Welcome to SubSim

SubSim predicts how a subwoofer system covers a venue before you rig it. You place the boxes - flown or on the floor, alone or in formations like end-fire and cardioid - and SubSim shows the level they make everywhere in the room: on the floor from above, in a slice from the side, in 3D, and as a response curve at any seat.

It is built for the questions that come up on a real job:

• Will the back of the room get as much bass as the front?
• Where is the power alley between a left and a right stack, and how wide is it?
• How loud is it on the stage behind a cardioid array?
• How high should the subs hang, and what does the floor bounce do?
• Do my flown and ground subs help each other, or fight?

Press F1 anywhere in SubSim to open this manual at the part you are working on.

What SubSim is - and is not

SubSim is a prediction: it adds up the sound of every box at every point, with its level, its timing and the floor's reflection. It is honest physics, taken from the textbooks (see The physics, and its limits and every formula in The formulas), and every part of it is checked against results worked out independently.

It is not a measurement. It has not been compared with microphone measurements of real rigs yet, and real rooms, stages and audiences do things a model leaves out. Use it to choose between designs and to find problems before the gig; then measure on site. SubSim can show a measurement next to its prediction - see Comparing with a measurement.

SubSim is free, and it works entirely offline.

Getting started

Your first prediction in five steps

1. Pick a sample. On the Design page, open Pick a sample and choose one - for example Center stacked cardioid: 3 stacks of 3. The boxes appear and the prediction runs.
2. Set the audience. In Audience, type the nearest and farthest seat (metres from the front of the boxes) and the width.
3. Choose what to look at. Single frequency shows one note (move the slider); Third-octave band adds up the whole subwoofer band, which is closer to what people hear.
4. Look. Switch between 3D, Plan, Side, Response and Timing above the map. Red is loud, blue is quiet.
5. Change something and press Predict (or Ctrl+Enter). With Auto predict on (top right) SubSim predicts by itself after every change.

The cards along the top of the Design page sum it up: how even the coverage is, how much it changes from seat to seat, the loudest and quietest seat, and the height of the array. See The result cards.

The screen

Along the top:

Wherever you see a small (i), hover over it: it explains the control beside it.

The Design page

The Design page is where you look at the result. On the left: the Setup card (and, with Full view on, more settings). Along the top: the result cards. In the middle: the views.

Setup: array, audience, frequency

1 Array - Pick a sample loads a ready-made rig (see The samples). To build your own, use the Cabinets page, or add single boxes with + Flown box and + Ground box above the map.

2 Audience - the area where people are, as a rectangle on the floor:

• Nearest edge and Far edge: metres from the front of the array. The first row is often 2 to 5 m out.
• Width: how wide the audience is.
• With Full view on, also Left / right of array (to move the audience sideways) and Ear height (1.2 m seated, 1.7 m standing).

The coverage and seat-to-seat figures are measured over this rectangle only; the rest of the floor is shown faded, for context.

3 Frequency

• Single frequency: the level of one note, set with Show the level at. Good for seeing why something happens - lobes and nulls are sharpest at one frequency.
• Third-octave band: the level summed over the subwoofer band (Passband from ... to, with Full view on), one third-octave at a time. Closer to what people hear, because music has every frequency at once.

The result cards

The 3D view

The floor is coloured by the level at ear height; the bright rectangle is the audience. The boxes are drawn to size.

Mouse:

• Click a box to select it; Ctrl+click adds boxes to the selection.
• Drag a box to move it over the floor - a flown box keeps its height.
• Drag a box's laser to turn the box (see Turning a box).
• Left-drag anywhere else to move the view (pan); right-drag to orbit around the point under the mouse; the wheel zooms at the mouse; double-click returns to the standard (isometric) view.

The View menu can add a vertical slice through the array (the Side view's map, standing up in 3D) and reset the view.

The Plan view

Looking straight down on the floor at ear height. This is the view for coverage, power alleys and the stage behind the array.

