LIVE SOUND SUBWOOFER SYSTEM DESIGN DR. ADAM J. HILL COLLEGE OF ENGINEERING & TECHNOLOGY, UNIVERSITY OF DERBY, UK GAND CONCERT SOUND, CHICAGO, USA 20 OCTOBER 2017
GOALS + CHALLENGES SINGLE SUBWOOFERS SUBWOOFER CLUSTERS HORIZONTAL ARRAYS VERTICAL ARRAYS SIGNAL PROCESSING TECHNIQUES PERCEPTUAL CONSIDERATIONS RECOMMENDATIONS
GOALS + CHALLENGES SINGLE SUBWOOFERS SUBWOOFER CLUSTERS HORIZONTAL ARRAYS VERTICAL ARRAYS SIGNAL PROCESSING TECHNIQUES PERCEPTUAL CONSIDERATIONS RECOMMENDATIONS
GOALS + CHALLENGES [01/19] COMMON ISSUES Inconsistent audience coverage Excessive sound energy outside audience area Smeared transients, muddy sounding system
GOALS + CHALLENGES [02/19] COMMON CHALLENGES Truck space Budget Sight lines Venue capabilities Power distribution Time!
GOALS + CHALLENGES [03/19] TYPICAL GOALS Consistent audience coverage Minimal energy outside audience area Sharp transient response, high quality sound
GOALS + CHALLENGES [04/19] ROOM-MODES These often come up in discussions with engineers Some questions: 1. What are room-modes? 2. Are room-modes an issue in live sound?
GOALS + CHALLENGES [05/19] ROOM-MODES 1. What are room-modes? Frequencies with integer multiples of ½ wavelength fit perfectly within one or more room dimension Results in a standing wave pattern Position-dependent listening experience
GOALS + CHALLENGES [06/19] AXIAL TANGENTIAL OBLIQUE
ROOM-MODES Theoretical room-mode frequencies calculated based on room dimensions, modal indices and speed of sound 2 2 2 2 + + = z z y y x x m L L L c f η η η GOALS + CHALLENGES [07/19]
ROOM-MODES 2 2 2 2 + + = z z y y x x m L L L c f η η η 5 M X 4 M X 3 M ROOM (10% ABSORPTION) GOALS + CHALLENGES [08/19]
GOALS + CHALLENGES [09/19] ROOM-MODES 2. Are room-modes an issue in live sound? Must define the modal region of a venue (where roommodes are perceptible) Conservatively defined using the Schroeder frequency RT 60 f s = 2000 V
GOALS + CHALLENGES [10/19] ROOM-MODES Medium venue (V = 8000 m 3, RT 60 = 1.5 s) Room-mode issues below 27.4 Hz Large venue(v = 25000 m 3, RT 60 = 2.5 s) Room-mode issues below 20.0 Hz Sports arena (V = 1500000 m 3, RT 60 = 6.0 s) Room-mode issues below 4.0 Hz
GOALS + CHALLENGES [11/19] ROOM-MODES Subwoofer band typically 20-100 Hz (but often subwoofers are crossed over around 70 90 Hz) With room-mode problems generally outside the subwoofer range, are they causing significant issues? Not likely comb-filtering between coherent sources + reflections is a more likely culprit
GOALS + CHALLENGES [12/19] COMB-FILTERING Principle of superposition Assumes two (or more) sources being combined are directly related to one another Called correlated addition
GOALS + CHALLENGES [13/19] COMB-FILTERING If two sine waves are exactly in-phase, then the resulting waveform will be twice as large as the originals (constructive interference)
GOALS + CHALLENGES [14/19] COMB-FILTERING If two sine waves are 180 out of phase (or reverse polarity), the resulting waveform will be zero (destructive interference)
GOALS + CHALLENGES [15/19] COMB-FILTERING This becomes a more complicated issue looking across the entire subwoofer band:
GOALS + CHALLENGES [16/19] COMB-FILTERING Position-dependent, causing inconsistent low-frequency responses across an audience area Can this be quantified?
