Soundproofing

Subsections that cover one component carry an Applies to tag; untagged subsections apply to all three. An Informs tag marks the subsections behind each decision above.

1 · Ratings: what each number measures and what it leaves out

Bottom line. Every single-number rating is a weighted fit over a band that starts at 80–125 Hz; traffic, rail and machinery put much of their energy below it, so ratings rank assemblies but do not predict low-frequency results.

Each rating fits a curve over its own frequency range

Informs D1

Table 2 and Figure 1 set out what each rating measures and what it leaves out.

Two consequences. Facades are compared on OITC and interiors on STC, and any assembly whose data stops at 125 Hz is unquantified below that. And a laboratory rating is a ceiling, not a forecast: the field result includes flanking and is lower.

A 3 dB change is the threshold of noticing

For broadband noise, 3 dB is the smallest change most listeners notice, 5 dB is clearly audible, and 10 dB is heard as roughly half as loud. These thresholds are the test for whether a measure registers at all, and they are used throughout this guide.

A-weighting, used in dBA and CNEL, discards 39.5 dB at 31.5 Hz and 16 dB at 125 Hz. An A-weighted single number is therefore nearly blind to the band where traffic rumble and machinery problems sit.

Field results run 5–10 points below laboratory ratings

Laboratory tests suppress flanking and seal the specimen perimeter. In a house the same assembly sits in a continuous floor deck and ceiling plane, with boxes, doors and ducts in it. Without a flanking analysis the usual allowance is field = lab − 5 to 10 points. The gap is the flanking (Part 3), and most of it is workmanship and path enumeration rather than physics.

2 · Leaves: mass, coincidence, resonance and coupling

Bottom line. A single panel gains 6 dB per doubling of mass. Two leaves on a cavity gain far more, but only above their resonance, and only as far as the connection between them allows.

Mass law: 6 dB per doubling of mass or frequency

Informs D2

m surface density, kg/m² · f frequency, Hz (field incidence)

a single limp panel blocks 6 dB more each time its weight per square foot or the frequency doubles.

For ⅝ in gypsum (2.2 psf, 10.7 kg/m²) at 500 Hz, one layer gives 27 dB, two give 33 and four give 39 (Figure 2). One layer to two costs one layer for 6 dB; two to four costs two more for the next 6. That fixed and poor exchange rate is why high-performing assemblies decouple instead.

Added mass is also structural load. A second ⅝ in layer on both faces adds about 4.4 psf of wall, which counts toward seismic weight (Framing · Roof weight drives seismic demand), and a topping on a floor counts against the beam schedule (Framing · Added floor dead load).

Each panel has a weak band at its coincidence frequency

Above its critical frequency a panel's bending wave matches the sound wavelength in air at some angle, the panel couples to the sound field, and transmission loss dips. The critical frequency falls as thickness rises: about 2,000 Hz for ⅝ in and 2,700 Hz for ½ in gypsum, about 2,000 Hz for 6 mm glass, near 115 Hz for 6 in concrete.

• Damping sets the depth of the dip. A viscoelastic compound between two boards (constrained-layer damping) typically adds 3–9 STC points, more on rigidly connected assemblies; laminated glass does the same for glazing.
• Dissimilar thicknesses put the two dips at different frequencies, so they do not fail together. This is why asymmetric glazing outperforms symmetric glazing of the same weight (Windows & Doors · Glazing build-ups).

A double leaf is a mass–spring–mass system with three regions

Informs D2

m₁, m₂ leaf surface densities, kg/m² · d cavity depth, m · γP₀ = 1.42×10⁵ Pa empty, 1.013×10⁵ Pa with fibrous fill

two leaves with air between them bounce together at one frequency, set by how light the leaves are and how thin the air spring is.

For a ⅝ in leaf against a double ⅝ in leaf on a 3½ in empty cavity, f₀ = 75 Hz (63 Hz with the cavity filled). The wall behaves differently in three regions (Figure 3):

• Below f₀ the leaves and the air move together, and the wall follows the mass law for the combined weight. The cavity buys nothing that laminating the boards face to face would not.
• Near f₀ the leaves move out of phase and the air pumps energy across. Loss falls below the combined-mass value, typically by 10–15 dB with a filled cavity and more with an empty one; a double wall at its resonance is worse than the same weight in one panel.
• Above √2·f₀ the leaves act independently and loss rises at about 18 dB per octave, until the cavity cutoff (c/2πd, 614 Hz here) where standing waves cut the slope to about 12 dB per octave.

