Below 200 Hz, the room is in charge
At low frequencies, wavelengths become comparable to the dimensions of the room. Sound no longer behaves as a diffuse field but as a set of standing waves, called modes, that settle between parallel surfaces.
Every mode has antinodes, where pressure adds up, and nodes, where it cancels. The consequence is immediate: at one point in the room a given frequency will be too loud; fifty centimetres away it may be barely audible. It is not the subwoofer that varies, it is where you listen.
The frequency of the first mode follows simply from the speed of sound and the distance between two opposing surfaces. For a room of 5.00 × 4.00 × 2.50 m:
| Dimension | Distance | 1st axial mode | 2nd harmonic |
|---|---|---|---|
| Length | 5.00 m | 34.3 Hz | 68.6 Hz |
| Width | 4.00 m | 42.9 Hz | 85.8 Hz |
| Height | 2.50 m | 68.6 Hz | 137.2 Hz |
What matters most is the geometry of these modes. The first length mode has its two antinodes against the front and rear walls, and its node exactly in the middle of the room, 2.50 m in this example. A seat placed there will barely hear 34 Hz, whatever you do with the subwoofer.
Above a certain frequency, in the order of 100 to 300 Hz depending on the room, the modes crowd together and overlap, the field becomes progressively diffuse, and subwoofer placement loses its influence. Everything is decided below that.
The point no setting can fix
This is the most useful idea on this page, and it decides the order in which you work.
| Measured defect | What it is physically | What an equaliser does with it | The right lever |
|---|---|---|---|
| Peak | an excess of energy | it removes what is in excess, cleanly | equalisation |
| Null | a cancellation: two paths in opposite phase subtract | nothing useful. Both contributions rise together and keep cancelling: you spend headroom, heat the driver, add distortion | position |
Hence the method: correct peaks with equalisation, and nulls with position. A deep null at the listening seat is a problem of geometry, and geometry is not adjusted from a menu. The guide on measuring with REW says the same thing at its correction step: peaks first, nulls almost never.
The crawl, and why it works
The technique is well known, often poorly explained, and its principle is elegant.
- Temporarily place the subwoofer at your listening position, at ear height if you can.
- Play a continuous low-frequency signal, or a music track rich in bass.
- Move around on all fours, ear at the height where the subwoofer will sit, along the walls and through the locations you are considering.
- Find the spot where the bass sounds most even, neither bloated nor hollow. That is where the subwoofer goes.
Why this works rests on a physical principle, acoustic reciprocity: in a room, swapping the positions of source and receiver does not change the measured response. What you hear at a given spot, with the subwoofer at your listening position, is therefore what you will hear at your listening position once the subwoofer sits at that spot.
Two cautions. The method tells you about one seat, the one where you put the subwoofer: with several seats it says nothing about how even they are between them. And the ear judges absolute level poorly in the bass, so look for evenness, not quantity.
The corner: most output, rarely the most even
A corner is an antinode for every mode in the room, since pressure is maximal there in all three directions. A subwoofer placed there excites everything, and delivers the most output for the same power. That is real and measurable, and it is why the position is so popular.
The downside comes from the same mechanism. Exciting every mode maximally also maximises the differences between them, so it produces the most uneven response. A corner gives plenty of bass, not necessarily good bass.
It is not a bad choice as a matter of principle: in a small room, or where placement is constrained, it is sometimes the only option, and the extra output is welcome. But it should be a decision, not a reflex.
One subwoofer or two?
With a single seat, one well placed subwoofer is often enough. The problem changes nature as soon as there are several seats, because a null at the middle seat may correspond to a peak at the side.
This is the subject of a landmark study: Todd Welti and Allan Devantier showed, by simulating a large number of configurations, that multiple subwoofers substantially reduce seat-to-seat variation, and that this reduction makes equalisation afterwards far more effective. Their simulations single out two or four subwoofers as particularly favourable configurations in a rectangular room.
The resulting order of operations matters, and it is often taken backwards: position first, then number, equalisation last. Equalising a response that varies widely from seat to seat amounts to improving one place while degrading the others, since a single filter applies to everyone.
What comes after position
Once the spot is found, three settings remain, and none of them replaces it.
Level. It is set relative to the other speakers, by measurement rather than by ear: bass gives an impression of level that depends heavily on the content being played.
Phase or delay. The subwoofer and the main speakers must add up in the crossover region, not subtract. A poor alignment digs a hole there that is often blamed, wrongly, on placement.
Crossover frequency. It decides the region where both sources work together. Too high, and you hand the subwoofer frequencies the room renders directional; too low, and you ask the mains for work they cannot do.
Where the answer is different
The room is not rectangular. The axial mode formulas assume parallel surfaces. With an L-shaped room, a sloping ceiling or a loft, the modes still exist but no longer compute so simply. The crawl, however, remains valid: it measures what is there, with no assumption about shape.
The room opens onto another. A large opening leaks in the bass, which weakens the modes but also reduces room gain. The volume to consider is no longer that of the listening room alone.
The floor is timber over a void. It radiates and absorbs in the bass, sometimes considerably. Decoupling the subwoofer then changes the result more than a few centimetres of movement.
You cannot move the subwoofer. That is common, and it is not a dead end. If the source is fixed, the seat becomes the variable. Moving the listening position thirty centimetres forward or back is sometimes enough to step out of a node.
What HTM predicts, and what you still need to verify
HTM models your room and computes the corresponding bass field. Two tools apply directly to this page: the bass field map, which shows where antinodes and nodes sit at a given frequency, and the subwoofer optimiser, which looks for the positions giving the most even response at the seats you have declared, including with several subwoofers.
The point is not to replace listening but to shrink the search space: rather than crawling the whole room, you start from two or three plausible candidates and compare them by ear.
This has to be said plainly: these are predictions, based on the geometry you entered and on assumptions about how the surfaces behave. They do not replace a measurement. The full approach is to predict, then verify with a real measurement, which is exactly what HTM can compare against its own predictions: see the guide on measuring with REW.
Sources
- Todd Welti and Allan Devantier, Low-Frequency Optimization Using Multiple Subwoofers, Journal of the Audio Engineering Society, vol. 54 no. 5, May 2006, pp. 347-364. AES record.
- The mode frequencies on this page are computed for a speed of sound of 343 m/s, the usual value at 20 °C. They vary slightly with temperature.
Further reading
The glossary defines the terms used here, and the guide on measuring with REW explains how to verify all of the above in your own room. The frequently asked questions cover how the application works.