A good acoustic panel specification describes performance and the evidence behind it, not a product. Ask for absorption as band coefficients or a sound absorption average (SAA), not NRC alone. Require an ASTM C423 report from an accredited laboratory, tested in the ASTM E795 mounting that matches the detail you will build, and fire results for the panel as assembled. Then add what public templates leave out: the ADA protrusion limit, a mock-up, approval samples and extra stock.
This guide works from two public templates, the Department of Veterans Affairs section 09 84 33 and the federal UFGS 09 84 20, and from a real accredited test report read line by line. It is written for architects, interior designers, specifiers and contractors. Fire classification gets its own article, what a Class A fire rating does and does not cover, and the list of documents to collect lives in the acoustic panel spec checklist. Here the focus is on writing the requirements and judging the evidence that comes back.
What public templates ask for, and what they leave out
Both public templates ask for an NRC range “per ASTM C423” and a Class A fire rating for the complete composite panel. Neither names the test mounting, asks for band data or SAA, sets a minimum laboratory edition, or calls for a mock-up or extra stock.
The VA master specification 09 84 33, Sound-Absorbing Wall Units, dated 01-01-21, covers fabric-covered wall panels. It sets thickness “as required to meet the indicated NRC range but not less than 25 mm (1 inch) nominal” and offers two NRC ranges, 0.50 to 0.60 or 0.80 to 0.90. For fire it asks for “Class A, 200 or less smoke density and flame spread less than 25” for the complete composite system. Its reference list still cites C423-17 and E84-20.
The UFGS 09 84 20, Acoustical Wall Panels, issued in August 2026 by the US Army Corps of Engineers, replaced UFGS 09 83 13 from August 2016. It covers fabric-covered, wood veneer, PET, felt and three-dimensional panels, each with bracketed NRC ranges. For felt panels the choices run from 0.30 to 0.75. For fabric panels it asks for Class A for the “complete composite system”; for wood, PET, felt and 3D panels, Class A “when tested as a full composite per ASTM E84,” with flame spread less than 25 in every case. It also asks for certificates of compliance from an independent laboratory accredited under the NIST laboratory accreditation program (NVLAP).
| Topic | VA 09 84 33 (01-01-21) | UFGS 09 84 20 (August 2026) | What to add |
|---|---|---|---|
| Absorption | NRC range 0.50 to 0.60 or 0.80 to 0.90, per ASTM C423 | NRC range per panel type, per ASTM C423 | SAA or band minimums, the E795 mounting, accepted editions |
| Fire | Complete composite: Class A, smoke 200 or less, flame spread less than 25 (E84 or UL 723) | Full composite: Class A, flame spread less than 25, smoke 450 or less (fabric) or less than 450 (other types), per E84 | What the report covered (core, fabric, adhesive, frame); the route for textile faces |
| Lab evidence | Manufacturer literature with acoustical data; fire certificate | Certificates from an NVLAP-accredited independent lab | Full C423 reports, report number and date |
| Samples | Four fabric samples, 18 in, full width of mill run, per color | Three samples per color range, 18 by 24 in minimum | A mock-up in the specified mounting |
| Accessibility | Not addressed | Not addressed | ADA 307.2 projection and 307.4 headroom |
| Extra stock | Not addressed | Not addressed | Quantity per type and color, same production run |
Two details in the fire wording deserve a second look. The IBC’s Class A band runs from a flame spread index of 0 to 25, so a product reporting exactly 25 is Class A under the code but fails a template that says “less than 25”. And the VA template’s smoke limit of 200 is stricter than the 450 in the code’s classes. If you keep either line, keep it on purpose.
Why templates stop where they do
A guide specification is written to be edited. The UFGS opens with a note telling the writer to adapt it “by adding, deleting, or revising text,” and the VA section asks the writer to delete what does not apply. Both assume the writer already knows which test conditions matter. The rest of this article fills in that knowledge.
Why NRC alone is a weak performance requirement
NRC is a four-band average rounded to the nearest 0.05, so two panels with the same NRC can behave very differently in the bass and low-mid range where speech energy and room boom live. Specify SAA or minimum coefficients in named bands, in a named mounting, and treat NRC as a summary.