A conventional 2×4 wall resonates at 63–75 Hz, inside the traffic band, with its useful region almost entirely above it. That is how a wall rated STC 50 for speech can be transparent to a passing truck. The same √2 crossover reappears in machine isolation (Part 6), because both are mass–spring–mass systems.

Cavity depth is the strongest lever on f₀

Informs D2

Table 3 and Figure 4 compare the three levers.

Fibrous fill does three things at once: it lowers f₀ by about 16%, damps the resonance dip, and absorbs standing waves above the cutoff, together worth 5–8 points in a decoupled wall. Any fibrous material filling at least 80% of the cavity behaves about the same; density is not the governing variable. Fill selection belongs to Insulation · Cavity fill; the acoustic requirement is only "fibrous, filling the cavity, not compressed."

Above about 200 Hz, the connection between leaves sets the result

Informs D2 D4

A stud screwed to both faces is not a layer that sound passes through; it is a rigid short across the air spring. Along the stud line the leaves move together as one panel of their combined weight, while each leaf keeps its own coincidence dip. The stud line acts as a narrow strip of low loss beside the decoupled field, and the two combine by area like a window in a wall (Part 3).

τ transmission coefficient, 10^(−TL/10) · φ stud area fraction, 1½ in ÷ 16 in ≈ 9%

the stud strip passes sound like a single heavy panel, and its share of the wall area decides how much of the cavity's benefit survives.

The stud strip still passes far less than an open path: in this model the rigidly connected wall sits about 10 dB above the combined mass law at 500 Hz, set by the stud fraction. The fraction used is the infinite-wall value; real wall lengths with end studs give slightly more stud area and about 0.5–1 dB less loss.

Resilient connections soften the short. Their mount resonance, about 100–200 Hz for resilient channel and 20–40 Hz for isolation clips, works like a machine mount: above √2 times the mount frequency the stud path is attenuated; below it there is no benefit and possibly some amplification (Figure 5). At 500 Hz the model gives 47 dB for rigid studs, 63 with resilient channel and 65 with clips, against 65 for an ideal unconnected wall. Below about 200 Hz all three give the same result in this model (28 dB at 125 Hz), because the air path through the cavity governs there; low-frequency gains come from cavity depth and fill, not from the connection.

A third leaf adds a second resonance

Furring over an intact existing wall makes three leaves: the two original faces and the new board. The new cavity is only the furring standoff, so its resonance is high (about 103 Hz for clips with a 1.6 in filled standoff, two ⅝ in layers over ½ in gypsum), not the low value of the full stud bay. Clips over an open bay with the old board removed give a cavity of stud depth plus standoff. The same hardware therefore gives very different results in the two cases.

With three leaves, equal cavities put both resonances at one frequency and deepen the dip; unequal cavities spread it into two shallower dips. For three-or-more-leaf assemblies, the ratio of the cavity depths affects the result more than leaf count does. Triple and secondary glazing apply this (Windows & Doors · Secondary glazing).

3 · Paths: composite, leaks, flanking and ducts

Bottom line. Sound reaches a room by every path at once and the energies add, so the result is set by the worst path weighted by its area; leaks, side paths and ducts routinely govern.

Paths combine by energy, weighted by area

Informs D3

Sᵢ area of element i · τᵢ its transmission coefficient

convert each element's loss to the fraction of sound it lets through, average the fractions by area, and convert back.

Reference case: a 100 sf exterior wall with 15 sf of glass (the 15% facade glass fraction), wall STC 39, window STC 28 (the existing class). The composite is 34.6 dB and the window, at 15% of the area, carries 69% of the transmitted sound (Figure 6).

The ranking holds while the window is materially weaker than the wall, which covers the existing glazing class (STC 26–28) against any typical stucco wall (STC 36–42). It inverts only once the glazing is near the wall's rating. Single-number composites approximate the band-by-band calculation; at low frequency the window's share is usually larger still.

An unsealed opening has a transmission coefficient of one

Informs D4

Table 4 and Figure 7 apply the composite equation to a single unsealed opening.

In descending order of how often they are missed: perimeter joints at plates and board edges (a non-hardening acoustic sealant; joint compound cures rigid and cracks); electrical boxes, especially back to back in one stud bay; door undercuts; recessed lights in a decoupled ceiling; the partition head under trusses; and envelope terminations such as ERV hoods and dryer vents on a treated facade.

