The Octagon has existed in near-original form for over 5 years...time for an update. The DIY community shares incredible knowledge in a very interesting hobby...and this page is offered in appreciation of that generosity.
A lot of detail is included here, so the whole write-up is a long read…read what interests you and blow off the rest! Some choices are unconventional, so you may see things you would do differently… heck, with 20/20 hindsight, there are things I would do differently! Beyond the differences though, the Octagon pages are intended to stimulate your ideas!
Without prior knowledge the listening environment, most speakers are designed for outstanding performance in free space…with that excellent performance mangled as soon as those speakers are placed in a real room. With a known environment, from the front wall with an eighteen driver subwoofer array to the limp membrane rear wall, the Octagon considers speakers and room an integrated system. However, the room has very real boundaries so perfection isn’t realistic…but we’ll not let perfect be the enemy of good.
In-room response irregularities can mask weaker signals, so smooth LP response can be key to revealing micro-detail. The objective here is to eliminate any audible distortion…and the spoken voice makes a great reference standard. If speech is not entirely natural, something can be improved…and all is fair game.
Here, SQ is valued far more than SPL, but only high output capability and low distortion can provide the life-like dynamics required for a realistic presentation. There is zero interest in dB drags, but reasonable headroom is a basic cost of entry. Acoustic problems deserve acoustic solutions so software is used only when acoustics become impractical.
The original Octagon plan hinted the waveguides were a “work in progress”…and they’ve actually become a journey in themselves. In v1.1, especially for the center channel, the original 30” WGs were replaced with 32”x26” elliptical profiles to eliminate the on-axis dip characteristic of a round OS. Version 1.2 teamed a 32”x38” WG with a focused and progressively shaded vertical array to reduce floor bounce.
Evolution of the system has reinforced the notion that reduced boundary interference yields more than just smoother measurements, but also a more detailed acoustic presentation. As you would expect, the first objective for v2.0 is to further reduced boundary interference to further improve detail retrieval.
Another clear opportunity is lower measured distortion through better driver-air coupling. So, while reducing boundary reflections, let’s improve the driver-air interface by loading the entire surface of the front wall, refine impulse response at the LP, and measure the result.
For Octagon, the strategy has always been more program material, less room. This time, to fully immerse listeners in the performance, large rectangular LCR “windows” should be no wider than necessary and no taller than necessary.
With subs there is “no replacement for displacement”; for horns and waveguides there is “no substitute for size”. Floor to ceiling and wall to wall, each of the new LCRs is 7’ wide by 8’6” tall.
Frequency response is intended to be within 1db of target from 10Hz to 20Khz. IACC (Inter Aural Cross Correlation, a measure of imaging quality) should be better than 90% for the first 10ms and at least 85% for the first 20ms. Good FR and tight IACC would help ensure satisfying movie reproduction as well as provide an excellent foundation for the enjoyment of 2-channel music. This would be a “stretch” project, so all goals and measurements will be from the MLP…more than fourteen feet from the drivers.
Six-wide Octagon seating calls for a horizontal window of about 70 degrees for the center and, firing diagonally across the listening area, 60 degrees for the left and right channels. The two outside seats are actually too close to the side surrounds and only occasionally occupied by kids more interested in their “devices” than a movie or concert. Given that reality, v2.0 spotlights the four-wide middle seats where the average should be +/-1db.
The easy choice for horizontal control remains waveguides.
Especially for the all-important center channel, floor bounce and ceiling slap are high priorities because they are the first reflections to arrive at the LP. Floor bounce is also the strongest. Radiating a tight vertical pattern to the lowest practical frequency would minimize vertical reflections.
Tom Danley’s paraline was considered but no DIY implementation I’m aware of exhibits smooth acoustic response. Synergy-style horns and coaxial CDs were scratched to eliminate a crossover and minimize diffraction. With floor bounce and ceiling slap high priorities, very strong consideration was given a multi-driver WMTMW with Peerless mids and Aurum Cantus AMT as implemented in Octagon 1.2. That configuration does provide vertical control superior to CDs in realistically sized WGs.
However, there is no crossover as good as no crossover, plus I really wanted to try big Beryllium diaphragms…so a two-way plus subs it would be! For the system to work well, Acourate must do a good job with floor bounce. The “low-n-slow” crossover envisioned would require a very robust tweeter, but that is within project scope.
The clarity and rock-solid soundstage of big vertically symmetrical speakers has always impressed me so, to control the widest practical frequency range, v2.0 implements focused WTWs at full room height. Each channel includes 12 woofers plus a wide-band compression driver, all in waveguides for smooth horizontals.
At the LP, each LCR sums to an acoustic origin at tweeter height, halfway between the woofer sources. Symmetrical arrays centered near screen height would help the brain integrate acoustics with on-screen visuals.
