-to-noise ratio are preserved while the center frequency for conversion can be positioned freely. The proposed converter is validated with behavioral simulations for a variety of filter orders, notch-filter frequencies, and oversampling ratios. Finally, robustness against component variations is demonstrated by Monte Carlo simulations.
2022
- arXivCalibrating Control-Bounded ADCs, Nov 2022
The paper considers the calibration of control-bounded analog-to-digital converters. It is demonstrated that variations of the analog frontend can be addressed by calibrating the digital estimation filter. In simulations (both behavioral and transistor level) of a leapfrog analog frontend, the proposed calibration method restores essentially the nominal performance. Moreover, with digital-filter calibration in mind, the paper reformulates the design problem of control-bounded converters and thereby clarifies the role of sampling, desired filter shape, and nominal conversion error.
- NorCASHigh-level comparison of control-bounded A/D converters and continuous-time sigma-delta modulatorsIn IEEE Nordic Circuits and Systems Conference, Oct 2022
In this paper, behavioural circuit simulations are used to compare the leapfrog control-bounded analog-to-digital converter to relevant continuous-time sigma-delta modulators in terms of nominal performance and sensitivity to component variations, clock jitter and finite gain-bandwidth product. Sim- ulations show that the nominal performance of the leapfrog is similar to that of a continuous-time sigma-delta modulator of the same loop filter order and with the same number of quantization levels. Component variations in the leapfrog’s analog system will introduce errors in the final output, unless the coefficients of the reconstruction filter are modified accordingly. Nevertheless, the simple, modular structure, analytical stability guarantee and single-bit quantizers make the leapfrog an interesting alternative to conventional continuous-time sigma-delta modulators.
2021
- CSSPControl-Bounded Analog-to-Digital ConversionHampus Malmberg, Georg Wilckens, and Hans-Andrea LoeligerCircuits, Systems, and Signal Processing, Sep 2021
A control-bounded analog-to-digital converter consists of a linear analog system that is subject to digital control, and a digital filter that estimates the analog input signal from the digital control signals. Such converters have many commonalities with delta–sigma converters, but they can use more general analog filters. The paper describes the operating principle, gives a transfer function analysis, and describes the digital filtering. In addition, the paper discusses two examples of such architectures. The first example is a cascade structure reminiscent of, but simpler than, a high-order MASH converter. The second example combines two attractive properties that have so far been considered incompatible. Its nominal conversion noise (assuming ideal components) essentially equals that of the first example. However, its analog filter is a fully connected network to which the input signal is fed in parallel, which potentially makes it more robust against nonidealities.
- ICASSPBinary Control and Digital-to-Analog Conversion Using Composite NUV Priors and Iterative Gaussian Message PassingRaphael Keusch, Hampus Malmberg, and Hans-Andrea LoeligerIn IEEE International Conference on Acoustics, Speech and Signal Processing, Jun 2021
The paper proposes a new method to determine a binary control signal for an analog linear system such that the state, or some output, of the system follows a given target trajectory. The method can also be used for digital-to-analog conversion.The heart of the proposed method is a new binary-enforcing NUV prior (normal with unknown variance). The resulting computations, for each planning period, amount to iterating forward-backward Gaussian message passing recursions (similar to Kalman smoothing), with a complexity (per iteration) that is linear in the planning horizon. In consequence, the proposed method is not limited to a short planning horizon.
2020
- ISCASAnalog-to-Digital Conversion using Self-Averaging Analog Hadamard NetworksHampus Malmberg, and Hans-Andrea LoeligerIn IEEE International Symposium on Circuits and Systems, Oct 2020
Control-bounded analog-to-digital conversion as described in the work of Loeliger et al. opens opportunities for entirely new analog circuit topologies. The structure of such a converter is shown in Fig. 1. In this paper, we propose such a converter where the analog linear system is a network of N fully connected identical integrators, with uniform sensitivity to noise and mismatch across the network. Nonetheless, the converter achieves a nominal conversion error similar to that of a ΔΣ converter with a N-th order loop filter.
