BackstageAn Engineer's Perspective

Focus SE- Behind the Design with Bill Dudleston

Consider this:  At 10kHz the wavelength is a bit more than one inch long, therefore, a 1” driver is adequate.  At 1kHz the wavelength is approximately a foot long, so pairs of 7” drivers can do justice.  But at 100 Hz the wavelength is more than 10 feet long… So where is the 10ft driver?  Some people look at Focus SE and think it is redundant in its piston area - not at all the case.  The benefit of the added piston area is quite audible to anyone with reasonable hearing. Focus SE maintains a quite linear directivity pattern, though not a constant directivity pattern as the Whisper XD

The Focus speaker is named from the acronym Field Optimized Convergent Source. More than a decade ago I was confident, as were other front runners like Ken Kantor ( who had left Acoustic Research to start up NHT), David Moran (from DBX) and Don Keele (the chief loudspeaker tester for Audio magazine), that the in-room power response painted a more accurate picture of what we were hearing from a loudspeaker than a simple curved spliced near-field measurement made at one meter distance.

We knew that reflections arriving within 5 msecs of the direct wave-launch arrival will tonally influence the sound character as these arrivals become fused.  Arrivals a bit later, 5-15 msecs, can affect the soundstage laterally while sounds arriving after 15 to 25 msecs virtually tell us about the character of the room we are in; its size, reflectivity, the distance of the source, etc..  Really late reflections, exceeding 25msecs, impede clarity of the source and eventually become discrete echoes. That’s why they are minimized in most recordings, except for large orchestral works, organ and chant pieces. Should we desire to hear dense echoes, we can add it in the mix.  What we don’t want is reverberant energy introduced by the playback, forcibly imprinting our room's dimensions onto the original venue. 

Where Whisper XD is virtually independent of the room for the first 20msecs, Focus SE works instead with the room.  In its design, we made some generalizations about your room, rather than model from an anechoic chamber.  We assumed you have a floor, so we spaced the midranges to interfere most constructively (work together) at seated listener height, and most destructively where the floor bounce occurs.  Focus SE exhibits less tonal shift from floor bounce in the upper bass and lower midrange because 4 drivers are contributing in the pass-band, 2 drivers are climbing in phase as frequency drops, while 2 drivers are falling in phase as frequency rises.   It is impossible for the floor reflection to synch up with both characteristics at four different physical distances, regardless of your seating position.

We also optimized the frequency response at 3 meters instead of the 1 meter distance that most manufacturers adopt.  The 1 meter measurement looks good in a printout, but does not translate in the far field unless the transducer is a coincident source (coaxial with time alignment), or constant directivity like Whisper.  Even then, it is subject to proximity effect.

Most of us have seen circular polar plots; 360 degree views of the pickup pattern of microphones for example.  The radial lines segmenting the circular diagram are indexed in decibels of attenuations so that an omni-directional transducer’s amplitude vs. radiation plot would appear as a circle. A dipole as a figure of 8, and wall mounted unit as a half circle.  In fact, the cardioid mic actually gets its name from the shape of its polar plot – a cardioid.

The problem with a typical front radiating loudspeaker is that it is an omni at low frequencies, and a cardioid at high frequencies.  If you add drivers to the rear to create a bipolar radiator, the speaker adopts the signature of that room, and the arrivals at the listener with not correlate well with the recording. Put it in a corner, it sounds like the corner.  Sit close to it in the middle of the room, and it will sound thin.

At best, such bipolar types can portray the “they are here experience”.  Unfortunately, it is the “you are there” experience for which most of us long.  We want to be transported to Albert Hall, or Ryman Auditorium, or Beale Street.  If one experiences a speaker with a uniform and controlled directivity index, and a strong inter-channel correlation,  the image will appear more vivid, the depth of stage more layered, and the sweet spot considerably wider.

Room effects of the recording venue should be included in the recording for playback. However, the more the playback environment is superimposed upon the listener, the more the naturally gathered ambient cues become masked.  The more you hear your own room, the harder it becomes to transport the listener.  That’s why movie theaters are designed acoustically dead today.   When room effects introduce a 9 dB dip at 300 Hz it is obviously detrimental.  Speakers that require lots of lateral reflections to disappear usually exhibit poor inter-channel correlation at listener position. The randomized reflections actually help puddle in holes in a ragged frequency response.