A flanking path caps the result regardless of the wall

Informs D4

the sound arriving by side paths adds to the sound through the wall, so the room hears whichever is larger.

Table 5 shows how one flanking path caps a good wall.

In a two-story wood-frame house several flanking paths exist whatever wall is chosen (Figure 8): the continuous second-floor deck, which is also the ground-floor ceiling; the ceiling plane over a partition; the partition-head gap under trusses; shared ducts; plumbing chases; and back-to-back boxes. Every adjacency needs each of these paths listed and checked against the wall it bypasses.

Where a partition head moves under trusses, the joint is a sound path

PartitionsExterior envelopeFloor–ceilings

Informs D4

Truss bottom chords rise in winter and settle in summer, so partitions under them connect through slip clips and the ceiling board is held back from the wall (Framing · Truss uplift). Two details are used. In the floating-corner detail the ceiling board is fastened to the partition plate and left unfastened to the trusses for 16–24 in, so the corner flexes and no slot forms unless it cracks. Where the board is held back from the wall instead, the head is a slot along every partition that crosses the trusses, and it must stay movable; the table below applies to that case.

Table 6 applies the leak arithmetic to the head gap.

The requirement looks contradictory: the joint must move (structure) and be airtight (acoustics). It is resolvable because airtightness and stiffness are separate properties. A seal needs to be continuous, not rigid: non-hardening sealant in a joint sized for the movement, sealant over backer rod, or a gasket held in compression, with the slip clips carrying no acoustic duty (Figure 9). Rigid tape, joint compound, trimmed foam, or board screwed to both partition and chord all crack and reopen. The joint is sized at framing, not caulked at the end.

A shared duct connects every room on it

Informs D4

Two terminals on a short common branch can pass sound with an effective loss of only 15–25 dB, far below any partition between the same rooms. A 6 in terminal is a 28 in² path in an STC 55 wall of 100 sf.

Table 7 shows the effect of one duct path.

The mitigations are decided when the duct layout is drawn, listed from the layout outward: a dedicated branch per room from a manifold rather than a daisy-chained trunk; the longest practical run between terminals that share a branch; lined or flexible final runs; and in-line attenuators on branches serving rooms to be isolated. Terminal velocity above about 2.5 m/s generates its own noise. A ducted ERV moves far less air than forced-air heating, so its paths are smaller, but not zero (HVAC · Ventilation).

4 · Exterior noise: from source to bedroom

Bottom line. Exterior noise reaching a room is the facade level, less the composite loss of the facade, adjusted for the room; for intermittent sources the event maximum, not the average, decides sleep.

Steady traffic decays as a line; a horn decays as a point

PartitionsExterior envelopeFloor–ceilings

d distance · d₀ reference distance

a single source loses 6 dB each time the distance doubles; a steady stream of traffic loses only 3 dB.

Table 8 and Figure 10 compare the two decay rates.

Source geometry matters as much as distance. A horn sounded along an approach to a crossing has its closest approach at the perpendicular distance to the track, which can be much shorter than the distance to the crossing. Federal rules require locomotive horns of 96–110 dB at 100 ft (49 CFR 229.129) and sound them on approach to public at-grade crossings unless a quiet zone is in effect. A measured or published level at the facade includes about +3 dB of facade reflection; it is counted once, either in the measured level or added, never both.

Typical spectral peaks: free-flow road traffic 63–125 Hz (tire roar above about 30 mph); stop-start traffic 50–100 Hz; passenger rail 50–250 Hz; horns 250–2,000 Hz and tonal. Any exterior prediction of consequence uses a measured facade spectrum, with third-octave data down to at least 50 Hz, since an A-weighted single number or a phone app says little below 125 Hz.

Facade treatment follows the composite: glazing first

PartitionsExterior envelopeFloor–ceilings

Informs D3

Table 9 compares the treatments available from the inside and through the existing openings.

Room correction and its low-frequency limit

Informs D1

S transmitting area · A room absorption, sabins · α absorption coefficient

a room with more absorption than transmitting area gains a little quiet; a hard room loses a little.

