Katz’s Corner Episode 12: Throwing Down the Gauntlet

This story originally appeared at InnerFidelity.com

Hello everyone! I’m sure some of you have noticed my absence from this column. I will try to be here more often, but I’ve been very busy, mastering some great music! I’m also designing a new mastering product that will be manufactured by a major pro audio manufacturer, which has been taking up my non-copious free time. I also do a bit of Facebook-based audio education and sharing as some of you may have noticed.

Recently I upgraded my loudspeakers. These new loudspeakers truly up the game. You might ask, “Why is Bob talking about loudspeakers on our headphone-based site?” Please bear with me while I lead up to the answer. My previous loudspeakers were no slouches—A-grade mastering quality. These were the fabulous Revel Ultima Gem 2’s, crossed over to a pair of JL Fathom F112 subwoofers with an 80 Hz Neville Thiele linear phase digital crossover implemented in Acourate Convolver. Response was flat to a target from 20 Hz to 20 kHz plus or minus 1 dB as measured in Acourate and also Room EQ Wizard!* It was flat to -3 dB at 16.7 Hz! I would never kick Revels out of bed; Revel makes full range loudspeakers which are even better and which I’ve never owned.

So, what did I just replace the Revel system with: A pair of Dynaudio Evidence M5P monitors, also supplemented by the JL subs, but this time crossed over at 60 Hz, extending their response from a native 27 Hz down to 15 Hz (-3 dB). Corrected response is flat to a target plus or minus 1 dB, 20 Hz to 20 kHz. Here’s a frequency response graph of the Acourate-corrected response in my room. Each horizontal line is 1 dB apart. That’s right, 1 dB/step!


12A Amplitude response, Dynaudios & JL woofers, corrected by Acourate Convolver. Blue = Left, Red = Right.

This ups my game: response is more linear, accurate and extended. The sound is smoother and rounder. But frequency response is just one small part of the improvement: These Dynaudios are taller than me! They produce a very large and impressive horizontal and vertical soundstage which reveals all the layers of depth in a good recording. Their specialized smoothed-edged cabinet reduces diffraction distortion which would smear an image and add an edge to the sound. Although the Revels are no slouch, with their own special tapered enclosure. Audiophile loudspeaker designers around the world have learned their lessons and have been focusing on reducing diffraction for a number of years: Bravo.

A special crossover design implements what they call “Dynaudio Directivity Control” (DDC). DDC reduces floor and ceiling reflections by, they claim, 75%. This serious development reduces the phantom images from floor and ceiling which would smear the sound quality. Lastly, and to my mind most importantly, Dynaudio implements a first order (6 dB/octave) crossover, since their drivers have the bandwidth to support such a gentle crossover. Very few loudspeakers today use first order slopes. Back in 1971 I owned a pair of Dahlquist DQ-10s which had a five-way, first order characteristic and were very transparent. Odd order crossover slopes permit all the drivers to be in the same polarity, and first order can potentially produce a perfect impulse and step response, unlike any other analog crossover design, unless the designer implements a phase-correcting all-pass filter, which can be an improvement. But to my mind, first order is the purest technique to approach a perfect impulse response.

With the right design and cabinet configuration, a multi-driver loudspeaker can approach the coherent sonics of a single point source and deliver a more linear polar pattern than planar loudspeakers like Quad and Magneplanar. The Revels needed considerable excess phase correction, which resulted in an impressive transient response and impact. But out of the box the Dynaudios are already phase-coherent so I did not need apply any phase correction for the main loudspeaker and just a bit of it to improve the performance of the JL subwoofers. With or without correction, the Dynaudios have the most impressive and accurate transient response I’ve heard from any speaker in my entire career! Here is a graph comparing the corrected step responses of the two loudspeakers.


12B Step Response, after excess phase correction. Green: Revels & JL woofers. Red: Dynaudios & JL woofers.

In both traces we can see a distinct single positive-going impulse followed by the recovery back to 0 volts. Even though the Dynaudio contains four woofers, two midranges and two tweeters! A remarkable achievement. It took a lot of correction to get there with the Revels and to my ears the sound is not as coherent or tight as the Dyns.

I’ve also improved my power amplifiers, using a pair of ATSAH 500s, by AcousticImagery, which use the Hypex NCore 500 modules, and can deliver instantaneous peaks up to about 1200 watts short term and continuous power of 700 watts into four ohms.

This Dynaudio loudspeaker is the most transparent loudspeaker that I’ve ever heard, probably beating or at least equaling Quad Electrostatics, Magneplanars and Apogee ribbons. But the Dynaudio’s transients and microdynamics exceed that of the other phase-coherent speakers: Snare drums have the impact of real life, orchestral dynamics are stunning.

Throwing Down The Gauntlet: Headphones versus Loudspeakers
This kind of performance leads us to this question: Can a pair of state-of-the-art loudspeakers placed in the far field of a room sound as transparent, defined, impacting and pure as the world’s best headphones? These Dyns give my Stax 007 MkII headphones a good run for their money. The Stax are driven by a KGSS amplifier. And what about dynamic headphones? I upped my game here, selling my Audeze LCD-X to invest in a pair of LCD4’s. And I borrowed a Deckard headphone amp on reviewer’s loan. No slouches here!

Regarding transient response, considering that headphones are placed directly on the ears, avoiding room interference, loudspeakers in the far field begin at a transient response disadvantage. I have found my Stax headphones to be slightly more revealing than my Revel loudspeakers, the Stax revealed sonic differences that were masked in the loudspeakers. For example, I was able to prove that a D/A/D chain was subtly lossy by auditioning through the Stax whereas in the loudspeakers I believed the D/A/D to be audibly transparent! Let’s try that test again comparing the Dynaudios to the Stax and Audeze. I’ll also throw my Oppo PM-3’s into the game as these are amazingly transparent, impacting, and pure. Compensation headphone EQ will be allowed. After all, I used compensation EQ on the loudspeakers! Let the games begin: Who is going to win the transparency race — state-of- the-art loudspeakers or state-of-the-art headphones?

As I write this column I honestly don’t know the answer to this question yet. As soon as I get to spike the speakers to the concrete slab I’ll begin the listening comparisons. May the best transducer win!

* In REW, for those who wish to follow along and perform measurements using the same criteria: I used a Tukey 0.25 window, left side 40 ms and right side 200 ms, for a good assessment of the perceived bass response in the room. I used my preference, the psychoacoustic weighting (instead of 1/6 octave) because it correlates very well with the psychoacoustic technique used in Acourate. I did not use the frequency-dependent-window in REW because I find it to be problematic.

Editor's Note: While not having to deal with room reflections and absorption, headphones do have the problem of internal reflections causing mid-treble resonances. These can disrupt the clean transient response needed for precise "localization." I put localization in quotes as headphones do disrupt the normal interaural time differences and pinna reflections heard with normal sounds and generally prevent headphones from possessing the type of out-of-head perception of acoustic images heard on speakers.

Though not step responses as shown by Bob above, impulse response is a related measure, and you can see in the two plots below that the impulse response from a Stax SR-007 and Audeze LCD-4 are not particularly clean.


I think Bob is going to find that these two types of acoustic presentation (speaker and headphone) may be so different that an apples-to-apples comparison may be very difficult. Still, I'm intrigued and look forward to reading his impressions.