At low frequency, all 12 woofers will share the load equally. RC shunting the outer drivers reduces source height with decreasing wavelength. Height reduction continues through crossover where a relatively small CD reproduces the shortest wavelengths. This continuous decrease with wavelength works in concert with wide crossover overlap to enable more consistent height of the WTW forward lobe.
Above and below seated listeners, the woofers produce vertical lobes and nulls. If it occurs on a listening axis, lobing is clearly detrimental. However, in this case nulls are used to reduce the energy reflected from the floor and ceiling. Differing angle, distance, or frequency all produce different lobe/null combinations so that at any individual point on the floor or ceiling there can be a lobe at one frequency and a null at another frequency. Averaging these infinite nulls and lobes reduces overall reflected energy. That reduction compliments tight-as-practical vertical radiation from the HF waveguide.
The woofers have the highest vertical directivity at the upper crossover point; where the HF WG looses vertical control. Therefore, smoothest system directivity is provided by a low-slope/high overlap crossover. Bottom line for the integrated woofer/tweeter…reduced floor bounce and ceiling slap.
SIMULATION & PROTYPING:
With failure not an option, this would be a very challenging project. After all, the existing Octagon sounded very good and equal or downgraded performance would be a huge disappointment! Since there are no plans to do this again, far more time was invested in simulations than the actual build.
All of the usual tweeter flares were evaluated in HornResp…Exponential, Tractrix, LeCleach, OS, Spherical, Conical, etc. Excessively narrow top octave response quickly eliminated all of the classics except OS, Conical, and very high “T” Hypex. The more promising tweeter profiles were exported to AxiDriver for detailed comparisons with custom profiles, including hybrids with standard throats tangent to various mouths. In the end, a continuous elliptical curve from throat to mouth provided the best overall match with performance goals favoring low diffraction and clean impulse response.
Vertical simulations in LspCAD, and HornResp were straightforward and processed nearly instantly. The time sink was AxiDriver which, even at low resolution with an i7 processor, typically took nearly half an hour to run. 1/48 octave simulations usually ran overnight. Since “the operator is the optimizer” it is easy to see why it took so long! (Note: ABEC appears to be a multi-threading evolution of AD so it runs much faster…but I had already endured one painful learning curve with AD. If you’re simulating your own waveguide profile from scratch, consider ABEC.)
All three channels share WG mouth sizes with different directivity achieved through differing depths and profiles. Listeners are below room center so, to help align acoustic centers with minimal room intrusion, the upper and lower woofers are slightly different. It took, literally, months of simulation before all of the WG designs were acceptable.
The weakest link in all of this is vertical directivity of the tweeter WG. The vertical dimension limits vertical pattern control at LF and achieving the desired aspect ratio (extremely narrow vertical by wide horizontal) would require an exceptionally wide and deep WG. Subwoofer mounting requires flat surfaces, but I was unable to identify a high-performance/low-diffraction tweeter profile with angled flats as the mouth termination. Unfortunately that results in tweeter WGs 4’ wide rather than 7’. At a listening distance of nearly 15 feet, mitigating LF tweeter floor bounce must be a task for Acourate.
Horns scale with frequency, but real drivers don’t. Several physical prototypes, some of them full-size, were built to validate plans before final construction began. The final tweeter “prototype” was built to become the center channel…and it did…whew!
TWEETER WAVEGUIDE DESIGN:
The idea of loading drivers with the maximum available HxV mouth area comes from pro sound. JBL has done it for years and still does , including the Everest and M2. Tweeter WGs more than a wavelength in all dimensions combat pattern flip by providing reasonable polar control into the region of crossover overlap. “Super-sizing” also reduces diffraction since the HF signal largely leaves the walls before reaching the mouth. A further benefit of size is lower driver distortion through improved acoustic loading via more optimal radiation resistance.
Freedom to choose any part of any ellipse provides great design flexibility. With a smooth continuous sweep from throat to mouth, the ellipse can be stretched, tilted, or offset to adjust performance trade-offs. By definition, diffraction “bumps” are avoided. Here, the long axis is the listening axis; the exact curve plotted in Excel with performance modeled in AxiDriver. Maximum SPL output and directivity was traded for smooth directivity and clean impulse response…sharp rise with very quick settling time. (Impulse response speaks, literally, volumes about real-world performance.)
The compression drivers used here have a negative 7 degree exit angle so, thanks to an engineering drawing from the driver manufacturer, phase plug output was included in the WG sims. For the converging/diverging CD/WG interface, a near-conical throat provided the most compact impulse response. With a tip-of-the-hat to Earl Geddes, a 1.5” cube of open cell foam absorbs diffraction generated at the driver-throat interface. The small block of foam is so effective at reducing destructive interference in the throat it actually increases output at 20kHz by more than 2db.
