In electronics, an analog-to-digital converter (ADC, A/D, or A-to-D) is a device that changes an analog signal, such as when fingers touch a touchscreen, sound enters a microphone, or light enters a digital camera, into a digital signal.

An ADC can also measure something separately, like a device that changes an analog input voltage or current into a digital number that shows the size of the voltage or current. Usually, the digital output is a two's complement binary number that is proportional to the input, but other options exist.

There are several ADC designs. Because of the complexity and the need for parts that match exactly, most ADCs are made as integrated circuits (ICs), except for very special ones. These are usually metal–oxide–semiconductor (MOS) mixed-signal chips that combine both analog and digital circuits.

A digital-to-analog converter (DAC) does the opposite; it changes a digital signal back into an analog signal.

Explanation

An ADC changes a continuous analog signal into a digital signal. This process includes quantization, which adds a small amount of error. Instead of converting the signal continuously, an ADC samples the input at regular intervals and limits the range of frequencies the signal can have.

The performance of an ADC is measured by its bandwidth, dynamic range, and signal-to-noise and distortion ratio (SNDR). Bandwidth is mainly determined by the sampling rate. SNDR depends on factors like resolution, noise level, linearity, and accuracy. Issues like aliasing and jitter can lower SNDR. SNDR is often described using effective number of bits (ENOB), which shows how many bits of the ADC's output are free from noise. An ideal ADC has an ENOB equal to its resolution. If the sampling rate is more than twice the signal's bandwidth, near-perfect reconstruction is possible. Quantization error limits SNDR even in ideal ADCs. If the ADC's SNDR is higher than the input signal's, the digital copy of the analog signal is nearly perfect.

The resolution of an ADC shows how many different values it can produce. Higher resolution reduces quantization error and improves the maximum signal-to-noise ratio. Input samples are stored as binary numbers, so resolution is usually measured in bits.

Resolution can also be measured in volts. The smallest voltage change that causes a change in the output is called the least significant bit (LSB) voltage. The resolution (Q) equals the LSB voltage. Voltage resolution is calculated by dividing the full-scale voltage range by the number of intervals:

where M is the number of bits and E FSR is the full-scale voltage range. E FSR is determined by:

where V RefHi and V RefLow are the highest and lowest voltages the ADC can measure.

The number of voltage intervals is:

where M is the number of bits. Each interval corresponds to a code level.

The useful resolution of an ADC is often limited by the system's signal-to-noise ratio (SNR) and other errors, expressed as ENOB.

Quantization error happens because the ADC rounds the analog signal to the nearest digital value. This error depends on the signal and is not uniform. In an ideal ADC, if the signal is uniformly distributed and the quantization error is evenly spread between -1/2 LSB and +1/2 LSB, the signal-to-quantization-noise ratio (SQNR) is:

where M is the number of bits. For example, a 16-bit ADC has a quantization error 96.3 dB below the maximum level.

Quantization error spreads from DC to the Nyquist frequency. If part of the ADC's bandwidth is unused, like in oversampling, some error moves out of the used range, improving SQNR. In oversampled systems, noise shaping can push more error out of the used range.

Adding a small amount of random noise (dither) to the input before conversion can improve ADC performance. Dither randomizes the least significant bit, allowing the ADC to capture signals more accurately, especially at low levels. However, dither increases noise slightly and does not improve linearity or accuracy.

Quantization distortion in low-level audio signals can make the sound unpleasant. Dither turns this distortion into noise, which can be averaged out over time. Dither is also used in systems like electricity meters to improve accuracy.

All ADCs have errors like quantization and nonlinearity. These errors are measured in LSBs. For example, an 8-bit ADC has an error of 1 LSB equal to 1/256 of the full signal range, or about 0.4%.

Nonlinearity errors occur because of physical imperfections in ADCs, causing the output to deviate from a linear relationship with the input. These errors can be reduced through calibration or testing. Key linearity parameters include integral nonlinearity and differential nonlinearity, which can lower the ADC's effective resolution.