• Drag a box to move it; Shift+drag (or drag the marker at 0, 0) moves the whole array.
• Drag a box's laser to turn it.
• The wheel zooms where the mouse is; drag empty map (or right-drag) to pan; double-click shows the whole map.
• Move the mouse over the map: the line under it shows the position and the level there.

Keep as B keeps this prediction so you can compare the next one with it: after a change, switch between A (now), B and A - B (red = louder now, blue = quieter now). The card under the buttons compares the two over the audience. Clear B forgets it.

The Side view

A vertical slice from the side: the array on the left, the audience to the right. This is the view for hang height, the floor bounce and what the sound does above the audience (tiers, balconies).

• Slice chooses where the slice goes: the centre, the left or right cluster, or the selected box.
• Drag the array up and down to change its hang height, or a single box to move it.
• The floor is the white line at 0 m. Below it, SubSim shows the sound as if the floor were not there - the wave the floor sends back up - dimmed and labelled. Nothing sounds there; it is the picture of the bounce.
• Point at the map: besides the level, the readout says what the floor reflection does at that spot - how many dB it adds or takes away, how much farther the bounce travels than the direct sound, and (at one frequency) how late it arrives in degrees.

The View menu holds this view's settings: the whole slice or the audience strip only, true scale or stretched taller (angles then look steeper), and how far below the floor to show (none, 2, 5 or 10 m).

The Response view

The level at the listener marker (the yellow dot on the maps), frequency by frequency, over the subwoofer band. The shaded band is the passband. Dips here are cancellations at that seat - often the floor bounce (SubSim marks the first floor null) or two groups of boxes arriving out of step.

Import measurement shows a measured response at the same seat next to the prediction - see Comparing with a measurement. Export prediction saves the predicted curve as text.

The Polar view (Full)

The level on a 30 m circle around the array, in the vertical plane: how the array radiates up, forward, down and back. The floor hides everything below the floor.

The Timing view

When does each box's sound reach the listener marker? One stem per path (the direct sound, and the floor bounce dashed), at its arrival time, and underneath the tone burst each part of the system produces there and their sum.

Click a box (or select a group): the view shows it against the rest of the array, and the box's settings card says how early or late it arrives. Delay it to line up adds the delay that makes it arrive together with the rest. With nothing selected and delay subwoofers switched on, it shows the delay subs against the main system.

The Height view (Full, flown)

For a flown array: how even the audience's level is (ripple and spread) as the hang height changes. The red zone is where the floor bounce puts a null inside the passband for the nearest seats. The Height advisor card (Full) suggests a height and can hang the array there.

The map tool bar

Above the map, for the 3D, Plan and Side views:

Building the rig: the Cabinets page

The Cabinets page builds the rig in three steps. The 3D view on the right shows it as you go, and you can click, drag, turn and copy boxes there as on the Design page.

Step 1: the box

One picker holds every box: the four generic types (single or dual 18", single or dual 21") and the real subwoofers from the Loudspeakers page. The line underneath sums it up: its level at 1 m, its pattern and its size.

Figures opens the details:

• Design: front-loaded, horn-loaded, tapped horn, bandpass, tapered / folded, or cardioid box (with a rear driver). The design sets a typical sensitivity and pattern.
• Sensitivity (1 W / 1 m), power per driver, impedance and boxes per amp channel - with the amplifier load that results (below 2 Ω most amplifiers cannot drive it).
• Full space / Half space: how the data sheet was measured. Most subwoofer data sheets quote half space (on the ground), which includes the floor's +6 dB. SubSim takes that off and adds its own floor, so it is not counted twice. Check yours.
• Level typed in / Level from power: type the level of one box at 1 m, or let SubSim work it out from sensitivity and power (continuous, program +3 dB or peak +6 dB).
• Width, height, depth and Box pattern (for a generic box: omni, cardioid, hypercardioid or anything between).

A real box brings the maker's published figures and says where they came from. Makers do not publish directivity in a form SubSim can read, so a real box uses the typical pattern of its design until you load measured data (see Directivity data).

