The scattering, diffusing action of distributed sound absorption has long been known (13). If absorbing or reflecting surfaces are in a regular pattern, the diffraction grating aspects of scattering are developed. Sound on a picket fence is split into two parts, one reflecting and the other transmitting. Both parts exhibit diffraction grating effects. The picket fence is a transmission type diffraction grating. If pickets are filled with a sound absorbing material, then only the reflective diffraction effect is developed. If instead the pickets are absorptive, then only the transmissive diffraction effects are observed. QSF rooms use the reflective component of the absorptive diffraction grating.

The sound traps that have been used for diffraction wall work are 1/2 round, tubular shaped. Their interiors are hollow; their curved surface is of highly compressed, fine filament fiberglass. The acoustically resistive surface (R) in conjunction with interior volume (C) establish an effective RC acoustical circuit. This is a “high pass” sound absorber whose lower frequency cutoff is set by the value of the RC time constant.

In addition to the two lumped acoustic parameters R and C, each trap has a “limp mass” reflector (L) buried in its outer surface. Thirty to fifty percent of the trap’s surface is covered with this strip that reflects 400Hz and above. The strip is usually centered on the trap. High frequencies are reflected off the strip while the lows pass through it, to be absorbed.

A series of tests easily show the results of the distributed reflective and absorbing surfaces. A Techron-12 Frequency Sweep from 100 to 30K gives a 13ms Time Window, in which the ETC is taken.

4.2A (see below) Test Setup shows the speaker mounted to the ceiling of a testing room and surrounded with 6 inches of absorption to a radius of 3 feet. This damps the ceiling image to give a sharp spike delivery. The 1/4 inch mic is 4 feet above and parallel to the floor, the ceiling is 8 feet.

4.2B (see below) shows the hard surface reflection. The floor return is nearly identical in timewise character to the direct signal, except that it is about 10dB lower in sound level. The direct signal passes by at 3.2ms and the floor bounce returns at 10.2ms. The ever present spike at 13.5 to 14ms is an extraneous reflection. The expanding point source wavefront accounts for a 20Log 12/4= 9.5dB reduction.

4.2C (see below) shows the floor bounce modified by a 2” of “703.” The soft bounce is 27dB below the hard concrete reflection. This reflection is typical of the common flat wall absorption panels used in dead rooms.

4.2D and 4.2E (see below) are for one trap below the speaker. Two reflections are seen. The first is 0.9ms ahead of the bare floor bounce and 10dB below it. The second is some 14dB below the floor bounce and delayed .01ms. The early signal (A) is a reflection off the limp mass diffusor surface. The delayed signal (B) is a reflection from the floor off either side of the trap. We see two dispersive actions, the first is specular from curved surface reflection and the second is diffractive from absorptive “edge effect” reflection.

4.2F and 4.2G (see below) show two traps added to either side of the central one. Signals (A) and (B) remain undisturbed. Between them appears the reflection off limp mass diffuser panels (C) of the two new traps.

Additional traps show no significant change in the signal.

As the incident angle increases, the protruding absorptive trap blocks a larger percentage of the wavefront. Full absorption occurs at 30 degrees off the surface. This set of oblique incidence tests used the setup as before with the mic 3 feet to the side.

4.3A shows bare floor reflection to be 2dB above the direct signal. This is due to the directional beaming of the speaker. The direct signal (D) is at 4.2ms and the reflection (1) at 10.5ms, both larger than before due to the angles involved.

4.3B shows the effect of one trap placed directly below the speaker as in 4.2D. The reflected signal (1) remains unchanged at 10.5ms because nothing was added where that reflection occurred. There is, however, an early signal (2) by about 0.3ms that is 9dB down. It is the acoustic glint off the sound scattering limp mass strip in the trap.

4.3C shows the effect of two traps being added on 18 inch centers on either side of the first trap as in 4.2F. The old floor bounce is now damped 9dB by the outer trap. Two new signals appear, one (3) is earlier than before at 9.7ms and the other later at 11.3ms. The first is the glint off the absorbing trap’s limp mass diffuser surface. The second is a diffracting hard floor bounce off the edge of the traps. The reflection bends back into the shadow zone cast by the absorbing sound trap. Again additional traps make negligible difference to the signature.