- arXivControl-Bounded Analog-to-Digital Conversion: Transfer Function Analysis, Proof of Concept, and Digital Filter ImplementationHans-Andrea Loeliger, Hampus Malmberg, and Georg Wilckens, Oct 2020
Control-bounded analog-to-digital conversion has many commonalities with delta-sigma conversion, but it can profitably use more general analog filters. The paper describes the operating principle, gives a transfer function analysis, presents a proof-of-concept implementation, and describes the digital filtering in detail.
- Control-Bounded ConvertersHampus MalmbergPhD Dissertation, ETH Zürich, Aug 2020
The need for analog-to-digital (A/D) and digital-to-analog (D/A) conver- sion is a ubiquitous part of many of today’s practical applications. The research fields of A/D and D/A conversion are multi-disciplinary, involv- ing topics such as discrete- and continuous-time signal processing, circuit theory, and circuit design. State-of-the-art achievements have refined the practical aspects of traditional converter architectures to a point where performance is reaching its physical limits and progress is stagnating. In this thesis, we present an alternative perspective of analog-to-digital and digital-to-analog conversion called control-bounded conversion. This new perspective utilizes standard circuit components to build up un- conventional circuit architectures through a novel theoretical framework between analog and digital. Ultimately, this versatile design principle allows less constrained analog and digital circuit architectures at the expense of a digital post-processing step. We demonstrate the control-bounded conversion principle by a selection of converter examples. First we consider the chain-of-integrators and the leapfrog analog-to-digital converters, which emphasize the division of the analog and digital parts of a control-bounded analog-to-digital converter. In particular, these examples reveal the global nature of the analog design task compared to the local digital part, which can be decomposed into independently operated, sub-circuits. Next, the chain-of-oscillators analog-to-digital converter shows how the control-bounded converter can be adapted for the problem of converting non-baseband signals as is common in communication systems. Specifi- cally, the modulation task (frequency shifting) is incorporated into the digital part of the circuit, removing the need for a pre-processing step. To suppress the influence of circuit imperfections, we introduce the Hadamard analog-to-digital converter that separates the physical and the logical signal dimensions of a control-bounded converter. This separation enables circuit architectures where the sensitivity to component mismatch and thermal noise can be distributed equally throughout the circuit architecture components, thereby minimizing its impact on conversion performance. The overcomplete digital control shows how the digital part’s complexity can be increased, resulting in better conversion performance, without substantially increasing the sensitivity to circuit imperfections. This idea relates to using higher-order quantization but partitions the analog part of the circuit in a novel way. We demonstrate that the control-bounded analog-to-digital conversion concept can provide improved conversion performance when converting multiple signals jointly as opposed to independent conversion. Finally, we show how the control-bounded conversion principle can be adopted for digital-to-analog conversion.
2018
- ISTCFactor Graphs with NUV Priors and Iteratively Reweighted Descent for Sparse Least Squares and MoreHans-Andrea Loeliger, Boxiao Ma, Hampus Malmberg, and Federico WadehnIn IEEE 10th International Symposium on Turbo Codes Iterative Information Processing, Dec 2018
Normal priors with unknown variance (NUV) are well known to include a large class of sparsity promoting priors and to blend well with Gaussian message passing. Essentially equivalently, sparsifying norms (including the L1 norm) as well as the Huber cost function from robust statistics have variational representations that lead to algorithms based on iteratively reweighted L2-regularization. In this paper, we rephrase these well-known facts in terms of factor graphs. In particular, we propose a smoothed-NUV representation of the Huber function and of a related nonconvex cost function, and we illustrate their use for sparse least-squares with outliers and in a natural (piecewise smooth) prior for imaging. We also point out pertinent iterative algorithms including variations of gradient descent and coordinate descent.
