Measurement rig¶
Part of How it was measured — optional reading. To build the speaker you only need the build guide.
Professional loudspeaker measurement doesn't need an anechoic room. It needs a reference, a repeatable setup, and the right time window. Everything here is affordable, off-the-shelf gear.
The setup: speaker on a chair (driver at 86 cm), every position marked on the floor. ① Far field: the ECM8000 at 50 cm, on axis. ② Near field: mic tip 5 mm from the dust cap. ③ Near field: mic tip in the plane of the port opening.
Equipment¶
| Item | Role | Why this one |
|---|---|---|
| Behringer UMC404 (USB audio interface, 4 in / 4 out) | plays the sweeps, records mic, loopback and the jig | four inputs: mic, loopback reference and the two impedance probes at once; class-compliant |
| Behringer ECM8000 (measurement microphone, omni, phantom powered) | the acoustic measurements | flat enough for comparisons, the standard budget measurement mic |
| Aiyima A70 (class-D amplifier) | drives the speaker | the amplifier the speakers are used with |
| Loopback cable (jack–jack, OUTPUT 2 → INPUT 2) | timing and level reference | cancels the interface's latency and frequency response |
| Impedance jig (50.1 Ω resistor + leads) | electrical measurement of the port tuning | build guide |
| Laptop with a Python measurement tool | sweep generation, deconvolution, analysis, plots | see Method |
| Multimeter | resistor value, wiring checks | |
| Masking tape, a ruler, a card spacer | repeatable positions | the cheapest and most important part |
Signal chain¶
Output 2 is wired straight back into input 2. Every recording therefore carries its own reference: the analysis divides by the loopback instead of the digital file, which cancels the interface's latency (≈ 260 ms on this Mac) and its own response (flat within ±0.4 dB, 30 Hz–16 kHz), and gives every measurement the same time zero — essential for adding driver and port near-field responses with the correct phase.
Microphone positions¶
| Position | Placement | Shows |
|---|---|---|
| Far field | 50 cm from the dust cap, on axis, at driver height (86 cm) | the direct sound (time-gated before the floor reflection), mids, treble, distortion; and the in-room response |
| Near-field driver | ≈ 5 mm from the dust cap (card spacer) | the cone alone, room-free |
| Near-field port | centred in the port opening, in the plane of the front panel (a ruler across the opening as the reference) | what comes out of the port — where box resonances show up |
Driver height sets the usable gate: at 86 cm the floor reflection arrives 3.8 ms after the direct sound, so the gated response is room-free above ≈ 290 Hz. Higher is better (1.2 m → ≈ 185 Hz).
Making it repeatable¶
- Floor marks for the speaker's chair and the mic, height marks on the stand.
- Taped knobs: mic gain, loopback gain, amplifier volume. Comparisons across sessions only hold if none of them moves.
- Fingerprints checked at the start of every session: mic distance from the arrival time (54 cm for the 50 cm position — the acoustic centre sits behind the dust cap), the sweep peak at the mic (−11.6 dBFS) and the impulse peak-to-noise ratio (84 dB).
- Re-placement tests: the port near field repeated after moving the mic away and back agreed within 0.1–0.3 dB.
Gain staging tip: set the mic gain on the sweep peak (aim −15 to −10 dBFS), not on pink noise. The sweep peaks about 12 dB higher than pink noise at the same setting; set by pink noise alone it can clip in the near field.
If a knob moved: reset the loopback with a tone sent only to output 2 (back to its reference level), and match the mic gain so the cone's output at 100–200 Hz equals an earlier session — a range no port or damping change can affect.