The lower center channel woofer WG sets minimum physical depth for the entire system. A classic FLH configuration would intrude too far into the room so side entry was mandated by available depth. Driver diameter/directivity limits HF output, so opposed three-high stacks of smaller drivers are positioned at the rear of each LF WG. This achieves maximum usable depth and effective throat size in the space available.
As with the tweeters, woofer WGs prioritize impulse response and directivity (they may look like horns but, since the purpose is directivity rather than increased output, we’ll call them waveguides.) Providing the distortion reducing acoustic lo-pass benefit of ports, the 2x3 woofers in the upper and lower WGs fire into angled reflectors. The reflectors keep in-band “front” and “back” radiation in-phase while acoustically attenuating the signal above crossover.
Not included in measurements presented here because the final ducts are not yet implemented, prototype experiments showed that each WG could be subdivided into six ducts, each duct having at least two acoustically translucent walls to damp internal reflections. These ducts pass the desired signal directly from the drivers into the room while soft walls absorb internal reflections. At a macro level, these partitions also absorb reflections between hard surfaces, particularly vertical reflections between the near-parallel horizontal WG walls.
Working as a system, natural driver directivity, reflectors, and duct walls combine to reduce HF breakup and harmonics by more than 20db. Since the reflectors and ducts were developed empirically, all are removable from the front of the WGs to facilitate continued experimentation as well as driver replacement. In fact, the reflectors and ducts remain a WIP to optimize acoustic response…a guy always needs a project, right?
To prevent horizontal waist-banding, woofer WG bodies are finished with Don Keele’s secondary mouth flare, or “lips”. The secondary flares also serve as subwoofer mounting plates as well as “rollback” for adjacent WGs. The outer lips flow onto the side walls.
The finite-sized woofer WGs have almost no horizontal control at 75Hz where they transition to the planar subwoofer array. However, at the 600Hz upper crossover, the upper and lower WGs provide great control and match tweeter horizontal directivity at XO.
Broad overlap requires drivers that are clean beyond the crossover region. For low distortion, the flares of each LCR are driven by twelve 7” Dayton RS-180 woofers. The RS-180 implements two copper shorting rings for improved linearity and reduced distortion, while the aluminum cone remains pistonic well above crossover. Drivers optimized for use in horns were considered, but the additional sensitivity wasn’t needed…so no need for the expense of replacing 36 drivers already on-hand.
In this configuration, HF distortion and hard-cone breakup physically generated within the drivers are reduced acoustically. The combination of high performance drivers and distortion reduction via the reflectors and ducts can yield distortion up to 60db below the F1 signal. Distortion at -60db is unlikely to yield audible improvement so no need to rush the ducts. For measured distortion, the modest RS180s could actually become the surprise star of the show!
A robust driver is necessary for a low crossover, so a tweeter upgrade was appropriate. The previous ND1460As are excellent, but the 3” compression drivers are outclassed by true SOTA 4” units. Measureable improvements can be found in both low-end distortion and clean top octave extension.
Higher radiating area = lower excursion related distortion. Larger 4” diaphragms have nearly twice the radiating area of a 3”, and more than 5x the area of high-quality 1” exit CDs. For the top end, extension can be improved by switching to exceptionally light and stiff Beryllium diaphragms.
Long story short, tweeters are the Radian 951Be-16. The 5-slot circumferential phase plug ensures diaphragm output is in-phase at the 1.4” exit. Materion Truextent domes are pistonic well into the top octave providing smooth, extended response. Powerful neodymium motors, inductance reducing shorting rings on the pole pieces, low excursion, poly surrounds, plus oversize WGs all reduce tweeter distortion. Measured HD, IMD, and sub-harmonics are quite low.
The planar array (high-density SBA) with limp membrane rear wall has thoroughly satisfied, so the array approach continues.
More “modern” subwoofer drivers were considered but inductance was higher, and sensitivity lower for the candidates examined. Be that as it may, mounting cutouts were sized to accommodate most available driver frames just in case drivers with better motors and stiffer cones become available at reasonable prices. Until then, the eighteen 15” Dayton DVC subs soldier on from the original Octagon. This relatively low-Q, low inductance driver uses a Kevlar impregnated cone and implements dual voice coils on a Kapton former. Motor design, relatively low excursion, and appropriate filtering all contribute to low distortion.
One of the great pleasures in DIY speakers has been the satisfaction derived from designing passive crossovers. However, the capabilities of DSP tools like Acourate now exceed what is even possible with passives. The performance of passives can get close, but never equal, so DSP plays an ever stronger role in Octagon 2.0. Too bad SOTA feels like cheating!
Primary crossover frequencies of 75 and 600Hz were finalized using measured FR, distortion, and directivity data. Low order/high overlap crossovers (Neville Thiele 2nd order) are used to compliment directivity and maintain more consistent height of the forward lobe.
