What Impact Noise Is, and Why It Needs Its Own Rating

Why this matters

Airborne sound has to get into a wall or floor before it can get through it, and most of it never does. An impact skips that step entirely: a heel, a dropped object or a rolling chair puts energy straight into the structure, and the assembly then radiates it into the space below as sound. That difference is why impact performance cannot be described by a sound pressure level at the source, why the rating had to standardize the source instead of the receiver, and why an assembly can gain 20 or more rating points from a soft finish while the resident below still reports a thump. The rating is a good tool used at the wrong end of the spectrum more often than it is used correctly.

Why no level at the source describes an impact

A machine has a sound power level, referenced to 1 picowatt, that belongs to the machine and travels with it. A footstep has nothing equivalent. What arrives below depends on the shoe, the floor finish, the mass of the person, the structural system, the span, the ceiling below and how the two are connected. Change any one and the same footstep produces a different result.

That is why impact test methods do not measure a source and predict a receiver. They fix the source by standard, measure what shows up in the room below, and rate the assembly on that. The comparison is only valid between assemblies tested with the same standardized source.

What the rating standardizes

The standardized source is a tapping machine: five small steel hammers lifted and dropped in sequence at a fixed rate, delivering 10 impacts per second onto the finished floor surface. Laboratory measurement is under ASTM E492 or ISO 10140-3; the single-number classification is ASTM E989 for Impact Insulation Class or ISO 717-2 for the weighted normalized impact sound pressure level. Field measurement is a separate method, ASTM E1007 or ISO 16283-2, and produces a separate field number.

Under ASTM E989 the classification fits a reference contour to the measured one-third-octave band levels, subject to deficiency rules that cap how far any single band may fall short and how much total shortfall is allowed, and the rating is 110 minus the fitted contour value at 500 Hz. That single relationship is worth carrying, because it is what lets you sanity-check a rating against a band table instead of taking it on faith.

Keep the imperatives around the machine in view: the hammers are a pinch point, so keep hands out of the hammer path and let the mechanism come to rest before lifting or repositioning the machine, and where a tapping run puts the operator at or above the 85 dB A-weighted eight-hour time-weighted average action level in 29 CFR 1910.95(c), hearing protection and the employer's hearing conservation program apply.

What the rating deliberately leaves out

This is the part that decides whether a number helps you.

Everything below 100 Hz. The ASTM E989 classification is built on one-third-octave bands from 100 to 3150 Hz. A heel strike on a lightweight framed floor puts substantial energy below that, and so does a dropped object. The rating is silent in exactly the region where most real complaints in wood-framed construction live. ISO 717-2 acknowledges the same gap by publishing a spectrum adaptation term and a low-frequency extended range alongside the base rating, which is the standards world admitting the base number is incomplete.

The character of a real footfall. The tapping machine is a hard, light, high-rate source and it is rich in mid and high frequency. A bare heel, a heel in a sock, a child landing off a couch and a rolling chair are all different from it and from each other. The machine was chosen for repeatability, not for realism, and it is honest about that.

Flanking. A laboratory rating is measured in a facility built to suppress every path except through the assembly. The field never reproduces that, which is why a field number exists as a separate quantity under a separate standard rather than as the lab number with a penalty.

The receiving room's own absorption. Field and laboratory impact levels are normalized to a reference receiving room condition so that a furnished room and a bare one produce comparable numbers. That normalization is part of the number; a raw sound pressure level measured in the room below is not a rating and cannot be compared to one.

Anything about airborne performance. A floor's airborne rating and its impact rating are independent numbers from independent tests, and a sibling article covers what happens when a design move raises one and lowers the other.

Which direction the number runs

Two conventions are in live use and they run opposite ways, so name the one you are quoting in the same clause as the number, every time:

• Impact Insulation Class rises as performance improves. Higher is better. It is 110 minus a level, so it inverts.
• Normalized impact sound pressure level falls as performance improves. Lower is better. It is a level in the room below.

A specification that says "impact rating of at least 50" and a report that says "weighted normalized impact sound pressure level of 50 dB" are not agreeing with each other. They are describing very different floors.

Worked example: the assembly that rated well and still generated a complaint

A two-storey building, concrete floor slab with a lightweight topping, hard finish, suspended ceiling below.

