Personal Psychoacoustics: A Journey towards Great Sound & Product Development

There have been a lot of articles and studies on the subject of psychoacoustics and the relative audibility of various acoustic phenomena and artifacts. Which measurements correlate most closely to the listening experience? Is phase really audible? Can we detect low levels of harmonic distortion and is it objectionable? What’s more important to a speaker’s sound—first arrival on-axis or total radiated far-field power response? What subwoofer crossover frequency range is the most critical for achieving a seamless sonic blend with the rest of the system? There are all kinds of questions like these and countless expert opinions, some supported by objective measurements and some totally subjective.

I was in the consumer electronics/musical instrument business for well over 40 years, with extended stays at companies such as Panasonic, Bose, Boston Acoustics, Atlantic Technology, Zildjian (the number one cymbal maker in the world), and inMusic (parent company to 17 professional music brands like Alesis keyboards, electronic drums and monitor speakers, Akai Professional keyboards, Alto Professional concert loudspeakers, and several others).

I have had a very direct hand in developing, voicing, and marketing some of the consumer electronics industry’s biggest-selling and most visible products, such as the Bose Wave Radio, the Roommate series of powered speakers, the AM-5 II, AM-3, and AM-7 speaker systems, virtually every Boston Acoustics home loudspeaker between 1992 and 2003 (the VR floorstanders, the CR bookshelf speakers, MR sub-sats, their in-walls and all-weather speakers, subwoofers, etc.), and likewise every single Atlantic Technology product between 2003 and 2013.

Bose AM-5 Series II

Do I have a so-called “Golden Ear?” I would never be so presumptuous or arrogant as to say something like that. Have I developed a reliable, consistent, and discerning ear over the years? Yeah. Do I know what I’m listening to and what to listen for? Yup. Have I had a major hand in or directly voiced a lot of big-selling products? Oh yeah. “My” products have sold in the millions. Millions. And along the way, those products have received more 5-Star, Editor’s Choice, Best in Class, Product of the Year, Top Rated, etc. awards than I ever dreamt possible. Very satisfying. Very satisfying indeed.

Quick aside—Before anyone jumps up and down and howls, “I wouldn’t be so quick to boast about having voiced Product XYZ from the Acme Audio Company! I hate that speaker. Everyone knows that it’s junk!,” let me add this: I worked for these companies, I didn’t own them. Not every single product was designed according to my personal philosophy. Some were but some were not. I did my development, voicing, and marketing work within the confines and context of that particular work/corporate environment and I tried to deliver the most credible, honest, and valuable results I could to the company. Some of the models I worked on over the years were completely and totally “mine,” right down to the brand of capacitor and the types of fasteners. Some were, shall we say, less so.

However, regardless of the degree of personal imprint that I put on the 1000’s of products I was associated with over the years, one thing is consistent: I listened to them all. In excruciating detail, for hours and hours on end. By themselves. Compared to other models in our product line. Compared to competitors. I listened and listened and listened. With all kinds of music, every genre, every kind of recording, every recorded format. I listened alone. I listened with other people and shared impressions and opinions. I listened in engineering labs, in perfectly-dimensioned listening rooms, and also in lousy acoustic settings. I took developmental prototypes and finished production products alike home and listened in my familiar surroundings, comparing them to whatever home speakers I had at the time and knew so well.

Sure, I looked at the frequency response graphs, the crossover schematics and voltage curves, the distortion curves, I analyzed the time-domain data and evaluated the off-axis behavior and the near-field/far-field response, the ‘with grille/without grille’ measurement comparisons, you name it.

But above all, I listened. And listened some more. In the end, that’s more important than anything.

So……what follows is my personal list of listening impressions and opinions about various psychoacoustic phenomena, all based on my extensive experience in this business, developing and evaluating new products. I am not presenting the following opinions as being ironclad, never-to-be-disputed truths. Rather, these are my personal opinions and impressions. One thing I’ve learned the hard way over many, many years of doing these things: Never take yourself too seriously and never stake out an inflexible position. New information and listening data always make themselves known. Some of it confirms your currently held opinion. Some of it shatters your previously held positions. (Along with your ego!) You gotta learn to roll with it and rise above all those petty emotional considerations.

Therefore, in no particular order, here are some of the most notable listening impressions I’ve formed over the years:

Frequency Response is the most important determinant of sound quality

There are a lot of audio measurement and listening factors, but to my mind—and ear—it all starts with frequency response. If the frequency response is off, you’re doomed from that point on and nothing can save the situation.

