Testing — how Chorrang! measures a cable

Conductor resistance and finished-cable resistance, four-terminal (Kelvin), converted to 20 °C. Finished cables are measured including their connectors, and we quote an average across samples rather than a best case.

Impedance against frequency, on the cable alone and on the earphone-plus-cable combination, on a bench LCR bridge used in four-terminal mode with a four-pin connection, so that lead resistance is excluded from the reading. We describe that instrument by what it resolves rather than by its brand and model: a list scan of ten points from 10 Hz to 40 kHz, a basic accuracy of 0.05%, a six-and-a-half-digit display, and resistance readable down to 0.001 mΩ. Putting a third party’s brand or model on this page would read as either an endorsement or a comparison we have not measured ourselves, so we do not do it.

No figure on this site comes from a listening test. Where a number depends on the earphone it was taken with, the page says which earphone; where it depends on the fixture, the page says that too.

A cable and an earphone form a voltage divider. The transfer function is

H(f) = Z_ear(f) / [ Z_ear(f) + Z_line(f) ]

with complex impedance, and the resulting response is expressed relative to 1 kHz.

For one measured pairing the total level loss came to about 0.27 dB at 1 kHz, and the deviation between frequencies stayed inside ±0.14 dB. That is a level loss applied across the band — it is not a tonal shift. The full result, with per-core data, charts, the frequency-by-frequency table and the limits of the measurement, is published as a research article: Does a cable change the sound? We measured it.

Two corrections we are recording rather than quietly fixing, because an earlier version of this page carried the old numbers:

• The 4 November 2026 session measured ten points from 10 Hz to 40 kHz with every frequency checked against the readout. An earlier round had one point set to 15 Hz when it should have been 15 kHz, and this page used to describe a peak in the earphone’s own impedance at 15 kHz (|Z| ≈ 14.6 Ω). That peak was the mis-set point: a low-frequency impedance rise read as if it were high-frequency. Measured at 15 kHz, the earphone’s own impedance is about 5.4 Ω, and the curve falls from 10 Hz to 20 kHz with no peak there.
• The same page used to quote about 0.30 dB of insertion loss and a spike near 15 kHz. Both came from that retired round. Nothing from it is quoted on this site any more.

A single number labelled “cable impedance” still says very little on its own. On this cable the loop opposition grows from about 0.243 Ω at 1 kHz to about 0.44 Ω at 20 kHz, and what that costs in signal depends on the earphone it is paired with: about 3% of the signal at 1 kHz and about 8.5% at 20 kHz in the pairing we measured.

Where these figures come from. They were taken on 4 November 2026 on a pre-production Purity sample and the reference earphone we keep in the lab — not on a production unit. The method and the physics carry over unchanged; the numbers will be re-run on the production build and republished with that build’s own figures in the specification table.

Capacitance (pF/m) and inductance (µH/m). On a cable this short, a bench LCR meter’s own lead residual makes the capacitance read as a negative value, and the inductance drifts across the 10 Hz–14 kHz range with return path, lay pitch and fixture. Publishing one figure would be publishing a number with no reproducible meaning. Measuring it properly needs a shielded fixture, open/short compensation and an impedance analyser — which is not proportionate to a cable that goes between a player and an earphone.

Other brands’ figures in sales copy. A comparison is only a measurement if the same method is applied to both sides, and then it belongs in a report with the conditions written down. It does not belong in a sales sentence.

• Finished-cable resistance measured on the production build. Until that exists, the specification table labels the figure as conductor-derived and pending, and it stays that way.
• The impedance and frequency-response results as machine-readable data, with charts drawn from that data rather than pasted in as pictures. Done for the 4 November 2026 round — the numbers live in data/research/cable-sound-20261104.yaml and the charts on the research article are generated from it. Re-running the same measurement on the production build is still open.
• Sample-to-sample spread. We intend to publish a spread across cables of the same build, not the reading of one lucky sample.