What Does “Analog FPGA” Mean?

The phrase analog FPGA is an intuitive way to describe programmable analog hardware. It is useful because many engineers already understand what an FPGA does for digital systems: the chip is manufactured first, while much of its application-specific behavior is defined later through configuration.

An FPAA extends a similar idea into the analog domain. Instead of fixing every amplifier, filter, comparator, integrator, and signal path when a PCB or custom IC is designed, an FPAA provides analog resources whose functions and connections can be configured after manufacturing.

This is not simply marketing terminology. A 1991 IEEE paper by E. K. F. Lee and P. Glenn Gulak described a CMOS field-programmable analog array built from configurable analog blocks connected through a programmable interconnection network.^[1] Later FPAA research expanded the concept toward larger analog and mixed-signal programmable systems.^[2]

That is why “analog FPGA” is a useful analogy, while FPAA is the more precise technical term.

FPGA vs. FPAA: Similar Idea, Different Signal Domain

An FPGA and an FPAA share an important principle: both allow application-specific hardware behavior to be configured after the silicon has been manufactured. What changes is the type of hardware being configured.

For a more detailed engineering discussion of the differences, tradeoffs, power considerations, and complementary roles of the two technologies, see FPAA vs FPGA: Power-Efficient Analog Flexibility for Embedded Systems .

What Changes When the Hardware Is Analog?

The most important distinction is not simply what blocks appear on a chip. It is where signal processing occurs.

A conventional embedded signal chain often conditions a physical signal with fixed analog circuitry, converts that signal into digital data with an ADC, and then performs more complex processing in an MCU, DSP, or FPGA.

An FPAA can make a substantial part of the analog front end programmable:

Filtering, amplification, comparison, integration, threshold detection, and other supported functions can therefore operate on the electrical signal itself before it is represented as digital samples.

Does an FPAA Replace the ADC?

Usually not. An ADC and an FPAA have different jobs. An ADC converts an analog signal into digital data. An FPAA performs configurable operations in the analog domain.

The two can work together. For example, an FPAA can provide programmable gain and filtering before an ADC, allowing the downstream converter and processor to receive a signal that has already been appropriately conditioned.

For a deeper discussion of this application, see FPAAs for Sensor Interfaces and ADC Signal Conditioning .

What Is Actually Programmable in an FPAA?

FPAA architectures vary, so there is no single block diagram that represents every device. That distinction matters: FPAA describes a class of programmable analog hardware, not one universal circuit architecture.

Most FPAA concepts nevertheless rely on two fundamental capabilities.

Research literature has explicitly compared this design philosophy with FPGA architecture. Hall, Hasler, and Anderson described an FPAA architecture analogous to FPGA architectures for configurable analog signal-processing systems.^[4]

A Commercial Example: The AN231E04

The Okika AN231E04 provides a practical example of a commercial FPAA.

The device contains four Configurable Analog Blocks arranged in a 2 × 2 matrix, programmable routing, configurable analog I/O resources, and a lookup table. Its resources can be configured to implement functions including filtering, amplification, summing, comparison, waveform generation, and nonlinear operations.

The point is not that every FPAA looks like the AN231E04. Rather, it makes the “analog FPGA” concept tangible: a manufactured IC contains analog hardware whose functional behavior and interconnections can be configured for a particular application.

Where the “Analog FPGA” Analogy Breaks Down

The analogy is useful for understanding programmability, but an FPAA should not be treated as an FPGA with analog replacements for digital logic gates.

An FPAA implements a real analog circuit. The resulting design is therefore governed by analog behavior, including:

• Signal amplitude and headroom
• Bandwidth
• Noise
• Offset
• Distortion
• Dynamic range
• Loading
• Routing parasitics
• Bias conditions
• Architecture-specific sampling behavior

Configuration software can simplify circuit implementation, but it cannot make those physical characteristics disappear.

Engineers moving from digital programmable logic into FPAAs should therefore expect a mixture of familiar concepts, including configurable blocks, routing, configuration, and reusable functions, alongside traditional analog engineering constraints.

Okika's Common First-Time FPAA Design Mistakes and How to Avoid Them explores this transition in more practical detail.

Is an FPGA Connected to an ADC an Analog FPGA?

No. A conventional FPGA can certainly be part of a mixed-signal system, and FPGA platforms can include or connect to ADCs, DACs, comparators, PLLs, and other analog or mixed-signal components.

The distinction is what happens inside the programmable fabric.

Neither approach is universally better. They solve different problems and are often strongest when used together.

FPAA + FPGA + MCU: Programmable Hardware Across Domains

Thinking of FPAAs and FPGAs only as competitors misses an important architectural opportunity. A system can assign work according to the domain where it makes the most sense.

The correct partition depends on bandwidth, latency, accuracy, power, data rate, available resources, and the requirements of the individual system.

For practical implementation considerations, see Integrating FPAA Designs with MCU and FPGA Systems .

Why Programmable Analog Matters at the Sensor Boundary

As embedded systems become more data-intensive, an important architectural question is not only how a signal should be processed, but where that processing should begin.

A digital-first architecture may sample a broad stream of sensor data and then perform filtering, detection, or feature extraction downstream. A programmable analog architecture creates the option to perform selected signal-facing operations before all of that information becomes digital data.

Depending on the application, this can be useful for filtering unwanted signal content, adjusting gain, detecting thresholds, extracting selected features, or otherwise preparing information closer to the source.

This does not eliminate the need for digital computing. Instead, it creates another point at which system architects can decide how work should be divided between analog and digital processing.

From Analog FPGA to System-on-Chip FPAA

The “analog FPGA” description becomes less complete as FPAA architectures grow into larger mixed-signal systems.

Large-scale FPAA research has explored architectures that combine programmable analog computation with configurable digital resources, embedded processors, memory, converters, routing, and software infrastructure.^[2]

A published FPAA system-on-chip has, for example, integrated programmable analog and digital resources with an embedded 16-bit MSP430 processor and mixed-signal interfaces.^[3]

At that scale, the architecture is better understood as a reconfigurable mixed-signal computing platform rather than simply an analog version of an FPGA.

Because Okika already maintains a dedicated technical resource for this subject, the detailed architecture, programming workflow, research history, design resources, and application material are not repeated here.

When Does an “Analog FPGA” Make Sense?

FPAAs are not intended to replace every fixed analog circuit. A simple, stable, high-volume function may still be best served by dedicated analog components or a purpose-built IC.

Programmable analog becomes particularly valuable when the ability to change the analog hardware has system-level value.

• Signal-conditioning requirements may change
• One platform must support multiple sensors
• Filter or gain settings vary between operating modes
• Analog prototypes require frequent iteration
• Products need field-configurable analog behavior
• Hardware must support multiple product variants
• Selected preprocessing belongs before digitization
• Analog and digital programmable hardware must work together

Potential application areas include sensor interfaces, instrumentation, robotics, industrial systems, control, audio, biomedical electronics, aerospace and defense systems, research, education, adaptive signal processing, and edge sensing.

The more useful engineering question is often not “Can an FPAA replace this analog circuit?” but rather:

One Idea, Different Reasons to Care

Explore FPAA Technology

References