Articles

One article per week, each one a self-contained tutorial with its tool embedded in the text. Titles link once the page is live — nothing here points at a page that does not exist yet.

Core series

• Published

  dB conversion, from dBFS to dB SPL

  All five reference levels, the 0.775 V history behind dBu, and the five conversion mistakes that survive a decade of experience.

• Published

  IEEE 754 floats and hex, decoded field by field

  Sign, exponent and mantissa by hand, subnormals, ULP, and reading a 32-bit capture from a logic analyser.

• Published

  Window functions: leakage, resolution and sidelobes

  Why the window is not optional, how to pick between Hann, Hamming, Blackman and flat-top, and the two corrections that depend on what you are measuring.

• Published

  FFT, RMS and peak levels in practice

  Why the tallest bin is not the amplitude, how far parabolic interpolation gets you, and how to turn a floor into a density.

• Published

  Magnitude-squared coherence between two channels

  How many averages a trustworthy coherence needs, why one average always reads 1.0, and what the cross-spectrum phase says that coherence cannot.

• Published

  FIR design with the windowed-sinc method

  Choosing tap count from transition width, predicting the passband ripple from the stopband attenuation, and why the group delay is not optional.

Extended series

Follow-ups that take one of the core subjects further, usually because the practical question turns out to be subtler than the theory.

• Published

  THD, THD+N, SINAD and ENOB

  Five distortion figures that are not interchangeable, the bandwidth caveat, and why harmonic percentages add in quadrature.

• Published

  Biquad design: the RBJ cookbook explained

  Eight filter types from one transfer function, why A = 10^(dB/40), and the Q-versus-bandwidth difference that no one mentions.

• Published

  Sensitivity, power and distance

  How loud a system really plays: the inverse square law, the two sensitivity conventions, and the crest factor that decides the amplifier.

• Published

  Adding decibels: +3, +6, or nothing

  Powers add, amplitudes add as vectors, and the answer depends on correlation — with the noise-floor correction table and the averaging bias.

• Published

  LUFS and true peak: what the meter measures

  The K-weighting filter, the two-stage gate, loudness range, and why every sample can be below full scale while the true peak is 3 dB above it.

• Published

  Room modes and the Schroeder frequency

  Axial, tangential and oblique modes, the coincidences that make a room sound wrong, and where the modal region ends.

Supporting calculators

• Published

  CRC calculator: parameter sets that actually match

  Poly, init, refin and refout — the settings that make two CRC-16 implementations disagree, and the check value that settles it.

• Published

  Nyquist, oversampling and alias folding

  Working out where an out-of-band tone lands before you spend a day chasing it on a scope.

• Published

  Acoustic delay, distance and temperature

  Speed of sound as a function of temperature, and why a five-degree error moves a five-metre alignment by four centimetres.