According to Intertek’s ASTM C423 summary, an accredited test lab, NRC is “the average of the absorption coefficients at 250, 500, 1000 and 2000 hertz, rounded to the nearest multiple .05.” SAA is “the average of the absorption coefficients at the 1/3 octave band frequencies from 200 to 2500 hertz, rounded to the nearest multiple .01.” The report we walk through later goes further: it defines SAA from ASTM C423-17 and describes NRC as “defined from previous versions of ASTM C423.”
What the average hides
The 125 Hz band is not part of NRC or SAA, and the 250 Hz band is one of four equal votes in NRC. A panel can score well on the average while doing little below 500 Hz. Rounding adds blur: any four-band average from 0.775 up to just under 0.825 is reported as 0.80.
NRC also says nothing about sound passing through a wall. That is measured by a different test family and rated as STC, a distinction set out in NRC vs STC.
Example (fictional). Two submittals each claim NRC 0.80 for a 1 in fabric-wrapped panel. Submittal A reports coefficients of 0.30, 0.85, 1.00 and 1.05 at 250, 500, 1,000 and 2,000 Hz. Submittal B reports 0.60, 0.80, 0.85 and 0.95. Both sum to 3.20, and 3.20 divided by 4 is 0.80, so the NRC is identical. At 250 Hz, though, B absorbs twice as much per square foot as A. In a meeting room with a booming male voice on every call, B is the better panel, and only a band requirement would have told the reviewer so. Both products are invented for this example.
The mounting moves the number too
An “NRC range of 0.80 to 0.90” says nothing about how the panel was mounted in the lab. In Armstrong World Industries’ published acoustical summary, one 1 in wall panel family carries a single-number rating of 0.80 when mounted flat (Type A) and 0.90 with a 20 mm space behind it (D-20). The same maker lists a 1 in wood fiber panel at NRC 0.40 flat and 0.80 in a C-20 mounting. A range that ignores mounting can be met on paper by data from a mounting you will never build.
Coefficients above 1.0 are not errors
ASTM C423 explains in section 5.3 that diffraction effects “usually cause the apparent area of a specimen to be greater than its geometrical area,” and that for highly absorptive specimens the coefficients “may exceed unity.” Do not reject a report for values above 1.00. Do keep them in perspective: section 5.4 says the coefficients “should be used with caution” because real installations are larger than test specimens and the sound field is rarely diffuse.
What to write instead
- Name the single number you want and its rounding: “SAA not less than 0.85” is more precise than “NRC 0.85”.
- Add band minimums where the room needs them: for speech rooms, a floor at 250 and 500 Hz; for rooms with amplified music, a floor at 125 Hz as well.
- Tie every number to a mounting: “in ASTM E795 Type A mounting” or the type that matches your detail.
- Ask for the full one-third octave table, not a summary sheet. The NRC calculator recomputes NRC and SAA from the table so a reviewer can check the arithmetic.
How to match the tested mounting to the installed detail
The lab result describes the panel in one mounting, and the specification should require data in the mounting that matches the drawing. ASTM E795 exists because, in its own words, the absorption of a material “depends not only on the physical properties of the material but also on the way in which the material is mounted.”
The scope page for ASTM E795-23 makes three points a specifier can use. The mountings “are intended to simulate in the laboratory conditions that exist in normal use.” Some require fixtures that are “for laboratory testing only” and “should not be specified for practical installations.” And “test reports may refer to these mountings by type designation instead of providing a detailed description.” That last point is why a report can say “Type C-25” and nothing else.
How to read a mounting designation
The letter names the arrangement and the number is a depth in millimeters. The Riverbank report discussed below spells it out: its Type C-25 mounting put the panel on 25 mm (1 in) wood furring “laid directly against the test surface,” producing a 25 mm cavity, which the lab filled with 1 in PET felt. For Type E, another Riverbank report, RAL-A22-451, quotes the definition from E795-16: the suffix is “the distance in millimeters from the exposed face of the test specimen to the test surface, rounded to the nearest integer multiple of 5.” So E-400 puts the face about 400 mm (about 16 in) from the test surface, like a lay-in tile in a suspended ceiling, and one major ceiling maker publishes its tile data in that mounting. Type A is the specimen laid directly against the test surface. Units such as baffles are tested as Type J and reported per unit, a case covered in our article on comparing ceiling clouds and baffles.