For electronic components like amplifiers, pre-amps, and CD players, this is obvious and should be automatic. The response of these units has to be dead flat, ruler flat, like within ± .25dB from 20 Hz to 20 kHz, flat. If it tilts up or down by more than 1 or 2 dB over a major part of the audible spectrum, you’ll hear it. This is where those vague but oft-used terms like “thin” or “dark” come from. And note that the component itself does not have to be at fault for there to be a frequency response error. If the input/output impedance or phase characteristics of the electronics don’t line up with each other properly, the result can be a somewhat mangled system FR, even if each individual unit is behaving correctly. Ditto the power amp (or power section) and speaker: Incompatibilities can result in FR variations. When those so-called “Golden Eared” publications wax on endlessly about Amplifier A or Amplifier B having a transparent, liquid midrange or a slightly gray, veiled quality, what they are often reacting to is a frequency response aberration caused by that particular system’s components’ interactions. Frequency response, baby. That’s what matters most. Not some mystical, unmeasurable quality. Just good ‘ole FR.

E. Brad Meyer (long-time audio expert and at the time—1991—President of the Boston Audio Society) wrote a fascinating, well-researched and meticulously documented article for Stereo Review (June 1991) showing how different speaker loads did indeed cause the frequency response of different expensive amplifiers to vary in a clearly audible manner. Nothing was malfunctioning; it was the interaction of the components that caused the audible anomaly.

 

Grossly audible amplifier frequency response variations caused by amp-speaker interaction

For speakers, this is even more true. A smooth, even response over the majority of the spectrum is the key to listening quality. And note this: In a speaker, response peaks are far more objectionably bad sounding than dips. Give a speaker a 4dB peak at, say, 5kHz and it’ll sound grating and annoying as all get-out. But if that same speaker has a dip of 4dB at 5kHz, it sounds “relaxed” and “unboxy.” Frequency response dominates the sound character, more than power handling, off-axis dispersion, distortion, deep bass extension, or anything else. Sure, some of what I just mentioned are actually part of a speaker’s frequency response to some degree, but you know what I mean. I’m talking about on-axis anechoic response 1w/1m, the speaker’s true, essential character. If it gets that right, it’ll sound good. If it botches that badly, it’s dead in the water. When I was voicing speakers, I never lost sight of the fact that smooth frequency response was essential and a peak was worse than a dip.

Arendal 1723 Frequency Response family of curves

A sub-sat speaker system is never as good as a well-designed full-range tower

I was the Product Development Manager behind what I immodestly consider to be the best “three-piece” sub-sat speaker ever done: The Boston Acoustics Micro Reference MR-90. At the time (mid-1990’s) BA did all their own driver design and manufacturing in-house, in Peabody Massachusetts. They had some truly outstanding transducer engineers and some great system engineers. As the person who conceived and defined the product, I wanted the 2-way sats to be as small as practical, and that requirement dictated a sat midrange driver no larger than 3 ½ inches. Tough to get a 3 ½-incher to deliver much in the way of low frequencies, below, say 180 Hz or so. If the sats only made it to 180 and the sub was going to be limited to around 120 Hz max (for localizability considerations), then we’d be left with that big 120-180 acoustic hole, just like you-know-who. Nope. Not acceptable.

However….Boston Acoustics had a secret weapon. He was Dave Cahill, our head Transducer Engineer. Dave was great, the best driver guy anywhere. He could design drivers that really sounded terrific, did exactly the job you needed them to do, and were inexpensive and easy to manufacture to strict tolerances in huge volume. Sure, those “boutique” driver companies have some nice stuff, with their cast frames and gold-colored cones, but they cost a friggin’ fortune. We needed a midrange with dead-flat response, bullet-proof reliability, and was simple to build for, like, $4 bucks.

Dave did it. Much more than that, the driver he designed for the MR-90 sat was a 3 ½-inch woofer, not a midrange driver. It had a really long throw, could take some serious juice, and had a nice, low Free Air Resonance while still maintaining pretty good sensitivity. (Ok, we’re going waaaay down into the weeds here, but you deserve the full story, and stuff like this never gets talked about these days, so I’m gonna tell you, so the story isn’t lost forever.):

Boston Acoustics’ production tolerances were really tight (driver-to-driver QC was ±1dB!), really great. Their meticulous production techniques meant that they could keep the voice coil gap very small, so the actual physical distance between the voice coil and the magnet was as small as possible. Other companies—more prone to slightly out-of-perfect-round voice coil formers or more wary of small metal burrs floating around the production environment (a very common thing) finding their way into the magnetic gaps—kept wider VC gaps, which would let these kinds of small mishaps just get by without any problem. But here’s the bottom line—with a tighter VC gap, you get more magnetic strength in the gap than you do with the same magnet and a wider gap. So, all things being equal, with more magnetic motor strength, you get better (higher) driver sensitivity. And with a tighter gap, you don’t need as big and expensive a magnet to get the magnetic strength you need.