- TBioCASEstimation of the Cardiac Field in the Esophagus Using a Multipolar Esophageal CatheterReto Andreas Wildhaber, Dominik Bruegger, Nour Zalmai, Hampus Malmberg, Josef Goette, Marcel Jacomet, Hildegard Tanner, Andreas Haeberlin, and Hans-Andrea LoeligerIEEE Transactions on Biomedical Circuits and Systems, Aug 2018
The rapid progress of invasive therapeutic options for cardiac arrhythmias increases the need for accurate diagnostics. The surface electrocardiogram (ECG) is still the standard of noninvasive diagnostics but lacks atrial signal resolution. By contrast, esophageal electrocardiography (EECG) yields atrial signals of high amplitude and with a high signal-to-noise ratio. Esophageal electrocardiography has become fast and safe, but the mechanical constraints of esophageal measuring catheters and the “random” motion of the catheter inside the subject’s esophagus limit the spatial resolution of EECG signals. In this paper, we propose a method to estimate the electrical field projected onto the esophagus with an increased spatial resolution, using commonly available esophageal catheters. In a first step, we estimate the time-varying catheter position, and in a second step, we estimate the projected electrical field with enhanced spatial resolution. The proposed algorithm comprises several consecutive optimization steps, where each intermediate step produces not just a single point estimate, but a cost function over multiple solutions, which reduces the information loss at each processing step. We conclude with examples from a clinical trial, where the fields of cardiac arrhythmias are presented as two-dimensional contour plots.
2017
- EUSIPCOUnsupervised feature extraction, signal labeling, and blind signal separation in a state space worldNour Zalmai, Raphael Keusch, Hampus Malmberg, and Hans-Andrea LoeligerIn 25th European Signal Processing Conference, Aug 2017
The paper addresses the problem of joint signal separation and estimation in a single-channel discrete-time signal composed of a wandering baseline and overlapping repetitions of unknown (or known) signal shapes. All signals are represented by a linear state space model (LSSM). The baseline model is driven by white Gaussian noise, but the other signal models are triggered by sparse inputs. Sparsity is achieved by normal priors with unknown variance (NUV) from sparse Bayesian learning. All signals and system parameters are jointly estimated with an efficient expectation maximization (EM) algorithm based on Gaussian message passing, which works both for known and unknown signal shapes. The proposed method outputs a sparse multi-channel representation of the given signal, which can be interpreted as a signal labeling.
2016
- ICASSPBlind deconvolution of sparse but filtered pulses with linear state space modelsNour Zalmai, Hampus Malmberg, and Hans-Andrea LoeligerIn IEEE International Conference on Acoustics, Speech and Signal Processing, Mar 2016
The paper considers the problem of joint system identification and input signal estimation of an unknown linear system from noisy observations of the output signal. The input signal is assumed to be sparse, and each individual input pulse may affect the system in its own (and unknown) way. Based on ideas from sparse Bayesian learning, we derive an efficient expectation maximization (EM) algorithm for jointly estimating all unknown quantities. Unlike related prior work, the proposed algorithm does not alternate between estimating the input signal and estimating the system parameters; instead, all unknown quantities are jointly updated in each EM step. We give closed-form expressions for these EM updates, which can be efficiently computed by Gaussian message passing.
- ITAOn sparsity by NUV-EM, Gaussian message passing, and Kalman smoothingHans-Andrea Loeliger, Lukas Bruderer, Hampus Malmberg, Federico Wadehn, and Nour ZalmaiIn Information Theory and Applications Workshop, Jan 2016
Normal priors with unknown variance (NUV) have long been known to promote sparsity and to blend well with parameter learning by expectation maximization (EM). In this paper, we advocate this approach for linear state space models for applications such as the estimation of impulsive signals, the detection of localized events, smoothing with occasional jumps in the state space, and the detection and removal of outliers. The actual computations boil down to multivariate-Gaussian message passing algorithms that are closely related to Kalman smoothing. We give improved tables of Gaussian-message computations from which such algorithms are easily synthesized, and we point out two preferred such algorithms.
2015
- ISITDeconvolution of weakly-sparse signals and dynamical-system identification by Gaussian message passingLukas Bruderer, Hampus Malmberg, and Hans-Andrea LoeligerIn IEEE International Symposium on Information Theory, Jun 2015
We use ideas from sparse Bayesian learning for estimating the (weakly) sparse input signal of a linear state space model. Variational representations of the sparsifying prior lead to algorithms that essentially amount to Gaussian message passing. The approach is extended to the case where the state space model is not known and must be estimated. Experimental results with a real-world application substantiate the applicability of the proposed method.