Analog to Digital Converters (ADC): Specs, Selection, and Legacy Sourcing

Analog-to-digital converters sit at the boundary between the physical world and the digital systems that process it. Every time a microcontroller reads a temperature sensor, a medical device captures an ECG waveform, or a radio receiver digitizes an RF signal, an ADC is doing the conversion work. They are not glamorous components, but they are often the most specification-sensitive part of a signal chain — and one of the first to create procurement headaches when a product has a long service life.

This article covers the essentials: how ADC conversion works, what the key specifications actually mean for real designs, how to choose the right architecture for a given application, and how to manage the sourcing challenges that arise when ADC parts reach end-of-life or become hard to find. Whether you are selecting an ADC for a new design or trying to keep an existing system running on a discontinued part, the goal here is practical guidance backed by clear technical context.

What Is an Analog-to-Digital Converter?

An analog-to-digital converter (ADC) is an electronic device that translates a continuously varying analog signal — voltage, current, sound pressure, light intensity, or any other physical quantity represented as a continuous electrical signal — into a discrete digital representation that a processor or digital system can read and manipulate. The real world is analog: temperatures drift, voltages ramp, audio pressure waves oscillate. Digital systems, by contrast, work in binary numbers. The ADC is the bridge between those two domains.

ADCs appear across virtually every segment of electronics. They are embedded inside microcontrollers for basic sensor reading, packaged as standalone precision ICs for industrial instrumentation, integrated into audio codecs, and designed into custom ASICs for high-speed communications. The specific implementation varies widely, but the fundamental function is always the same: take an analog input, produce a digital output.

How ADC Conversion Works: Sampling, Quantization, and Encoding

Every ADC conversion involves three sequential operations: sampling, quantization, and encoding. Understanding each step is important because each one introduces a distinct source of error that shows up in the ADC’s datasheet specifications.

Sampling is the process of capturing the instantaneous value of the analog input at a specific moment in time. A sample-and-hold (S/H) circuit freezes the input voltage so the rest of the conversion can proceed on a stable value. The rate at which samples are taken — the sampling rate — must be at least twice the highest frequency present in the input signal (the Nyquist criterion). Sampling too slowly produces aliasing: high-frequency components fold back into the digital output as spurious low-frequency artifacts that are indistinguishable from real signal content.

Quantization maps the held sample voltage to the nearest available digital level. An ADC with n bits of resolution has 2ⁿ possible output codes, and the voltage span between adjacent codes is one LSB (least significant bit). Any analog value that falls between two codes gets rounded to one of them — this rounding is quantization error, and it is unavoidable. More resolution (more bits) means smaller steps and less rounding error, but also more circuit complexity, more power, and often a lower achievable sampling rate.

Encoding converts the quantized level into a binary output word that the downstream digital logic can use. The output is typically formatted for a standard interface — SPI, I²C, parallel bus, or LVDS — depending on the ADC’s intended application and speed. The complete sequence of sample, quantize, and encode defines the ADC’s conversion time, which feeds directly into its maximum throughput rate.

ADC Architectures and Their Tradeoffs

No single ADC architecture wins across all dimensions. Speed, resolution, power, cost, and latency all trade off against each other, and the dominant architectures each occupy a different region of that tradeoff space. Choosing the right architecture is the first and most consequential ADC selection decision.

SAR (Successive Approximation Register)

SAR ADCs convert by running a binary search: they test the most significant bit first, then refine bit by bit down to the LSB, comparing the input to an internal DAC at each step. The approach is efficient — a single comparator resolves all n bits over n clock cycles — and the result is an architecture that delivers a strong combination of resolution, speed, and power efficiency. SAR ADCs most commonly range from 8 to 16 bits of resolution with sample rates up to several megasamples per second (Msps), and their low power consumption makes them well suited to portable instruments, battery-powered sensors, industrial controls, and data acquisition systems.

One practical note: SAR ADCs present a dynamic input impedance that changes during the acquisition phase, which means the signal source driving them needs adequate drive strength and settling time. For high-impedance sensors, a buffer amplifier at the input is often necessary.

Sigma-Delta (ΔΣ)

Sigma-delta ADCs take the opposite approach to speed: they oversample the input at a rate many times higher than the signal bandwidth, then use noise-shaping techniques to push quantization noise toward higher frequencies where a digital decimation filter removes it. The result is very high effective resolution — 16 to 24 bits is common — at the cost of reduced throughput speed. Sigma-delta converters require no special trimming or calibration to achieve those high resolutions, and because they oversample so heavily, they do not require steep anti-alias filters at the analog input. These converters dominate in audio applications, precision instrumentation, weigh scales, and scientific measurement systems where accuracy matters more than speed.