This kind of thing doesn’t show up in the lit. It doesn’t show up when a person picks up the driver with its impressive cast basket and goes, “Oooooo….” But this is what makes a true high-quality driver, one that gets the absolute maximum performance out of a given dollar expenditure. This is what good engineering and production are all about. That was BA in the 1990’s.

So, that 3 ½-inch sat driver, its woofer, went down to 120Hz no sweat. It met up with the subwoofer with no gap, no acoustic hole. The sat used BA’s best tweeter, their one-inch aluminum VR tweeter with their Amplitude Modification Device. (The AMD was a small plastic ‘bridge’ over the dome with ducts of varying lengths to filter out the aluminum dome’s resonant peak. It really worked, too.)  The result? It was the best sub-sat system ever done up until that time and it may still be the best. It was the only three-piece system I’ve ever heard that could compete with truly great conventional speakers.

Boston Acoustics Micro 90 sats

It could compete, but it was not better. As a matter of fact, it wasn’t quite as good as a fully integrated speaker (a medium-sized tower) that had roughly the same frequency range. To my ear, any of our tower speakers at the time (like, say the excellent 2 ½-way VR-30, which went for around $800/pair in 1994, compared to the MR-90 system, which I think was $700 bucks when introduced in 1996) sounded a little better, more unified, more coherent and “whole.” The VR-30 and MR-90 used the exact same 1-inch aluminum VR tweeter and they were both designed by the same engineer (Dave Fokos, a brilliant guy), and they were both voiced by me (and Dave).

 

Boston Acoustic VR-30 tower speaker

So, things were as consistent and apples-to-apples as possible, right? A perfectly valid comparison. The VR-30 wins, hands down. Not by a huge amount, but it wins. Same thing a few years later. The MR-90 sub-sat is still in the line (because it was selling so well) and now there was also an MR-100 with 5 ¼-inch 2-way sats at a higher price point.

Same result. Even though you think “by the numbers” that the MR-100 and its 5-incher would completely fill in any gap and sound perfectly seamless, it just doesn’t work that way in the real world. You can measure till the cows come home and try to “prove” to me that at the listening position, there are no gaps, no holes, no phase issues with a well-designed 3-piecer. Annnnnnntt! (Game show buzzer sound.) You lose! To my ears, the measurements just don’t capture everything that’s going on. A full-range tower sounds better, more unified, than the best three-piece. And the MR-90 and MR-100 three-piece systems were the best of their class.

The subwoofer crossover has to be lower than 50Hz to be truly seamless

This topic is related to the above. Let’s say you’re adding a sub to a critical-listening music system. Full-range music, intended to be realistic and lifelike, not a home theater, where the LF effects have no real-life analogue. (An exploding Death Star does not exist in real life, so there is no way to evaluate the realism of the speaker’s reproduction. You can evaluate its clarity or its articulation or its freedom from obvious distortion, but you cannot evaluate its realism if it doesn’t exist.) However, music exists, it’s real, especially acoustic music, like jazz or classical or vocals.

My 40+ years’ experience tells me this: If you’re adding a sub to a high-caliber music system, you better cross it over below 50Hz (with a sharp LPF slope, preferably 24dB/oct) or it’ll sound like a 3-piece system. I have had countless subjective experiences that confirm this to my total satisfaction. The latest example is my current system—When I moved my Legacy Signatures into a different room, their extreme low bass suffered. Recordings that I knew very well, with content below 30-35Hz, were much weaker in the new room. I added an SVS-3000 SB sealed sub. I experimented with the LPF setting and not until I settled on 40Hz was the sub truly inaudible as a separate entity. At 60Hz, it was awful. At 50, it was ok. At 40, it was perfectly integrated. Let me repeat: awful at 60, not 80 (THX standard).

Editorial Note on Blending Subs with Tower Speakers by Gene DellaSala

Steve raises valid points as the overlap between large speakers and subs playing bass can become localizable and non fluidic sounding IF the crossover is set too high and the system isn't properly time-aligned and EQ'ed. In my experience, the crossover point can be raised higher when using full-range towers IF you're running two subs up towards the front of the room in close proximity to to main speakers or corners. Then you can also use EQ to flatted the boost you get with the mains and subs exciting room modes at overlapping frequencies. In most cases, without the proper measurement tools and know-how at your disposal, it's usually better to use bass management and let the bass duties fall on the responsibility of your subs with a HPF applied to your speakers.