Pick the mounting from the drawing
| Installed detail | Mounting data to require | What to watch |
|---|---|---|
| Panel glued or clipped flat to drywall | Type A | Clips that hold the panel off the wall add a small air space; Type A data is then the conservative choice |
| Panel on furring with an empty cavity | Type D with the cavity depth (for example D-20) | Match the depth on the drawing, not the deepest test available |
| Panel on furring with insulation in the cavity | Type C with the cavity depth (for example C-25) | The fill material must match the one tested |
| Ceiling panel or cloud with a deep air space above | Type E with the stated depth, or a unit test | Ask how the area in any coefficient was computed |
| Baffles or other free-hanging units | Type J, reported as absorption per unit | Require the spacing and number of units in the test |
In the published summary above, the spaced mountings of the wall panel and the wood fiber panel absorbed more at 125 and 250 Hz than their flat mountings did. That is the usual reason a supplier offers spaced-mounting data. It is fine to accept it, as long as the drawings show the same air space and the installer builds it.
One product, two mountings
The same panel often appears twice on a project: flat on the walls and hung as a cloud overhead. Those are two installed details and need two data sets. Wall data in Type A says nothing reliable about a cloud with air on both faces. And a cloud figure reported per square foot of face may already include the back face and edges, so it is not directly comparable with a wall panel’s coefficient. Ask for each mounting separately, and give each its own line in the performance requirement.
Writing the requirement
A workable sentence reads: “Submit ASTM C423 test reports for the construction specified, tested in ASTM E795 mounting Type [letter and depth], matching the mounting detail on drawing [sheet number].” Add: “Where no report exists for the specified mounting, submit the nearest tested mounting and state the difference.” The second sentence keeps an honest supplier in the running and makes a gap visible instead of hidden.
Reading an accredited test report line by line
A real report answers six questions: who tested it, to which edition, in what mounting, on what specimen, with what results, and with what limits. The example below is Riverbank Acoustical Laboratories report RAL-A23-086, which its sponsor publishes, for a wood slat panel over PET felt.
We chose it because it is complete and public, not to recommend the product. Open the report PDF alongside this section.
The header and accreditation
The first page states that the lab is accredited by the National Institute of Standards and Technology under NVLAP “as an ISO 17025:2017 Laboratory (NVLAP Lab Code: 100227-0) and for this test procedure.” That final phrase matters. Accreditation is granted for specific procedures, so a lab accredited for one test is not automatically accredited for another.
Method, edition and dates
The test “conformed explicitly with ASTM C423-22,” and the mounting followed ASTM E795-23. It was conducted on April 10, 2023, and the revisions appendix shows the original report issued on April 18, 2023. ASTM’s store now lists C423-23e1 as the active edition, so this report predates the current one. That is normal; the question is whether your specification accepts it.
The specimen
The sponsor named the product, and the lab verified what it could see: a PET felt base layer 25.11 mm (0.9885 in) thick, a face layer of wood slats over 9.21 mm PET felt, and an overall specimen 108 by 96 in. The calculation area was 72 sq ft, the 8 by 9 ft size Intertek describes as the recommended sample for products that cover an extended plane surface. The report adds that “the accuracy of such sponsor-provided information can affect the validity of the test results.”
The mounting
The report describes Type C-25 in full: furring 24 in on center, a 1 in cavity filled with PET felt panels, and perimeter edges “sealed with metal framing and tape.” The sealed perimeter is a laboratory condition. On site, ask whether your panel edges will be exposed, butted or trimmed, and whether the cavity fill matches.
The results table
The table lists 18 one-third octave bands from 100 to 5,000 Hz, each with total absorption in square meters and sabins and a coefficient. The coefficient is the sabins divided by the specimen area. At 500 Hz the report shows 70.44 sabins; 70.44 divided by 72 sq ft is 0.978, which rounds to the 0.98 shown.
