Speaker Enclosure Calculator (Thiele-Small)
A bare driver produces no bass: it is the box that decides how low the bass reaches, how tight it sounds and how loud it can play. The driver's Thiele-Small parameters — Fs (resonance frequency), Qts (total quality factor) and Vas (volume of air equivalent to the suspension compliance) — are all it takes to design that box.
This calculator sizes a sealed enclosure and/or a bass-reflex: internal volume, tuning frequency, port length, simulated frequency response, −3 dB cutoff, cabinet dimensions and maximum output levels.
The three required parameters appear on the driver's data sheet. The optional fields (Qes, Sd, Xmax, Re, power) do not change the computed volume, but they unlock the EBP diagnostic, the sensitivity, the maximum output level and the air speed in the port.
Driver and box
The Thiele-Small parameters: what Fs, Qts, Qes, Qms and Vas measure
Neville Thiele and Richard Small, in the early 1970s, boiled a driver's low-frequency behaviour down to a handful of numbers. Manufacturers still print them. DIY builders still live by them. Five of them already tell you how the cone will sit in a sealed box or a ported one.
- Fs — free-air resonance, in hertz. The frequency at which the moving mass and the suspension's spring cancel. In free air the impedance peaks there.
- Qts — total Q at Fs. Electrical losses and mechanical losses in parallel. Low Qts is well damped; high Qts is peaky and “lively”.
- Qes — electrical Q, set by the motor (Bl, Re). Usually the dominant half of Qts.
- Qms — mechanical Q, the surround and spider. High Qms means a low-loss suspension.
- Vas — the volume of air whose compliance matches the suspension, in litres. A large Vas is a soft suspension; it wants a large box.
1 / Qts² = 1 / Qes² + 1 / Qms² · Qts = Qes · Qms / (Qes + Qms)
The second form is the usual product-over-sum. Re, Sd, Xmax and Mms matter too — they just do not pick the box volume on their own.
Reading a driver data sheet and spotting dubious figures
Spec sheets lie, sometimes by optimism, sometimes by a sloppy test jig. A few red flags save a wasted cabinet.
- Qts below 0.2 paired with a huge Vas — often an understated Qts, to make the driver look “tight” in a small box.
- Fs that does not fit the size — a 6.5-inch midwoofer at 30 Hz is a unicorn unless the magnet and surround say so too.
- No impedance curve, no on-axis response — a sheet with only five numbers is a brochure.
- Sensitivity that fights the motor — 94 dB from a small ferrite and a light cone is a favourite inflation.
A honest 6.5-inch hi-fi driver typically sits around Fs 40–60 Hz and Qts 0.30–0.50. Compare with a known reference of the same size before you trust a stranger's PDF.
Measuring Fs, Qts and Vas yourself (added-mass method, test box)
If the sheet smells optimistic, measure. You need a sine source, a way to read voltage and current (or a cheap USB interface), and patience around resonance.
Added mass. The classic bench method:
- Measure Re with a DC milliohmmeter.
- Sweep impedance in free air, find Fs and the −3 dB points that give Qes, Qms, Qts.
- Stick a known mass (10 or 20 g of modelling clay) on the dust cap, centred.
- Measure the new resonance Fsm. Vas follows from the shift, Sd and Mms.
Test box. Mount the driver on a sealed jig of known net volume Vb. The new resonance Fc and the new Qtc, plus the free-air pair, give Vas through the usual Thiele–Small identities. Either method is fine if the mass sits in the middle and the box does not leak.
EBP: what the criterion really says, and its limits
Efficiency Bandwidth Product is just a ratio:
EBP = Fs / Qes
Folk lore: EBP < 50 → sealed; EBP > 100 → bass-reflex; 50–100 → either. It is a starting hint about how “lossy” the motor is at resonance, not a building code. An EBP of 80 can be a lovely sealed monitor (higher Qtc) or a modest ported stand-mount. It does not know your room, your target F3, or whether you care more about group delay than about the last five hertz.
Treat it as a first glance at the data sheet. Then run the actual alignment.
Sealed or bass-reflex: efficiency, power handling, transient response
Two different filters, not two religions. A sealed box is a second-order high-pass: 12 dB/octave below Fc, cone excursion that falls as 1/f², plenty of damping from the trapped air. A bass-reflex (ported box, Helmholtz vent) is fourth-order: 24 dB/octave below Fb, more output near the tune, and a cone that is unloaded below Fb — excursion shoots up, so power handling collapses under the tuning frequency.
| Sealed | Bass-reflex | |
|---|---|---|
| Low-end extension | Gentle; F3 often near Fc | Lower F3 for the same driver |
| Excursion / power | Air spring protects the cone | Risky below Fb; need a high-pass |
| Group delay | Lower, “faster” bass | Higher near Fb; some hear it as loose |
| Box size | Usually smaller | Often 1.5–2× the sealed volume |
Want deep and loud in a given litre count? Ported. Want a studio nearfield that stops cleanly? Sealed is still the default for a reason.