Step 2: the formation

First Flown, Ground stack or Free. Free means no formation - only the boxes you add yourself.

Then the formation, as tiles:

Only the settings of the chosen formation are shown: how many columns, stacks or groups; rows deep; spacing; row distance; steering.

Stacked cardioid tuning. How long the reversed box waits sets the pattern:

• Cardioid - it waits exactly its own depth: silent straight behind.
• Supercardioid - 0.58 of its depth: the quietest back half overall, silent about 125 degrees either side. The arena choice.
• Hypercardioid - one third of its depth: narrower forward, silent about 110 degrees either side, a small lobe straight back.

Balanced: with two forward boxes against one reversed, the back can only cancel properly when the forward boxes are turned down 6 dB so the reversed one matches them. SubSim turns the forward boxes down rather than the reversed one up, so no box is asked for more than it can give. The Front-to-back card (Full) shows the result both straight behind and over the whole back half.

Step 3: where it hangs

• Centre or Left + right. Left + right puts a copy of the formation on each side, Apart metres centre to centre; the two sides can be the same or set separately.
• Height of the array centre (flown), with the underside shown beside it.

A left + right system always makes a power alley - a loud strip down the middle and alternating loud and quiet fingers either side - because the two sides arrive in step on the centre line and out of step elsewhere. A centred array (flown or on the floor) avoids it.

Delay subwoofers (Full)

Subs placed out in the venue to reach the back. Set their distance from the main system, their position left / right, how many and how high. Their delay is worked out from the 3D distance and the air temperature, so their sound leaves just as the main system's arrives; Extra delay (a few ms) keeps the sound seeming to come from the stage.

Front-to-back (Full)

How much quieter the whole system is behind than in front: straight behind, and over the whole back half on average, with every delay, polarity and facing included. Use it to compare cardioid tunings.

Every box (Full)

A table of every box: its loudspeaker, level, delay, polarity, facing and position. Type a value and press Enter. Level, delay and polarity are the totals for that box - a formation's own delays are already in them. Edited rows are blue; Reset all edits puts every box back to what its formation makes. Clicking a box on a map and using its settings card is usually quicker.

Working with boxes

Selecting and the box's settings card

Click a box on the 3D, Plan or Side view. Its settings card opens on the map:

• X across, Y along, Z height - where it is, in metres. Type a value to move it.
• Level and Delay - the totals for this box.
• Aim - where it points, in degrees (0 = at the audience, + to the right, 180 = back). Easier: drag its laser.
• Faces front / back and Inverted (polarity).
• At the listener - whether it arrives early or late against the rest of the array, and Delay it to line up.
• Copy, Paste, Duplicate, Select all, Remove.

Ctrl+click adds boxes to a group; the card then shows the group's centre, and typing moves the whole group. Click empty map (or the X on the card) to select nothing.

Moving boxes

Drag a box on any map. On the Plan and 3D views it slides over the floor at its own height; on the Side view it moves along and up / down. With Snap on it moves in 10 cm steps. Every selected box moves together. Moving a box does not change its delay.

Turning a box

Grab a box's laser (the aim line) on the Plan or 3D view and drag: the box turns on the spot to point at the mouse, all the way round. With Snap on it turns in whole degrees. For an exact angle, type it in the card's Aim field.

Turning changes the sound only of a box that has a pattern - a cardioid box, a horn, or a box with measured directivity data. A generic front-loaded box is modelled as omni (the same in every direction), so turning it changes nothing.

Copy and paste

Select one box or a group and press Ctrl+C, then Ctrl+V - or use Copy and Paste on the box's card. The copies keep their level, delay, polarity, aim and loudspeaker, and land on the first free spot beside the originals, already selected so you can drag them into place. Ctrl+D (or Duplicate) does both at once. Each paste lands one step further along. While you are typing in a field, Ctrl+C and Ctrl+V copy and paste text instead.