Wide-range frequency response is far more important than spatial effects

This topic was really important in the 1990’s when “Home Theater in a Box” (HTiB) systems first became popular. As the home theater craze started to take off, lots of low-end, mass-market companies began to offer multi-channel “surround systems” with a little shoebox-sized subwoofer and four or five small plastic-enclosure speakers that the user could place around their room. By virtue of having speakers widely spaced around the room, these systems did deliver some degree of three-dimensional spatial character, with a fair amount more acoustic depth than just an L-R stereo system.

But did they sound good? No. They sounded lousy.

Samsung “Home Theater in a Box” system

When I was asked what was the best, least expensive way to get into home theater, I’d tell people just get a hi-fi VCR (this was before 1996, before the DVD), rent Dolby Surround VHS tapes and run the VCR’s audio outputs into their stereo receiver’s Aux inputs, then play the movie soundtrack through their stereo system. Even if they were still using that Pioneer SX-780 receiver and Advent speakers from their 1980’s college days, the actual quality of the sound—both the LF/HF frequency extension and the accuracy/smoothness of the frequency response—would be incomparably better than any so-called Home Theater in a box with their anemic 5 ¼-inch “subwoofers” and cheapo plastic satellite speakers. Many friends of mine took my advice and were very happy with the results. The dialog intelligibility and bass impact of a well-recorded movie sounded extremely good when played through a solid two-channel stereo system. Really good. Those inexpensive HTiB systems were pure junk, and they used the sonic trickery of spatial  “depth” to fool prospective customers into buying them.

 

Home Theater, pre-1996: Hi-Fi VCR + Dolby Surround VHS movie

At any price level—even with a lower-end but totally legit home theater component system—it’s my contention that wide-range, tonally accurate sound trumps three-dimensionality and directional cues every day of the week.

Bass THD in loudspeakers is inaudible for the most part

This one always amazed me. Way back in the 1960’s, the industry’s leading speaker company--Acoustic Research—published very detailed and credible technical data about their speakers, including frequency response, off-axis response, total radiated power response and bass THD at different drive levels. AR was the only speaker company to publish such thorough and complete specifications and their example of honestly in both revealing their speakers’ actual performance and showing the industry as a whole how it should be done remains a model even to this day. Very few speaker companies follow suit and share such intricate data. They should, but they don’t.

One very interesting thing about AR’s published bass THD measurements was that at higher drive levels (10 and 20 watts, as opposed to the more commonly cited 1-watt level), the distortion in the really deep bass  (below 30Hz) rose to levels that one would never tolerate in an amplifier—over 5%, approaching 10%. Yet, when listening to their speakers playing regular program material—even at high SPL in the living room, 100dB or better—one never noticed any objectionable distortion. The bass always seemed clean and articulate, even though at those high SPLs, it was undoubtedly well over 5%. (And just to be clear here—AR’s low bass distortion was the best in the business in those days, by more than a fair margin.)

(For more information on this topic, see Audioholics’ article “The Audibility of Distortion at Bass Frequencies.”)

This phenomenon was confirmed to me many years later when I was at Boston Acoustics—using distortion analyzers that were light years more advanced than anything AR had in the 1960s—we measured the bass THD of BA’s best speakers when playing at high SPLs. Once again, it was well over 5% in the deep bass, even more than 10% when the going got really loud. And once again, it was not audibly objectionable or specifically noticeable as distortion.

How come? Well, two reasons come to mind:

1. The human ear is nowhere near as sensitive to bass as it is in the midrange, around 1000-2000Hz. In the heart of the midrange, people are sensitive to the most minute level changes, FR aberrations, phase shifts, and the like. But in the bass, your hearing is not as sharp and perceptive and it tends to just gloss small things over.
2. A well-designed, correctly behaving speaker produces mostly lower-order THD, mainly 2nd and 3rd-order distortion. 2nd-order is an octave above the fundamental and will usually not be perceived as “distortion” because the pitch is the same. Most Western music is based on 3rds, so again, a 3rd is harmonically related to the fundamental tone and therefore may not be thought of as being musically out of place.

If a speaker’s mechanical suspension (its surround and spider) is extremely non-linear (especially at high power input levels), then that speaker might produce higher levels of upper-order THD, especially odd-order THD, and those distortion products are extremely dissonant-sounding and will be readily perceived by the human ear as distortion. This is a theoretical advantage of true acoustic suspension speakers over vented, because the restoring force that they provide to the woofer cone is the perfectly linear compression and expansion of the air inside the sealed enclosure. In contrast, vented speakers rely on the mechanical linearity of the woofer’s surround and spider.