The single numbers, checked
The report gives SAA 0.88 and NRC 0.90. You can check both. For NRC, add the four octave bands: 0.55 + 0.98 + 1.06 + 0.92 = 3.51, and 3.51 divided by 4 is 0.8775, which rounds to the nearest 0.05 as 0.90. For SAA, the twelve bands from 200 to 2,500 Hz sum to 10.54, and 10.54 divided by 12 is 0.878, which rounds to the nearest 0.01 as 0.88. The two numbers differ by 0.02 here only because of rounding rules.
Four bands, 630 through 1,250 Hz, exceed 1.00, peaking at 1.10 at 800 Hz. As C423 explains, that is diffraction at the specimen edges, not a mistake.
The appendix and the fine print
Appendix A extends the data from 31.5 Hz to 12.5 kHz and labels it “non-accredited,” “unofficial” and “intended for research & comparison purposes.” Some of its lowest values are negative. Do not let a summary sheet quote those bands as performance.
Every page carries two limits worth copying into your review notes: “The results reported apply only to the specific sample submitted for testing,” and the report “must not be used by the client to claim product certification, approval, or endorsement” by the lab, NVLAP, NIST or any government agency. A test report is evidence about a sample, not a product approval.
What this report supports
It supports a specification for this construction in Type C-25 mounting: slats over felt, on 1 in furring, with the cavity filled. It does not describe the same face glued flat to drywall, the same furring with an empty cavity, or a different backer. If your detail differs, ask for the matching test or plan with the difference stated.
A ten-point check for any absorption report
- The lab is named, with its accreditation and a statement that it covers this procedure.
- The method and edition are stated (for example ASTM C423-22).
- The mounting is named by E795 type and described in words.
- The specimen is described by layer, with thickness and density, and matches the product submitted.
- The specimen area is stated, so you can divide sabins by area yourself.
- Every band from 100 to 5,000 Hz is shown, not only a single number.
- NRC and SAA recompute from the band table.
- The test date and report number match the submittal cover sheet.
- Any extended or non-accredited data is labeled as such and kept out of the performance claim.
- The report is complete, with every page present, rather than an excerpt or a reformatted summary.
A report that fails one of these is not necessarily wrong. It is incomplete, and the submittal review should say so in writing.
Laboratory accreditation, report age and editions
Building codes do not require acoustic test certification, so the specification is the only place the absorption evidence is enforced. Say who may test, which editions you accept and whether summary sheets count.
Intertek’s summary is plain about it: “acoustical certification is currently not required by building codes.” The C423 scope notes that the method itself includes information on laboratory accreditation in an annex. The UFGS template asks for certificates “from an independent laboratory accredited by the National Laboratory Accreditation Program,” and the VA template is silent on the lab. If you edit from the VA text, add the accreditation sentence.
Summary sheets versus reports
Makers publish summary tables because full reports are long, and the tables are useful for shortlisting. They are not evidence of the product you are buying. The summary cited earlier says so in its own footnote: “These are representative data for the product family. Sound absorption test data for specific item tested. Data and test reports for most products available upon request.” Take the invitation. The specification should require the report itself for the exact construction submitted.
Editions
ASTM lists C423-23e1 as active. Reports to C423-17 and C423-22 are common, and the VA template still cites C423-17 while the UFGS cites “ASTM C423 (2023; E 2024).” Decide which you will accept and write it down. A practical clause names the current edition and accepts earlier ones listed by number, so the reviewer has a rule rather than a judgment call.
International data
Data measured to ISO 354 uses a larger specimen; Intertek notes that ISO testing requires “a larger, 10 m2 (107.6 ft2) specimen.” European data sheets may summarize results as a weighted coefficient rather than NRC. Ask for the band table and compare like with like, and do not convert one single number into the other.
Field checks
If the owner wants proof in the finished room, the specification can call for a field measurement of reverberation. C423 says its own method can be used for field measurements of room absorption and points to ASTM E2235 for a field method. Write the target, the occupied or unoccupied condition and the bands, and keep in mind that a field result tests the room, not the panel.
Fire results for the assembly as installed
Require fire test results that cover the panel as built: core, facing, adhesive and frame, in the way it will be mounted. A core’s rating or a fabric’s rating on its own does not describe the finished panel.