Pipeline

Pipeline ADCs pass the input through a series of stages, each resolving a few bits before amplifying the residual error and passing it to the next stage. Because the stages operate simultaneously on successive samples, pipeline ADCs achieve high throughput despite the multi-stage architecture. They are typically used in applications requiring 10 to 14 bits of resolution at speeds from around 10 Msps to over 100 Msps — video processing, wireless base stations, radar receivers, and high-speed data acquisition. The tradeoff is pipeline latency: results emerge several clock cycles after the input is sampled, which matters in closed-loop control applications. Pipeline ADCs also consume more power than SAR designs and are sensitive to non-linearities in the inter-stage amplifiers.

Flash (Parallel)

Flash ADCs use a bank of 2ⁿ-1 comparators operating simultaneously, each comparing the input to a different reference level. The output is available in a single clock cycle — no iteration, no pipeline stages. This makes flash converters the fastest available, capable of gigasample-per-second (Gsps) conversion rates used in oscilloscopes, RF digitizers, and optical communications. The cost is silicon area and power: every additional bit of resolution roughly doubles the number of comparators, the die area, and the power consumption. For this reason, flash ADCs are typically limited to 6 to 8 bits of resolution. Higher-resolution designs that need flash-like speeds usually use pipeline or folding architectures that embed flash sub-ADCs internally.

Integrating / Dual-Slope

Integrating ADCs measure voltage by observing how long it takes an integrator circuit to ramp and discharge. The dual-slope variant — widely used in digital multimeters — integrates the input for a fixed period, then discharges using a precision reference and counts the discharge time. Because the measurement averages the input over an integration interval, these converters naturally reject noise and power-line interference. They achieve 12 to 16 bits or more of resolution at very low conversion rates (typically below a few hundred samples per second), with low power and low cost. They are the right choice for slow, high-accuracy measurement tasks — panel meters, precision balances, environmental monitoring — where speed is irrelevant.

Key ADC Specifications Explained

ADC datasheets contain a dense set of parameters. Some describe the fundamental conversion capability; others characterize real-world imperfections. The table below covers the most important specifications and what they mean in practice.

A few of these deserve extra attention. ENOB is arguably more informative than nameplate resolution: a 16-bit ADC with an ENOB of 13.5 bits is delivering 13.5 bits of useful accuracy, not 16, and any design margin should be calculated against the ENOB figure. Similarly, INL and DNL are not interchangeable — DNL describes code-to-code step uniformity (and a DNL worse than −1 LSB means missing codes), while INL describes the cumulative shape of the transfer function. Both matter, but they matter differently depending on whether you are measuring absolute values or tracking relative changes.

ADC Selection: A Practical Decision Framework

Selecting an ADC starts with characterizing the signal, not browsing a catalog. The following questions define the decision space.

What is the signal bandwidth? The input signal’s highest frequency component determines the minimum required sampling rate. Add margin above the Nyquist minimum to simplify the anti-alias filter, and confirm the ADC’s analog input bandwidth (not just its sample rate) covers the signal range.

What accuracy is actually needed? Resist the temptation to over-specify resolution. For general sensor monitoring, 10 to 12 bits is usually adequate. Medical instruments, precision measurement, and scientific equipment typically justify 16 to 24 bits. The ENOB figure should comfortably exceed the accuracy requirement with margin for noise, temperature drift, and reference error.

What does the power budget allow? In battery-powered or thermally constrained designs, power consumption often constrains the architecture choice before resolution or speed do. SAR and sigma-delta converters are inherently more power-efficient than pipeline or flash designs. High-speed pipeline ADCs, while capable, can consume significant power and may require active thermal management.

What is the latency requirement? Pipeline ADCs introduce multi-cycle latency between input and output — acceptable in open-loop data acquisition but potentially problematic in closed-loop control. SAR ADCs offer essentially zero pipeline latency, which is why they appear in motor control, power conversion, and servo systems where response time matters.

What interface does the host processor support? Slow, precision applications often work well with I²C or SPI. Higher-speed designs may require LVDS or JESD204 serial interfaces. The interface choice also affects PCB layout complexity and host processor load.

The table below summarizes where each major architecture fits:

One often-overlooked consideration is whether to use a microcontroller’s built-in ADC peripheral or a discrete external ADC. An MCU’s on-chip ADC is convenient and costs nothing extra, but it typically tops out at 12 bits with moderate noise performance. Go external when the design requires 16 bits or more, more than 2 Msps per channel, multi-channel simultaneous sampling, deterministic latency, or input isolation. The performance gap between an on-chip 12-bit ADC and a quality external 16-bit SAR is substantial for precision applications.

