What an STC Rating Covers and What It Misses

Why this matters

STC is the number that gets specified, quoted, argued about and warranted, and almost nobody who uses it has seen how it is produced. It is not measured. It is a contour fitted to sixteen third-octave transmission loss values from a laboratory test, using two allowance rules that let a partition hide a real weakness and keep its number. Two assemblies carrying the same STC can differ by more than 10 dB in the one band your job depends on, and neither rating is wrong. If you sell partition work, this is the arithmetic behind the number on your proposal, and knowing it is the difference between meeting a specification and solving a problem.

Two documents, not one

The transmission loss data comes from a laboratory test, ASTM E90 in North America or ISO 10140 internationally, in the edition the report states. The rating is computed from that data by a separate classification standard, ASTM E413, also in a stated edition. Neither is a regulation. They bind you because a specification, a contract or a building code as adopted and amended by your authority having jurisdiction points at them, and that authority is a named role with power to interpret and enforce, not a synonym for whichever inspector arrives.

The field counterparts are different documents again. ASTM E336 measures airborne sound attenuation between rooms in an actual building, and the classification applied to field data yields field descriptors, not STC. A field result routinely lands several points below the laboratory rating for the same construction, because flanking, penetrations and workmanship are all present and none of them were in the test opening. A specification saying "STC 50" without saying whether that is a laboratory rating for the assembly or a field-verified requirement for the finished wall is ambiguous in an expensive direction, and that ambiguity gets settled after the ceiling is closed.

The contour and its two allowance rules

The reference contour has three segments across sixteen third-octave bands from 125 to 4000 Hz: it rises 15 dB in 3 dB steps from 125 to 400 Hz, rises a further 5 dB in 1 dB steps from 400 to 1250 Hz, then runs flat from 1250 to 4000 Hz. Expressed against its own 500 Hz value, the contour is:

The contour is raised as high as it will go subject to two rules. The sum of all deficiencies, meaning every amount by which the measured transmission loss falls below the contour, must not exceed 32 dB. And no single band's deficiency may exceed 8 dB. The STC is the contour's value at 500 Hz once it has been raised as far as those two rules permit.

Everything this rating cannot see follows from that construction. It starts at 125 Hz, so the 63 Hz and 31.5 Hz octaves are outside it entirely. It permits one band to sit 8 dB below the curve without penalty. And it is a laboratory result for a specimen with flanking suppressed by design.

Worked artifact: two assemblies, one rating

Two partitions, both tested to ASTM E90, both fitted per ASTM E413. All values are transmission loss in dB, and all sixteen bands are shown.

Assembly A. Total deficiency 27, which is inside the 32 dB allowance with 5 dB of headroom. Largest single deficiency 8, at 125 Hz, exactly at the limit. Try to raise the contour one more decibel and the 125 Hz deficiency becomes 9, which breaks the single-band rule, so the fit stops. STC 50, and the binding constraint is one band at the bottom of the range.

Assembly B. Total deficiency 25, inside the allowance with 7 dB of headroom. Largest single deficiency 8, at 2000 Hz, again exactly at the limit. Raise the contour one decibel and the 2000 Hz deficiency becomes 9 and breaks the same rule. STC 50, and the binding constraint is one band in the middle of the speech range.

Two partitions, the same rating, and look at what the rating averaged away. At 125 Hz, B is 6 dB better than A. At 2000 Hz, A is 14 dB better than B. That 14 dB is the difference between a conversation you cannot follow and one you can, because consonants carry most of the intelligibility in speech and most of their energy sits at 2 kHz and above. For a speech-privacy job, assembly A is the right wall and assembly B is a poor one, and the specification that asked for STC 50 accepted either.

Assembly B's shape is worth naming: a local minimum at 2000 Hz in a curve that is otherwise climbing is the coincidence signature, and the sibling card covers where it comes from and what moves it. The 8 dB single-band allowance is precisely what let that dip survive into a respectable rating.

There is one more thing neither column contains. If the complaint is a rooftop unit at 63 Hz, or a compressor at 40 Hz, no row above describes it. Both walls are unrated below 125 Hz and the difference between them down there could be anything.

Blocking and absorbing are different jobs

The most common and most expensive category error in this subject is buying an absorption product for a transmission problem. Three ratings, three different questions.

• STC answers how much airborne sound is stopped from getting THROUGH an assembly. Decibels, higher is better, from ASTM E90 data classified by ASTM E413.
• NRC answers what fraction of incident sound a surface ABSORBS rather than reflecting. A dimensionless number from 0 to 1, the arithmetic average of the 250, 500, 1000 and 2000 Hz coefficients rounded to the nearest 0.05, from ASTM C423, and it is meaningless without the mounting designation printed beside it, because the same material tested flat against a wall and tested with an airspace behind it gives different numbers.
• IIC answers how well an assembly resists structure-borne impact from a standardised tapping machine, as 110 minus a fitted contour value, so it RISES as performance improves. A different source, a different path, a different standard again.

Hanging absorption panels in a room does not change a partition's transmission loss. What it changes is the reverberant level in the room they are hung in, and that is worth about 3 dB per doubling of the room's total absorption, in the source room and the receiving room alike. That is a real effect and it is modest: doubling the total absorption in an already-furnished room is difficult, and it will not deliver what a partition upgrade delivers. Sell it as what it is, a room-comfort and reverberation control, and do not let it be bought as a wall.

For sources rich in low-frequency energy, particularly transportation and heavy plant, there is a rating designed for the job. OITC, from ASTM E1332, reaches down to 80 Hz and weights the source spectrum differently, and it is the number to ask for when the complaint is a highway or a rooftop chiller rather than a conversation.

What to write in a specification instead

Name the bands, not just the rating. If the complaint is speech privacy, state a minimum transmission loss at 1000, 2000 and 4000 Hz alongside the STC, and assembly B is excluded on the spot. If it is plant noise, state the bands the plant occupies, which will usually mean 63 and 125 Hz and therefore a figure the STC cannot supply.

Say whether the number is a laboratory rating for the assembly or a field-verified requirement for the finished construction, and if it is the latter, say which field method will be used and who pays for the test.

Require the report, not the number. A submittal that quotes a rating without the tested band data, the test method, the edition, the specimen description and the mounting detail has given you a claim rather than evidence, and it is the same claim for both assemblies above.

How to check a rating you have been handed

Take the reported band data and re-run the fit yourself. Write the contour at the claimed STC using the offsets in the table above, subtract band by band, add the deficiencies, and confirm both that the total is at or below 32 dB and that no single band exceeds 8. Then try one decibel higher and confirm that it breaks. If the claimed rating passes but the next one up also passes, the rating was computed conservatively or the data is not the data it was computed from.

While you have the deficiency column open, read where the deficiencies sit rather than what they total. Deficiencies concentrated at the bottom of the range describe a partition that is light or resonant low down. Deficiencies concentrated in one mid or high band describe a coincidence dip. A rating alone tells you neither, and the column takes two minutes.

References

• ASTM E90, laboratory measurement of airborne sound transmission loss, and ASTM E413, classification for rating sound insulation, each in the edition stated on the report you are reading; ISO 10140 and ISO 717 are the international counterparts
• ASTM E336, field measurement of airborne sound attenuation between rooms, which produces field descriptors rather than a laboratory STC
• ASTM C423, sound absorption in a reverberation room, which is the source of NRC and requires a mounting designation to mean anything
• ASTM E1332, classification for rating outdoor-indoor sound attenuation, for transportation and low-frequency-rich sources
• See related: What Transmission Loss Actually Measures; What the Coincidence Dip Is and Why It Lands Where It Does