Phase, polarity and interference

Phase is one of the most abused words in audio. A phase value has meaning only with frequency and a reference. Polarity is a separate binary operation. Interference is the physical consequence when waves meet. Keeping those three ideas separate makes system alignment dramatically clearer.

1. What does phase mean?

For a sinusoid, phase describes position within the cycle relative to a reference. A pure delay τ creates frequency-dependent phase:

Symbols: φ(f) is delay-induced phase shift in radians, f frequency in hertz and τ delay in seconds. The minus sign indicates phase lag, and a fixed delay produces proportionally more phase rotation at higher frequency.

Thus a 1 ms delay is -36° at 100 Hz, -180° at 500 Hz and -360° at 1 kHz. Saying “the speaker is 90° out of phase” without frequency/reference is incomplete.

2. Polarity

Polarity inversion multiplies the complete waveform by -1. It swaps positive and negative excursion instantly and at all frequencies. It is not equivalent to adding a fixed time delay, except in the narrow sense that a single sinusoid appears 180° inverted.

3. Constructive and destructive interference

When coherent waves overlap, complex pressure sums. Equal-amplitude in-phase waves can double pressure (+6 dB SPL at the point); equal-amplitude opposite-phase waves can cancel ideally. Real systems exhibit partial addition or cancellation because amplitude and phase differ with position and frequency.

4. Comb filtering

A delayed copy of a signal creates regularly spaced peaks and nulls. For a delay Δt:

Symbols: f_null,n are cancellation frequencies, f_peak,n reinforcement frequencies, n an integer index, Δt the delay between two equal-amplitude paths and Δf the spacing between adjacent comb-filter features. Perfect nulls require ideal coherence and equal amplitudes.

The pattern is spatial as well as spectral: move the microphone or listener and path differences change.

5. Why subwoofer placement is critical

Subwoofer wavelengths are long enough that multiple cabinets often interact strongly across the entire audience area. Separating left and right subs can create a centre “power alley” where path lengths are equal, with alternating off-axis lobes and nulls where path difference approaches fractions of a wavelength.

6. Delay alignment

Delay can compensate a real propagation-time offset. If two acoustic centres differ by distance d, first approximation is τ=d/c. Alignment should be performed around the intended crossover/overlap band, where both sources have useful amplitude. Matching arbitrary phase at a frequency where one source is already 30 dB down has little meaning.

7. Without measurement it is easy to move in the wrong direction

An RTA magnitude trace cannot distinguish many causes of a dip. Transfer-function measurement with a time reference provides magnitude, phase and coherence, allowing delay and polarity to be separated from room reflections. Alignment by ear can be useful as a final perceptual check, not as the sole diagnostic tool.

8. Phase wrapping and unwrapping

Displayed phase is usually wrapped into a ±180° or 0-360° interval, so a smooth delay appears as repeated jumps. Unwrapped phase removes those 360° display discontinuities and makes slope easier to interpret. The underlying system has not suddenly jumped in time; only the plot representation has.

9. All-pass filters

An all-pass filter is designed to alter phase while maintaining essentially flat magnitude. First-order all-pass sections can rotate phase over roughly 180°; second-order sections can provide up to roughly 360° around a chosen centre frequency/Q. They are useful when two acoustic slopes need phase shaping without a corresponding magnitude EQ.

10. Choosing a frequency for sub-top alignment

Work in the region where both sub and top contribute meaningfully and where coherence is good. Compare phase slopes, polarity and delay over a band, not at one cursor frequency. A single perfect-looking phase crossing does not guarantee robust summation across the crossover region.

11. Polarity check versus transfer-function measurement

A polarity tester can verify whether a driver initially moves in the expected direction. It cannot tell you whether two loudspeakers are time-aligned at crossover. Transfer-function measurement addresses the frequency-dependent relationship.

12. Comb filtering is spatial

A deep geometric null cannot generally be “filled” with EQ. Boosting the source raises both interfering paths and the cancellation remains. Change geometry, delay, source count, coverage or listening position instead. This is a classic example of a non-minimum-phase problem where amplitude EQ alone is the wrong tool.

13. Why we do not need “0° phase” everywhere

Absolute displayed phase is less important than the relative phase of sources that overlap. Minimum-phase magnitude errors can often be corrected with EQ because magnitude and phase are linked through the system transfer function. Pure delay, many reflections and spatial cancellations are non-minimum-phase components and cannot be repaired by ordinary magnitude EQ.