This was far more of an issue 50 or 60 years ago when design technology and manufacturing methodology and precision were so rudimentary compared to today. The linearity of compressing/expanding air made up for a lot of design and assembly sloppiness in those days, so acoustic suspension speakers like AR, KLH and Advent had bass that was far superior to most ported speakers of that time. That advantage has pretty much vanished these days, because today’s woofers are designed and built so much better than before and computer-aided design has done away with essentially all of the rough “guesswork” that plagued many vented speakers in the 50’s, 60’s and 70’s.

AR technical FR and THD data from 1971. AR published comprehensive, unretouched curves and graphs that no other speaker company of that time period had the nerve to do. Even today, you don’t get this kind of data from the manufacturer.

Note that at a 10-watt drive level (upper 90’s dB SPL in a normal listening room, which is quite loud), the woofer’s THD is barely 3% at 30Hz. Vinyl LPs didn’t go lower than 30Hz. At a 2-watt drive level (90dB in a normal listening room), THD is under 3% all the way down to 20Hz. 3-5% THD at 20Hz is great for a speaker, but we wouldn’t tolerate it in an amplifier.

This leads me to my next observation. It’s related to the relative inaudibility of bass THD but it’s different enough to merit its own section. So, here it is:

Mechanical noise is the most audible bass distortion artifact

As we’ve discussed above, the human ear easily tolerates relatively high levels of THD from a woofer.

But…the ear is pretty darned unforgiving when it comes to hearing and recognizing things that shouldn’t be there. Yeah ok, maybe 2nd-order THD gets a pass because it’s just an octave above and doesn’t necessarily sound “bad.” It’s just harmonizing with the fundamental, like a singer vocalizing in perfect tune with another singer.

However, toss in the hard <<thwack!!>> of a voice coil bottoming against a backplate and, man! You’ll sit up and take notice. How about when the woofer is straining to move too far and it overexerts, causing a sharp ˆ<<snap!!>> to emit from the surround. And ya gotta love those non-flared ports and the chuffing of the air that sounds like an out-of-shape runner gasping for breath as he struggles up Heartbreak Hill at the 20-mile mark of the Boston Marathon.

 

Flared and unflared bass ports

At BA, we discovered every type of mechanical noise there was, especially in subwoofers. That aforementioned MR-90 sub-sat system? It used a 75-watt powered 8-inch sub. The amplifier in early prototypes wasn’t sealed off from the rest of the interior of the enclosure. At high excursions when the woofer was really moving air, we heard a “whistling” that we couldn’t identify at first. Know what it was? It was the air being forced through the unused set of RCA connectors! Unintentional small “ports.” Not only did they whistle, they hurt the bass response a little with additional untuned venting. The very first production run of that sub had silicon goop all over the inside of those connectors to seal them from the outside air. Later production units had a nice small plastic housing covering the entire amp board to seal it off.

PCB buzzing, amp plate rattling, enclosure panel “singing,” you name it. Low frequency reproduction can set all kinds of mechanical noises into motion, and they are BAD!!  Mechanical noise is a far worse type of bass distortion than modest amounts of musically related THD.

Editorial Note on Mechanical Noises by James Larson:

Agreed. Flutter noises, mechanical noises, and port noises are all much more audible than THD in normal circumstances. However, harmonic distortion can be surprisingly audible if you are listening for it specifically. As an experiment, run REW on your sound system. Run some deep bass sine wave test tones, and raise the signal until it has some tonal changes. You might be surprised at how quick those tonal changes crop up. Use REW's RTA screen to see the levels of the distortion products while you do this.

But again, in real world listening, even relatively high levels of THD would be nearly impossible to hear, unless you knew exactly what to listen for.

One of the advantages of a rear-mounted port or bottom-mounted port is that port noise is nicely masked, since, as a higher frequency noise, it is more directional, yet the bass produced by ports is omnidirectional and so can still be heard clearly.

Related to flaring, JBL published a paper that examined port turbulence, and they found that the overall port shape mattered more than different types of flaring. Basically they found that its best when nearly the entire tube is flared or has an elliptical shape.

It turns out that their solution was close to one that Bose had already come up with, so Bose sued them for patent infringement.

How high in frequency is really needed for true high-fidelity sound?

In the 1990’s, Boston Acoustics had three models of outdoor speakers. The Voyager was the best one. A 5 ¼-inch 2-way in a rugged, heavily ribbed plastic enclosure, the Voyager was the real deal. Bona fide hi-fidelity performance in a bullet-proof enclosure. One famously survived after being thrown several hundred feet away when the porch it was attached to was destroyed by Hurricane Andrew in Florida in 1992 or 1993. The owner retrieved the speaker from down the street, hooked it up and it played perfectly. We used it in a magazine ad.