ADC vs. DAC: Core Differences

ADCs and DACs (digital-to-analog converters) serve inverse functions and appear together in most complete signal-chain designs. An ADC digitizes an analog input; a DAC reconstructs an analog output from a digital code. In an audio system, for example, a microphone signal passes through an ADC for processing and storage, while the playback path sends digital audio through a DAC to drive a speaker. The two converters share many specification parameters — resolution, sampling rate, reference voltage, INL, DNL — but their performance characteristics differ in a few meaningful ways.

DACs generally achieve better linearity than ADCs of comparable resolution because the conversion direction (digital to analog) is inherently more controlled: the input is a known digital code, not a noisy analog signal. ADCs, by contrast, must deal with input noise, aperture jitter, and reference uncertainty that all degrade the effective accuracy below the nameplate resolution. ADCs also tend to have more complex internal architectures than DACs, particularly at high speeds, which contributes to higher power consumption in fast designs. When specifying either device, it is worth matching the resolution, speed, and dynamic range of the ADC and DAC in a system — a 24-bit DAC downstream of a 12-bit ADC cannot reconstruct signal content that the ADC never captured.

Common ADC Applications by Industry

ADCs are used across essentially every industry that involves measurement, control, or signal processing. A few sectors illustrate how the architecture and specification requirements vary by application context.

• Industrial automation and process control: SAR ADCs at 12 to 16 bits handle pressure, temperature, flow, and position sensors. Isolation, wide temperature range operation, and long-term calibration stability are key requirements. Many industrial ADC designs have service lives measured in decades, which creates strong legacy sourcing pressures.
• Medical devices: ECG, EEG, ultrasound, and patient monitoring equipment require high-resolution sigma-delta or SAR ADCs with very low noise, good common-mode rejection, and regulatory-grade traceability. Device approval cycles are long, making re-qualification of a replacement part expensive.
• Communications and RF: Pipeline and flash ADCs digitize IF and RF signals in base stations, software-defined radios, and radar systems. Speed and spurious-free dynamic range (SFDR) are the dominant specifications. These parts are often high-value and subject to allocation pressure during supply crunches.
• Consumer electronics: Microphone interfaces, touch controllers, and camera sensors all incorporate ADCs, typically integrated within larger SoCs rather than as discrete components. Volumes are high and product cycles are short.
• Automotive: Engine management, battery management for EVs, lidar, and radar systems each have specific ADC requirements. Automotive-grade parts carry AEC-Q100 qualification and wide-temperature ratings, and they must remain available across platform lifetimes of 10 to 15 years.
• Test and measurement: Bench instruments — oscilloscopes, spectrum analyzers, data loggers — push ADC performance to its limits. A precision bench instrument may use a 24-bit sigma-delta for DC accuracy or a 10-bit Gsps flash ADC for bandwidth, sometimes both in the same instrument.

Legacy ADC Sourcing: Managing EOL and Obsolescence Risk

ADCs used in industrial equipment, medical devices, defense systems, and automotive platforms frequently outlast the component’s production lifecycle. A precision SAR or sigma-delta ADC specified into a design in 2005 may still be running in production equipment in 2025, long after the original manufacturer discontinued the part. The transition from active production to end-of-life (EOL) typically follows a predictable sequence: the manufacturer issues a product discontinuance notice (PDN), a last-time-buy (LTB) window opens (commonly around six months from the PDN), a last-time-ship date follows, and after that the part is no longer available from the original manufacturer through any channel. Once a part passes that last-time-ship date, the options narrow considerably.

The scale of this challenge is significant. Legacy analog front-end components — precision ADCs among them — are among the component categories most commonly affected by long-tail obsolescence in industrial and defense applications. For programs with multi-decade service requirements, a reactive approach to obsolescence (waiting until stock runs out to start looking) is a reliable path to production disruption and expensive redesigns. Proactive lifecycle management — tracking PDN notices, building structured safety stock, and pre-qualifying functional replacements before they are needed — is the better strategy.

Once a part is past last-time-ship, purchasable stock can still exist across a number of supply channels, and locating it becomes a sourcing exercise rather than a catalogue order. What decides whether that stock is safe to buy is how the supplier manages quality, and the questions worth asking are the same whoever is quoting: recognized quality certifications (ISO 9001, and AS9120 for aerospace-adjacent work), a documented counterfeit mitigation process, ERAI membership or reporting history, and clear policies on inspection, date-code disclosure and returns. Inspection should be matched to the risk the material carries — a converter destined for a medical or airborne program warrants more verification than one going into a benchtop fixture. Unusually low prices on genuinely scarce ADC parts are a counterfeit risk indicator, not a procurement win.