GOALS + CHALLENGES [17/19] SPATIAL VARIANCE (SV) Measure of average seat-to-seat difference in frequency response (measured in db) SV = 1 N f f hi ( p L ( ) ( )) p p, i Lp i 1 i= f p p= 1 lo N 1 N 2
GOALS + CHALLENGES [18/19] MEAN OUTPUT LEVEL (MOL) It s also essential to inspect system efficiency MOL is the measure of average SPL across an audience MOL = N f 1 N p f hi i= f p= 1 lo N p L p ( p, i)
GOALS + CHALLENGES [19/19] LOW FREQUENCY FACTOR (LF) Used to take into account number of sources in system Results in an adjusted MOL LFF = 20log10 ( ) N S MOL (w/lff) = MOL 20log10 ( ) N S
GOALS + CHALLENGES SINGLE SUBWOOFERS SUBWOOFER CLUSTERS HORIZONTAL ARRAYS VERTICAL ARRAYS SIGNAL PROCESSING TECHNIQUES PERCEPTUAL CONSIDERATIONS RECOMMENDATIONS
SINGLE SUBWOOFERS [1/6] Is it correct to assume a perfectly omnidirectional response? Any deviation could impact array/cluster performance
SINGLE SUBWOOFERS [2/6] Virtual investigation with d&b ArrayCalc + B6 subwoofer 40 HZ 63 HZ 80 HZ 100 HZ Predicted sound energy coverage at 40 Hz (grid line spacing = 5m, color contours spaced at 6 db) Predicted polar response
SINGLE SUBWOOFERS [3/6] d&b B6 subwoofer measured in hemi-anechoic chamber Speaker centered in room, measured at 2 m (30 increments) Not quite as expected why? 40 HZ 63 HZ 80 HZ 100 HZ Predicted polar response Measured polar response
SINGLE SUBWOOFERS [4/6] Centering the speaker in the room skews the polar response measurement due to the shift in acoustic center Better to position the speaker so that the acoustic center coincides with the room center
SINGLE SUBWOOFERS [5/6] Subwoofer re-measured in hemi-anechoic chamber Acoustic center (approx. 37 cm) positioned at room center Results much closer to predictions! 40 HZ 63 HZ 80 HZ 100 HZ Predicted polar response Measured polar response
SINGLE SUBWOOFERS [6/6] It s essential to account for the acoustic center when measuring subwoofers! Seems trivial in practice is this actually important? 40 HZ 63 HZ 80 HZ 100 HZ Measured polar response (w/o aco ) Measured polar response (w/aco )
GOALS + CHALLENGES SINGLE SUBWOOFERS SUBWOOFER CLUSTERS HORIZONTAL ARRAYS VERTICAL ARRAYS SIGNAL PROCESSING TECHNIQUES PERCEPTUAL CONSIDERATIONS RECOMMENDATIONS
SUBWOOFER CLUSTERS [01/32] Subwoofer cluster = compact arrangement of multiple individual sources to achieve desired coverage pattern
SUBWOOFER CLUSTERS [02/32] GRADIENT LOUDSPEAKERS Zero-order ¼ λ spacing ½ λ spacing Second-order First-order (cardioid) First-order (dipole)
SUBWOOFER CLUSTERS [03/32] GRADIENT LOUDSPEAKERS Physical separation + electronic delay must be correct! ¼ wavelength spacing + delay = expected polar response Deviation from this gives unwanted behavior Some examples
SUBWOOFER CLUSTERS [04/32] GRADIENT LOUDSPEAKERS 1 st order gradient loudspeaker Configured to 60Hz (1.43 m spacing, 4.17 ms delay) PROPERLY CONFIGURED NO POLARITY REVERSAL ½ WAVELENGTH SPACING/DELAY
SUBWOOFER CLUSTERS [05/32] GRADIENT LOUDSPEAKERS So gradient loudspeakers operate nicely when the physical spacing, electronic delay + polarity are correctly chosen for a specific frequency Is the resulting polar response the same for all frequencies in the subwoofer band (20 120 Hz)? Some more examples
SUBWOOFER CLUSTERS [06/32] GRADIENT LOUDSPEAKERS Inconsistent low-frequency coverage Configured for 60 Hz 60 HZ 30 HZ 120 HZ
SUBWOOFER CLUSTERS [07/32] GRADIENT LOUDSPEAKERS These were all simulations, though Does this apply to the real world?