Choosing your Qtc: what you hear between 0.5 and 1.1
For a sealed alignment, Qtc is the shape of the knee. Volume sets it: smaller box, higher Qtc, higher Fc.
- Qtc = 0.50 — overdamped, early roll-off, very controlled. Studio monitors like it.
- Qtc = 0.707 — Butterworth. Flattest amplitude, −3 dB at Fc. The textbook all-rounder.
- Qtc = 1.0 — a 3–4 dB hump around Fc. More “body”, less accuracy.
- Qtc = 1.1 — a real bump. Punchy in a tiny box; one-note if you push it further.
Above 1.1 the bass rings. Below 0.5 it simply thins out for most living-room listeners. 0.7 is not a moral obligation — it is the flat one.
The classic bass-reflex alignments (SBB4, QB3, C4, B4) and their trade-offs
Ported boxes have a pair of numbers to pick: Vb and Fb. The old named alignments are recipes that freeze that pair for a given driver.
- SBB4 (super-boom-box / extended) — lowest F3, large box, low tune, smooth roll-off.
- QB3 (quasi-Butterworth 3) — the common hi-fi compromise: smaller than SBB4, a hint of ripple, this calculator's default recipe.
- C4 (Chebyshev 4) — smaller box, higher tune, a deliberate hump for efficiency.
- B4 (Butterworth 4) — flattest at Fb, the purist's fourth-order high-pass.
WinISD, VituixCAD and the rest will draw all four. The names are a language, not a cage: once you can see the curve, you can slide Vb and Fb off the named point to miss a room mode or to fit a cabinet you already own.
Sizing the port: minimum cross-section, chuffing, flared ports, multiple ports
Too small a vent and the air turns turbulent. That noise is chuffing — a “pfff” on every kick drum. A working floor is about 0.3 × Sd of port area. For a 6.5-inch (Sd ≈ 140 cm²) that is 42 cm², a tube near 7.3 cm in diameter. Go larger if you play loud; the length then grows, because Helmholtz wants area and length together.
- Flares — round both ends. Even a sanded PVC cut is quieter than a sharp lip.
- Several ports — same total area, shorter individual tubes, easier to fit. Keep them apart so one does not steal the other's end correction.
- Length — past about 40 cm the tube becomes furniture. Fold it, or split the area, before you build a trombone inside the box.
The wavelength at Fb is huge compared with the port diameter. That is why a vent can be a short tube at 40 Hz and still be a Helmholtz resonator rather than a quarter-wave pipe — unless you make it long enough that it becomes one, which is a different alignment entirely.
Box losses (Ql): why the real response is gentler than the model
The textbook curve assumes a rigid, airtight cabinet. Panels flex, joints weep, stuffing leaks energy. Ql is the lumped number for that.
- Ql ≈ 15 — dense walls, good seals, little stuffing in the vent path.
- Ql ≈ 7 — a typical careful DIY box. Use this unless you have a reason not to.
- Ql ≈ 3 — leaky or floppy. The port peak sags and Fb becomes vague.
Lower Ql knocks the tuning peak down and spreads it. The measurement is almost always softer than the lossless simulation. That is not the driver “lying”; it is the box being real.
Stuffing and internal damping: wool, felt, effect on apparent volume
Stuffing does three jobs: it soaks up midrange reflections off the rear of the cone, it damps panel ringing a little, and — if you fill rather than line — it can push the compression toward isothermal, which raises the apparent volume by something like 10 to 20 %.
- Polyester wadding — clean, easy, enough from a few centimetres on the walls.
- Mineral wool / fibreglass — stronger absorption; gloves, mask, and a cloth wrap so fibres stay off the cone.
- Bitumen or MLV on the panels — mass-damping, not absorption. Different tool.
In a ported box, leave a clear path to the vent. Stuffing the port mouth is how people accidentally lower Ql to 3 and then wonder where the bass went.
Cabinet standing waves, proportions and anti-vibration bracing
A rectangular box has axial modes at f = c / (2L). A 50 cm internal height puts the first one near 343 Hz — right in the midrange, where a coloration is obvious. Avoid integer ratios (cubes, 1:2) so the modes do not stack. Golden-ish sets such as 1 : 1.618 : 0.618, or a modest 1 : 1.3 : 1.6, spread them.
Braces between opposite panels raise the panel resonance and cut the “boxy” slap. 18 mm MDF is a floor for a 6.5-inch; 22 mm (or a constrained layer) once the driver gets larger or louder. A cube with no brace is a drum, whatever the T/S model says.
Room gain: placement against a wall, in a corner, in-room measurement
The alignment you just calculated is a half-space, anechoic idea. The room is not. A speaker out in the floor gets little boundary lift. Park it near a wall and the bass rises 3–6 dB. Stick it in a corner and you can see 6–12 dB, plus a forest of modes.