For a 12 × 12 × 8 ft furnished bedroom (ᾱ 0.15), S = 96 sf and A = 101 sabins, so the correction term 10·log₁₀(S/A) is -0.2 dB and NR ≈ TL. Absorption inside a room controls reverberation, not transmission; acoustic panels do not stop sound passing through a wall.

The correction assumes a diffuse field, which holds above the Schroeder frequency, f_S = 2,000·√(RT/V) with RT in seconds and V in m³. For the same bedroom (RT ≈ 0.56 s) f_S ≈ 262 Hz. Below it the field is dominated by room modes and level varies 10–20 dB with position, so the whole traffic band is in the region where a single room level is only indicative.

An average level and an event maximum are different design cases

PartitionsExterior envelopeFloor–ceilings

Informs D1 D3

Table 10 and Figure 11 compare the average and the event case for the reference facade.

Every row meets the 45 dB CNEL interior level that California general plans and CBC 1206 apply to new residential buildings, including the existing facade. CNEL is a 24-hour energy average with +5 dB for 7–10 pm and +10 dB for 10 pm–7 am. In the reference case, 40 horn events a day of 5 s effective duration add 67 dB CNEL to the 70 from traffic, a total of 71.7: the average moves by under 2 dB while each event reaches 52 dB indoors.

Under a vented attic, the ceiling is the boundary

PartitionsExterior envelopeFloor–ceilings

With a vented attic the roof covering is not a leaf of the room's boundary: the attic is a large ventilated volume, not a sealed cavity, and is not modeled as a mass–air–mass system. The ceiling assembly and its airtightness are the boundary for the upstairs rooms, and for elevated or distant sources the roof path can rival the walls. Deep attic insulation and a sealed ceiling plane serve both heat and sound (Insulation · Attic strategies); the partition-head gaps are holes in that plane.

Rain noise is a structure-borne excitation of the outer skin and depends on its mass and damping. Concrete tile on battens is not a significant rain source; thin metal roofing and skylights over a room are.

5 · Interior isolation: partitions, doors and targets

Bottom line. Interior isolation is decided by the connection between the leaves, the cavity depth and the weakest path through the wall, usually the door.

Partition assemblies compared

PartitionsExterior envelopeFloor–ceilings

Informs D2

Table 11 and Figure 12 compare the common partitions.

Only the double-stud wall puts its resonance near the bottom of the traffic band (about 34 Hz filled, so decoupling works from about 48 Hz up), and its cavity doubles as a chase that needs no notching. Its cost is floor area: 10½ in across a 12 ft wall is about 6 sf. A wall that cannot be rebuilt (a braced panel, a wall under a girder truss) is limited to mass, damping and furring on its faces (Framing · Which walls are structural).

Other options in brief: mass-loaded vinyl is a limp mass, costlier per pound than a second gypsum layer; "sound" gypsum panels are option B with damping pre-applied; steel studs are inherently somewhat resilient and suit non-bearing walls, but change fastening and blocking throughout.

A door is the weakest path in most partitions

PartitionsExterior envelopeFloor–ceilings

Informs D4

Table 12 shows the door's effect on a good partition.

An STC 55 wall with an unsealed hollow-core door performs in the mid-20s; a solid-core door with full perimeter seals and a drop seal raises the same wall by about 14 dB. Jamb depth and hardware are fixed at framing.

What a given rating sounds like

PartitionsExterior envelopeFloor–ceilings

Informs D1

Table 13 maps ratings to what is heard.

Indicative steady background targets, for orientation only: bedrooms NC 25–30, living rooms NC 30–35, media rooms NC 20–25, kitchens and baths NC 35–40; for sleep, single events at or below about 45 dB L_max. NC and RNC curves differ at the same number and are not interchangeable.

6 · Impact and structure-borne noise

Bottom line. Force injected into the frame travels through wood with little loss and is radiated by every connected surface; it is reduced at the source by isolation, and transmission loss does not apply.

Impact is rated by the level produced below, not by a loss

PartitionsExterior envelopeFloor–ceilings

Informs D5

Table 14 gives typical ratings for wood floor–ceilings.

Carpet's divergence, almost no airborne effect and a large impact effect, is the clearest evidence that the two mechanisms are different. IIC uses a tapping machine and covers 100–3,150 Hz; heavy footfall and dropped objects put energy below 100 Hz, so an assembly rated IIC 55 can still pass a thump.

Rotating machines force the frame below every rating band

PartitionsExterior envelopeFloor–ceilings

Informs D5

a machine shakes the floor at its turning speed in cycles per second, and at multiples of it.