SUBWOOFER CLUSTERS [08/32] GRADIENT LOUDSPEAKERS Measured in a hemi-anechoic chamber (not ideal, but 40 HZ 63 HZ 80 HZ 100 HZ still gives indication of agreement with simulations) Good results at 63 Hz, other bands not so good Why? # Spacing Polarity Delay 1 1.5 m Normal 0.0 ms 2 (front-tofront) Reverse 4.37 ms Front-to-rear rejection (db) Delay 40 Hz 63 Hz 80 Hz 100 Hz 4.37 ms 2.71 8.82 0.69 0.04
SUBWOOFER CLUSTERS [09/32] GRADIENT LOUDSPEAKERS Electronic delay is incorrect! Each source s acoustic center will shift ~37 cm This adds 74 cm to the front-to-front spacing Effective source spacing now 1.5 m + 0.74 m = 2.24 m Delay must be set to 6.53 ms
SUBWOOFER CLUSTERS [10/32] GRADIENT LOUDSPEAKERS Now including acoustic center shift 40 HZ 63 HZ 80 HZ 100 HZ Ideal performance predicted at 38.38 Hz Results still not great Why? # Spacing Polarity Delay 1 1.5 + 0.74 m Normal 0.0 ms 2 (front-tofront) Reverse 6.53 ms Front-to-rear rejection (db) Delay 40 Hz 63 Hz 80 Hz 100 Hz 6.53 ms 2.94 5.41 3.82 4.14
SUBWOOFER CLUSTERS [11/32] UNIDIRECTIONAL (CARDIOID) Issue #1: Cardioid clusters require the same forward/backward moving sound energy sources facing opposite directions don t allow for this!
SUBWOOFER CLUSTERS [12/32] UNIDIRECTIONAL (CARDIOID) Issue #1: Cardioid clusters require the same forward/backward moving sound energy sources facing opposite directions don t allow for this! Issue #2: 40 Hz performance isn t great due to reduced source output and limitations of test environment
SUBWOOFER CLUSTERS [13/32] UNIDIRECTIONAL (CARDIOID) Issue #1: Cardioid clusters require the same forward/backward moving sound energy sources facing opposite directions don t allow for this! Issue #2: 40 Hz performance isn t great due to reduced source output and limitations of test environment Issue #3: 80 Hz & 100 Hz vaguely figure 8 patterns (due to ~½ wavelength spacing in this range) Orienting both sources in the forward direction and reducing the physical spacing will hopefully solve all issues
SUBWOOFER CLUSTERS [14/32] GRADIENT LOUDSPEAKERS Ideal performance predicted at 100.88 Hz 40 HZ 63 HZ 80 HZ 100 HZ No spacing compensation needed aco shifts are equal + in same direction Results seem good! Poor 40 Hz due to spacing # Spacing Polarity Delay 1 0.85 m Normal 0.0 ms 2 (front-tofront) Reverse 2.48 ms Front-to-rear rejection (db) Delay 40 Hz 63 Hz 80 Hz 100 Hz 2.48 ms -0.69 9.85 8.86 8.28
SUBWOOFER CLUSTERS [15/32] END FIRE End-fire configuration was also tested (1 st source 40 HZ 63 HZ 80 HZ 100 HZ delayed, no polarity reversal) Very similar results to gradient approach Better transient response? # Spacing Polarity Delay 1 0.85 m Normal 2.48 ms 2 (front-tofront) Normal 0.0 ms Front-to-rear rejection (db) Delay 40 Hz 63 Hz 80 Hz 100 Hz 2.48 ms -2.63 3.63 10.19 8.37
SUBWOOFER CLUSTERS [16/32] POLAR RESPONSE OPTIMIZATION To review, clusters with omni units, 2 options: 1. Delay + polarity reversal to rear unit (gradient) 2. Delay to front unit (end fire) Simulated with 8 stack system Configured for optimal rejection at 60 Hz Configured for array coupling up to 60 Hz
SUBWOOFER CLUSTERS [17/32] POLAR RESPONSE OPTIMIZATION 1. Delay + polarity reversal to rear unit (gradient) 30 HZ 60 HZ 120 HZ Excellent rear rejection (at most frequencies) Slightly impaired audience coverage Poor forward transients?
