D.R. Frey

Bio: D.R. Frey is an academic researcher from Lehigh University. The author has contributed to research in topics: Filter (signal processing) & Voltage-controlled filter. The author has an hindex of 1, co-authored 1 publications receiving 711 citations.

Log-domain filtering: an approach to current-mode filtering

Abstract: A novel approach to filter design, based on Adams' [1] ‘log-domain’ filters, is proposed that yields a truly current-mode circuit realisation. Adams' idea, which was introduced in a limited context, is generalised to permit a complete distortionless synthesis procedure, which results in circuit implementations readily realisable using complementary bipolar processes. It is shown that, by introducing an exponential map on the state-space description of the desired linear system, a log-domain filter can be fully realised with transistors configured in current mirror-type groupings, current sources and capacitors. Owing to the mapping, the state variables are intrinsically related to current, and not voltage, in the resulting circuits, a fact that emphasises the current-mode nature of the design. A general biquadratic filter section is designed, and, following discussion of cascading sections, a seventh-order Chebychev lowpass filter is designed. All designed circuits are shown to be tunable over a two-decade range in frequency while their characteristics are accurately preserved, even for biquad sections whose f0Q product is greater than fT/10. The Chebychev filter is shown in simulation to possess nearly 60 dB dynamic range relative to 0.9% THD, with a cutoff frequency of nearly 5 MHz, using transistor models from AT&T's CBIC-R 300 MHz complementary bipolar process.

Neuromorphic Silicon Neuron Circuits

Abstract: Hardware implementations of spiking neurons can be extremely useful for a large variety of applications, ranging from high-speed modeling of large-scale neural systems to real-time behaving systems, to bidirectional brain-machine interfaces. The specific circuit solutions used to implement silicon neurons depend on the application requirements. In this paper we describe the most common building blocks and techniques used to implement these circuits, and present an overview of a wide range of neuromorphic silicon neurons, which implement different computational models, ranging from biophysically realistic and conductance-based Hodgkin-Huxley models to bi-dimensional generalized adaptive integrate and fire models. We compare the different design methodologies used for each silicon neuron design described, and demonstrate their features with experimental results, measured from a wide range of fabricated VLSI chips.

Neurogrid: A Mixed-Analog-Digital Multichip System for Large-Scale Neural Simulations

Abstract: In this paper, we describe the design of Neurogrid, a neuromorphic system for simulating large-scale neural models in real time. Neuromorphic systems realize the function of biological neural systems by emulating their structure. Designers of such systems face three major design choices: 1) whether to emulate the four neural elements-axonal arbor, synapse, dendritic tree, and soma-with dedicated or shared electronic circuits; 2) whether to implement these electronic circuits in an analog or digital manner; and 3) whether to interconnect arrays of these silicon neurons with a mesh or a tree network. The choices we made were: 1) we emulated all neural elements except the soma with shared electronic circuits; this choice maximized the number of synaptic connections; 2) we realized all electronic circuits except those for axonal arbors in an analog manner; this choice maximized energy efficiency; and 3) we interconnected neural arrays in a tree network; this choice maximized throughput. These three choices made it possible to simulate a million neurons with billions of synaptic connections in real time-for the first time-using 16 Neurocores integrated on a board that consumes three watts.

Exponential state space filters: a generic current mode-design strategy

Abstract: A new method of analog filter design is proposed where nonlinear active components are a natural part of a filter realizing an overall linear transfer function. This approach to design is articulated in the generation of the class of so-called "Exponential State Space" (ESS) filters. An outgrowth of "log filters", ESS filters are created via a mapping on the state space of a linear filter. Specifically, the state variables are equal to simple functions of exponentials or node voltages. The use of exponential, hyperbolic tangent, and hyperbolic sine mappings is shown to produce realizable nodal equations that result in "log", "tanh" and "sinh" filters, respectively. Aspects of interpretation and realization of the transformed state equations are discussed. It is shown that sinh filters are class AB filters, which is an intriguing extension of a concept introduced by Seevinck (1990). The different filter types are compared to an analogous standard transconductance-C filter via simulation of a band-pass filter with a Q of 5. All filters demonstrate tunability over a 100 to 1 range with excellent frequency response stability and accuracy over the tuning range, which extends to 5 MHz. These results and IMD and noise performance are given using both ideal transistors and transistors whose cutoff frequency equals approximately 300 MHz.

Externally linear, time-invariant systems and their application to companding signal processors

Abstract: This paper considers externally linear, time-invariant systems which can be internally nonlinear and/or time-varying. Several examples of the synthesis of such systems are presented, and it is argued that they offer advantages in comparison to internally linear, time-invariant systems. In particular, their use in signal processing is considered, and it is shown that they can be designed to reduce the undesirable effects of overloading and noise corruption through the use of companding. It is shown that a variety of previously proposed schemes for this purpose are all related. Several practical problems are discussed, as are performance criteria and types of measurements needed for the evaluation of the systems discussed.