There was also a series of speakers just below the Voyager called the Runabout. There were two models—one with a 4 ½-inch full-range driver, and one that was a 4 ½-inch 2-way, with one of those inexpensive ¾-inch hard dome tweeters. The Runabouts were weather-resistant, not weatherproof like the Voyager, but they were perfectly fine little outdoor speakers.

The full-range Runabout sounded ok. Not great, just ok. Its full-range driver was pretty good, but there’s a limit to how high up a single driver is going to go. This one had a curved cone, which gave it somewhat better HF extension than a flat-cone driver would have (because the curved profile stiffens the cone a bit, so it stays more pistonic to a higher frequency with less cone flex and breakup), but the difference between the full-range and 2-way Runabouts was quite stark. One was real hi-fi and the other was pleasant but well below the true quality sound threshold.

BA’s president wanted to know why, quantitatively, where that dividing line was. So we put the two speakers side-by-side in mono (always, so as not to be distracted by stereo “imaging” considerations) and played a wide variety of program material through them, A-B’ing constantly. On the 2-way, we had a very sophisticated sharp-slope filter and we could progressively limit its HF extension until we found the point at which we no longer considered the sound to be real “hi-fi.”

Wide open (no filter) the 2-way Runabout made it to around 17 or 18kHz on axis, no problem. The full-range Runabout was pretty good up to around 9 or 10kHz. Good performance for a single 4 ½-inch driver, but you’d never mistake one speaker for the other.

We replayed the program material and reduced the HF limit in 500Hz steps: 18kHz, 17.5kHz, 17kHz, and so on. Know where the “high fidelity limit” was? It was at 13kHz. That doesn’t mean that we couldn’t hear above 13k or that we couldn’t tell the difference between 13k and 17k. We definitely could. (At the time, we were all guys in our mid-30’s-early 40’s and our HF hearing was just fine.), But listened to by itself, the 13kHz limit sounded just fine. Perfectly believable and convincing. The 9-10k limit of the full-range Runabout sounded compromised, all by itself, no A-B needed. But at 13k, things opened up, there was sparkle and sheen and any other audiophile adjective you’d care to haul out. Not at 11k, not at 12k, but at 13k.

I’ve always remembered that and I consider it to have been one of the most interesting, informative, and well-controlled audio experiments I’ve ever taken part in. We all heard it and we all agreed: 13kHz was the point where the high frequency extension became truly high-fidelity in nature. Fascinating info.

Higher-order THD is really audible, lower-order THD is surprisingly benign

This somewhat related to the audibility of bass THD topic, but it expands on it. When I was in college in the 1970’s, I was a member of the Boston Audio Society (the BAS), a group of audiophiles who met monthly to discuss audio matters, share discoveries/observations and hear presentations from manufacturers and other industry notables.

One topic that was popular in those days was, “Do amplifiers really sound different from each other?” One of our more technically adroit members (the late Peter Mitchell, an incredibly insightful audio researcher and physicist) said that he found a hard, repeatable factor that caused amps to sound markedly different from one another. That factor was the spectral makeup of the amp’s distortion products at clipping. Remember, this was 1975, and things that are well understood and accepted now were just being discovered then.

Mitchell took three popular receivers of the day, listened to them at normal levels (below clipping) and then pushed them to the brink of clipping and then further, into “hard” clipping. He had a scope, on which he could observe and analyze each unit’s distortion makeup. He repeated the demonstration for us at the next BAS meeting, using the same three receivers and Large Advent speakers. (I distinctly remember, it was October 1975. My friends all went to a very hip concert featuring Santana and Tower of Power. Mr. Nerd here went to a BAS meeting.) The speakers were chosen because Advents were very widely used at that time, had a reasonably neutral tonal signature that wouldn’t call attention to themselves and distract the listeners, and most important, they were 8-ohm speakers, very easy to drive, so below clipping, you could have confidence that the receivers under test were performing correctly.

Pioneer SX-828 2CH Receiver

Here’s what we heard: The Pioneer SX-828 (50 watts RMS/ch) sounded great below clipping and pretty harsh—but not horrible—once into clipping. The scope showed that its THD at clipping had a good deal of 4th and 5th-order components, in addition to the usual 2nd and 3rd-order.

The AR receiver (also 50 watts RMS/ch) sounded very smooth even at clipping, hardly distorted at all. Its distortion components were almost exclusively 2nd and 3rd-order (lower-order), which are musically related to the program material and therefore not necessarily perceived by the human ear as “distortion.”

The last unit under test was the Marantz 2270 (70 watts RMS/ch). The Marantz line at that time was regarded as a “step up” line, a bit better than your average garden-variety receiver. They had an elegant appearance and feel, with that cool horizontal thumbwheel FM tuning knob and beautiful script lettering.