The lab number. The assembly's ASTM E492 report shows a fitted reference contour value of 58 dB at the 500 Hz one-third-octave band, so the ASTM E989 classification is 110 minus 58, which is Impact Insulation Class 52. The specification called for a minimum of 50, so the submittal was approved.

Correction printed, mounting and flanking. That 52 already contains the laboratory's own mounting with flanking suppressed. It is not a prediction of the building. The building number has to come from a field measurement under ASTM E1007, and that measurement is a different quantity, not the lab number with a deduction. The field test on this floor returned a field impact class of 45.

Correction printed, normalization. Both the 52 and the 45 are normalized to a reference receiving room condition. The raw one-third-octave levels measured in the unit below were higher than the reported numbers before normalization, and quoting a raw level as if it were a rating is the most common way a dispute gets started with two correct numbers on the table.

Correction printed, band limits. Both figures describe 100 through 3150 Hz. The complaint from the unit below was a thump, described as felt as much as heard, arriving with each footfall in the hallway above. A supplementary measurement at the 50 and 63 Hz one-third-octave bands, outside the rated range, showed the energy that occupants were reacting to. It cannot appear in either rating by construction.

The finish change, and what it bought. Carpet with a pad was installed in the hallway. Published assembly tests commonly show carpet with pad adding 20 or more points of Impact Insulation Class over the same structural floor with a hard finish, and route the specific number to the tested assembly's own ASTM E492 report rather than assuming it. The retest confirmed a large gain in the rated range.

And what it did not buy. The complaint persisted, reduced but present. The soft finish reduces the high-frequency content of the impact, which is where the rated bands are, and it barely changes the low-frequency force the footfall delivers into the structure. The rating moved a long way, the occupant's experience moved a little, and both observations are true. Anyone who had promised the resident an outcome proportional to the rating gain had just spent their credibility.

What actually would have moved it. The low-frequency path is structural, so the fix is structural: a discontinuity between the walking surface and the structure below, or between the structure and the ceiling. That is the subject of the resilient layer article and it is not a finish decision.

Sibling-rule check. Every rating above carries its test method and its rated band range, and the lab and field numbers are named as separate quantities under separate standards rather than one derived from the other. Both direction conventions are stated in the same clause as their numbers, and the higher-is-better convention is not silently switched in any heading. No level appears without quantity, weighting, reference, bandwidth and time basis, and where a raw level is discussed it is explicitly disqualified as a rating. The finish gain is routed to the tested assembly's own report rather than given as an invented figure, and the residual complaint is reported as a partial success rather than narrated as a pass.

How to specify so this does not happen to you

Name the standard, the edition and whether the number is laboratory or field. "Impact Insulation Class of not less than 50 under ASTM E492 and ASTM E989" and "field impact class of not less than 45 under ASTM E1007" are two different requirements and a project can carry both.

Ask for the band data, not just the rating. The one-third-octave table is what tells you whether the assembly is uniformly good or is passing on a fitted contour with a deficiency parked in the band you care about.

Where low-frequency footfall is the known risk, say so explicitly. Lightweight framed floors and long spans are where it bites. The rating will not catch it, so either extend the requirement to the lower bands, using the ISO 717-2 spectrum adaptation term or an explicitly stated extended range, or accept the risk in writing before the floor is built.

Check what the code actually adopts. Impact requirements for dwelling unit separations come from the model building code as adopted and amended by the authority having jurisdiction, which is a named role with authority rather than a synonym for the inspector who shows up. The edition the jurisdiction adopted decides which test methods and which minimum ratings are enforceable on your job, and the specification may exceed them by contract.

References

• ASTM E492 for laboratory impact sound transmission using a standardized tapping machine, ASTM E989 for the Impact Insulation Class classification, and ASTM E1007 for field impact measurement; the tested assembly's own report in the edition tested owns the number.
• ISO 10140-3, ISO 16283-2 and ISO 717-2 for the international counterparts, including the spectrum adaptation term and the low-frequency extended range.
• The model building code as adopted and amended by the authority having jurisdiction, which sets whether an impact rating is enforceable on a given occupancy and in which edition.
• 29 CFR 1910.95 for the occupational noise action level and hearing conservation program during tapping machine runs.
• See related: Why a Floor That Blocks Airborne Sound Can Fail on Footfall; What a Resilient Layer Is Actually Doing.