SUBWOOFER CLUSTERS [18/32] POLAR RESPONSE OPTIMIZATION 2. Delay to front unit (end fire) 30 HZ 60 HZ 120 HZ Maintained strong + even audience coverage Moderate stage rejection, but not perfect Better forward transients?
SUBWOOFER CLUSTERS [19/32] POLAR RESPONSE OPTIMIZATION Subwoofer clusters Great work done by Dave Rat (Rat Sound, USA) How do we cover non-standard audience areas? Let s try to achieve 270 of coverage with adequate rejection on stage
SUBWOOFER CLUSTERS [20/32] POLAR RESPONSE OPTIMIZATION Subwoofer clusters Let s start by configuring two omni subwoofers to give a cardioid polar response Optimized for 60 Hz Spacing = 1.43 m Electronic delay = 4.17 ms Polarity = sub 1 (normal), sub 2 (reverse)
SUBWOOFER CLUSTERS [21/32] POLAR RESPONSE OPTIMIZATION Subwoofer clusters Let s start by configuring two omni subwoofers to give a cardioid polar response
SUBWOOFER CLUSTERS [22/32] POLAR RESPONSE OPTIMIZATION Subwoofer clusters Not let s add a third subwoofer to cover the side audience area Physical spacing (from sub 1) = 1.43 m Electronic delay (from sub 1) = 4.17 ms Polarity = sub 3 (reverse)
SUBWOOFER CLUSTERS [23/32] POLAR RESPONSE OPTIMIZATION Subwoofer clusters Not let s add a third subwoofer to cover the side audience area
SUBWOOFER CLUSTERS [24/32] POLAR RESPONSE OPTIMIZATION Subwoofer clusters There isn t great rejection on stage let s fix that! Add a 4 th subwoofer Physical spacing (from sub 2/3) = 1.43 m Electronic delay (from sub 1) = 0 ms Polarity = sub 4 (reverse)
SUBWOOFER CLUSTERS [25/32] POLAR RESPONSE OPTIMIZATION Subwoofer clusters There isn t great rejection on stage, let s fix that!
SUBWOOFER CLUSTERS [26/32] POLAR RESPONSE OPTIMIZATION Subwoofer clusters (gradient) How does this operate at other frequencies? 30 HZ 60 HZ 120 HZ
SUBWOOFER CLUSTERS [27/32] POLAR RESPONSE OPTIMIZATION Subwoofer clusters (end fire) How does this operate at other frequencies? 30 HZ 60 HZ 120 HZ
SUBWOOFER CLUSTERS [28/32] POLAR RESPONSE OPTIMIZATION Again, let s have a look at the measurements
SUBWOOFER CLUSTERS [29/32] GRADIENT END-FIRE # Spacing Polarity Delay Polarity Delay 1 0.85 m Normal 0.00 ms Normal 2.48 ms 2 (front-tofront) Reverse 2.48 ms Normal 0.00 ms 3 Reverse 2.48 ms Normal 0.00 ms 4 100 Hz Reverse 0.00 ms Reverse 0.00 ms
SUBWOOFER CLUSTERS [30/32] 40 HZ 63 HZ 80 HZ 100 HZ Gradient configuration End-fire configuration
SUBWOOFER CLUSTERS [31/32] POLAR RESPONSE OPTIMIZATION Cardioid subwoofer clusters Cardioid subwoofers (or stacks) give more precise polar response control Similar to Dave Rat s approach
SUBWOOFER CLUSTERS [32/32] POLAR RESPONSE OPTIMIZATION Cardioid subwoofer clusters Delay: D 1 = 0 D 2 = 0.5d D 3 = 2d D 4 = 0 Delay: D 1 = 0 D 2 = d D 3 = 0 D 4 = d Delay: D 1 = 0 D 2 = 0 D 3 = 1.5d D 4 = 0.5d Delay: D 1 = 0 D 2 = 2d D 3 = 2d D 4 = d d = propagation delay between units (ms)
GOALS + CHALLENGES SINGLE SUBWOOFERS SUBWOOFER CLUSTERS HORIZONTAL ARRAYS VERTICAL ARRAYS SIGNAL PROCESSING TECHNIQUES PERCEPTUAL CONSIDERATIONS RECOMMENDATIONS
HORIZONTAL ARRAYS [01/35] SUBWOOFER SPACING Typical left + right configuration, 20 m spacing