Our goal with the Focus SE is a rhythmic, authentic and articulate presentation.  It is not for the musically timid or uninspired listener. I want full orchestral swells as the massed strings couple together and the tympani rolls. When I listen to the Neville brothers, I love how the percussion hangs back on the bass line.  There is this anticipation- then it hits with authority.  It punctuates the musical phrase in front of it …Same way with Bob Marley.  The music rolls in waves; it’s solid, visceral, yet clear even at soft levels.  I love the blattiness of a tuba, the resonance of the reed and the wind noise in a softly played, yet well blown sax.  At our studio across from our manufacturing facility, we experience kick drums, toms, crash and ride cymbals, harmonicas and pianos on a daily basis.  Our workers know what real is - as does the audiophile for whom we are building.  As Focus owner producer Rick Rubin says – I don’t set out to make recordings that just sound right, I make them sound real. Anyone who has heard his work with the late Johnny Cash knows exactly what Rubin means by this.

Over the years the better driver manufacturers come to know that when a product designer is fanatical about performance, you are likely to help them to improve their wares.  Over the years I have had good experiences working with Beyma of Valencia, Spain, Eton of Ulm, Germany, Audax of France, B&C of Florence, Fountek of Taipei, Focal of France, Seas of Norway, Vifa and Peerless of Denmark, Foster of Japan, Eminence of Kentucky, MISCO of Minnesota and numerous others.  I still receive regular shipments of developmental prototypes from most of these manufacturers despite changes in ownership of some.

Sometimes what you want does not yet exist and you need to start from scratch.  And since midrange frequencies are ‘where we live’ in the musical spectrum, a great deal of work went into the 7” Rohacell backed Silver/Graphite driver. What is so unusual about this cast frame driver is that there is a hefty neodymium slug under the front phase plug.  This improves transients, reduces distortion and increases efficiency.  This driver’s useable range extends over seven octaves.  This minimizes the in-line components of the crossover network, thereby reducing power loss. Driver manufacturing is a cooperative effort involving three suppliers.

The Eton 7” midrange we used for years was compromised of a Nomex honeycomb structure sandwiched between two layers of Kevlar. This driver remained the finest available for more than ten years.  It offered tremendous strength for its mass, though it was not light by any means. It had some drawbacks; to overcome the added mass and yet retain sensitivity, the voice coil gap was the tightest I had ever seen.  We rejected about 20% of what we received, despite the great precision afforded by the Germans, as it is impossible to fabricate a perfectly round voice coil. Over the years, I continued to work with Eton and other vendors to provide a superior alternative.  I tried magnesium, ceramic, carbon fiber and they all had significant drawbacks.

Then I received a sample of a new, woven composite material comprised of graphite and metal fibers.  Experimentation found we could alter the ratio of metal to graphite, and the type of metal strands to balance the trade-offs of each material.  Make the diaphragm out of too much metal and it would ring. Make it out of too much graphite and it would lose definition.
We found that silver strands sounded much better than aluminum, copper or magnesium when woven with the graphite.  After optimizing the silver to graphite ratio, we bonded the material to a thin layer of rohacell for stiffness.

This is a white, crusty, open cell material that looks like fake snow or crunchy styrofoam. Low in mass - and stiff as heck.

We then worked to optimize the suspension.  I went with a more compliant surround to lower the frequency of resonance out of the vocal range, and added a phase plug to reduce the cavity effect present in all cones.  But the real key is the remarkable motor structure on this thing. It is in essence a push/pull electro-dynamic motor.  The reverting field generated by the front focusing magnet also concentrates the field normally straying from the pole piece.

The new diaphragm outperforms the Kevlar. It has more bandwidth, and better transient behavior.  So much so, I am going to add some downloadable sound files to our website to give relative comparison.  Even with headphones, you’ll can hear an obvious difference in the A/B.

With treble, the best way to evaluate a drive unit is to record a Zildjian or Sabian (cymbal) with a calibrated mic and listen to the shimmer after the ting.  It comes off with a natural, rich ‘sh-shing’ quality with our ribbons. With metal domes, all I hear is the initial ‘ting’.  We have a 24 bit studio and can record up to 96 kHz so the reference samples are dead on.

The upper accordion-pleated ribbon tweeter is a third generation influence of Oskar Heil.  New magnetic materials are applied to improve force factor, the diaphragm is the latest Kapton variety and the front pole piece, acoustic slots and back chamber are optimized for our applications.  Like all of the drivers we use today, it is a proprietary house design. The steel, the neodymium, the diaphragm are all sourced separately to our spec, then meticulously assembled.  To date we have rejected over 5% of our own assemblies.  This adds to the cost, but assures a strict level of quality.  This driver can move air with 4 times the volume velocity of a flat driver. This translates to greater speed of attack and a shorter settling time.