Sure enough, below clipping, the 2270 sounded great. But once pushed into clipping—even just on the edge of clipping—it quickly devolved into the worst of the three, by a considerable and clearly audible margin. The scope laid it all out: The Marantz had a preponderance of upper-order distortion, a lot of 5th, 6th and 7th-order distortion components and that made the unit almost unlistenably harsh and strident when pushed past its comfort zone. Now, 70 gutsy watts RMS (it could actually put out close to 90 without clipping) was more than enough juice to push any normal speaker to extremely loud SPLs in a regular-sized living room or college dorm room.  But if you asked more from the 2270 than it wanted to give—even just a little—it would howl in screeching protest.

Marantz 2270 receiver

What a great presentation that was. The presence of even modest amounts of un-harmonically related upper-order THD components is far worse than numerically greater amounts of lower-order THD. Take that to the bank.

 (The BAS in 1975 was onto something, which you can read about in our article- Human Hearing - Distortion Audibility Part 3)

 

THD spectrum representation

Far-field power response is much more important to speaker listening quality than currently thought

In the ‘old days’ of the 1950s and ‘60s, the leading speaker company of the time (Acoustic Research or AR) posited that a speaker’s far-field power response was mainly responsible for listening quality. AR designed all of its speakers according to that dictum, believing that the power response at the listener’s ears sitting 10-15 feet away from a set of stereo speakers in a normal living room was what mattered most. AR’s speakers didn’t pay particular attention to near-field response, so things like grille/cabinet edge diffraction, driver interference, etc, were of little importance in their designs. AR’s most famous and biggest-selling models of the day—the AR-3a, AR-2ax, and AR-4x—all had what we’d call today major “errors” in their basic design: interference-prone side-by-side midranges and tweeters, baffles set back such that there was significant cabinet edge diffraction, intrusive grille frame diffraction and interference, etc. This led one noted audio critic to say that the AR-3a was a “cacophony of near-field interference and phase cancellations.”

Perhaps it was, but that didn’t matter to AR. Smooth on-axis near-field system response was not their goal. Their goal was a smooth far-field power response at the listener’s ears. They felt that interference artifacts that might be audible four feet from the speakers would be “averaged out” and totally inaudible 10 or 12 feet away.

Since then, of course, the speaker industry, led by Floyd Toole and his extensive investigations into listener preferences and how those preferences correlate with measurements, has come to the position that a speaker’s first-arrival, on-axis frequency response and the behavior of its resultant early reflections are what’s most important, especially at mid-to-higher frequencies, regardless of the listening distance.  Off-axis response and far-field power response do indeed contribute to overall listener perception and preferences, but not nearly as much as on-axis first arrival. That’s the current thinking.

However….While first-arrival may indeed be the most important contributing factor to loudspeaker’s sound quality, it’s not as dominant a factor as some corners would have you believe. Not by a long shot.

You can prove this at home, in your own listening space. You’ll need some really sophisticated test equipment, available at your local Bed, Bath & Beyond. Dig deep, because this equipment is likely to run you as much as $5.00 bucks. Maybe even $6.00. But to a committed audiophile, such an extravagant expenditure in the pursuit of Audio Truth is well worth it, right?

You’re gonna need a sofa pillow. A nice, generously stuffed sofa pillow, roughly 15” square. If your listening room is fairly large (you’re about 10 feet or more away from the L/R music speakers), here’s what you do: Play some music that’s rich in upper-midrange and treble. Lots of cymbals and percussion, repeating, steady. Well-recorded jazz is perfect for this.

Listen to your system like normal. Sounds great, right? Great definition and detail, lots of sparkling high end, very tonally realistic. You’ve assembled a terrific system, and it sounds wonderful—punchy, crisp but not harsh, beautifully three-dimensional and meaty. The word that comes to mind is “organic.” Your system sounds like it’s alive, living and breathing.

Now, from your 10-12-15 foot listening distance, take your ultra-sophisticated test equipment (that sofa pillow!) and hold it out about 18 inches or so in front of your face, so it’s blocking your sightline to the speakers. It’s a thick, soft, cushy pillow, a high-frequency sponge. Held out in front of your face in this manner, the pillow effectively prevents the direct on-axis “first-arrival” upper-mid/treble sounds from your speakers from reaching your ears.

 

Author sitting in far field, using ultra-sophisticated test equipment

Now lower the pillow and lay it on your lap. Now raise it in front of your face again. Guess what? The sound changes, but nowhere near as much as you thought it was going to, right? It just sounds like you turned the treble control down to about “11:00,” but it’s not anywhere near awful. Know why? Because in the far-field (where you’re sitting) the majority of the sound reaching your ears is the reverberated, reflected sound from all around the room. In other words, what you’re hearing from 12 feet away in a large room is mostly the speaker’s total power radiation, not predominantly its on-axis first-arrival sound.