Adding and removing boxes

+ Flown box hangs a box on top of the highest flown box; + Ground box stands one beside the rightmost ground box. Remove (on the box's card) takes the selected boxes out. A box removed from a formation can be put back with Reset all edits. Everything can be undone.

Lasers

A laser shows where each box points: from the front of the box to the far edge of the audience. A box facing away gets a short dashed arrow backwards. Laser on the tool bar hides them. Grab a laser to turn its box.

Air, floor and rooms

The floor

With Floor reflection on, the floor reflects the sound, as in real life: every box gets a mirror copy below the floor. Near a box on the floor its copy is right beside it, so the floor adds about 6 dB and nothing cancels. Under a flown array the copy is far below, so the direct sound and the bounce arrive at different times: they add in some places and cancel in others - the stripes on the Side view. Turn it off to see a flown rig in open space.

The floor surface (Full, Air and floor) sets how much it reflects:

Air

Air temperature sets the speed of sound (331.3 × √(1 + T / 273.15) m/s), so it moves where things add and cancel, and the delays SubSim works out. Humidity sets air absorption (ISO 9613-1), which is tiny at subwoofer frequencies.

Rooms and walls (Full)

Room with walls adds walls, a ceiling and a floor as reflecting surfaces. Set the room's size, and for each wall its distance, thickness and material. Walls reflect by the mass law: heavy, thick walls reflect almost everything at low frequencies; thin plasterboard lets a lot through. Parallel walls make the room ring at its standing-wave frequencies (n × c / 2L), listed under the walls. Reflections followed sets how many bounces are added up - more is slower. The maps stop at the walls.

Comparing and exporting

Comparing with a measurement

On the Response view, Import measurement reads a measured frequency response saved as text or CSV from Sonasync, REW or Smaart (two columns: frequency and dB). Put the listener marker where the microphone stood. Match level shifts the measurement to the prediction's average over the passband, because most measurements are relative: compare the shape. A Sonasync export also says which range it trusts; outside it the curve is dotted. Choose smoothing from none to 1/3 octave.

Exporting

• Export prediction (Response view): the predicted response at the listener marker, as text.
• Export cabinet list (CSV) (Full, Project): every box with its position, level, gain, delay, polarity and aim - for the system processor or a rigging plan.

Saving designs

Save writes a .subsim file: the whole design - boxes, formation, edits, audience, environment, loaded directivity data and an imported measurement. View settings (which slice, the Snap switch) are not part of it. New starts empty; Reset to defaults (Full, Project) goes back to the starting design.

The Loudspeakers page

The real subwoofers SubSim knows, from their makers' published figures: max SPL, low-frequency limit, weight, sensitivity, power and size. Nothing is guessed - what a maker does not state is shown as not published, with the source for everything.

• Choose a maker on the left, search, or show single / dual drivers only.
• The card shows the box's figures and Use for the whole array.
• The directivity balloon shows how the box radiates in 3D (right-drag to orbit, as in the 3D view). Without measured data it is the typical pattern of its design.
• To scale draws the box next to a 1.8 m person; Compared with ranks it against the other boxes.

Directivity data

Load directivity data (CSV) replaces the typical pattern with a measured balloon for the whole array. The file is text: optional # name:, # manufacturer:, # absolute: (1 if the levels are absolute SPL at 1 m) and # size_m: lines, then columns freq_hz, azimuth_deg, elevation_deg, spl_db and optionally phase_deg, on a regular grid (azimuth 0 = the front, positive to the box's right; elevation positive = up). Without phase the box is treated as zero-phase, which the program says.

The samples

Reading the results

Colours and levels

The colour scale runs from red (loud) to blue (quiet) over the Map range (30 dB by default). On the Plan and 3D views the top of the scale is the loudest seat in the audience; on the Side view it is set by the loud part of the slice. Two maps on different scales cannot be compared by colour - compare the numbers (point at the map), or keep one as B and look at A - B.