HORIZONTAL ARRAYS [02/35] SUBWOOFER SPACING Typical left + right configuration, 12 m spacing
HORIZONTAL ARRAYS [03/35] SUBWOOFER SPACING Typical left + right configuration, 8 m spacing
HORIZONTAL ARRAYS [04/35] SUBWOOFER SPACING Typical left + right configuration, 4 m spacing
HORIZONTAL ARRAYS [05/35] SUBWOOFER SPACING Typical left + right configuration, 2 m spacing
HORIZONTAL ARRAYS [06/35] SUBWOOFER SPACING Typical left + right configuration, 1 m spacing
HORIZONTAL ARRAYS [07/35] SUBWOOFER SPACING Single central subwoofer
HORIZONTAL ARRAYS [08/35] SUBWOOFER SPACING Spatial variance dropped once subwoofers were within 2 m of one another why?
HORIZONTAL ARRAYS [09/35] SUBWOOFER SPACING Spatial variance dropped once subwoofers were within 2 m of one another why? Upper analysis frequency = 100 Hz Wavelength @ 100 Hz = 3.42 m ½ wavelength @ 100 Hz = 1.72 m Sources must be within ½ a wavelength of each other in order to properly couple (line array theory)
HORIZONTAL ARRAYS [10/35] SUBWOOFER SPACING Most events need more than 2 stacks of subwoofers 4 subwoofer stacks with 2 m spacing
HORIZONTAL ARRAYS [11/35] SUBWOOFER SPACING Most events need more than 2 stacks of subwoofers 8 subwoofer stacks with 2 m spacing
HORIZONTAL ARRAYS [12/35] SUBWOOFER SPACING Most events need more than 2 stacks of subwoofers 16 subwoofer stacks with 2 m spacing
HORIZONTAL ARRAYS [13/35] SUBWOOFER SPACING Subwoofer spacing vs. SV, MOL and MOL (w/lff)
HORIZONTAL ARRAYS [14/35] COVERAGE CONTROL Processors allow for electronic delay on each channel Can this be used to alter the coverage pattern?
HORIZONTAL ARRAYS [15/35] COVERAGE CONTROL PERFECT DELAY Delay all subwoofers to the edge of the audience Result = wider coverage? Example: 8 stacks, spaced for coupling up to 60 Hz Calibration point distance = 25 m
HORIZONTAL ARRAYS [16/35] COVERAGE CONTROL PERFECT DELAY POST-ADJUSTMENT PRE-ADJUSTMENT Didn t exactly improve the situation why not?
HORIZONTAL ARRAYS [17/35] COVERAGE CONTROL PERFECT DELAY Perfectly delaying the system to the edge of the audience only optimizes the response at that point If the audience is wide, this won t work! Instead, we need to find a good compromise an imperfect delay approach
HORIZONTAL ARRAYS [18/35] COVERAGE CONTROL IMPERFECT DELAY My approach 1. Calculate propagation delay from each source to calibration point 2. Determine time difference of arrival (TDOA) for each source 3. Apply TDOAs to opposite pairs in array (i.e. 1 to 4, 2 to 3, 3 to 2, 4 to 1 for half of an 8 sub array)
HORIZONTAL ARRAYS [19/35] COVERAGE CONTROL IMPERFECT DELAY IMPERFECT DELAY PERFECT DELAY Much better, but still isn t perfect from 60 100 Hz
HORIZONTAL ARRAYS [20/35] COVERAGE CONTROL This is the exact problem I encountered this past summer
HORIZONTAL ARRAYS [21/35] COVERAGE CONTROL SPACING COMPENSATION We re still having issues above 70 Hz or so This is likely due a slight increase in the effective unit-to-unit spacing due to the electronic delay on top of too wide spacing for > 60 Hz Lack of coupling in the upper frequency range What if we adjusted the physical spacing of the subwoofers, to maintain consistency and coupling?
