What Transmission Loss Actually Measures

Why this matters

A partition's published transmission loss is a laboratory result for a specimen sealed into a test opening between two rooms built specifically so that sound cannot get around it. Your building is not that. The number is still useful, but only if you know what it is a property of, because two of the most expensive mistakes in this work come from misreading it: treating a lab number as a prediction of the level difference you will measure on site, and assuming that a wall's rating survives the door, the gap and the plenum that were not in the test opening. This card is about reading the report correctly and then doing the two pieces of arithmetic that turn it into a building.

What the quantity is

Transmission loss starts with a ratio. The transmission coefficient of a surface is the sound power transmitted through it divided by the sound power incident on it. Transmission loss is ten times the base-10 logarithm of one over that coefficient.

So TL 30 dB means a coefficient of 0.001: one part in a thousand of the incident power goes through. TL 45 dB means about three parts in a hundred thousand. Three things follow immediately, and every practical use of TL in this card rests on them.

• It is a power ratio, not a level. It has no reference, no weighting and no time basis, because it is not a level at all. It is a property of the specimen at a stated frequency.
• It is normalized to area. A bigger specimen transmits more power, and TL divides that out, which is what lets a lab result be applied to a wall of a different size.
• Because it is built on a coefficient, surfaces combine by area-weighting the coefficients, not by averaging the decibels. That is the whole of the composite arithmetic below, and it is where holes get their power.

The report, field by field

A transmission loss report has seven fields worth reading before you accept the number on the cover.

Test method and edition. ASTM E90 for the laboratory airborne test in North America, ISO 10140 internationally, in the edition the report states. The method is not a formality: what counts as an acceptable specimen mounting and what flanking limit the laboratory must demonstrate are both defined in it, and both changed across editions.

Specimen description. Every layer, every thickness, every fastener spacing, every cavity fill. Change any of them on site and you no longer have this specimen. Stud spacing and screw pattern are part of the assembly, not installer discretion.

Test opening dimensions. The specimen area the coefficient was normalized against. A large wall built to the same recipe does not automatically match, because the specimen was a specific size with specific edge conditions.

Perimeter and mounting detail. How the specimen was sealed into the opening, and whether the perimeter was resiliently mounted or rigidly built in. This is the field most often skipped and it moves results.

The band data. Transmission loss at each third-octave band, usually 125 through 4000 Hz. This is the actual result. Everything below 125 Hz is outside the standard test range, so the report is silent about the octave most of your complaints live in.

The single-number rating. A contour fitted to the band data by a separate classification standard. It is a summary of the field above it, computed rather than measured, and a sibling card covers what it can and cannot carry.

The flanking statement. The laboratory's declaration that sound reaching the receiving room by paths other than through the specimen was below a stated limit. That declaration is the reason a lab number exists at all, and it is the one thing your building will not reproduce.

What the report cannot contain

   source room  |  partition  |  receiving room
                |             |
  ceiling plenum over the top of the partition
  ============================================
        \                                /
         \---- flanking path: plenum ---/
                |             |
   ------> direct path through partition ---->
                |             |
         /----- flanking path: floor --\
        /                               \
  ============================================
   structural floor slab under the partition

The laboratory measured the middle arrow. Your building offers all three, plus the ones this drawing cannot fit: a continuous window mullion, a shared duct, a back-to-back electrical box, a common ceiling grid. Flanking is why a field level difference almost never reaches the lab number, and the gap is usually a path rather than a material. When you go above a ceiling to look for one, set the ladder on a firm level surface and work from it rather than standing on the ceiling grid, treat everything above the ceiling as containing energized wiring until proven otherwise, and in a building of unknown vintage do not disturb suspect thermal system insulation or surfacing material: presume it contains asbestos until it is sampled. OSHA's asbestos standards are 29 CFR 1910.1001 for general industry and 29 CFR 1926.1101 for construction work including maintenance and renovation, and the presumed-asbestos-containing-material provisions sit in the construction standard.

Worked example: a submittal against a real wall

A partition between a plant office and a shop area. The submittal gives TL 45 dB at 500 Hz for the wall construction, tested to ASTM E90 in the edition printed on the report, with a stated flanking limit. The wall as built is 10 ft by 20 ft, so 200 sq ft, and it contains a 3 ft by 7 ft door, 21 sq ft, whose own report gives TL 26 dB at 500 Hz. Measured with a rule at the threshold, the door is undercut by 0.5 in across its 3 ft width, which is 0.125 sq ft of open gap.