What the patterns mean

• Stripes fanning out from a flown array on the Side view: the floor bounce. Their spacing depends on the hang height and the frequency.
• Fingers down the room on the Plan view of a left + right system: the power alley and its side lobes.
• Quiet behind a cardioid or end-fire: the rear cancellation working. Compare straight behind and back half on the Front-to-back card.
• One tone against the band: at a single frequency the pattern is sharp; over the band (third-octave band) the stripes smear, which is closer to what people hear.

Flown and ground subwoofers together

Two sources several metres apart that play the same frequency will be out of step somewhere. A delay moves where that happens; it cannot remove it. Measured on sample 13: at ear height the flown and ground boxes mostly add up (the ground line is much louder near the stage, and farther out they arrive nearly together); they cancel high up near the rigs - where tiers and balconies are - by up to 14 dB at 80-100 Hz. The only way to stop them fighting is to let them play different frequencies (a crossover between them), which SubSim does not model yet.

The physics, and its limits

How SubSim computes. Each box is a point source: its sound falls as 1/r and arrives after the time it takes to travel, r / c. The sound of every box at a point is added as a complex number, so level and timing count - that is what makes things add or cancel. The floor is a mirror copy of every box below the floor, scaled by how much the surface reflects (the image-source method). Rooms add a copy for every wall bounce, up to the number of reflections you choose. Air absorption follows ISO 9613-1; the speed of sound follows the air temperature. Cardioid and supercardioid tunings follow first-order gradient theory.

Checked. The tests compare the engine with results worked out independently: the array factor of a line of boxes, the notch in front of a wall, explicit mirror sources for a wall, room standing waves spaced c / 2L, and the cardioid formulas. A second, independent implementation of the formulas in The formulas, fed only with the boxes the program exports, agrees with the program's level to within 0.01 dB at every point tested - lines, left + right, stacked cardioid, end-fire, flown supercardioid, flown over ground, turned boxes, grass floor, hot and dry air, one tone and the band sum. Air absorption matches the published table of ISO 9613-2 (Table 2) to its printed precision.

What it leaves out:

• A box is a point: within about one box size of it the near field is not modelled.
• A generic box has a simple pattern (a front-loaded box is omni). Load measured directivity data for the real one.
• The floor is one number (the share reflected); real ground reflects a little differently at each angle and frequency.
• No diffraction around the stage or the boxes, no audience absorbing sound, no wind or temperature layers bending it.
• Rooms are rectangular, with mirror-like walls.
• No crossovers or filters per box yet.

SubSim has not yet been compared with measurements of real rigs. Its maths is right and tested; whether a model matches a particular venue is what a measurement shows.

The formulas

Every figure SubSim shows comes from the formulas below. They are the formulas the program uses - copied from its source code, not a simplified description - so an engineer can check any number by hand. Symbols: f frequency (Hz), ω = 2πf, c speed of sound (m/s), k = ω / c the wavenumber, λ = c / f the wavelength, r distance (m), p_ref = 20 µPa.

Speed of sound

c = 331.3 × √(1 + T / 273.15)        T = air temperature, °C

At 20 °C, c = 343.2 m/s. Every delay SubSim works out itself (a formation's delays, the delay subwoofers', "Delay it to line up") and every distance-to-time conversion uses this c - so changing the temperature moves where sound adds and cancels, and those delays follow it. A delay you add to a box by hand is kept as an addition on top of what its formation gives.

One box, one path

Each box is a point source. The complex sound pressure it makes at a point, along one path (the direct sound, or one reflection), is

p = s × g × p1m × 10^(G/20) × D(θ) × A(r) / r × e^(−j(k·r + ω·τ))

  p1m  = p_ref × 10^(L1m / 20)    the box's level at 1 m, on axis

  G    = the box's gain (dB)

  s    = −1 if the box is inverted, else +1

  τ    = the box's delay (s)

  D(θ) = the box's pattern (below)

  A(r) = 10^(−α·r / 20)           air absorption, α in dB/m (below)

  g    = 1 for the direct sound,

         the reflection coefficient for a reflection

r is never taken below 5 cm (the point-source formula has no meaning closer).