HORIZONTAL ARRAYS [22/35] COVERAGE CONTROL SPACING COMPENSATION W/COMPENSATION W/O COMPENSATION Not too bad!
HORIZONTAL ARRAYS [23/35] COVERAGE CONTROL AMPLITUDE SHADING We re in pretty good shape now If you want to further optimize your array, you can apply amplitude shading I won t go into the specifics now, but we can examine a simple example Amplitude attenuated linearly from 0 db to -6 db across the array (from the center to outside) Much better approaches available, but that s a talk for a another day!
HORIZONTAL ARRAYS [24/35] COVERAGE CONTROL AMPLITUDE SHADING W/AMP SHADING W/O AMP SHADING Slight improvement in SV, but note the -2.5 db MOL
HORIZONTAL ARRAYS [25/35] COVERAGE CONTROL COVERAGE WIDTH Our coverage is on the narrow side This can be easily addressed by switching your calibration point Closer = wider coverage Further = narrower coverage
HORIZONTAL ARRAYS [26/35] COVERAGE CONTROL COVERAGE WIDTH CALIBRATION @15m CALIBRATION @25m Widened coverage, as desired. Slight increase in SV
HORIZONTAL ARRAYS [27/35] COVERAGE CONTROL HOW DID WE DO? POST-OPTIMIZATION PRE-OPTIMIZATION Consistent tonality, natural SPL loss with distance
HORIZONTAL ARRAYS [28/35] STAGE EFFECTS The stage can destroy the polar response of subwoofers! The reflection(s) from under the stage prevent the desired pattern to form
HORIZONTAL ARRAYS [29/35] STAGE EFFECTS Key consideration: When using cardioid subwoofers/clusters the stage can significantly affect the polar response Things to remember: Placing directional subwoofer systems under or on a stage can destroy directionality Simple solution: Place subwoofers in front of the stage. Avoid acoustically non-transparent stage skirts
HORIZONTAL ARRAYS [30/35] STAGE EFFECTS Subwoofer placement Free-field Under stage On top of stage In front of stage Front-to-rear rejection (db) 11.76 db 3.91 db* 8.48 db 12.67 db * modelled data due to stage height restrictions Hill, A.J.; J. Paul. The effect of performance stages on subwoofer polar and frequency responses. Proc. IOA Reproduced Sound, Southampton, UK. November 2016.
HORIZONTAL ARRAYS [31/35] STAGE EFFECTS Most simulation software ignores the stage Many engineers place subwoofers under the stage How does this affect polar response?
HORIZONTAL ARRAYS [32/35] STAGE EFFECTS Stage is essential for accurate simulations! NO STAGE IN SIMULATION STAGE INCLUDED IN SIMULATION
HORIZONTAL ARRAYS [33/35] AUDIENCE EFFECTS We ve thus far ignored any acoustic effects of the audience on low-frequency coverage Does the audience affect ground-based subwoofer system performance?
HORIZONTAL ARRAYS [34/35] AUDIENCE EFFECTS YES! Excellent work has been done by Elena Shabalina (now at d&b audiotecknik) Summary of findings: Tightly-packed audiences causes an acoustical impedance mismatch at their boundaries Results in the audience behaving as a room (of sorts), causing room mode-like behavior!
HORIZONTAL ARRAYS [35/35] AUDIENCE EFFECTS Shabalina, E. The propagation of low frequency sound through an audience. PhD dissertation (p 38), RWTH Aachen University, 2013.