A washer spinning at 1,000–1,400 rpm forces the floor at 16.7–23.3 Hz; direct-drive fans at 15–30 Hz; pumps near 29 Hz; compressors 29–58 Hz, and inverter-driven compressors across a wide range. These fundamentals sit below the IIC and STC ranges, so neither rating speaks to them. The force travels as bending waves through joists and studs and is radiated most efficiently by large, light surfaces such as the gypsum ceiling below (Figure 13).

Mount deflection sets isolation; below √2 the mount amplifies

PartitionsExterior envelopeFloor–ceilings

Informs D5

δ static deflection of the mount under load, in · fₙ mount natural frequency, Hz · T fraction of force transmitted

a softer mount (more deflection) has a lower natural frequency; it isolates only when the machine runs well above it, and amplifies when the two are close.

Table 15 and Figure 14 give the isolation of common mounts at three spin speeds.

The model assumes a rigid base. On a wood floor, whose own natural frequency is often 15–25 Hz, the isolation holds only where the floor deflects little compared with the mount; a soft floor erodes it.

The mechanisms for a machine over a quiet room, ranked by how directly each acts on the cause: moving the machine off a sensitive room (total, and available only while the layout is open); isolating it with mounts of at least ¼ in deflection, the machine level and touching nothing (about 15–20 dB); breaking the path with a floated floor or a decoupled ceiling below; stiffening the floor to move its own resonance away from the forcing; and adding mass to the ceiling below, which acts only on radiation. This ranks mechanisms by directness, not a sequence of work.

Rigid connections bypass the mounts

PartitionsExterior envelopeFloor–ceilings

Informs D5

A mount works only if it is the sole connection. Rigid supply and drain lines, a machine touching cabinetry or the wall, two machines touching each other, and rigid conduit each carry force around it. Flexible supply connectors and a resiliently supported drain are inexpensive; built-in casework that touches the machine re-couples it.

Floor build-ups for isolation compete with radiant floors for dead load

PartitionsExterior envelopeFloor–ceilings

Informs D5

A floated floor and a radiant-floor topping both add dead load to the second-floor framing, and both use the same build-up depth. A 1½ in gypsum-concrete topping can move a beam up a size (Framing · Added floor dead load). A decoupled ceiling below a machine room cannot also be a thermally coupled radiant ceiling, so the choice of ceiling below fixes the radiant method in that zone (HVAC · Radiant delivery by floor). Recessed lights in a decoupled ceiling are holes in the leaf plus rigid penetrations (Lighting · Recessed fixtures).

7 · Code: what binds and what does not

Bottom line. No acoustic standard binds an interior remodel of a detached house; structural, fire and energy rules still limit what can be built, and the work is inspected for those.

No acoustic standard binds an interior remodel of a detached house

Every acoustic requirement in the California codes is scoped to separations between dwelling units, or to nonresidential buildings (Figure 15). Within one detached house there is no code STC, IIC or interior level for a plan checker to enforce.

Table 16 lists each provision and why it does or does not reach this work.

With no inspection to fail there is also no backstop. A structurally deficient beam is caught at plan check; an acoustically inert wall is found after the drywall is finished, when the only remedy is demolition. Numerical targets per adjacency (Part 5) take the place of the missing standard.

Structural, fire and energy rules constrain acoustic assemblies

Table 17 collects the provisions that limit what may be built.

Rebuilding a partition is framing work and needs a building permit; inspection takes place at framing (including fire-blocking in double-stud and staggered walls), at insulation where required, and at drywall nailing where a rated assembly is involved.