The level at 1 m. Typed in, or worked out from the box's figures:

L1m = S + 10·log10(W × n) + C

  S = sensitivity (1 W / 1 m)     W = power per driver

  n = drivers per box             C = 0 continuous, 3 program, 6 peak

If the figures are half space (measured on the ground), SubSim uses L1m − 20·log10(2) = L1m − 6.02 dB, because it adds the floor itself.

The box's pattern

Without measured data, a box has a first-order pattern:

D(θ) = a + (1 − a) × cos θ

  θ = the angle between the box's axis and the direction to the point

  a = 1      omni (front-loaded box)     a = 0.5    cardioid

  a = 0.366  supercardioid               a = 0.25   hypercardioid

The box's axis points along its aim φ (0 = at the audience, + to the right) and elevation ε: (sin φ·cos ε, cos φ·cos ε, sin ε). The pattern is signed: behind a hypercardioid it is negative, which is the rear lobe's reversed polarity.

With measured directivity data, D is the file's value at that frequency and direction: interpolated bilinearly in azimuth and elevation, and linearly in log-frequency, with its phase when the file has one.

Adding everything up

The pressure at a point is the complex sum over every box and every path:

P(f) = Σ_boxes Σ_paths p

At one frequency the level is

L = 20·log10(|P| / p_ref)       (dB SPL)

Over the band (third-octave band), the band powers are added:

L = 10·log10( Σ_b |P(f_b)|² / p_ref² )

where f_b are the ISO third-octave centre frequencies (1000 × 10^(n/10) Hz) from Passband from to Passband to, each with weight 1. Because it is a sum of powers, a cancellation at one band lowers the total less than it lowers that band - which is why the band view is smoother than a single tone.

The floor

The floor is an image source: each box is mirrored below the floor (height z → −z) and its sound is added with the reflection coefficient γ:

p_floor = the one-box formula for the mirror box, with g = γ

  γ: concrete 1.0      wooden stage 0.9     packed earth 0.8

     short grass 0.6   long grass 0.4       snow / carpet 0.2

The mirror box radiates as the mirror image of the real one (its pattern is reflected too). Only points above the floor get the reflection.

Where the floor bounce cancels. For a box at height H, a listener at distance d and ear height h, the bounce travels farther by

Δ = √(d² + (H + h)²) − √(d² + (H − h)²)

and the first null is at f = c / (2Δ) - where the bounce arrives half a wavelength late. The Side view's readout shows Δ, Δ / c in ms and 360° × Δ × f / c for the box nearest the mouse, and the level with the floor minus the level without it, for the whole system.

Below the floor (Side view), with the floor reflecting, the map shows the direct sound only - the sound as if the floor were not there. No sound is there; it is the wave the floor sends back up.

Rooms and walls

Walls are image sources too, up to the chosen number of reflections: every combination of mirrorings across the six surfaces, each image scaled by the product of the reflection coefficients along its path. Images whose total reflection is below 0.001 are left out.

A wall's reflection coefficient follows the mass law of a limp panel:

R = jx / (1 + jx) × √(1 − α_m)

  x = ω·m / (2·ρ·c)

  m = density × thickness (kg/m²)

  ρ = 1.204 kg/m³ (air)

α_m is the material's own absorption at low frequency. A heavy wall (large x) reflects almost everything; a light one lets low frequencies through. Opposite walls a distance L apart ring at their axial modes, f_n = n × c / (2L).

