Signal-to-Noise Ratio (SNR) is the ratio of the power of a desired signal to the power of background noise in the same channel or measurement, typically expressed in decibels. When working with amplitudes or RMS voltages, the ratio is squared (equivalent to using 20 log10 in the dB form) under the standard assumption of equal impedances. A higher SNR means the signal is more distinguishable from noise, while a lower SNR indicates noise is more dominant relative to the signal.
In practice
SNR appears constantly in ADC and DAC system design. The theoretical peak SNR of an ideal N-bit ADC is approximately 6.02N + 1.76 dB, derived from quantization noise alone. In practice, real converters fall short of this figure due to thermal noise, clock jitter, and nonlinearity, so datasheets often specify a lower effective number of bits (ENOB) derived from measured SNR. When selecting an amplifier to drive an ADC input, the amplifier's own noise contribution must be budgeted carefully so it does not degrade the ADC's SNR — the blog post "Specifying the Maximum Amplifier Noise When Driving an ADC" addresses this calculation directly.
In digital communications, SNR (or its close relatives Eb/N0 and carrier-to-noise ratio) governs achievable bit error rate for a given modulation scheme. A BPSK link tolerates a lower SNR than 64-QAM for the same BER target. When building simulations of receivers or ADC front-ends, SNR must be set deliberately in the noise model — see "Modelling a Noisy Communication Signal in MATLAB for the Analog to Digital Conversion Process" and "Setting Carrier to Noise Ratio in Simulations" for practical approaches. The blog post "Understanding and Relating Eb/N0, SNR, and other Power Efficiency Metrics" clarifies how SNR, Eb/N0, and related metrics convert into one another.
SNR is also relevant in sensor interfaces: strain gauges, thermocouples, current sensors, and similar transducers produce low-amplitude signals that must be amplified before digitization. Every stage in the analog chain — cable, anti-aliasing filter, op-amp — adds noise that degrades SNR. Grounding strategy, shielding, and component selection all affect the final SNR seen at the converter input.
A common pitfall is conflating SNR with dynamic range or SINAD. SNR strictly counts only noise power in the denominator, while SINAD (Signal-to-Noise-and-Distortion) also includes harmonic distortion. Datasheets may use these terms inconsistently, so always check the test conditions and definition used by the specific vendor.
Discussed on DSPRelated
Frequently asked
How is SNR calculated in decibels?
SNR (dB) = 10 * log10(P_signal / P_noise), where P_signal and P_noise are power quantities. If working with amplitudes or RMS voltages rather than powers, the formula becomes SNR (dB) = 20 * log10(V_signal / V_noise), which holds under the standard assumption of equal impedances. Make sure both quantities are measured over the same
bandwidth to get a meaningful result.
What is the relationship between SNR and ADC resolution?
For an ideal N-bit
ADC with a full-scale sinusoidal input, the theoretical maximum SNR is approximately 6.02N + 1.76 dB. Each additional bit of resolution improves the ideal SNR by about 6 dB. Real devices achieve less than this due to thermal noise, clock jitter, and distortion; the ENOB metric back-calculates the effective resolution from the measured SNR.
How does SNR differ from Eb/N0?
SNR is a general ratio of signal power to noise power, typically measured in a given
bandwidth.
Eb/N0 (energy per bit to noise spectral density) normalizes for data rate and bandwidth, making it more useful for comparing modulation schemes and link budgets in communications systems independent of occupied bandwidth. The blog post 'Understanding and Relating Eb/N0, SNR, and other Power Efficiency Metrics' covers the conversion between the two in detail.
How is SNR affected by oversampling?
Oversampling spreads quantization noise across a wider frequency band. If the signal
bandwidth is kept constant and the
sample rate is doubled, digital low-pass filtering removes noise above the signal band, improving SNR by approximately 3 dB per doubling of oversampling ratio — assuming white quantization noise and a fixed signal bandwidth. This is a useful rule of thumb rather than a universal guarantee. This principle underlies noise shaping in
delta-sigma converters, which can achieve very high effective SNR from low-resolution comparators by sampling at rates far above the
Nyquist frequency of the signal.
Why does SNR matter when choosing an op-amp to drive an ADC?
The op-amp in a signal-conditioning stage adds its own voltage and current noise, which appears directly at the
ADC input. If the op-amp's integrated noise over the signal
bandwidth is comparable to or larger than the ADC's internal noise floor, the converter's effective SNR degrades regardless of its resolution. The op-amp noise must be budgeted as part of the overall signal chain, with a target typically set well below the ADC's own noise contribution. The blog post 'Specifying the Maximum Amplifier Noise When Driving an ADC' provides a worked methodology for this calculation.
Differentiators vs similar concepts
SNR is often confused with SINAD (Signal-to-Noise-and-Distortion Ratio) and with dynamic range. SNR counts only noise power in the denominator; SINAD adds harmonic distortion products, so SINAD is generally equal to or worse than SNR for the same signal under consistent measurement conditions. Dynamic range describes the ratio between the largest and smallest signals a system can handle — though the precise definition varies by context, ranging from largest-to-smallest detectable signal to the usable output range before hitting the noise floor or clipping. SNR is also distinct from
Eb/N0 and carrier-to-noise ratio (CNR/C/N): those are
bandwidth-normalized metrics used in communications link budgets, whereas SNR is a general measurement applicable to any analog or digital signal chain.
