What Are Field Programmable Analog Arrays?

Field programmable analog arrays (FPAAs) are integrated circuit devices containing configurable analog processing blocks and programmable interconnects that allow users to implement analog circuits without fabricating a custom chip. Analogous in concept to the field programmable gate array (FPGA) in digital electronics, an FPAA provides a reconfigurable fabric whose functional behavior is determined by configuration data stored in on-chip memory rather than by the physical layout of a fixed design. The term was first used in 1991 by Lee and Gulak, who proposed the arrangement of computational analog blocks (CABs) connected through a routing network that could be configured digitally.

FPAAs address a persistent challenge in analog design: the cost and delay of producing a dedicated analog ASIC makes it impractical to prototype or iterate on analog circuits through silicon fabrication alone. An FPAA lets designers implement filters, amplifiers, oscillators, and analog-to-digital converter front ends on a single reconfigurable device, revising the configuration in software rather than ordering new masks. Commercial FPAAs have been produced by companies including Anadigm, whose products use switched-capacitor architectures, and by research groups at Georgia Tech and other institutions developing large-scale floating-gate FPAA platforms.

Architecture and Programmability

An FPAA consists of an array of configurable analog blocks, each built around one or more operational amplifiers with surrounding programmable passive networks. In switched-capacitor FPAAs, capacitor arrays replace resistors and are sampled at a system clock rate; the ratio of capacitor values determines the effective transfer function of the block. In continuous-time FPAAs, transconductance amplifiers (Gm cells) whose bias currents are digitally adjustable implement filters and signal processing functions without a sampling clock, making them suitable for radio-frequency front-end applications. The routing fabric connecting the blocks uses analog switches, typically implemented as transmission gates, to form the desired signal paths. Configuration data is loaded into non-volatile or volatile memory, enabling the same device to serve as a lowpass filter in one application and a programmable gain amplifier in another. A detailed description of continuous-time FPAA architectures is available in IEEE conference publications on reconfigurable Gm-cell arrays.

Design Flow and Tools

Designing an FPAA circuit typically begins with a schematic capture or block-diagram tool provided by the device vendor, in which the designer specifies the functional blocks, their parameters, and the interconnect topology. The tool maps this description onto the available CABs and generates a configuration bitstream that is loaded into the device. Unlike FPGA design flows, which rely on hardware description languages such as VHDL or Verilog, most FPAA tools use graphical block-level interfaces because analog circuit design relies heavily on visual topology and component value intuition. After configuration, in-circuit tuning is possible by adjusting bias currents or capacitor ratios through reprogramming, which supports iterative optimization without hardware changes. Large-scale floating-gate FPAAs developed at Georgia Tech use charge stored on floating gates to set transconductor biases, enabling very fine analog tuning, and their design methodology is described in Georgia Tech research publications on large-scale FPAA systems.

Comparison with ASICs and FPGAs

FPAAs occupy a middle position between fully custom analog ASICs and digital FPGAs. An analog ASIC provides the best possible performance for a specific function, achieving the lowest noise, highest bandwidth, and smallest die area, but requires months of design and fabrication at high mask cost. An FPGA handles digital signal processing tasks and can implement many analog functions through oversampling and digital filtering, but at higher power and with limited precision for strictly analog signals below the conversion boundary. The FPAA trades some performance relative to a custom ASIC for the ability to configure and reconfigure in the field, making it particularly useful in rapid prototyping, sensor interface development, and low-volume production where ASIC economics do not apply. The commercial landscape for FPAAs, including Anadigm product families, is described on the Anadigm FPAA product pages.

Applications

Field programmable analog arrays have applications in a wide range of disciplines, including:

• Sensor signal conditioning and analog front-end prototyping
• Audio signal processing for hearing aids and consumer audio
• Software-defined radio analog preprocessing stages
• Robotics and control system analog feedback networks
• Low-power IoT edge computing with reconfigurable analog inference