GOALS + CHALLENGES SINGLE SUBWOOFERS SUBWOOFER CLUSTERS HORIZONTAL ARRAYS VERTICAL ARRAYS SIGNAL PROCESSING TECHNIQUES PERCEPTUAL CONSIDERATIONS RECOMMENDATIONS
VERTICAL ARRAYS [1/4] GROUND-BASED VS. FLOWN ARRAYS Flown arrays (L + R hangs) Advantages Even front to back coverage Better sightlines Disadvantages Rigging (time + equipment) Less stage rejection Uneven horizontal coverage Loss of Waterhouse effect
VERTICAL ARRAYS [2/4] GROUND-BASED VS. FLOWN ARRAYS Flown arrays (center hangs) Advantages Even front to back coverage Even horizontal coverage Better sightlines Disadvantages Rigging (time + equipment) Less stage rejection Video sightlines? Loss of Waterhouse effect
VERTICAL ARRAYS [3/4] FLOWN ARRAY STEERING Cardioid pattern can be achieved in similar manner to ground-based systems Do we want the pattern to be focused directly forward?
VERTICAL ARRAYS [4/4] FLOWN ARRAY STEERING Flown cardioid arrays can be steered to focus sound energy on the audience, w/still limited stage SPL
GOALS + CHALLENGES SINGLE SUBWOOFERS SUBWOOFER CLUSTERS HORIZONTAL ARRAYS VERTICAL ARRAYS SIGNAL PROCESSING TECHNIQUES PERCEPTUAL CONSIDERATIONS RECOMMENDATIONS
SIGNAL PROCESSING TECHNIQUES [01/10] EQUALIZATION Use sparingly! You can t EQ for comb-filtering Excessive EQ = poor transients
SIGNAL PROCESSING TECHNIQUES [02/10] TIME-ALIGNMENT A number of options available 1. Use software to get necessary delay time Tricky to get 100% correct as propagation paths to audience differs greatly Much easier for flown subwoofer arrays
SIGNAL PROCESSING TECHNIQUES [03/10] TIME-ALIGNMENT A number of options available 2. Run a sine wave (set to crossover frequency) through main left/right and subwoofer system simultaneously Adjust delay to subwoofers (assuming subwoofers are in front of the main PA) Listen for peak SPL in sine wave
SIGNAL PROCESSING TECHNIQUES [04/10] POLAR RESPONSE OPTIMIZATION Many companies use DSP algorithms within their processors to achieve cardioid polar responses across the subwoofer frequency band Uses a form of all-pass filter Works very well!
SIGNAL PROCESSING TECHNIQUES [05/10] HEADROOM EXTENSION - EQ Old trick Apply very high-q parametric EQ at 20 Hz (or so) and attenuate as much as possible Much more headroom for the subwoofer system!
SIGNAL PROCESSING TECHNIQUES [06/10] HEADROOM EXTENSION VIRTUAL BASS Psychoacoustical effect Principle of the missing fundamental Can be used in live sound Perceived boost in low-frequency content Less risk of driver over-excursion Decreased amplification requirements
SIGNAL PROCESSING TECHNIQUES [07/10] DIFFUSE SIGNAL PROCESSING (DISP) Extension of old trick used by many in the industry Aimed at source decorrelation to avoid comb-filtering
SIGNAL PROCESSING TECHNIQUES [08/10] DIFFUSE SIGNAL PROCESSING (DISP) Originally developed for distributed mode loudspeakers (DMLs) Applies imperceptible, low-level noise-like decay onto signal (different for each loudspeaker) Frequency-dependent decay of noise (to avoid unwanted signal coloration) Static or dynamic applications
SIGNAL PROCESSING TECHNIQUES [09/10] DIFFUSE SIGNAL PROCESSING (DISP) Has been shown experimentally to reduce spatial variance across an audience by up to 50% (or so) Presentation tomorrow! P18 Sound Reinforcement & Acoustics, 2pm, Room 1E11
SIGNAL PROCESSING TECHNIQUES [10/10] DIFFUSE SIGNAL PROCESSING (DISP) Advantages: Moderate spatial variance reduction No calibration necessary Avoids perceptible signal coloration Disadvantages: Correct implementation of phase noise essential Overuse = perceptually noticeable
GOALS + CHALLENGES SINGLE SUBWOOFERS SUBWOOFER CLUSTERS HORIZONTAL ARRAYS VERTICAL ARRAYS SIGNAL PROCESSING TECHNIQUES PERCEPTUAL CONSIDERATIONS RECOMMENDATIONS
PERCEPTUAL CONSIDERATIONS [1/5] HOW TO DRIVE THE SUBWOOFERS 1. Via the stereo bus (main left + right outputs) Advantages No extra work, easy to set up Certain level of signal decorrelation Disadvantages No control over what s fed to subwoofers Potential subwoofer overload Potential muddy signal
PERCEPTUAL CONSIDERATIONS [2/5] HOW TO DRIVE THE SUBWOOFERS 2. Via a mono auxiliary send Advantages Full control over what goes to the subwoofers Still fairly simple set up Disadvantages No signal decorrelation Loss of stereo image?