Convert each surface to a coefficient, area-weight, then convert back. All at 500 Hz.

An open gap is taken as a coefficient of 1.0, which is exact enough for a gap large compared with the wavelength and conservative below that.

Composite coefficient: 0.18341 divided by 200 sq ft, which is 0.000917. Composite transmission loss: ten times the logarithm of one over that, which is 30.4 dB. The 45 dB wall is behaving as a 30 dB wall.

Now read the transmitted-power column, because that is the finding:

• Wall: 0.00566 of 0.18341, or 3.1 percent.
• Door leaf: 0.05275, or 28.8 percent.
• Undercut gap: 0.12500, or 68.2 percent. Those three round to 100.1 percent.

That gap is 0.125 sq ft in a 200 sq ft partition, which is 0.0625 percent of the area, and it is carrying over two thirds of the transmitted power. No material on that wall matters until it is closed.

Fix it in the order the arithmetic dictates. Seal the undercut alone and the total drops to 0.05841, giving a composite coefficient of 0.000292 and a composite transmission loss of 35.3 dB, a gain of 4.9 dB for a threshold seal. Then upgrade the door leaf to a rated door with TL 36 dB: the door term falls to 0.00528 and the composite reaches 42.6 dB. A hypothetical perfectly blocking door gives 45.5 dB, marginally above the bare wall figure because only 179 of the 200 sq ft is transmitting at that point, which is a useful sanity check that the arithmetic is behaving.

The step nobody prints: from transmission loss to a level difference. The number you will measure on site is the difference in level between the two rooms, and that is not the same quantity. It is related by the partition area and the receiving room's total absorption: the level difference equals the transmission loss plus ten times the logarithm of the absorption divided by the partition area. This is a re-basing of the level difference into a property of the wall, not an allowance added to it.

Take the receiving office as 12 by 20 ft with a 9 ft ceiling, carpeted, with an acoustic ceiling, giving roughly 200 sabins of total absorption at 500 Hz. Partition area is 200 sq ft, so the logarithm term is zero and the expected level difference equals the composite transmission loss: 30.4 dB. Strip the carpet and put in a hard ceiling, halving the absorption to about 100 sabins, and the same wall now yields a level difference of 30.4 minus 3.0, which is 27.4 dB. The wall did not change. Doubling a receiving room's absorption raises the measured level difference by 3 dB at constant transmission loss and constant partition area, and none of that 3 dB belongs to the partition.

Sibling-rule check. Transmission loss is handled throughout as a coefficient-based power ratio with no reference, weighting or time basis, and it is never summed with levels. Every figure is stated at 500 Hz and no value is carried across bands. The composite arithmetic area-weights coefficients rather than averaging decibels. The room term is named as a re-basing of the level difference, and the comparison between the two office finishes holds both partition area and transmission loss constant, so a corrected figure is not being compared with an uncorrected one. Leaks are treated as a high-frequency-dominant loss, consistent with the low-frequency card.

How to verify you got this right

Two checks, and they catch different failures.

Add your area-weighted column and confirm it equals the total you divided by, and confirm your areas sum to the gross partition area with nothing double-counted. The most common arithmetic slip is leaving the door's area in the wall row, which inflates the wall term and quietly flatters the result.

Then check the field. Measure the level difference band by band with the source room driven, then repeat with the door open. If opening the door changes the receiving level by less than a couple of decibels, the door was never the dominant path and something else is: go looking above the ceiling and under the floor before you buy a better door. If the measured level difference is far below what the composite arithmetic predicts even with everything closed, you have a flanking path, and no product installed on the face of that wall will reach it.

References

• ASTM E90, laboratory measurement of airborne sound transmission loss of building partitions, in the edition stated on the report you are reading; ISO 10140 is the international counterpart
• ASTM E336, field measurement of airborne sound attenuation between rooms, which is the method that produces a field level difference rather than a specimen property
• 29 CFR 1910.1001 and 29 CFR 1926.1101, asbestos in general industry and in construction work including maintenance and renovation, for presumed asbestos-containing material above ceilings
• See related: What an STC Rating Covers and What It Misses; Why Low-Frequency Noise Is the Hard Problem