4.4 Technical Discussion

The overall result of this diffraction grating technique is that the single sharp hard wall reflection is splintered into a set of 3 to 4 lower level reflections whose strength is about 10dB down. The splintered reflections are distributed out in time at 1/3 ms intervals and are within 3dB of each other. A balance has been struck in the dispersion of sound between the higher frequency, diffracting edge effects distributed absorption.

As a basis for comparison, recall the strength of a reflection off a 2” high density fiberglass. About 28dB of cut compared to the hard surface reflection is produced by this ever-so-common flat wall “acoustic treatment” for sound rooms.

The overall strength (Ld) of a multiple reflection signal is determined by the mean signal level (Lo), the number of signals level (Ln) and the fraction of time signal level (Lt).

Ld = Lo + Ln + Lt where Ln = 10LogN and Lt = 10Log (L1 + L2 + . . . ) / T

The perpendicular reflection off a hard surface was split into three reflections (N=3) each some 12dB below the single hard surface reflection strength (Lo= -12). The time width of each reflection was 0.15ms, 0.2ms, and 0.15ms over a (T=1.2ms) period. The perceived strength of this composite is calculated:

Ld = - 12 + 10Log3 + 10Log((0.15 + 0.2 + 0.15) / 1.2) = - 12 + 4.2 – 3 = - 10.8dB

The diffraction grating reflection is 10.8dB down from the hardwall bounce. It is spread out in time by a factor of 10.

The oblique reflection off the grating produced 4 spikes each down 9dB and spread over a 1.8ms time smear. The discrete reflections are 0.2, 0.1, 0.1 and 0.15ms long. The resulting splintered reflection has a calculated level:

Ld = - 9 + 10Log4 + 10Log(0.2 + 0.1 + 0.15) / 1.8 = - 9 + 6 – 5.1 = - 8.1dB

The oblique, angled reflection off the absorption grating is down 8.1dB compared to the hard wall bounce and is spread out over time by a factor of 15.

5 Fresnel Diffraction Grating—Polar Plots

The most general diffraction grating is the Fresnel which allows for a spherical wavefront, the source being near the grating. A subclass is the Fraunhoffer diffraction, which requires parallel wave fronts. The absorptive diffraction gratings presented here are of the complex, Fresnel type. The sound source in small rooms is necessarily close to the diffraction grating and Fresnel diffraction occurs.

5.0A shows test positions for a 4 x 8 sheet of plywood in the open, with and without a grid of the 1/2 round sound traps. The speaker is at 8 feet and mic positions are every 5.6 degrees on a 5 foot radius about the center of the panel (14).

5.1 Test Setup and Measurements

5.1A and 5.1B (see below) show the ETC and EFC (Energy Frequency Curve) of the flat panel for perpendicular reflection. The ETC is 6dB per division with a 30ms window. The reflection is 12 feet behind the direct signal and down 15dB, due of course to the expansion of the wavefront (the mic faces the panel, giving the reflection a small directional boost). The frequency scale (B) is linear, to see comb effects. An unimpressive, but realistic speaker frequency response is seen.

5.1C and 5.1D (see below) show the diffraction grating effect. Notice the early double peak return off the reflectors of the center trap and the pair aside. The 1.1ms time difference between the first reflection and the surface reflection produces a 1/1.1 sec or 900Hz comb effect, characteristic of diffraction gratings.

5.1E and 5.1F (see below) are the 32 frequency sweeps at 5.6 degree intervals that compare the smooth, specular reflection (E) with the very irregular, diffraction grid reflection (F), similar to that of 5.1D. The frequency sweep is 200 to 8K, linear scale. Throughout the angles measured, dramatic diffraction grating effects are obvious.

5.2 Polar Plots

The following is a set of polar plots taken at 5.6 degree intervals at specific frequencies. Data compares specular reflection to diffraction grating reflection of a spherical wavefront. Each plot is normalized to the strength of the reflection, perpendicular to the panel. Absolute levels are not displayed. The lobing near the 90 degree axis is erroneous, due to the direct signal leaking into the time window that is centered on the reflection.

