Spectral Distribution
Frequency resolved noise power measurements quantify how random electrical fluctuations distribute across the spectrum per unit of bandwidth. In signal conditioning and sensor qualification, noise power spectral density determines the baseline noise floor and ultimate resolution limits of measurement channels. Units are expressed in volts squared per hertz or amperes squared per hertz, with square root values representing spectral density in volts per root hertz.
The metric separates white noise floors from low frequency flicker noise profiles. Metrological characterization requires specifying measurement bandwidth and source impedance conditions.
Noise Composition
Thermal noise provides a flat, frequency independent spectral profile across all practical sensor operating bands. Flicker noise exhibits an inverse frequency power slope, dominating the low frequency measurement spectrum below the corner frequency. Shot noise generated across semiconductor junctions adds broadband current spectral density proportional to direct bias current.
Amplifier input stages introduce both voltage and current noise sources that interact with sensor source impedance. Total integrated noise represents the integral of spectral density across the active measurement bandwidth.
Acquisition Design
Preamplifier selection requires matching equivalent input noise voltage and noise current to the specific sensor impedance. Low frequency measurement applications necessitate chopper stabilized or auto zero amplifiers to shift signal bands away from one over f noise knees. Shielding and balanced differential topologies minimize external electromagnetic interference pickup that corrupts spectral plots.
Sourcing specifications mandate maximum allowable noise spectral density at reference frequencies like one hertz and one kilohertz. Digital signal processing filters eliminate out of band noise power to optimize signal to noise ratios.
Metrological Verification
Dynamic signal analyzers compute fast Fourier transforms of amplified sensor outputs under shielded, zero input conditions. Calibration involves measuring Johnson noise from precision reference resistors to verify analyzer noise floor and gain accuracy. Spectral density plots confirm that sensor noise conforms to theoretical modeling without spurious discrete interference spikes.
Acceptance criteria establish strict thresholds for corner frequency placement and white noise plateaus. Incomplete noise characterization leads to overstated sensor resolution ratings in precision commercial instruments.
