Digital vs Analog Signal

Definition: A digital signal represents information as discrete steps, typically binary high/low levels, while an analog signal represents information as a continuously varying quantity.

How It Works

Digital signals are read as one of a fixed set of states, usually two: a “high” voltage band representing 1 and a “low” band representing 0
Because only the band matters, not the exact voltage, small noise or drift gets rounded back to the correct state on read-out
A receiving circuit uses voltage thresholds to decide whether an incoming level counts as high or low, ignoring everything in between as invalid or transitional
Analog signals can take any value within a continuous range, capturing phenomena like sound pressure, temperature, or light intensity directly and proportionally
Analog circuits must preserve the exact shape and amplitude of a waveform, so any noise picked up along the way becomes part of the signal permanently
Digital signals are typically clocked, meaning values are only meaningful at specific sampled instants, while analog signals are continuous in time as well as amplitude
Multi-level (non-binary) digital signaling exists too, such as PAM4 in high-speed data links, trading noise margin for more bits per symbol
Digital logic families define specific voltage thresholds, for example TTL treats below 0.8V as low and above 2.0V as high, with the gap between called the forbidden or transition region
Amplitude quantization and time sampling are two separate steps: sampling fixes when a value is read, quantization fixes how precisely that value is represented
Analog signal chains accumulate noise additively at every stage, amplifier, cable, connector, while digital chains can regenerate a clean signal at every stage with a comparator or logic gate
Oversampling, sampling well above the Nyquist minimum and averaging, is a common trick to trade sample rate for extra effective bit depth without a more expensive ADC
Digital-to-analog conversion at the output stage also needs a reconstruction filter to smooth the stepped output back into a continuous waveform

Illustration

Under the Hood

Nyquist sampling theorem, the minimum sample rate to capture a signal without aliasing:

fsample ≥ 2 × fmax

Quantization step size for an N-bit ADC over a voltage range:

Δ = Vrange / 2^N

Signal-to-noise ratio improvement per additional bit (approximate, for a sine wave input):

SNR(dB) ≈ 6.02 × N + 1.76

Worked Problem 1: Minimum sample rate Given: An analog audio signal contains frequencies up to 20 kHz (the edge of human hearing). Step 1: fsample ≥ 2 × 20,000 Step 2: fsample ≥ 40,000 Hz Answer: A sample rate of at least 40 kHz is required; CD audio uses 44.1 kHz to leave margin for anti-aliasing filters.

Worked Problem 2: Quantization step Given: A 10-bit ADC (2^10 = 1024 levels) reads a 0-5V range. Step 1: Δ = 5V / 1024 Step 2: Δ ≈ 0.00488V (4.88 mV per step) Answer: The smallest voltage change the ADC can distinguish is about 4.9 mV.

Worked Problem 3: SNR from bit depth Given: A 16-bit audio ADC, common in CD-quality audio. Step 1: SNR ≈ 6.02 × 16 + 1.76 Step 2: SNR ≈ 96.3 + 1.76 = 98.1 dB Answer: The theoretical dynamic range is about 98 dB, meaning the quietest resolvable signal is roughly 98 dB below full scale.

Worked Problem 4: Bits needed for a resolution target Given: A sensor needs to resolve 1 mV steps over a 0-3.3V range. Step 1: Number of steps required = 3.3V / 0.001V = 3300 steps Step 2: 2^N ≥ 3300, so N ≥ log2(3300) ≈ 11.7, round up to N = 12 Answer: A 12-bit ADC (4096 levels, 0.8 mV resolution) is the smallest standard bit depth that meets the requirement.

Why It Matters

Real-world signals are analog, so converting between the two is central to how computers sense and produce the physical world
Digital signals tolerate noise, attenuation, and long transmission distances far better than analog, since a receiver only needs to distinguish high from low, not preserve an exact value
Digital data can be stored, copied, compressed, and error-corrected without any generational loss, unlike analog recordings that degrade with each copy
Analog circuits remain essential wherever a physical quantity must be sensed, amplified, or actuated at the boundary with the real world, microphones, sensors, motor drivers, and antennas are inherently analog
Digital error correction, checksums, parity bits, forward error correction, has no real analog equivalent, letting digital systems recover from corruption that would permanently distort an analog signal
Mixed-signal design, combining analog front ends with digital processing on the same chip, is now the dominant approach in nearly every modern sensor and communication device
Digital communication links can be regenerated at each repeater or switch, meaning a signal can travel across a continent with essentially zero accumulated degradation, something no purely analog link can achieve

