Baffle step

This is the commonest reason a speaker measures thinner than its simulation promised, and almost every box simulator leaves it out.

What happens

A loudspeaker cabinet is an obstacle in the air. How it radiates depends on how its size compares with the wavelength.

Well above the frequency whose wavelength matches the cabinet, the baffle is large compared with the sound, and it acts like a wall: the driver radiates forward into a half space of 2\pi steradians.

Well below it, the sound simply wraps around the box as though it were not there, and spreads into the full space of 4\pi.

Spreading the same energy into twice the space costs 6.02 dB. That is the baffle step, and it is entirely real.

Where the transition sits

At the frequency whose wavelength is comparable with the cabinet's size, so bigger boxes step lower:

| Cabinet | Step centered near | | --- | --- | | 5 L | 510 Hz | | 30 L | 280 Hz | | 120 L | 180 Hz | | 250 L | 140 Hz |

A small standmount loses its bass from a few hundred hertz down. A large floor stander holds on much lower. This is a real advantage of a big box that has nothing to do with the alignment inside it.

Why your simulation looked better

Baffle solves the enclosure in half space at every frequency, as lumped-element enclosure simulators conventionally do. That is the right assumption for a driver in a wall or a very large baffle, and it is optimistic by up to 6 dB for anything else.

Tick the box and Baffle models the finite cabinet. The bass drops, the maximum SPL curves drop with it, and what you see is closer to what you will measure. Excursion and impedance do not change: those are the driver's business, not the room's.

Expect F3 to leap, and do not panic. A 30 L sealed box that reported F3 at 49 Hz will report about 264 Hz once the cabinet is modelled. Nothing got worse; the reference moved. F3 is quoted relative to the passband, and the passband is now the midrange level rather than the optimistic half-space one. Saying you are 3 dB down at 264 Hz relative to your midrange is exactly what a measurement would tell you, and exactly why the crossover needs baffle step compensation. The low-frequency shape (the alignment, Qtc, the rolloff) is unchanged, as the unchanged Qtc and fc in the summary show.

What to do about it

• Expect it. A two-way with a 6-inch driver typically needs several dB of broadband compensation, usually built into the crossover as a baffle step circuit or as a deliberate tilt.
• Room gain gives some of it back. Against a wall or in a corner the room restores some of the lost radiation space at low frequencies. How much depends entirely on placement, and Baffle does not model it.
• Do not fix it with equalization without checking excursion. Boosting 6 dB at 40 Hz asks the cone for twice the travel. Look at the excursion plot.

What Baffle does and does not model

There are two models, and the menu names the trade-off.

Step only (exact). The cabinet is a sphere of the same volume: the exact solution for a point source on a rigid sphere, no fitted constants, correct at both ends. A sphere has no edges, so you get the step and no ripple.

Step and ripple (fitted). Every element of the baffle's rim is treated as a secondary source. This follows your actual outline, so it shows the ripple and how moving the driver changes it. The price is a rim reflection coefficient that was fitted to measurements rather than derived, and a first-order treatment whose own author puts its accuracy at around 5 dB on rectangular baffles.

One detail worth knowing, because it is the kind of thing that gets copied between tools without checking. The usual fitted coefficient is −0.60, and it is what most software uses. But that coefficient also decides the model's low-frequency limit, and −0.60 puts it at −7.96 dB where the physics says −6.02. Baffle uses −0.5, which gets the limit exactly right.

Where the ripple lives

For a 300 × 450 mm baffle the ripple sits above about 800 Hz, which is off the right-hand edge of the default 10–1000 Hz grid. Widen the grid to see it.

Centered, that baffle ripples about 5.3 dB peak to peak. Moving the driver 55 mm across and 125 mm up brings it to 4.3 dB, because the four edges stop being equidistant and their contributions stop arriving together. That is the whole reason tweeters are so often mounted off center, and you can watch it happen by dragging the offset.

What neither model claims

Neither models your room, the driver's own directivity, a rounded edge, or a cabinet that is not a simple box. If you need diffraction to a decibel, measure it. If you need to know that your bass will be 6 dB quieter than an infinite baffle promised, and roughly how much your baffle will ripple, this is enough.