Listening Comparison Of Comb Filtering in Boundary Mounted & Free Air Mics

One of the criticisms of placing microphones next to boundaries is that it produces objectionable comb filtering. This comparison test demonstrates that any comb filtering produced is very subtle and occurs at very high frequencies. The use of moving pink noise  for detecting the phenomenon was suggested on the Nature Recordist list. I performed many "3 O'clock" to "12 O'clock" pans with a boom box speaker around the mics in an arc at a distance of 5 ft. and 10 ft. I selected one sweep from both distances where all three mics exhibited smooth amplitudes.  I used a 744T recorder with 45 dB of gain to record all three mics at the same time.  I had to use a mix of AT3032 and AT 4022 mics which are known to be very, very similar in terms of performance. 

A blind version of the QuickTime movie comparison is available for downloading (3mb .zip)  to see if you can match the samples with the right mic configurations. I'll soon post the ID's below. 

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

Here's the comparison test the samples identified [ download .zip 3mb]
Here's a larger image of the sonogram with more detail.


Follow-up discussion can be found on the NatureRecordist list as well as in Paul Jacobson's assessment below. Curiously, the Perpendicular to Boundary capsule positioning produced the least amount of comb filtering in this test. 

The guesses I received  suggest that any artifacts pink noise sweeping might reveal are difficult to accurately link to the arrays:


Sample A Guesses:  Free Air, Free Air, PBB, Perp2Air, Perp2Boundary
Sample B Guesses:  Perp2Boundary, PBB, Perp2 Boundary, PBB. PBB
Sample C Guesses PBB, Perp2Boundary, Free Air, Free Air, Free Air

PBB3 - Portable Stereo Boundary Mic Array Plan

Finally got around to seeing how PBBx components would scale for a baseball cap. Worn backwards, would it be a little less conspicuous?   I wanted a rig I could use hands-free and hear what the mics are "hearing" without needing headphones. The stereo imagery should have impressive depth and reach. Recordings should play well on speakers as well as headphones. There are only a few small compromises in quality as best as I can tell.

In solving the structural integrity aspects, make sure that the leading edges of the boundaries end where shown. Add no hard, reflective surfaces in between them, even behind the foam.  I'd consider adding a "V" wedges behind the boundaries that attach to the cap sides.  The angles can flex and change a bit, no problem.   You might also want to clip off any part of the brim that sticks out beyond the foam baffle that is under the capsules-- that would create another boundary. Off-set center of mic capsules in the vertical  boundaries .75" from top on one channel and .75 from bottom on the other. For wind-screening, I'd start with stretchable fabric like a sheer Lycra? pulled across all of the mic components. No obstacles allowed on the boundaries themselves. Make sure  the diaphragms of the mics are flush with the boundaries.

Partially Baffled Arrays - Small vs. Medium Size Boundaries

The tree that was holding-up my outdoor S1 array collapsed on it last month and I decided to replace the ad hoc "Perp2Sphere" array with a Partially Baffled Boundary array. There are two qualities in my PBB2 array I'd like to improve upon if I can: (1) More deep bass < 125Hz and (2) A little less "sharpness" in the frequency response between 3-4KHz. A logical attribute to experiment with is the size of the boundaries. 

The boundary area of my new rig, PBMB1 (Partially Baffled Medium Boundary ver 01), is about 9 times as large as the PBB2's (10.25" high X 12" wide vs. 4.75" x 3.6"). The middle section where the baffle is located is identical-- its just twice the height. The PBB2 has AT3032 mics and the medium boundary array has AT4022 mics. The AT4022 mics have more bass response on their own accord so some of the increased bass response will be from the mics themselves and some from the medium sized boundary. To get some sense of the former, see Paul D's 3.4 mb QuickTime movie comparison you can download and assess with QuickTime player.

Note: Paul Jacobson astutely detected that I had a 80 Hz -18dB "roll-off" filter enabled on my 744T recorder thus affecting the small boundary array recording. Even with this error, I felt the test was still useful so I added matching roll-off filtering to the PMMB1's recording. As a result, comparing the arrays in terms of bass performance will have to be resumed in another test.

