Why a Floor That Blocks Airborne Sound Can Fail on Footfall
Why this matters
A floor upgrade that measurably improved airborne performance and measurably wrecked impact performance is not a contradiction, a bad install or a measurement error. The two ratings come from different mechanisms and nothing makes them move together. In the building below, one weekend of work bought 5 points of field airborne performance and cost 12 points of field impact performance, and the specification that authorised it named only the airborne number. If you only specify one, you have only bought one, and the tenant who calls is usually calling about the other.
The upgrade that worked, and the complaint that replaced it
A converted mid-rise, offices below, apartments above. The original complaint was speech: the office below could make out conversation and television from the unit overhead. Field airborne performance measured to ASTM E336 came in at 48.
Three things were done in one weekend.
- A 1.5 in normal-weight concrete topping poured over the existing 4 in slab.
- The carpet replaced with a hard finish, which was a design decision made by the owner and not by anyone thinking about sound.
- A second layer of 5/8 in gypsum board added to the ceiling below, screwed through the existing resilient channel, with longer screws than the first layer used.
Retest: field airborne performance 53, up 5 points. The speech complaint was gone and stayed gone.
Six weeks later the office below filed a new complaint about footfall, describing it as a thud with every step near the window wall. Field impact performance under ASTM E1007 measured 39, against 51 on the same assembly before the work.
What airborne performance is actually made of
For a single limp panel in the frequency region between its stiffness-controlled low end and its coincidence dip, transmission loss rises about 6 dB per doubling of surface mass, holding frequency constant, and about 6 dB per doubling of frequency, holding surface mass constant. That is the mass-law prediction and it is an upper bound; measured field-incidence data in that region commonly comes in nearer 5 dB per doubling, which is the figure used below.
That is a punishing relationship. Mass is the most expensive decibel in the business.
- Existing slab: 4 in of normal-weight concrete at about 150 lb per cubic foot is 12.5 lb per square foot per inch, so 50 lb per square foot.
- With the topping: 5.5 in, so 68.75 lb per square foot.
- Ratio 1.375. The base-2 logarithm of 1.375 is 0.459, so at 5 dB per doubling of surface mass at field incidence, in the mass-controlled region, holding frequency constant, the topping is worth about 2.3 dB.
The measurement gained 5 points, so roughly 2 of those belong to the topping and the rest belong to the added ceiling leaf. Three changes went in at once, so that split is an estimate from the mass law rather than an attribution the measurement can support. Doing them one at a time would have bought the attribution and cost another two site visits, which is a real trade and worth naming out loud when a client asks why you cannot say which change did what.
What impact performance is made of
An impact does not have to get into the floor. It is already in it. What decides the outcome is how much energy the strike injects into the structure, how efficiently the assembly radiates it into the room below, and whether there is a discontinuity anywhere along that path to interrupt it.
Mass helps only weakly here. Doubling the surface mass raises the impedance the hammer sees, which reduces injected energy somewhat, but it also gives you a heavier radiating plate, and neither effect is close to the 5 dB per doubling that airborne gets. What actually moves impact performance is a break in the path: a soft finish at the top, a resilient layer inside the build-up, or a resilient suspension at the ceiling. A sibling article covers what a resilient layer is doing and how it gets defeated.
Why the two do not track
This table is the spine worth keeping.
| Design move | Effect on airborne | Effect on impact |
|---|---|---|
| Add mass to the structural floor | Up, about 5 dB per doubling of surface mass at field incidence, in the mass-controlled region, frequency held constant | Small gain at best; low-frequency footfall largely unchanged |
| Replace a soft finish with a hard one | No meaningful change | Large loss in the rated bands |
| Add a second board layer to the ceiling leaf | Up | Up only if the resilient path stays intact |
| Fasten the ceiling through to the framing | Small change either way | Large loss |
| Seal a penetration or a perimeter gap | Up, sometimes by a lot | Little to none |
| Add a resilient layer under the topping | Up modestly | Up, and it reaches the lower bands |
Two rows in that table are what happened here. The finish change cost impact performance and did nothing for airborne. The long screws reaching framing cost impact performance and did very little either way for airborne, because a stiff path in parallel with a resilient one sets the floor on what the resilient element can achieve, which the rigid-connection article covers in full.