The larger ribbon operates below 8kHz and acts as a transition driver to the cone midranges. It is phase locked and time aligned with the pleated ribbon which sets back a bit.  They are positioned closely to minimize vertical comb filtering and unify them. The Kapton diaphragm is flat with an etched leaf tracing.  We definitely worked to maintain the precise 7” driver spacing also found on the 20/20.

I might also note that both ribbons have much greater power handling than the other ribbons I have used in the past, or recently tested.  This is due to Kapton over foil and greater magnetic flux, which increases efficiency, so there is less heat to dissipate.  Foil diaphragms stretch out of shape very easily and won’t retain an accordion fold very well. We had worked with Fountek on their folded ribbons for several years but never achieved the durability needed to keep place with the Aura woofers.

The soft dome used previously on the Focus was a very good driver, but like all domes exhibited a natural resonance in the vocal range. This required an additional compensation network to control. Mistune the network even a little bit and a nasal coloration would appear. The new planar mid is a purely resistive load.  The plot of its impedance vs. frequency is a straight line, whereas the dome’s impedance would increase by 400% at 750Hz.  Eliminating resonances is a good thing for sure. The greatly reduced mass of the ribbon diaphragm also improves transient behavior.

All of the updated drivers have wider bandwidth than in the Focus 20/20. The bass drivers used in the 20/20 were heavy enough that we added a front firing mid-bass driver to improve snap.  We have no need for this driver now, as the totally enclosed magnetic structure of the new 12” provides high electrical damping, with a broader frequency spectrum and greater linear X-max.  The extra cabinet volume made available also translates to more efficient bass and also deeper bass.

The motor on the Aura 12” is a feat of engineering and fabrication.  This encapsulated design provides a highly uniform magnetic field with a lot less mechanical noise than the inefficient long throw super woofers that abound. The maximum linear displacement is a full 20mm peak to peak with an excursion of 44mm without concern of bottoming.  The 12” drivers resonance is 21 Hz and the cost frame incorporates a large heat sink.

Though several of our designs appear symmetrical, none of our speakers are actually D’Appolito configurations.  Joe’s configuration requires that the mid and tweeter center to center distance not exceed ¼ of a wavelength at the crossover frequency.  For a 3500Hz crossover, the drivers would have to be 4” from center to center.  So you can’t go larger than a 5” mid and meet the criteria. This type of configuration also results in three lobes in the forward profile at the crossover frequency, mirrored above and below the horizontal plane through the tweeter.  There is a dark null you pass through when you stand up, which also hinders presence. Many people mistakenly refer to M-T-M arrays as D’Appolito.  Interesting, the competent designer that Joe is, merely described the arrangements behavior.  To my knowledge has not used this configuration in most of his designs in recent years.

In any closely spaced symmetrical array of identical drivers you can generate a beneficial low frequency mutual coupling which counters diffraction loss, and affords terrific steering capability.  Off-axis, a useful destructive interference occurs (desirable assuming you are able to also steer the upper bass frequencies also) as with V and Whisper.

We wind every coil and hand-solder every crossover in house. We make up our wire harnesses from 12 gauge Signature Cable for the bass section, and 99.999% OFC for the midrange and treble.  We use approximately one million feet of copper wire per year. We assemble our rear panel terminal plates and install the binding posts, gasket and install each driver. Besides testing each driver, we test each assembled speaker individually, then again as a pair with swept sine waves, LMS, pink noise and finally, and most importantly, by ear.  The last 15 minutes of voicing is where the pair is truly made.

For our binding posts, the oversized posts and plungers are turned from solid brass, plated with silver then nickel. We staggered them to allow more finger room access and angles of entry for the variety of audiophile terminations today. The jumpers are punched from solid copper.

The Focus speaker system has been continually refined since 1992. That’s approaching twenty years! The reason the design holds up so well is the design criteria were very thoughtfully established – a flat power response optimized at listener position.  With every generation of Focus, the crossover seems to become a bit simpler. By controlling the design and fabrication of the drivers, unnecessary complexity is avoided.  One thing I do that many designers don’t, is factor in the natural acoustic roll-off of each driver before applying the actual electrical network. This makes phasing the network much more precise.  For example, our midranges acoustically exhibit a 12 dB/octave roll-off below 85 Hz in their sealed enclosure, climb 1 dB/ octave over the pass-band, while rolling out at 6 dB/octave above 4kHz. The optimal electrical network compliments these characteristics as well as the drivers above and below.

Evaluating a crossover is more than just tonal. One must assure that the image does not wander laterally as the frequency changes.  A flute is excellent for uncovering such flaws in lateral localization, as the low frequency wind noises lie beneath the notes, they can modulate and reveal phase problems.