Ahh, now grab a kitchen chair and sit four feet in front of one speaker. You’re in the near field now, not the far field. Here, the speaker’s on-axis first-arrival sound will dominate. When you hold that pillow out in front of your face, blocking your sightline to the speaker, the sound character changes dramatically, doesn’t it? The upper mids and highs are drastically reduced. When you lower the pillow to your lap, all those mids and highs return in full force.

You just did the test yourself. Don’t let anyone tell you that far-field power response isn’t important in loudspeakers. It’s critically important. You just did the test.

The interesting thing about speaker design, however, is that accurate, well-controlled on-axis/first arrival response and smooth, uniform far-field power response are not mutually exclusive. A good loudspeaker can and should have both. And in fact, the really great loudspeakers do. For clean near-field response, designers need to minimize cabinet/grille diffraction, utilize vertically-aligned drivers, employ the drivers in the frequency ranges where they’re not drastically directional or beamy and just generally follow good, time-proven design practices. If this is done, then chances are the speaker will exhibit both good near-field first-arrival response and a smooth far-field power response. Speakers that measure well in both cases tend to be the best-sounding speakers, in my experience.

Asymmetrical single-sub placement works best

Subwoofer placement, of all things, has become somewhat of a controversial subject. The late, highly respected reviewer Tom Nosaine did a comprehensive analysis for Stereo Review magazine in the January 1995 edition, complete with comprehensive in-room FR measurements, where he put forth the contention that for a single subwoofer, placement in the corner of the room works best. So placed, he said,

Driving all the room modes (length, width, height, oblique, and tangential) simultaneously delivers the smoothest, deepest frequency response with maximum boundary reinforcement. The response of a subwoofer improves steadily as you move it into a corner of a room.

But very often, corner placement is not practical or possible. Maybe there’s furniture in the way. Maybe you do not have (or want) a wireless sub connection and your sub cable isn’t long enough to reach the corner. Non-corner single sub placement is probably the most common scenario. What to do then?

In the 1960s, Acoustic Research (AR) offered these real nice walnut speaker stands for their “full-sized” bookshelf speakers, the AR-2 series, the AR-3 series, and the AR-5. The stands themselves were about 12 inches tall, so the 25-inch-tall AR-3a stood around 37 inches high. Looked good, sounded good. AR gave some very specific placement instructions for those stands: They told the user to make certain that the height off the floor (which was already known, 12 inches), the distance from the speaker to the wall behind it and the distance from the speaker to the nearest sidewall should all be different distances. Even back in the 1960’s, AR understood that an asymmetrical speaker placement with respect to the room’s boundaries would result in the smoothest in-room bass response, with the least amount of destructive dimensionally-based frequency reinforcement and cancellation.

   

AR’s “X” speaker stands from the 1960’s; AR-3a sitting on those stands

If Nosaine’s recommendation of a single-sub corner placement isn’t possible, I’ve always found that an asymmetrical sub placement scheme—consistent with AR’s findings—works best. If the room is, say, 18 x 24 feet, locate the sub (if possible) at non-factor distances along the wall. Try for 5 feet in on the 18-foot wall or, say, 7 feet in on the 24-foot wall. That gives you 5/13 on the 18-foot dimension or 7/17 on the 24-foot dimension. All prime numbers, nothing is an even factor multiplier of the other (the way 6/12 or 6/18 would be) and the placement is nicely asymmetrical. I have found this works quite well, and I’ve set up 1000’s of systems this way over the decades in hotel rooms for training/demo sessions, in dealer listening rooms for retail display, in customers’/friends’ houses, etc.

All the foregoing is said with the realization that a single-sub system is a compromise and optimally placed multiple subs consistently deliver far superior in-room results. Nonetheless, single-sub systems are by far the most common, so it’s useful to look at that scenario.

Forget Absoute Phase

This one is quick. Unless there is destructive phase cancellation in your system that is somehow really messing up that all-important frequency response, you ain’t gonna hear or be bothered by absolute phase. It don’t matter a whit whether the silver or copper speaker wire is connected to the + or - terminals. As long as they’re consistent and the L/R speakers are connected in the same way.

Conclusion

So, there are some of my favorite and well-established personal biases and psychoacoustic beliefs and habits. As always, YMMV (Your Mileage May Vary). I’m fairly certain that all of the foregoing is true and repeatable/confirmable, but, who knows? I could’ve been stubbornly holding onto an irrational belief or two for the last 50 years. Wouldn’t be the first time!

Share your thoughts and opinions of this article in the related forum thread below.