The IBC treats paneling applied for “acoustical correction” as interior finish that must comply with Section 803, as published in the 2021 Seattle Building Code, Chapter 8, which adopts the 2021 IBC with local amendments. ASTM E84 warns in section 1.4 that supporting materials under a specimen can lower its flame spread index, and that results “do not necessarily relate” to the material tested without that support. That is why both templates say “complete composite” or “full composite,” and why you should keep that phrase.
Textile faces
Fabric-faced products can fall under the code’s textile wall covering provisions, which require testing “in the manner intended for use, using the product-mounting system, including adhesive,” and then either a room-corner test or Class A with sprinklers. The full route, by occupancy and edition, is laid out in acoustic panel fire ratings. For the specification, the practical step is to require the submittal to state which code section the evidence answers.
What the fire report should state
- The method and edition, such as ASTM E84-26a, and the lab.
- Every layer in the specimen: core, facing, adhesive, backing and any edge treatment.
- How the specimen was supported or mounted in the test, and any specimen preparation practice used.
- The flame spread index and the smoke-developed index, as numbers, not only a class.
- Any note that the material melted, dripped or delaminated. ASTM E84 section 1.5 warns that such behavior “results in low flame spread indices that do not relate directly” to materials that stay in place.
Staying attached
IBC 803.14 requires interior finish to be fastened so it “will not readily become detached where subjected to room temperatures of 200°F (93°C) for not less than 30 minutes.” Put the fastening method on the drawings and ask the supplier to confirm it is suitable. Adhesive-only mounting deserves a direct question.
How Fibson reads and lists fire reports for its own planning work is set out on materials and safety.
Accessibility and installation limits that belong in the section
Panel thickness plus mounting hardware is a projection from the wall, and the ADA Standards limit it. Clouds and baffles over walking routes have a headroom limit. Put both in the acoustic section, because the acoustic section is where thickness gets chosen.
The US Access Board’s guide to protruding objects summarizes section 307.2: wall-mounted objects with leading edges above 27 in and below 80 in are “limited to a 4 inches maximum projection.” Objects with leading edges at or below 27 in are within cane sweep and “can protrude any amount,” and so can objects above the 80 in headroom line. Section 307.4 requires “headroom clearance of at least 80 inches high” along circulation paths.
Suppose a specification calls for a 2 in panel on clips that hold it 1/2 in off the wall. It projects 2.5 in and meets the 4 in limit anywhere. Now suppose a designer upgrades to a 4 in panel on the same clips: 4.5 in of projection, which fails along a corridor unless the panel’s bottom edge sits at 27 in or lower or the panel moves out of the circulation path. Both cases are hypothetical, and the ADA protrusion checker runs the same check for any thickness and height.
Tolerances and conditions
The VA template sets plumb and level within 1/16 in in 48 in and joint variation within 1/16 in. The UFGS asks for an area temperature of at least 60 degrees F, or the maker’s recommendation, for 72 hours before and during installation and until the adhesive dries. Both are worth keeping. Add a line requiring the installer to follow the maker’s written instructions for the mounting type named in the acoustic requirement, so the mounting that was tested is the mounting that gets built.
Samples, mock-ups and extra stock
Samples confirm color and texture; a mock-up confirms the installed result, edges, joints and mounting depth included. Extra stock lets the owner replace a damaged panel with one from the same production run. Neither template includes the last two.
The VA section asks for fabric samples 18 in in size, “four (4) each, full width of mill run for each color specified.” The UFGS asks for finished swatches at least 18 by 24 in, three samples of each color range. Keep the sample requirement and add three things.
- A mock-up. One full panel, or a wall bay of two or three, in the specified mounting, edge detail and fabric, reviewed under the room’s own lighting. Once approved, it becomes the standard for the rest of the work and may stay in place if it meets the specification.
- Extra stock. A stated quantity per panel type, size and color, from the same production run as the installed panels, packaged and labeled with location and color. Set the number by the risk of damage: a lobby wall at chair-back height needs more spares than a cloud at 10 ft.
- Care data. The UFGS already asks the installer to submit the manufacturer’s “recommended cleaning instructions.” Make that a closeout item for every fabric and felt, and see choosing and cleaning acoustic panel fabric for what those instructions should cover.