Conclusion

We have illustrated methods by which a specific type of acoustic signature can be developed. It is characterized by the direct signal being immediately followed by a dense group of signals that rapidly decay out in time. This timewise signature is objectively distinct but that alone provides insufficient basis upon which to draw conclusions. The content of the trailing signal group remains to be resolved and its impact on the direct signal established.

The direct signal is a voice which can be colored by lower level signals that are both derived from the direct signal and received by the listener within 10ms following the direct signal. Reflected sound is obviously a signal derived from the incident sounds, correlation between the two is very high. The reflected signal may not have the same spectral content as the incident sound, depending on the absorption characteristics of the reflecting surface. The direct signal can be colored by spectral characteristics of its nearby reflections.

An instrument has directional properties in the sound field it produces. Its total sound is desired to be presented to the mic. Acoustical containment resulting in multiple reflections is a means by which the divergent components of the instrument’s sound field become redirected to pass by and be captured by the recording mic. In order for the multiple reflections to compliment and develop the voice of the instrument, they must fill the first 10ms time window. A small room is in order as the reflections are too time delayed in larger rooms.

There are two very different types of sound dispersive reflecting surfaces. The absorptive reflection method is signal coherent while the resonance reflection systems are signal incoherent. By definition the resonant reflection panels available today ought not to be able to faithfully develop the voice of an instrument. Both reflecting systems can produce comparable ETC records that look healthy but the quality of coherence or incoherence in the diffuse reflections is the issue. Incoherent diffuse early reflections should create distracting room ambience effects that mask the presence of the instrument’s voice. Collection of instrumental ambience requires retention of coherent, diffuse reflections that have good correlation to the direct signal. There remains both subjective and objective exploratory work to be done in the area of coherent vs. incoherent diffuse early reflections.

Epilogue

There once was a great singer who was accompanied by an excellent local choir. They were quite successful and hired an agent to schedule a world tour. This fella was very creative and decided the choir needed a more worldly air. He proceeded to thank, then discharge each of the local choir members. He replaced them with singers from many foreign countries. Each was to sing in their own native tongue. With this complete, the great singer and his newly formed entourage left on tour. They were know as the “Choir of Babbel” and were never heard of again.

1) A.H. Benade, “From Instrument to Ear in a Room: Direct or Via Recording,” presentation Audio Engineering Society 74th Convention, Oct 1983 New York, Preprint Number 2042 (C-2).

2) D. Davis and C. Davis “LEDE Concept of Acoustic and Psychoacoustic Parameters in Recording Control Rooms,” J. Audio Eng. Soc. Vol. 28 No. 9, September 1980.

3) J. Strawn, “Orchestral Instruments: Analysis of Performed Transistions,” presentation Audio Engineering Society 78th Convention, Anaheim, May 1985, Preprint Number 2229 (B-10).

4) Benade, preprint 2042

5) Private communication with recording studio engineer, consultant Sam Lynn of Chicago, October 1987.

6) A. Noxon, “Studio Applications for TUBE TRAPS,” Information Bulletin, Acoustic Sciences Corp., August 1984.

7) H. F. Olson, “Acoustical Engineering,” pg. 505, Van Nostrand, 1957.

8) A. Noxon, “Listening Room-Corner Loaded Bass Traps,” presentation Audio Engineering Society 79th Convention, New York, October 1985.

9) U.S. Patent No. 4,548,292 issued October 22, 1985.

10) “Sound Absorption Test on 1/4 Round TUBE TRAPS,” Report No. 1262-NV, Acoustic Section, Institute for Research in Construction, National Research Councel, Canada, May 5 1987.

11) L. E. Kinsler and A. R. Frey, “Fundamentals of Acoustics,” pg. 421 John Wiley and Sons, 1950.

12) A. Noxon, “Room Acoustics and Low Frequency Damping,” presentation Audio Engineering Society 81st Convention, Los Angeles, November 1986.

13) Olsen pg. 543.

14) P. D’Antonio and J. H. Konnert, “The Acoustic Properties of Sound Diffusing Surfaces: The Time, Frequency and Directivity Engergy Response,” presentation Audio Engineering Society 78th Convention, New York, October 1985, Preprint 2295 (B-6).