Air absorption

ISO 9613-1, in dB per metre, with T in kelvin (T0 = 293.15 K, T01 = 273.16 K), relative humidity h_r and air pressure p_a (p_r = 101.325 kPa):

C   = −6.8346 × (T01 / T)^1.261 + 4.6151

h   = h_r × 10^C / (p_a / p_r)      (water vapour, molar %)

frO = (p_a/p_r) × (24 + 4.04e4 × h × (0.02 + h) / (0.391 + h))

frN = (p_a/p_r) × (T/T0)^(−1/2)

        × (9 + 280 × h × e^(−4.170 × ((T/T0)^(−1/3) − 1)))

α   = 8.686 × f² × [ 1.84e−11 × (p_r/p_a) × (T/T0)^(1/2)

        + (T/T0)^(−5/2) × ( 0.01275 × e^(−2239.1/T) / (frO + f²/frO)

                          + 0.1068  × e^(−3352.0/T) / (frN + f²/frN) ) ]

At subwoofer frequencies this is a few thousandths of a dB per 100 m.

Formations

End-fire and steering. For a target direction u, every box is delayed so all their sound arrives together along u:

τ_i = (s_i · u − min_j s_j · u) / c      s_i = the box's position

The box nearest the target waits longest. For an end-fire with rows a distance D apart, the front row waits D / c (1.4 m → 4.08 ms at 20 °C) and each further row forward another D / c.

Cardioid rows and pairs. The rear box stands a distance D behind, faces back, is inverted and waits D / c.

Stacked cardioid. The reversed box radiates from its back face, one box depth d behind the stack, is inverted and waits

τ = k × d / c

  k = 1 (cardioid), 1/√3 (supercardioid), 1/3 (hypercardioid)

For a pair of equal sources this gives the first-order pattern a = k / (1 + k): 0.5, 0.366 and 0.25. Balanced turns the h − 1 forward boxes of an h-high stack down by 20·log10(h − 1) dB (6.02 dB for 3 high), so the reversed box's sound matches theirs behind the stack.

Delay subwoofers wait for the main system's sound: τ = d_3D / c + extra, with d_3D the 3-D distance between the main system and the delay subs.

The result figures

The audience is sampled on a regular grid of up to 160 × 160 points at ear height.

• Coverage: σ = the standard deviation of L over the audience points (in dB); within 3 dB is the share of points with |L − mean| ≤ 3 dB.
• Seat to seat (after Welti): for every third-octave band b of the passband, σ_b = the standard deviation of L_b over the seats; the card shows the mean of σ_b over the bands and the worst band.
• Level min / mean / max: over the same audience points.
• Front-to-back: L(80 m in front) − L(80 m behind) at the array's centre height, in open air; and the back half: 10·log10(Σ_front |P|² / Σ_back |P|²) every 5° around a horizontal 80 m circle.
• A − B: L_now − L_B at each point of the map.
• Time alignment (delay subs, Timing): the phase between the two systems' arrivals at the listener - within 45°: aligned; within 90°: partly; beyond: misaligned; more than 10 dB apart: one system dominates.

Measurements

• Smoothing at 1/N octave: at each frequency f, the power average of the measured points between f / 2^(1/2N) and f × 2^(1/2N).
• Match level: the offset added to the measurement is the mean of (predicted − measured) over the measured points inside the passband and inside the measurement's trustworthy range.

How the grid is chosen

The maps are computed on regular grids whose size is set by a time budget of about 12 seconds of computing, from the number of boxes, bands and reflection paths. A small rig gets a fine grid (up to 420 points across); a room with many reflections gets a coarser one rather than a prediction that runs for minutes. Levels between grid points are drawn by bilinear interpolation.

Reference

Keyboard

Mouse

Questions

The map says OUT OF DATE. The design changed since the last prediction. Press Predict (Ctrl+Enter), or switch Auto predict on.

Turning a box changes nothing. It is an omni box - see Turning a box.

The level looks 6 dB too high. Check Full space / Half space in the box's figures: most data sheets are half space.

The prediction is slow. Rooms with many reflections cost the most. Lower Reflections followed, or switch Auto predict off and predict when ready. SubSim keeps a prediction to about 12 seconds by using a coarser grid for very big jobs.

I want the old layout of a setting back. Every change can be undone; Reset to defaults (Full, Project) starts again.

Glossary