Where it goes wrong

1. Treating three mechanisms as one. Mass on a ceiling for a spin cycle, or a rebuilt wall behind an old window, fails because the treatment does not act on the mechanism (Table 1, Part 3).
2. Checking the wall and not the paths. An STC 60 wall with a hollow-core door performs near 25; an ERV branch or an unsealed box does the same (Part 3).
3. Leaving the truss-head gap open, or sealing it rigidly. Open, it concedes about 20 dB; rigid, it cracks within a season (Part 3).
4. Using resilient channel for low-frequency problems. It gives no isolation below about 210 Hz, and one long screw shorts it (Part 2).
5. Buying mass for a low-frequency problem. Mass gains 6 dB per doubling at every frequency and leaves the f₀ dip where it was (Part 2).
6. Decoupling under the sheathing of a braced panel, which destroys the shear connection (Part 7).
7. Fire-blocking across both leaves of a double wall: it fails the acoustics or the inspection (Part 7).
8. Stiff rubber pads under a machine. At 0.05 in deflection a pad does almost nothing at 20 Hz and can amplify at lower speeds; useful isolation starts near ¼ in (Part 6).
9. Machines touching casework, walls or each other, or rigidly plumbed (Part 6).
10. Specifying windows on STC alone. For traffic, STC flatters by 5–10 points (Part 1); symmetric glazing shares one coincidence dip (Part 2).
11. Judging a facade by CNEL alone where intermittent sources wake people (Part 4).
12. Acoustic panels bought to stop transmission; they change reverberation in the room (Part 4).
13. Clipping over intact drywall and expecting the stud-bay f₀; the cavity is only the standoff (Part 2).

Reference

Glossary

Leaf

One independent panel of an assembly. Studs fastened to both faces make a rigidly coupled double leaf, which behaves closer to one panel.

Decoupling

Separating two leaves so vibration in one is not conducted into the other; changes the slope of loss with frequency.

f₀ (mass–air–mass resonance)

The frequency at which two leaves bounce on their air spring; the benefit of a cavity begins above √2·f₀.

Coincidence (critical) frequency

Where a panel's bending wave matches the sound wavelength in air and loss dips.

Constrained-layer damping

A viscoelastic layer between two boards that turns bending energy into heat.

Flanking

Any path that bypasses the assembly being rated.

τ (transmission coefficient)

The fraction of incident sound energy transmitted, 10^(−TL/10); the quantity that adds across paths.

Transmissibility T

For a mount, the fraction of force passed to the structure.

Static deflection δ

How far a mount compresses under its operating load; sets its natural frequency.

Schroeder frequency

Below it a room's sound field is dominated by modes and a single room level is not meaningful.

CNEL

Community Noise Equivalent Level: 24-hour A-weighted average with +5 dB evening and +10 dB night penalties.

L_max, L_eq

The maximum level during an event, and the energy-average level over a period.

NC, RNC

Noise criteria curves for steady background sound (ANSI/ASA S12.2); not interchangeable at the same number.

Sabin

One square foot of perfect absorption; room absorption A = Σ(S·α).

STC, OITC, IIC, NNIC, NISR

Single-number ratings; see the ratings table in Part 1.

Formula sheet

Figure color legend

Red: deficient, or fails the stated criterionAmber: marginal, or a caveated bandGreen: meets the criterionGray: existing or baselineTeal: neutral quantity, reference line or annotation; never a grade

Line charts of transmission loss use a teal family for the reference curves and red, amber and green only where a curve carries a grade, as stated in each caption.

Code points and inputs to confirm

• Whether the local jurisdiction adopts CRC Appendix BG
• CALGreen 5.507.4 values in the adopted 2025 text
• Typical STC and IIC ranges in this guide are published typical values, not test data for a specified assembly; the OITC reference spectrum is from ASTM E1332-10a (the current edition is E1332-22)

Sources

• Soundproofing Calculation Reference & Specification v1.7 (2026-09-26), the computational companion to this guide: §§12–17, 21, 25–27, 31–33.
• 2025 California Building Code §1206 and 2025 California Residential Code Appendix BG, via GoCodeBook.
• CALGreen §5.507.4, as quoted in City of Morgan Hill and City of Santa Cruz acoustical reports.
• 49 CFR 229.129 (locomotive horn levels) and 49 CFR 222 (horn use and quiet zones).
• ASTM E90, E336, E413, E492, E989, E1007, E1332; ANSI/ASA S12.2-2019.
• American Wood Council TR-15; NRC Canada IRC-IR-761 (gypsum wall transmission loss); WHO night-noise guidance (L_max about 45 dB indoors).
• Companion guides: Framing, Windows & Exterior Doors (set v2); Insulation and HVAC (set v2); Plumbing (set v2); Lighting (set v2).

Set v2 draft 1, 2026-10-23: rewritten to spec set v2 from the v1 guide (archived) and calculation reference v1.7, revised after review; values from one script. 2026-10-23: Insulation and HVAC references now link to their set v2 guides. Draft 1a, 2026-10-23: Plumbing and Lighting links repointed to the v2 guides.

↑ Back to top