Example (fictional). A designer specifies 2 by 4 ft fabric-wrapped panels for a 30-seat training room. The mock-up clause asks for two panels on the specified clips, beside the room’s window wall. At review, the fabric reads warmer under daylight than on the sample board, and the clip depth leaves a visible gap at the panel’s top edge. Both are corrected before 40 panels are made. The extra stock clause adds two panels in each color, labeled by wall. The room and project are invented for this example.
Emissions and sustainability documents
Name the emissions method and its version, list the documents you need, and leave material choice to the material comparison. The federal template already sets clear requirements for PET and felt panels.
UFGS 09 84 20 says “PET and felt acoustical wall panels must be formaldehyde-free and low-emitting and must comply with CDPH SECTION 01350,” and that felt panels “must be free of formaldehyde, heavy metals, and halogenated flame retardants.” Its reference list names the CDPH method as “(2017; Version 1.2).” The California Department of Public Health has since published Version 1.3 (June 2026), which supersedes 1.2 and brings new emission criteria into effect on January 1, 2027. A specification issued now should say which version a report may use.
The UFGS also carries bracketed recycled-content defaults, such as 50 percent for fabric-covered panels and 40 percent for felt panels, which the writer is told to verify against regional availability before keeping. Require the document that supports any recycled-content figure. How the common cores and felts compare on these points is a material question for the core comparison in this series rather than the specification. If the project is pursuing a WELL rating, the acoustic features add their own documentation, covered in what WELL asks for in sound.
A performance clause you can adapt
The text below turns the points above into specification language. Brackets mark choices for the writer, and every line can be checked against a submittal.
Acoustic performance.
- Sound absorption: SAA not less than [0.xx], with coefficients not less than [0.xx] at 250 Hz and [0.xx] at 500 Hz, when tested to ASTM C423 [edition, or editions accepted] in ASTM E795 mounting Type [letter and depth], matching the mounting detail on drawing [sheet number].
- Reports: full test reports from a laboratory accredited under NVLAP for ASTM C423, for the construction specified (core, thickness, density, facing and backing), showing every one-third octave band from 100 to 5,000 Hz, the mounting description, the specimen area, the report number and the test date. Summary sheets alone are not acceptable.
- Units: for baffles and other free-hanging units, absorption per unit in sabins at each band, with the number and spacing of units in the test.
- Deviations: where no report exists for the specified mounting, submit the nearest tested mounting and state the difference in writing.
Fire performance.
- Interior finish class: [A / B / C] as required by the code in force for the occupancy and location shown.
- Evidence: test reports covering the complete panel as manufactured and mounted, including core, facing, adhesive and edge treatment, stating the method and edition, the flame spread and smoke-developed indices, and the code section the evidence is offered under.
Accessibility.
- Wall-mounted panels with leading edges above 27 in and not more than 80 in above the floor along circulation paths: total projection, including mounting hardware, not more than 4 in.
- Suspended units over circulation paths: lowest point, including hardware, not less than 80 in above the floor.
Quality control and closeout.
- Mock-up: [one panel / one wall bay] in the specified mounting, approved before fabrication.
- Extra stock: [number] panels of each type, size and color, from the same production run, labeled.
- Closeout: maker’s cleaning instructions for each facing; emissions reports naming the CDPH Standard Method version.
Edit the clause to fit the project, and keep the habit behind it: every number in a specification should name the test, the mounting and the edition it comes from.
What to do next
Before you issue the section, run through this list.
- Replace “NRC range” with SAA or band minimums, each tied to a mounting type that matches the drawing.
- Add the accreditation sentence if your template lacks it, and state which C423 editions you accept.
- Keep “complete composite” in the fire paragraph and ask which code section the evidence answers.
- Add the 4 in projection and 80 in headroom limits, and check thick panels with the ADA protrusion checker.
- Add a mock-up and extra stock, and make care instructions a closeout item.
- Name the CDPH method version for emissions.
- At submittal, recompute NRC and SAA from the band table with the NRC calculator and compare the report’s mounting with your detail.
The acoustic panel spec checklist lists the documents to request from any supplier. Fibson Acoustics designs and makes acoustic panels to order and has not yet tested its finished panels, so its spec sheets cite the core and fabric makers’ reports by lab, method, edition and mounting. Architects and designers can request samples, spec sheets and draft specification text through design trade, or send a project through the contact page.