The alternative to sourcing legacy stock is a functional replacement or redesign. Many obsolete ADCs have parametric equivalents from current production lines — same resolution, similar INL/DNL, compatible pinout or package, compatible interface. Cross-reference tools from the original manufacturer and third-party parametric search databases can identify candidates. The caution is that “functionally equivalent” on paper may not be electrically identical in a validated design, and industries with qualification requirements (medical, aerospace, automotive) will need to run the replacement through their approval process regardless of how close the datasheet match appears — final approval belongs to the customer’s engineering team, not to the cross-reference tool. Pre-qualifying a drop-in alternate before the original part goes end-of-life avoids the qualification timeline pressure when a shortage actually hits.

WIN SOURCE is a global independent distributor of electronic components, in business since 1999, supporting component supply, shortage sourcing, lifecycle management and alternative solutions — including precision ADCs and other analog ICs from manufacturers such as Texas Instruments, Analog Devices, STMicroelectronics and Microchip. That coverage spans the product life cycle — current-production parts for new builds, bridge stock and cross-references as a design ages, and lifecycle status kept visible on the BOM throughout — which is why early engagement on a lifecycle-risk part beats a call placed after the shortage has already hit.

Conclusion

Analog-to-digital converters are technically diverse components with performance characteristics that vary dramatically by architecture. SAR converters dominate mid-speed, mid-to-high resolution applications. Sigma-delta converters handle audio and precision measurement. Pipeline and flash ADCs cover high-speed communications and RF. Integrating converters serve slow, very precise measurement tasks. Matching the architecture to the application — before looking at specific parts — is the most important selection decision an engineer can make.

Beyond the initial design selection, ADC procurement carries a lifecycle dimension that is easy to underestimate. For long-service-life programs in industrial, medical, automotive, and defense sectors, the component you specify today may need to be sourced for 10 to 20 years. Proactive lifecycle management, pre-qualified functional alternates, and relationships with qualified distributors who can navigate EOL and shortage scenarios are supply-chain assets that pay off over the life of a product, not just at launch.

Frequently Asked Questions

Which ADC architecture should I choose?

Start from bandwidth and resolution, not from part numbers. SAR covers mid-speed, mid-to-high resolution work with low power and near-zero latency, which suits industrial sensing and closed-loop control. Sigma-delta gives the highest resolution at modest speed for audio, medical and precision measurement. Pipeline handles 10–500 Msps for video, radar and wireless. Flash is the fastest and the most power-hungry, at low resolution. Integrating converters are for very slow, very precise measurements such as bench multimeters.

What is ENOB, and why does it matter more than the bit count?

ENOB is the effective number of bits — the resolution you actually get once the converter’s own noise and distortion are accounted for, and it typically lands one to two bits below the nameplate figure. A 16-bit converter delivering 14 bits of ENOB gives you 14 bits of usable accuracy. Design your error budget against ENOB, because the nameplate number is a description of the output word width, not of the accuracy.

How fast does my ADC need to sample?

At least twice the highest frequency present in the signal, which is the Nyquist criterion, and in practice comfortably more than that so your anti-aliasing filter has somewhere to roll off. Note that the requirement is set by the highest frequency actually reaching the converter, not the highest frequency you care about — anything above Nyquist that is not filtered out will fold back into your band and cannot be removed afterwards.

When should I use an external ADC instead of the one in my microcontroller?

An on-chip ADC is usually fine up to about 12 bits at moderate noise levels, and it is free. Go external when you need 16 bits or more, sample rates beyond a couple of megasamples per second, galvanic isolation, better noise or linearity than the MCU’s shared-die environment allows, or a specific interface such as LVDS or JESD204 for high-speed capture.

What happens when an ADC reaches end-of-life?

The manufacturer issues a product discontinuance notice, a last-time-buy window opens — commonly around six months — and a last-time-ship date follows, after which the part is no longer available from the manufacturer. Your options are a last-time buy sized against remaining production, a parametric cross-reference to a current part, or a redesign. A datasheet match is not the same as electrical equivalence in a validated design, and final approval of any replacement belongs to your engineering team.

© 2026 Win Source Electronics. All rights reserved. This content is protected by copyright and may not be reproduced, distributed, transmitted, cached, or otherwise used without the prior written permission of Win Source Electronics.