Method: I set-up and oriented both arrays as close as I could get them in space about 6' off the ground. Both arrays received 60dB of gain which is typical for recording ambience in quiet locations. The PBB2 was connected directly to a SD744T and the PMBM1 is connected to a Symetrix SX 202 external pre and Hammerfall RME interface A-D.

Here's the comparison of the mic arrays as a 30 mb QuickTime Movie to download and play with QuickTime. You can also open the movie in any Quicktime compatible sound editing app which will place the 48K/16bit soundtrack in waveform view. The PBMB1 sample always comes before the PBB2 sample. 

= = = = = = = = = = =

Assessment: There is less "sharpness" in the recording made with the medium size boundary (compare the close sparrow calls at 27 secs and 35 secs)-- so that is good news. Another mentionable plus is animals with lower pitched voices like the distant trees frogs (2:21) and distant coyotes (0:42) receive improved presence or separation from background sounds when recorded to the side with the medium boundary array. Unfortunately, these improvements come with an overall degradation of stereo imaging. The smaller boundary array has better depth imaging, better phantom-imaging between the speakers and better reproduction of sound reflections.  If I can understand how these traits are produced in the small boundary array, I might be able to adapt them to the special circumstances of capturing distant subjects in quiet settings.


The bass content in the medium boundary samples tends to emanate directly from the speakers more so than from the middle of the stereo field as it does with the small boundary array. This creates more horizontal "spread" in the medium boundary's stereo image. One can hear this quite readily in the first section with the distant motorcycle where the location of the motorcycle jumps from hard right to just left of the right speaker in the stereo field.  All small and medium sized boundary arrays suffer poorer very low frequency response compared to omni mics in free air.  According to Crown's boundary mics application guide (pdf p13), this effect is less pronounced for sound waves that are parallel to or "rake" the plane of the boundaries.  Curiously, looking at a map of the setting,  the sound waves from the motorcycle do seem to be arriving on the ~4 o'clock "raking" angle.  An increase in volume is similarly confirmed by louder impulses coming from the 3-4 o'clock range in the localization tests for the medium size boundary.  The Crown literature explains that the bass "lift" from a 12" boundary becomes effective at about 188 Hz (188/D where D= 1 Foot). The lift from the smaller, 3.6" boundary becomes effective considerably higher at 564 Hz.  These different lifts can be seen in a sonogram of the first pair of motorcycle samples and also confirmed when equalization is used on the PBMB1 recording to make it sound, tonally, more like the PBB2.

Its my opinion that small boundary array exhibits less muddiness in the 200-500 Hz range than the medium boundary, and that its higher pitched lift accentuates ~800 Hz sound reflections from a bluff at 2 o'clock thus making this additional spatial detail audible.  A hill blocking direct sound passage along the raking angle probably contributes to the muddiness of the medium boundary array, but even if these sounds waves were uninterrupted, the stereo image would still have exaggerated width. This explanation seems consistent throughout the samples where there is more volume and reverberation nuance in the center of the field with the small boundary and more bass emphasis on the sides with the medium boundary.

I was tempted before I made this new array to only extend the width-- not the height of the boundaries.  That may have been good intuition. A boundary width of around 9-10" should create emphasis around 1300 -1500 Hz compared to the 3.8 K Hz fundamental added by the 3.5" dimension in the small boundary array. This compromise might reduce sharpness and increase warmth with less muddiness. An array with lateral boundaries that start at 4-1/4" with hinged  extensions to about 10 inches would allow me to test the variable more easily. If 10" is too wide, I can trim down the extensions a little at a time.

 

PBBx Sample Recording by Paul Dickinson

Paul Dickinson, a sound artist based in Chicago, is sharing a beautiful stereo recording he made with his DIY PBBx rig in Big Island Lake Wilderness Area of the Hiawatha National Forest in Upper Michigan.  Like some of the beverages Paul prefers to hand-make, he has posted the recording for us to evaluate, "raw and unfiltered."  The ability to do this is a testament to his micing skills and conditions favoring strengths of the array. Thanks, Paul!