Common Pitfalls

Assuming digital is “more accurate” in general, when in fact digital accuracy is capped by bit depth and sample rate, and a poorly designed digital system can be worse than a good analog one
Undersampling a signal below the Nyquist rate, producing aliasing where high frequencies masquerade as false lower frequencies in the digital output
Ignoring anti-aliasing filters before an ADC, letting frequencies above fsample/2 corrupt the sampled data
Treating a digital signal’s voltage levels as exact, when real logic families define a valid “high” and “low” range with an undefined region between them
Forgetting that digital signals still travel as analog electrical waveforms on the wire, so signal integrity issues like ringing and reflection still apply at high speeds
Assuming more bits always means a better ADC, when sample rate, input noise, and reference voltage stability often limit real-world accuracy well below the theoretical bit-depth resolution
Clipping an analog signal that exceeds the ADC’s input range, which produces a hard, irreversible flat-top distortion in the digital capture
Overlooking jitter, small timing variations in the sampling clock, which introduces noise into a digitized signal even when amplitude quantization is perfect

Comparison

Property	Digital	Analog
Value range	Discrete steps	Continuous
Noise tolerance	High, noise below threshold is ignored	Low, noise directly corrupts the signal
Storage/copying	Lossless	Degrades with each copy/generation
Processing	Easy with software/logic	Requires dedicated analog circuitry
Precision limit	Set by bit depth and sample rate	Set by component noise floor
Real-world interface	Needs ADC/DAC to interact with physical world	Direct
Example device	Digital thermometer with LCD readout	Analog mercury or dial thermometer
Error correction	Possible (checksums, ECC)	Not possible after corruption
Long-distance transmission	Regenerable at repeaters	Degrades with distance/attenuation

Example

A microphone produces an analog voltage that varies continuously with sound pressure. An analog-to-digital converter (ADC) samples that voltage thousands of times per second, turning it into a stream of binary numbers a computer can store, process, or transmit; a digital-to-analog converter (DAC) later reverses the process to drive a speaker.

History

Early telephone, radio, and television systems were entirely analog, transmitting continuously varying electrical waveforms that directly mirrored sound or light.
Pulse-code modulation, the basis of modern digital audio, was proposed by Alec Reeves in 1937, decades before the electronics existed to make it practical.
The shift to digital accelerated through the late 20th century as ICs made ADCs, DACs, and digital signal processors cheap enough for consumer devices.
CDs, launched in 1982, were among the first mass-market products to prove digital audio could match or beat analog vinyl and tape in practice.
Claude Shannon’s 1948 paper “A Mathematical Theory of Communication” laid the theoretical groundwork for treating information itself, analog or digital, as a quantifiable, transmittable signal.
Broadcast television’s transition from analog to digital signals, completed in most countries between the 2000s and 2010s, freed up radio spectrum and improved picture quality within the same channel bandwidth.

FAQ

Why do digital signals resist noise better than analog? A digital receiver only needs to decide which of a few discrete bands a voltage falls into, so noise smaller than the margin between bands gets ignored. An analog receiver has no such margin, since every voltage value is meaningful.

Is analog sound actually “better” than digital? Not inherently. A digital system with sufficient bit depth and sample rate can reproduce audio indistinguishable from the original analog source; perceived differences often come from mastering choices, playback equipment, or genuine bandwidth limits in the digital conversion.

What is aliasing, in plain terms? When a signal is sampled slower than twice its highest frequency, high-frequency content folds back and appears as a false, lower-frequency signal in the digital output, an unrecoverable error unless filtered out beforehand.

Can a signal be both analog and digital at different points? Yes, this is the normal case. A guitar signal is analog at the pickup, gets digitized for recording or effects processing, and is converted back to analog to drive the speaker, crossing the digital/analog boundary multiple times.

Why do digital systems still need analog circuitry at all? Every sensor and every physical actuator interacts with continuous real-world quantities, and every wire carrying a digital signal is still governed by analog electrical physics, so a purely digital system with no analog interface cannot sense or affect the physical world.

What determines audio quality more, sample rate or bit depth? Sample rate sets the highest frequency that can be captured, bit depth sets the dynamic range and noise floor. For typical music, 44.1-48 kHz and 16-24 bits are well beyond the limits of human hearing, so further increases give diminishing returns.

Why do vinyl records and tube amplifiers still have fans if digital measures better on paper? Analog media and equipment can introduce distinctive, often pleasant distortion, saturation, and compression characteristics that some listeners prefer aesthetically, even though they are technically deviations from the original signal rather than improvements in accuracy.

Do digital displays use analog signals internally anywhere? Yes, the backlight driver, power supply, and often the final drive to individual pixels involve analog voltage or current control, even though the image data itself is processed and stored digitally.

Referenced by

Dig deeper

Spot an error in this note?