Measure in the chair, not in the brochure. A cheap measurement mic and REW will tell you whether the 38 Hz tune is a gift or a boom. Placement is the first EQ. A long T60 at 80 Hz is a room treatment problem; it will not be fixed by retuning the port two centimetres.
From bare driver to finished speaker: crossover, baffle step, phase alignment
A driver in a box is still not a loudspeaker. Three more jobs sit between the T/S alignment and a listenable two-way.
- Crossover — splits the band to the tweeter. You need the impedance curve and the on-baffle response, not just Fs and Vas.
- Baffle step — below a frequency set by the baffle width, radiation goes toward 4π; above it, 2π. That is a 6 dB shelf. Leave it uncompensated and the midrange shouts.
- Phase / time alignment — a few millimetres of acoustic offset at the crossover can cut a hole in the summing. Recess, tilt, or delay until the two drivers add rather than notch.
The low-frequency calculator stops at the woofer's box. The rest is a different afternoon.
A complete worked example: a 6.5-inch driver in a 34-litre bass-reflex
Take a typical 6.5-inch: Fs = 45 Hz, Qts = 0.40, Qes = 0.45, Vas = 35 L, Sd = 140 cm², Xmax = 5 mm.
- EBP = 45 / 0.45 = 100. Borderline. Ported, for extension.
- QB3-ish — 34 L net, Fb ≈ 45 Hz, F3 ≈ 40 Hz. That is this page's recipe, give or take stuffing.
- Port — 0.3 × 140 = 42 cm² minimum. An 8 cm tube (50 cm²) wants about 16 cm of length at 45 Hz, before flares.
- Box — internal 25 × 33 × 41 cm is ~34 L. 18 mm MDF, a cross-brace on the large panels.
- Stuffing — line rear and sides, keep the vent path open. With Ql ≈ 7 and ~15 % extra apparent volume, the lossless 1.2 dB peak sags toward 1 dB.
- Baffle step — 30 cm width puts the shelf near 340 Hz. A low-pass around 2.5 kHz to a dome, plus a baffle-step network, is a separate design; do not skip it and then blame the 34 L alignment for a shouty mid.
Near a wall, expect another 3–4 dB of bass. If that is too much, lengthen the port a couple of centimetres (Fb down a hertz or two) before you reach for the equaliser.
When the lumped Thiele-Small model stops being honest
T/S parameters are small-signal, linear, isothermal-ish numbers taken at one voltage. Turn the volume up and Bl sags at large excursion, the suspension stiffens (Cms is not constant), the coil heats (Re rises, Qes rises), and the port can go turbulent. The beautiful QB3 curve is then a sketch.
For a living-room 6.5-inch at civilised levels, the sketch is excellent. For a PA sub at Xmax, you design with excursion plots, a steep high-pass below Fb, and a grain of salt on every simulated decibel.
The sealed box as a second-order high-pass
Electrically, a sealed alignment is the same animal as an RLC high-pass. Fc is the pole frequency, Qtc is the Q. That is why 0.707 is “Butterworth”: maximally flat, 12 dB/octave, −3 dB at Fc. A Linkwitz-Riley 12 dB electrical high-pass stacked on a Qtc = 0.5 box is a favourite trick for a fourth-order acoustic roll-off with a sub — two second-order halves, Q = 0.5 each.
Once you see the box as a filter, the “sealed versus ported” argument becomes “12 dB versus 24 dB, and where is the group-delay bump”. Less mystique, better choices.
End correction: why the port is always a bit shorter than the formula
Air at each end of the tube does not stop dead at the timber. It piles out as a lumped mass, so the effective length is Lv + k·Dv. k is around 0.73 for a flush, unflanged end on a thin wall, less once you flare, more if the port sits hard against a wall of the cabinet. This calculator already subtracts that term. Cut the tube to the number it prints, then trim after a measurement — wood does not get longer, and a flare eats a few millimetres of the physical tube.
Passive radiators: a port that does not chuff
A passive radiator is a cone with no motor, tuned by its mass on the box compliance. Same Helmholtz idea as a vent, no air rushing through a tube, so no chuffing and no 40 cm pipe to hide. You pay with the PR's own Fs, Qms and Sd, and with a slightly higher group delay. Small boxes that cannot fit a large-diameter, short port — think a compact 6.5-inch stand-mount — are where PRs earn their keep.
The T/S worksheet still starts with the active driver's Fs, Qts and Vas. The PR is a second moving mass you have to include before you trust the F3.
Net volume is the air the driver actually sees: external dimensions minus wall thickness, minus the volume of the driver, port, braces and crossover. A “25 L” box built to outside measurements can be 18 L inside, and the alignment you simulated is then someone else's speaker.