Five Types of High Density "Baffle" Foam Compared

-- See bottom of this post regarding the influence of  high density foam thickness on HF absorption --

In the process of studying the frequency absorption influences of five types of open cell, high-density foam, I may have confirmed my suspicion that some sound waves above 500 Hz can pass through the baffle of a PBB2 and interact with the short setback distances of the boundary mounted mic capsules. Surprisingly, the standard reference foam I measured transmitted almost all of the sound energy under 2000 Hz. That comparison is the last pair on this test:

 View the QuickTime movie Comparisons in web browser. Download movie for playing with QuickTime.

So which foam would work best for the PBB2 array?  I'm not sure. The dark gray acoustic foam (1.7 lbs per cubic foot with medium size pockets) accentuates frequencies between 500-1600Hz to take full advantage of the "partial baffling" effect:

But is this truly a beneficial "effect;" does it have a role in the increased "airyness" the PBB2 rig has exhibited?    In theory, sounds above 1200 Hz rely more on timing differences between the ears, so passing more off-axis sound above 1200Hz through the baffle could obscure critical amplitude differences. If this is the case, then more left-right isolation above 1200Hz  might work better and the "Luxury" foam might perform better.  I have a much better idea of what the foam can do, but not what it should do. Simply matching the reference foam either by ear or by sonogram is not enough.  Its also very possible that foam type is moot because a much larger percentage of the sound reaching the capsules does not pass through the baffle.  Thanks to Paul Dickinson for the source of three of the higher density samples tested in this test.  FoamForYou.com.

So, its looks like the next step is to compare the different types of foam in the PBB2 array within a stereo field to see if I can hear the indicated Hz absorption differences and impacts on stereo imaging. Please chime in if you ascertain different conclusions from mine! [For the above test, I used an AT 3032 mic pointed directly between the woofer and and tweeter of a two-way speaker a distance of 12 inches away.  The foam was positioned between the mic and the speaker 1/16" from the front of the mic.]

 = = = = = = = = = = The follow-up test below added 09-2010 = = = = = = = = = = = =

High Densiy Foam Thickness Affect on >1000 Hz Absorption

Click on image to see enlargement.

Michael Billingsley specifies that the SASS's baffle should attenuate frequencies between the sides above 1000 Hz by at least 9 dB. I tested the absorption rates of different thicknesses of high density foam to see how much foam is needed to satisfy this and found that a thickness of 4" -5" should be sufficient. I used 1.5  pound per cubic foot dark gray high-density acoustic foam for this test.

The test also confirms that foam even 14" thick does not appreciably attenuate frequencies below 1000Hz. It remains likely that the off-side passage of sound waves between 500-1000Hz  through the baffle are playing a role in the spatial imaging of this array.  Wavelengths under 500 Hz are shown to be passing around the foam unaffected as expected.

 

Partially Baffled Boundary Array with Improved Tone & Spatiality?

Impressed by the tonal and spatial imaging advantages demonstrated in Andrew Skeoch's and Paul Jacobson's comparison, I made an array combining small boundaries and short setback distances with the partial baffle principle employed in Andrew's array. I set-up this new array with Audio Technica AT3032 mics installed next to another stereo array I've been using for recording ambience and made some simultaneous recordings. The reference "Perp2Shphere" array uses two Rode NT2000 mics in omni mode mounted perpendicularly to a 7" diameter coated foam sphere. I've been happy with the tonality and spatial imaging of this array using two them daily for the past year but, I've kept hope alive that I might find something better.

Here's a 9mb QuickTime movie for browser viewing or .zip download  that alternates clips from both arrays starting with the Perp2Sphere. (The play cursor will line-up better with the change points in the graphic if downloaded and viewed with QuickTime Player. In the sample pairings, the PBB1 always comes second.)

The sound clips cover a wide variety of typical "far-field" recording subjects ranging from very faint, distant events at varied positions around the arrays to very delicate small mammal sounds to louder animal calls and human-produced effects.