When converting a sine wave, clock jitter (uncertainty in sampling times) introduces noise. The error caused by jitter depends on the signal's frequency and amplitude. For high-frequency or high-amplitude signals, jitter significantly reduces the effective number of bits (ENOB). Jitter is linked to phase noise and limits ADC performance for high-bandwidth signals. For lower-bandwidth applications, like audio sampling at 44.1 kHz, jitter has less impact.

Analog signals are continuous in time and must be converted to digital values. The rate at which new digital values are sampled is called the sampling rate. A continuous, bandlimited signal can be sampled and reconstructed using a reconstruction filter. The Nyquist-Shannon sampling theorem ensures accurate reconstruction if the sampling rate is at least twice the signal's highest frequency.

Types

There are several common methods for creating an electronic ADC (Analog-to-Digital Converter).

Resistor-capacitor (RC) circuits have a predictable pattern for how voltage increases and decreases over time. This pattern can be used to find the value of an unknown analog signal.

The Wilkinson ADC is a type of linear ramp converter, first described by Denys Wilkinson in 1950. In this design, an input signal is compared to a voltage that increases steadily over time, created by charging a capacitor with a constant current. A comparator generates a signal that stays active until the increasing voltage matches the input. The length of this signal is proportional to the strength of the input. During this time, a clock sends regular pulses, and the number of pulses counted becomes the digital output.

Because the voltage ramp can be made very precisely, the output codes are evenly spaced and consistent, which helps the converter work accurately. The time it takes to complete the conversion depends on the input strength and the clock speed. Stronger signals take longer to measure. Wilkinson converters were widely used in the 1960s and 1970s for analyzing nuclear signals because of their accuracy and ability to create histograms.

If the analog value to measure is a resistance or capacitance, these can be determined by placing them in an RC circuit with fixed components. By measuring how long it takes for a capacitor to charge from one known voltage to another, the value of the unknown resistance or capacitance can be calculated using the formula:

V_capacitor(t) = V_supply × (1 – e^(-t/(R×C)))

This method is similar to the Wilkinson ADC but works in reverse: it measures resistance or capacitance with a known voltage, while the Wilkinson ADC measures voltage with known resistance and capacitance.

For example, the time it takes for a capacitor in a 555 Timer IC to charge from 1/3 of the power supply voltage to 2/3 of the power supply voltage determines the width of the output pulse. If this pulse is sent to a microcontroller with a precise clock, the pulse duration can be measured and used with the capacitor charging formula to find the unknown resistance or capacitance.

Larger resistances or capacitances take longer to measure than smaller ones. The accuracy of this method depends on the microcontroller’s clock precision and the time available to measure, which could change during the process or be affected by outside factors.

A flash ADC, also called a parallel search ADC, uses many voltage comparators that check the input signal at the same time. Each comparator has a different voltage threshold, set by a resistor network. When the input matches a threshold, the comparator sends a signal, and a priority encoder combines these signals into a binary number that represents the input voltage.

Flash ADCs use a lot of space on a chip and consume significant power. They are used in applications like video, radio communications, and other areas that need fast signal conversion.

This type of ADC is fast because all comparisons happen at once, not one after another. The conversion time is usually less than 100 nanoseconds. However, each additional bit in the output doubles the number of comparators needed, making the circuit more complex.

A successive-approximation ADC uses a comparator and a step-by-step process to narrow down the input voltage’s range. It starts by comparing the input to the midpoint of the possible range, using a digital-to-analog converter (DAC). After each step, the result is stored in a register, and the DAC adjusts its output for the next comparison.

