Le Petit

Jean Hiraga's Le Petit from a single sheet of OSB, with an affordable Aiyima 4″ full-range driver, tuned by measurement, one change at a time.

All files, drawings and data: https://github.com/1704audiolabs/le-petit · part of 1704 Audio Labs — Measure & Taste

One-sheet · A4

The whole project on one page — what was measured, what was heard and how to build it. Open the one-sheet →

In short

• One 2000 × 1000 mm board of 22 mm OSB for the pair, a drain pipe as the port tube, one sheet of sheep's wool fleece per speaker for damping.
• Bass flat within ±1 dB from 55 to 150 Hz, down to 47 Hz — the box with this driver and the original bare port had a +8 dB boom at 75 Hz.
• Box resonances about −5 to −6 dB with a damping layout that fits on the back panel.
• Every step measured and documented, with all raw data as CSV — build it, or check it.

Contents

The result · 3D model · Build it · Port tube · Damping · Stand · The driver · How it was measured · What I'd do differently · Repository map · Credits

The result

Left speaker, 30 Hz–20 kHz, room-free. Terracotta: the untreated box. Blue: damped, with the 175 mm port. Below 400 Hz the near-field driver + port sum, above it the time-gated far field.

	Untreated box	Damped + 175 mm port
Port tuning Fb	74.4 Hz	≈ 48 Hz
Bass hump at ~75 Hz	+8.1 dB	gone (flat ±1 dB, 55–150 Hz)
Bass extension (−3 dB)	54 Hz	47 Hz
Box resonance at 350 Hz (port output)	+13 dB spike	6.3 dB lower
"Honk" at ~950 Hz / ~1.5 kHz	clear peaks	4.7 / 6.0 dB lower
Direct sound 300 Hz–4 kHz	±2–3 dB	unchanged

The resonance rows were measured after damping, before the tube went in (step 1, L-10).

±1 dB bass, 55–150 Hz

47 Hz −3 dB extension

≈ 48 Hz port tuning, 175 mm port

27.0 l inside volume

It doesn't sound like more bass — it sounds more balanced.

The boom that read as "lots of bass" is spread evenly down to 47 Hz. Listening note, left speaker, after step 2.

The two treble peaks at ~6 and ~10.5 kHz remain; EQ them to taste.

3D model

Turn it, look inside, cut it open, explode it, play the bass — open the 3D model.

see-through	cut open	exploded

Build it

	Specification
Box	530 × 340 × 232 mm outside (254 with the doubler), 486 × 296 × 188 mm inside = 27.0 l, removable back
Material	one 2000 × 1000 mm board of 22 mm OSB (square edges), beech cleats, HT DN75 drain pipe
Driver	Aiyima 4″ full-range, 8 Ω version (fits the Ø 94 mm cutout)
Joints	butt joints with biscuits (size 20) and wood glue — no screws; only the back panel is screwed, on cleats with gasket and M4 inserts

Both cabinets from one board: two edge trims, five cross cuts (530 · 530 · 300 · 296 · 300), then each piece is ripped or cut to length.

The building plan — three drawing sheets with every dimension, plus cut list, holes, biscuit positions and assembly order: the building plan.

The build guide in 14 steps — all of it here:

1. Cut the board (cutting plan above)
2. Drill the holes — Ø 94 driver, Ø 70 port in the front and the doubler, Ø 12 terminals
3. Seal the inside faces — OSB isn't airtight
4. Glue the cleats 23 mm from the rear edge
5. Biscuit slots and carcass — glued, no screws
6. Front and doubler — glued with biscuits
7. Seal the corners
8. Finish
9. Inserts and gasket
10. Fit the back panel at 292 × 482
11. Driver and wiring
12. Port tube, 175 mm total
13. Damping
14. Close it and listen

Port tube

An HT DN75 drain pipe extends the port inside the box: 175 mm total, measured straight through from the tube's inner end to the doubler's outer face. That tunes the box to ≈ 48 Hz. The original 43 Hz tuning is meant for the lean Fostex FE103NV2 next to a wall; with the Aiyima it would droop early.

Target Fb	Total port
55 Hz	112 mm	fuller, less deep
50 Hz	146 mm
≈ 48 Hz	175 mm	built — flat bass
43 Hz (the plan's tuning)	216 mm	doesn't fit the 188 mm box depth

Mounted from the inside: heat-bent flaps screwed to the front panel, joint sealed. Details, prediction and measurements: Step 2 — Port tube.

Damping

The box has three resonances that leak out of the port, and its dimensions predict them closely:

Resonance	From	Predicted	Measured
Height mode ("boxy")	486 mm inner height	353 Hz	350–360 Hz
Depth mode ("honk")	188 mm inner depth	912 Hz	~945 Hz
Port pipe resonance	44 mm port + end corrections	≈ 1.5 kHz	~1.5 kHz

Damping only works where the air moves. The layout that came out of eight measured variants:

• one layer of sheep's wool fleece (≈ 10 mm) covering the inside of the back panel (against the depth mode), and
• a wool pocket with 50 g of loose polyester fill across the middle third of the height (where the height mode moves the most air).