The correction, and the arithmetic behind it
Correction printed, band range. Both impact figures describe 100 through 3150 Hz one-third-octave bands under the classification method. The complaint was a thud, and a supplementary look at the 50 and 63 Hz bands, outside the rated range, showed content the ratings cannot represent. The rating drop was real and it was also not the whole story.
Correction printed, laboratory versus field. Nothing above is a laboratory rating. The 48, 53, 51 and 39 are field measurements under ASTM E336 and ASTM E1007, which are separate quantities from the laboratory ASTM E90 and ASTM E492 numbers on any submittal, not the same numbers with a deduction.
What was done. The ceiling fasteners that had reached framing were backed out and replaced with the correct length, which meant overhead work from a portable ladder set per 29 CFR 1910.23, eye protection per 29 CFR 1910.133 against falling grit, and a respirator under a written program meeting 29 CFR 1910.134 for the gypsum dust released by backing screws out of a finished ceiling, since that is an inhalation route and a glove does nothing for it.
A resilient underlayment was then installed under a new topping in the corridor and the two worst rooms, which required coring the existing topping in three places to confirm the build-up. Coring or cutting concrete releases respirable crystalline silica: use a wet method or on-tool extraction and work under the exposure controls of 29 CFR 1910.1053 for general industry, or 29 CFR 1926.1153 including its Table 1 specified controls where the work is construction, and do not dry-cut it.
Retest. Field impact performance came back to 50, one point below the pre-work carpeted value and 11 points above the low point. Field airborne performance measured 55, up 2 more from the resilient layer, because a break in a rigid path helps both quantities even though it only helps one of them a lot.
Sibling-rule check. The mass law carries the variable it holds constant in the same clause, in both directions, and it is stated at field incidence with the mass-controlled region named. Every rating carries its test method and whether it is laboratory or field, and the two are never differenced. The rated band range is stated and the low-frequency content is explicitly placed outside it, matching the impact-rating article rather than contradicting it. The direction conventions agree with the siblings: impact class rises as performance improves, and a soft finish moves the rated bands far more than it moves low-frequency footfall. The 2.3 dB mass law estimate is presented as an estimate against a measurement that cannot separate three simultaneous changes, rather than as an attribution. Rounding on the mass law ratio was carried to one decimal rather than rounded up to a friendlier figure.
What the specification should have said
Both numbers, both methods, both stages. A minimum airborne rating under ASTM E90 and ASTM E413 for the submittal and a field verification under ASTM E336; a minimum impact rating under ASTM E492 and ASTM E989 for the submittal and a field verification under ASTM E1007. Naming the edition on each keeps the acceptance argument short.
A finish clause. Impact ratings are almost always tested with a stated finish. A specification that approves an assembly at a rating achieved with carpet, and then allows a hard finish as a tenant option, has approved two different floors. Say which finish the rating is contingent on, in the same sentence as the rating.
A fastener note on any resilient element. Screw length and pattern are acoustic requirements on a resilient ceiling, not just structural ones, and they belong in the specification text rather than in a detail nobody reads on site.
A check on what the jurisdiction actually enforces. Dwelling unit separation requirements come from the model building code as adopted and amended by the authority having jurisdiction, which is a named role with authority rather than the inspector who arrives. The adopted edition decides which test methods and which minimum values are enforceable; a specification can and often does exceed them by contract, and only one of those two is a legal obligation.
References
- ASTM E90 and ASTM E413 for laboratory airborne transmission loss and its classification; ASTM E336 for field airborne performance between rooms.
- ASTM E492 and ASTM E989 for laboratory impact testing and classification; ASTM E1007 for field impact performance.
- 29 CFR 1910.1053 for respirable crystalline silica in general industry and 29 CFR 1926.1153 including Table 1 for construction; 29 CFR 1910.134 for the written respirator program; 1910.23 for ladders and 1910.133 for eye protection.
- The model building code as adopted and amended by the authority having jurisdiction, which sets enforceable separation ratings and the edition they come from.
- See related: What Impact Noise Is, and Why It Needs Its Own Rating; What a Resilient Layer Is Actually Doing; Why a Rigid Connection Defeats an Isolator.