PERCEPTUAL CONSIDERATIONS [3/5] HOW TO DRIVE THE SUBWOOFERS 3. Via a stereo auxiliary send Advantages Full control over what goes to the subwoofers Moderate signal decorrelation Disadvantages (Slightly) more involved set up
PERCEPTUAL CONSIDERATIONS [4/5] HOW TO DRIVE THE SUBWOOFERS So what s the best option? Previous research conducted at University of Derby: Research indicates that perceptually, there is little difference between mono and stereo subwoofer systems So does it matter for large-scale live sound?
PERCEPTUAL CONSIDERATIONS [5/5] HOW TO DRIVE THE SUBWOOFERS Actually, yes... The experiments demonstrated reduction in LF nulls in the audience using stereo reproduction Decorrelated left + right signals cause less destructive interference A seemingly obvious conclusion that appears to have been overlooked by many in the industry
GOALS + CHALLENGES SINGLE SUBWOOFERS SUBWOOFER CLUSTERS HORIZONTAL ARRAYS VERTICAL ARRAYS SIGNAL PROCESSING TECHNIQUES PERCEPTUAL CONSIDERATIONS RECOMMENDATIONS
RECOMMENDATIONS 1. SUBWOOFER SPACING No more than ½ wavelength Keep array as wide as possible for best coverage 2. POLAR REPONSE OPTIMIZATION 3. GROUND-BASED VS. FLOWN ARRAYS 4. SIGNAL PROCESSING 5. PERCEPTUAL CONSIDERATIONS
RECOMMENDATIONS 1. SUBWOOFER SPACING 2. POLAR REPONSE OPTIMIZATION Cardioid pattern w/2 omni subwoofers More precise control = subwoofer clusters Be careful of polar response vs. frequency! Delay, positioning compensation and/or amplitude shading can be used to optimize coverage 3. GROUND-BASED VS. FLOWN ARRAYS 4. SIGNAL PROCESSING 5. PERCEPTUAL CONSIDERATIONS
RECOMMENDATIONS 1. SUBWOOFER SPACING 2. POLAR REPONSE OPTIMIZATION 3. GROUND-BASED VS. FLOWN ARRAYS Flown arrays = better front-to-back consistency Flown arrays = loss of Waterhouse effect Both varieties exhibit same spacing issues Sightlines + rigging capabilities must be considered 4. SIGNAL PROCESSING 5. PERCEPTUAL CONSIDERATIONS
RECOMMENDATIONS 1. SUBWOOFER SPACING 2. POLAR REPONSE OPTIMIZATION 3. GROUND-BASED VS. FLOWN ARRAYS 4. SIGNAL PROCESSING Don t over EQ! Consistent polar response w/all-pass filter Always be sure to time-align system Various DSP algorithms available to help with headroom and coverage consistency 5. PERCEPTUAL CONSIDERATIONS
RECOMMENDATIONS 1. SUBWOOFER SPACING 2. POLAR REPONSE OPTIMIZATION 3. GROUND-BASED VS. FLOWN ARRAYS 4. SIGNAL PROCESSING 5. PERCEPTUAL CONSIDERATIONS Stereo subs might not be essential, perceptually Stereo subs help with consistent audience coverage Sharp transients are essential!
THANKS FOR LISTENING! ANY QUESTIONS? The simulation software used in this tutorial is free to download at: www.adamjhill.com