 
The difference in the tonal response between the two arrays is immediately apparent to me. As hoped, the PBB1 exhibits a high frequency "lift" as Andrew's array does. From initial indications, the lift appears to begin around 500 Hz and extend to around 4K Hz.  I feel this lift contributes a sense of greater detail in distance sound sources and their reflections, In this stereo mic array comparison, its remarkable that the chorus of Peeper frogs a 1/2 mile away on the right channel "jumps out" in the PBB1 array's image but is very restrained in the Perp2Sphere's image.  The impact on the coyote and owl exchanges is not as profound but it helped me identify the number of callers and because the echoes are brighter, locate them with more accuracy.  There are several dramatic contrasts that occur with closer sounds. Consider the very delicate, high frequency rodent vocalizations at 2:12 where the PBB1 array provides much more relief. The audible differences in the passing deer at :30 and a small branch that drops on the right at 3:20 are also quite telling.  Both arrays were 7' from the ground, placed right next to each other and at very similar distances from the sounds included in the comparison.

 Spatially, the performance differences are equally striking. I detect additional sound reflections and more distinct echoes in the PBB1's samples. Take a listen to the additional slap back echo from a bluff on the left in the coyote bark at 1:30 and the additional "ring" and tree fluttering in the echoes from the PBB1's hand claps starting at 4:41. With the louder events like the passing trucks, cars and helicopter, I'm much aware with the PBB1 that these sources are coming from behind the mic arrays and that much of the sound is reflected from hillsides in front of the mic.  There's more upper harmonic content in the decay of all of the echoes once you are alert to it.

The width of the stereo image feels "tighter" on first impact than most spaced omni arrays I've studied, but with repeated listening, I think its because the the bass is very centered and in phase. The higher register sounds localize and spread very similarly to that of spaced omni Perp2Sphere's stereo image. I choose the Perp2Sphere over other popular arrays because I felt it improved "airyness" and depth, but when I listen to the difference in the distance imaging in the snipe calls from 1:32-1:47, it suggests to me realize the PBB1 could be in a different league.  I wanted to think that the PBB1 renders sounds more precisely across the stereo field after hearng the hand claps. but I think in many cases, the additional high frequency content aids localization more.  The shift of the owls from left to left center at :54 seems to be more the result of the tonal lift and increased definition of the sound reflections from the hillside than an improvement in sheer "stereo" abilities. (Its also helps to know that the direct sound of the owls is blocked on the left!) The additional high frequencies can also aid in localizing close sounds such as with the deer passing. 


As for the "narrowness" that the centered bass tends to produce, I found the array very responsive to using M-S processing in post. This allowed me to spread the image by lowering the volume of the center 1-2 dB and to use inserted EQ to lessen the "dirty" roar of the exaggerated traffic and human-made sounds that exist with the bass more towards in the center of the field without affect side content. I tried the same trick with the Perp2Sphere array, but it required a lot less EQ with the PBB1 and the result on the Perp2Sphere seemed "hollower."  There is less bass impact in the PBB1 array's "sound," but it seems smoother and less jumbled.

There is no question that the PBB1 array has a profound effect on tonality and spatial imaging. How much of it is the mics and not the arrays? The frequency response of the NT2000 in free air is actually brighter than that of the AT3032 in free air. The tonal difference in the lowest octaves in open air are very modest compared to that heard when the mics are mounted in the arrays.  I prefer the lower mid-range response (125-700 Hz) of the large diaphragm NT-2000's mics when mounted in free air as with most spaced omni arrays, but the PBB1 array gives a tonal advantage in this crucial range to the AT3032's. [I would install the NT2000's in a PPB1 array in a flash, but the capsules must be flu-mounted and the Rode's have "side address" capsule orientation. I may try to flush- mount some NT1-A capsules using the forward-facing mod, but I fear I'll get lots of RFI and noise without full shielding that will tend to obstruct the boundary. And they're cardioids.]


The heightened brightness that the PBB1 array seems to be especially useful in "far field" recording because it makes up for some of the loss of high frequencies absorbed by the air.  This pre-emphasis can mean less HF boosting in post to hear sounds clearer and less end-result audible hiss from mic self-noise.  I did notice, however, that robust, treble-rich sound sources like morning birds were a little brighter than I would have preferred when they took up perch within about 30 feet of the array. Others may feel this is an advantage too. I'm prompted to think of Walter Knapp's tall light stand that he has adapted to get his mic arrays further away from amphibians and to improve its "vista" of more distant subjects. 