The photo shows the test setup. The damping shown here is the experimental rig built to compare materials and placements: fleece cut to size and stapled in, a hand-made pocket, material added and taken out again between measurements. Materials: sheep's wool garden fleece — the unbleached fleece sold as a plant-protection or mulch mat, ≈ 10 mm, 350 g per 5 × 0.5 m roll (≈ 140 g/m²) — and polyester fibre fill, the synthetic stuffing found in ordinary pillows. The finished speakers carry the same layout, made neatly: fleece cut to the panel, a sewn pocket, fixed securely.

One sheet of fleece per speaker, all on the back panel, about −5 to −6 dB on every resonance, direct sound unchanged. All eight variants: Step 1 — Damping.

Stand

Four legs and an H of bars in Ø 6 mm bright round steel, silver-soldered, 300 mm high; the legs screw into M6 inserts glued into the box bottom. The driver sits at 715 mm.

Cut list, hardware and the build steps: Stand · drawn by make_stand_plan.py

The driver: Aiyima instead of Fostex

	Original plan (FE103NV2)	This build (Aiyima 4″, 8 Ω)
Power rating	5 W	25 W
Bass	lean; needs a wall or corner	more bass of its own
Tuning	43 Hz (139 mm port, the plan's calculation)	≈ 48 Hz (175 mm port)
Damping	box modes come from the box — the same for both

The driver changes the port tuning, not the damping. Note: the datasheet shipped with the Aiyima (Re 3.4 Ω, Fs 73 Hz, Qts 0.52) is for the 4 Ω version; the impedance measurement showed the 8 Ω version. Everything known about the driver — the listing, the manufacturer's lab sheet, what was measured: Driver. More: Aiyima vs. FE103NV2.

How it was measured

The three microphone positions: ① far field, 50 cm on axis · ② near field, 5 mm from the dust cap · ③ near field, in the port opening.

Item	Role
Behringer UMC404	plays the sweeps, records mic, loopback reference and the impedance jig
Behringer ECM8000	measurement microphone: far field at 50 cm, near field at the cone and the port
Aiyima A70	the amplifier
Loopback cable (OUT 2 → IN 2)	timing and level reference: cancels the interface's latency and response
Impedance jig — 2 × 100 Ω in parallel + jack leads	measures the port tuning to ±1 Hz, independent of the room

Synchronized exponential sweeps, loopback-referenced deconvolution, a 3.4 ms gate before the floor reflection, near-field driver + port summed with the correct phase, one change per round, repeatability checked to 0.1–0.3 dB.

Measurement rig · Impedance jig · Method · Raw data

What I'd do differently

• Measure the impedance first. A 50 Ω resistor and two jack leads show the driver version and the real port tuning in minutes.
• Cut the back panel 2 mm smaller than the opening (292 × 482) and fit it with the gasket and screws from the plan. A friction-fit back leaks and shifts the bass every time it's refitted.
• Don't trust the port calculator for a bare hole. The box tuned to 74 Hz instead of the predicted 61 Hz. Scale the tube from a measurement, cut long, trim.
• Put the damping by the mode, not by the panel. One thin layer on its own did almost nothing against the 950 Hz "honk"; the pocket at mid-height did.

Repository map

docs/
├── one-sheet.md             the whole project on one A4 page
├── 3d/index.html            the interactive 3D model (three.js), as built
├── 01-build.md              build guide, 14 steps, parts list, cutting plan
├── building-plan.md         the building plan: drawings, cut list, holes, joints, assembly
├── stand.md                 the steel stand: drawing, cut list, hardware, steps
├── driver.md                the Aiyima 4″: listing, lab sheet (T/S), measured values
├── 02-measurement-rig.md    equipment, signal chain, microphone positions
├── 03-impedance-jig.md      the 50 Ω jig: parts, wiring, calibration, verification
├── 04-method.md             sweeps, windows, near-field sum, impedance estimator
├── 05-step-1-damping.md     box resonances, eight damping variants, result
├── 06-step-2-port-tube.md   port tuning from 72 to 48 Hz, result, listening test
├── 07-aiyima-vs-fe103.md    what changes with the Aiyima driver
├── 08-lessons.md            recommendations and conclusions
├── plans/                   building-plan sheets, stand sheet and cutting plan (SVG) + their generator scripts
├── data/csv/                every measurement as CSV
├── data/                    raw recordings (.npz) and overview plots
└── assets/                  photos, charts, diagrams, 3D stills, site styles
tools/
├── shoot_3d.py              stills of the 3D model for this README
└── svg2png.py               PNG previews of the plan sheets (headless Chrome)
overrides/                   page templates: home cover, one-sheet

Folder	Contents
docs/plans/	sheet 1 views and sections · sheet 2 panels and cut list · sheet 3 assembly · cutting plan · generators: sheets, cutting plan
docs/data/csv/	frequency responses, impedance and bass sums of every session
docs/data/	raw impulse responses and impedance (.npz), format description
docs/assets/	photos, charts, diagrams

Credits and license

Le Petit is Jean Hiraga's design (L'Audiophile no. 29, 1983); hole positions after R. Berchtold (2022). This project stands on the plans and write-ups of everyone who has built it before — this is my way of giving something back. Documentation, drawings, photos and data: CC BY-SA 4.0; the plan scripts in docs/plans/: MIT. Details: Credits.