With Focus SE, we have included typical low frequency room gain into the response.  If you don’t have a floor in your room, you will not benefit.  If you sit less than 4 feet away, the midrange null at the floor will not be beneficial.  If you already have stereo subwoofers, you may not need the low frequency extension Focus HD affords.  But for the rest of the world …

Focus SE is utilizing the floor and the room impedance to achieve its low frequency output.  Pardon the analogy, but Focus SE is like bathing in bass, whereas Whisper HD is like showering.  Whisper doesn’t wait for the tub (room) to fill.  It comes and goes.  Whisper is able to propagate a wave naturally, without waiting for the room to load. The wave actually travels past you as in the real, open world. The faithfulness in this is unique. Whisper is allowed full excursion down to its free air resonant frequency of 22 Hz, then we progressively roll it out electrically.  We have to do this because for each octave the frequency falls, the cone excursion increases by a factor of four.  Simply stated, the 15’” diaphragms would have to travel 16 times farther at 8 Hz than at 32 Hz.   The only way to increase useful low frequency output in a differential alignment below 20 Hz is to make the diaphragms larger.

Wave propagation models are often oversimplified, as if ping pong balls we bouncing off mirrors.  When, in fact, the wave is constantly expanding spherically, so when it glances off a boundary, energy is absorbed,  with some compression loss, while expansion of the wavelet radius continues through the reflection. When designing a loudspeaker to work in the real world, these are important things to grasp.

My educational background is in chemical engineering with a minor in mathematics, actually, from the University of Illinois.  Though I supervised a research lab for an electronics company, the real benefits of my education were the expertise in fluid dynamics.  Air, being a gas, offers resistance (impedance). The higher the acoustical impedance on a speaker diaphragm, the move effectively it can do work on the air.  Did you ever ponder the concept that when you are rowing a boat quickly the water seems to be as stiff as taffy? Row slowly and there is very little resistance in the water. Viscosity is a dynamic parameter.  That’s why tweeters are inherently more efficient than woofers.

Boundary reinforcement can be used beneficially if well applied. At low frequencies wavelengths are longer, so Focus SE has great attack due to four 12” woofers operating in the pair and coupling at these low frequencies.  In home theater situations most people merely route the subwoofer LFE channel to the Focus via the bass management software.

The Focus SE is not a traditional vented alignment. It utilizes the ports to lower the tuning frequency of the enclosure. Electrically, the impedance curve does not resemble a vented box.   A vented box curve looks like a camel’s back with 2 humps.  With Focus HD we tune very low, then transform the potential energy of the upper hump to a more linear impedance, thus stepping up the voltage in this region.  The driver receives the added current, the load has less capacitive and inductive reactance and the cone is still stabilized at enclosure tuning.

When setting up, the key is to not position your head and the speakers the same distance from the nearest wall boundaries.  If you pull the speakers out 1.5 ft from the wall behind them, try to sit at least 4.5 ft from the wall behind you.  Focus SE works well when spread 10 feet apart sitting 12 feet away, or 8 feet apart, sitting 10 feet away. By adjusting toe-in, you can widen them more.  I like Focus SE firing over my shoulders, and Whisper crossing their axis about six inches in front of me.  Focus is used in dozens of recording mixing rooms that are 10’ x 12’ and smaller. Keep in mind that while bass gain may be greater in smaller rooms, the treble balance also shifts upwards with closer proximity.  Studios consoles can EQ the bass parametrically in 24 bit, right off the monitor output.

In the early years of building Focus it was a bit of a wild mustang in the bottom end.  We did not yet have the experience of hearing it in dozens of rooms. So we asked 300 owners to make some low frequency measurements for us with a handheld SPL meter at listener position.  After the first 20 surveys were returned, it was obvious what we needed to do to smooth things out in general. So we developed an outboard adjustable analog unit with a level control (Steradian).  Rather than put a bunch of circuitry in the signal path.  We had a removable module to adjust frequency for specific needs.  As time went by, we were able to incorporate this into the crossover and adjust with the rear panel switch.

Can we still improve on Focus SE?  Sure, by controlling directivity at all frequencies.  Just as Harry Olson described and mathematically modeled the differential microphone and its low room sensitivity, we can do the same for its analogous transducer – the loudspeaker.  All we need to do is widen the baffle a bit, transplant the same mids and treble units, and tuck two figure- of- eight  bass radiators one behind the other at the top and bottom.  Toss in an adaptable digital processor - I think we should call it Whisper.

-Bill Dudleston

Listen to Legacy Legacy Backstage Building a Legacy