It occurs to me that one could  place this mic array anywhere where one expects subtle, high frequency sounds and probably get amazing results. Insects, frogs and other sounds that have lots of mid and upper range frequency content will probably sound extraordinarily close with proximate micing.


Lots more to be learned about this array. hopefully, this test and my comments provide a useful start!


Addendum A
Some recordists have wondered if the AT3032 mics have an inherent treble lift compared to other omni mics. Paul Jacobson suggested that we could take another listen to the comparison recordings Walt Knapp made a few years ago featuring AT3032's and MKH-20's mounted in a modified SASS arrays. To facilitate this comparison, I've made a movie with short excerpts from outdoor presence Walt recorded at his rural home. The recordings were made back-to-back, not at the same time, so I selected the two parts which appeared most similar using sonograms. The clips are a bit boring, but I believe a potential difference in tonality and potential impact on spatial clarity will be easier to evaluate without close, discrepant bird songs.  

Blind Comparison

MKH-20 mics vs AT3032 in SASS (2mb) QuickTime Movie for browser viewing and for download.

I renamed the audio files with random strings and recorded the identities with screen captures before making the comparison movie so it would be a blind test for me as well. I'll going to wait until we have discussed our impressions before we find out which is which.

= = = = = = = =

A group of group of five field recordists proved to have pretty good ears. At least Sennheisier would like to think so:
Array                                        Selected as Preferred           Correctly Identified 
  A.   AT3203  mics /SASS                  1 time                                3/4 times       
  B.    MKH-20 mics /SASS                  3 times                              3/4 times 
The fifth recordist described the performance of the two arrays as a "tie" and didn't venture a guess as to which  was which.  

The recordists were not able to determine whether 3032's have an inherent "brightness" from this test, but if it does, it would probably be audibly small in comparison to the "lift" of the PBB1 array. Here's a QuickTime movie [2mb download] with a sonogram comparing an AT3032 in free air  to one in a PBB1 Array.

My thanks to Paul Jacobson, Curt Olson, Rich Peet and Allan Haighton for input and Walt Knapp for the original comparison files.

Frequency Response Differences Between SASS & Head-Space Parallel Boundary Mic Arrays

Following the A/B comparison of Andrew Skeoch's and Paul Jacobson's comparison of their SASS/MKH-20 and Head-Spaced Parallel Boundary/Audio Technica AT-3032 mic arrays, I was curious to see if I could get a better sense of the tonal difference between the two rigs. It proved to be possible to approximate the difference fairly well using parametric equalization. This QuickTime movie contains that assessment:

(A) QuickTime movie comparing clips from the two rigs for viewing with a web browser (13mb).  Requires QuickTime Plug-in.

(B) QuickTime movie comparing clips from the two rigs for downloading and viewing with QuickTime, MPPEG Streamclip and other Media players.  (13mb .zip)  

I chose sound clips with distant sound sources because they pose a greater challenge for stereo arrays to render.  Be sure adjust the playback volume of the movie to a comfortable level.

Both arrays use omni mics with comparable self-noise and frquency response, especially in the lower midrange where the response difference is showing here. Here's a sonogram showing the Hz response and self-noise of an AT-3032 mic in comparison to MKH-40 and MKH_80 mics.

Asssessment:

"Array C" is the Head-Spaced Paralle Boundary Array with AT-3032 mics and "Array D" is the SASS array modified with MKH-20 mics by Walter Knapp. 

I feel that the SASS array's impact on spatial clarity over the HSPB mic array is positive. Evaluated on my Mackie HR824MKII speakers, there seems to be more acoustic "airyness" without excessive resonance and more audible detail in distant sound sources.  I was able to adjust the EQ of the HSPB array to resemble that of the SASS array using using two parametric curves effective between 400 Hz and 1900 Hz. To match the tonal response of the HSPB array with EQ adjustments to the SASS's recording required two more curves.  It was revealing to hear details in the backgrounds become less audible as I made those adjustments to the SASS's recording.
 

The HSPB array has a 1" setback or distance from the front edge of the boundaries to the mic capsule.  Rob D.