A ramp-compare ADC creates a saw-tooth voltage that rises or falls, then quickly returns to zero. When the ramp starts, a timer begins counting. When the ramp voltage matches the input, a comparator stops the timer, and the recorded time becomes the digital output. This method is cost-effective, but the ramp time can be affected by temperature because the ramp is often created by a simple analog circuit. A more accurate version uses a clock to control a DAC. One advantage of this system is that measuring a second signal only requires another comparator and timer. To reduce errors, a sample-and-hold circuit can store the input voltage on a capacitor, and the time to discharge it with a constant current is measured.

An integrating ADC (also called a dual-slope or multi-slope ADC) applies the unknown input voltage to an integrator, which allows the voltage to rise for a fixed time. Then, a known reference voltage of the opposite polarity is applied until the integrator’s output returns to zero. The input voltage is calculated based on the reference voltage, the fixed time, and the time it took to return to zero. The longer the integration time, the higher the resolution. This type of ADC is used in digital voltmeters because it is accurate and flexible.

A delta-encoded or counter-ramp ADC uses an up-down counter connected to a DAC. The input signal and the DAC’s output are compared, and the comparator adjusts the counter until the DAC matches the input. This method can handle a wide range of signals with high accuracy, but the conversion time depends on how quickly the input changes. These converters are good for real-world signals, which usually change slowly. Some designs combine delta and successive-approximation methods, which works well when the input signal has few high-frequency components.

A pipelined ADC (also called a subranging quantizer) uses multiple steps to convert the input. First, a rough estimate is made, then the difference between the estimate and the input is converted more precisely. The results are combined for the final output. This method improves on the successive-approximation ADC by using a feedback signal that covers a range of bits instead of just one. By combining the speed of flash ADCs and the precision of successive-approximation ADCs, this type is fast, accurate, and efficient.

A delta-sigma ADC (also called a sigma-delta ADC) uses a loop with an analog filter and low-resolution components. This design reduces noise and improves accuracy over time, making it suitable for applications requiring high precision.

Commercial

The pins on an integrated circuit are often the most expensive part because they increase the size of the package and each pin must connect to the silicon inside. To reduce the number of pins, ADCs often send data one bit at a time using a serial interface. Each bit is sent when a clock signal changes. This method uses fewer pins on the ADC package and usually doesn't complicate the design.

Many commercial ADCs have multiple inputs that connect to the same converter through an analog multiplexer. Some ADC models include additional features like sample and hold circuits, instrumentation amplifiers, or differential inputs that measure the difference between two signals.

Applications

Soundstream was likely the first digital audio recording system with commercial significance. Analog-to-digital converters (ADCs) are essential in modern music reproduction and digital audio workstation-based sound recording. Music created on computers often begins with analog recordings, requiring ADCs to generate pulse-code modulation (PCM) data used in compact discs and digital music files. Today’s ADCs used in music can sample signals up to 192 kilohertz. Many recording studios record in 24-bit 96 kHz PCM format, then reduce the sample rate and adjust the signal for Compact Disc Digital Audio (44.1 kHz) or for 48 kHz use in radio and television broadcasts.

ADCs are necessary in digital signal processing systems that handle, store, or transmit analog signals in digital form. For example, TV tuner cards use fast video ADCs. Slower 8-, 10-, 12-, or 16-bit ADCs are common in microcontrollers. Digital storage oscilloscopes require very fast ADCs, which are also important for software-defined radio and related technologies.

Digital imaging systems use ADCs to convert pixel data into digital form. Some radar systems use ADCs to transform signal strength into digital values for processing. Many in situ and remote sensing systems also rely on similar technology.

Many sensors in scientific instruments produce analog signals, such as temperature, pressure, pH, and light intensity. These signals can be amplified and sent to an ADC to create digital versions.

Flat-panel displays are digital by nature and require ADCs to process analog signals like composite or VGA inputs.

Testing

Testing an analog-to-digital converter (ADC) needs an analog input source, equipment to send control signals, and tools to record the digital data output. Some ADCs also need a precise reference signal source.

The key parameters to test an ADC are: