Spurious-Free Dynamic Range (SFDR)

SFDR is the ratio of the RMS amplitude of the fundamental carrier signal to the RMS amplitude of the highest spurious component in the frequency spectrum, expressed in dBc (relative to carrier) or dBFS (relative to full scale).

• Formula: SFDR = 20 × log₁₀(V_fundamental / V_max_spur)
• Measurement: Performed via FFT analysis of a pure sine wave input
• Key distinction: Unlike THD, SFDR captures the single worst artifact, not the aggregate
• Typical values: High-performance ADCs achieve 90-100 dBc SFDR

The specific frequency, amplitude, and phase of spurious components are direct manifestations of device-specific hardware imperfections, making SFDR analysis a powerful fingerprinting tool.

• Interleaving spurs: Time-interleaved ADCs produce deterministic spurs at Fs/N intervals due to gain, offset, and timing mismatches between sub-ADCs
• Integral Non-Linearity (INL) spurs: Static transfer function curvature generates harmonic and intermodulation products unique to each die
• Clock coupling artifacts: Parasitic coupling of clock energy into the signal path creates spurs at the sampling frequency and its sub-harmonics
• Power supply modulation: Poor PSRR allows switching regulator noise to appear as sidebands around the carrier

While SINAD aggregates all noise and distortion into a single figure of merit, SFDR isolates the most prominent non-linearity, which is often the most robust and repeatable fingerprint feature.

• SFDR advantage: Identifies a single, high-SNR artifact that persists across varying channel conditions
• SINAD limitation: Averaging masks individual spur characteristics that are critical for device discrimination
• Complementary use: SFDR identifies the strongest fingerprint feature; SINAD provides a holistic quality metric
• Practical impact: A device with 80 dB SFDR may have a dominant 3rd harmonic spur that serves as a reliable identifier even when the noise floor fluctuates

SFDR is not a static parameter; it exhibits predictable drift patterns over temperature and device lifetime that must be tracked for robust long-term fingerprinting.

• Temperature coefficient: INL-induced spurs typically shift in amplitude by 0.1-0.5 dB per 10°C due to changing bias conditions
• Aging effects: Hot carrier injection and NBTI in CMOS converters gradually alter transistor threshold voltages, causing slow spur amplitude migration
• Compensation requirement: Fingerprinting systems must implement drift compensation algorithms that update reference signatures over time
• Forensic value: The rate and pattern of SFDR degradation itself becomes a meta-fingerprint indicating device age and operating history

The 'spurious-free' range is defined by the worst-case spur, which may not be a harmonic of the input and can originate from unexpected sources.

• Non-harmonic spurs: Interleaving artifacts, clock feedthrough, and digital crosstalk produce spurs unrelated to the input frequency
• SFDR frequency dependence: The worst-case spur often changes with input frequency as the converter's non-linear transfer function is exercised differently
• Multi-tone testing: Applying two or more tones reveals intermodulation spurs that single-tone SFDR measurements miss
• Fingerprinting strategy: Characterizing SFDR across a swept frequency range creates a multi-dimensional spur map unique to each device

Dithering—the intentional injection of noise before quantization—decorrelates quantization error from the input signal, fundamentally altering the SFDR profile and the device's fingerprint.

• Spur-free improvement: Proper dithering can increase SFDR by 10-15 dB by breaking up coherent quantization spurs
• Noise floor trade-off: The dither signal raises the overall noise floor, potentially masking weaker fingerprint features
• Subtractive dithering: Dither is injected before the ADC and digitally subtracted after, preserving SFDR improvement without noise penalty
• Fingerprint modification: A dithered converter presents a different SFDR signature than an undithered one; fingerprinting systems must account for the dither state