The compression driver XO includes a series DC blocking cap, plus an oversize inductor (high L low R) directly across driver terminals. The cap protects the Beryllium CDs in case of AC line transients or amp failure while the inductor helps ensure low IMD by preventing subwoofers from modulating the HF diaphragms. Relative LF damping provided by the large inductor is improved by using 16ohm voice coils. High impedance CDs also reduce power amplifier distortion as well as voltage sensitivity to eliminate any potentially audible hiss generated in the electronics. Finally, before digital-to-analog conversion, the HF signal is Lo-passed with a steep slope at 20k to reduce generation of IM products in low-level electronics, power amps, and drivers.
From 75-600Hz, three series-parallel four-woofer “sets” operate in series. At 75Hz, all twelve RS-180s share the load equally. The two outer sets are, in turn, RC shunted to bypass higher frequencies toward the four central 7” drivers. In addition to increasing height of the forward lobe at the upper crossover, this approach maintains a distortion friendly 8-ohm minimum load as well as high sensitivity. Note, photos and measurements show frequency shading disabled in order to focus on duct development.
The only passive crossover components are high-quality series capacitors and parallel inductors for the compression drivers, plus a few resistors and poly caps for shading the woofers. To ensure easy access for adjusting component values, woofer connections are brought inside the lower WG flares. (Crossover components mounted in the woofer WGs are acoustically invisible because their physical dimensions are a fraction of a WL at LF.) Also for convenience, tweeter passive components are located at the amplifier outputs.
ENOUGH TALK….LET’S MAKE SAWDUST!
TWEETER WG CONSTRUCTION:
Tweeter waveguides were built over boatbuilding-style bucks with physical measurements transcribed from Excel in 10mm increments. Metric dimensions were very handy for tweeter simulations and for the bucks. (Lower scale on the small stick is metric...the larger stick is simply clamped to the WG centerline for reference.)
Steady as she goes!
“Disposable” OSB table-tops were screwed to a permanent table-top, enabling WG construction around various rigid forms or plugs, plus enhancing utility of a new 4’ table for other projects. A “peak” fixed to the buck (not shown) is a steel washer machined 0.050” over the 1.4” tweeter exit diameter to ensure proper throat clearance. Different disposable bucks were used for the center channel vs the L+R flares.
WG walls are 1/8” hardboard skins attached to a female skeleton. Flare construction begins by cutting oversize triangular sections of hardboard. The rough side of the hardboard is wetted with nearly-saturated wet towels for 24 hours. Each opposite pair of wetted hardboard walls are clamped onto the curved male OSB buck and covered with two dry towels each to dry for at least 24 hours.
Put water in, then take it out.
Scraps cut from the male OSB buck become templates and ribs for a female reinforcing skeleton behind the flare. After drying to the final profile, the initial triangular shape is fitted at the corners. HxV WG walls do not touch so as not to disturb the profile dictated by the buck. The female skeleton is then permanently attached to the WG walls with PL Premium. Following that, the backside corners are “stitched” together with PL. Support at the throat and mouth, along with four central lines of skeleton attachment, plus four 90* corners make the WG extremely rigid.
Too bad there is no room to store this stuff. This first WG was built to test construction technique…it served the purpose and met a sad end!
Final versions are plywood framed.
Additional structure is provided by rigid polyurethane foam backing (be sure to mist the hardboard to kick-off the poly cure). The foam ranges from about 3” to 6” thick, depending on other local strength…curves, ribs, etc.
The 0.7” radius front-side fillets running from throat to mouth are simple caulk, smoothed by dragging the end of an appropriate sized tube along the corner.
WOOFER WG CONSTRUCTION:
Lower woofer WGs were built using the same basic approach as the tweeters except, with flat walls, they were assembled without a buck. Since the straight walls do not have the added strength of curved surfaces, reinforcing ribs are added in a 2.5”over 2.5” lattice pattern. The lattice structure is then impregnated with spray foam.
Foam backed hardboard worked very well for the heavily curved tweeter WGs but, by the time extra bracing was added, it didn’t reduce woofer WG weight enough to be worth the time and effort.
I was growing antsy to get the system running, so the upper woofer WGs were built using more conventional ¾” plywood construction. Since the plywood WGs were much faster and cost less, I do not recommend spray foam for flat-sided waveguides. Triangular low-Q driver back chambers running back to front are integral to each WG.
Interior separators running back to front within the woofer enclosures stiffen both the sidewalls and back chambers.
UltraTouch denim insulation fills the chambers. Upper and lower “internal" WG walls (next to the tweeter WGs) are braced with standard full depth strakes. Top and bottom “external” walls are immobilized by adhering them with foam directly to the concrete floor and multi-layer OSB ceiling of the subwoofer enclosure.
Reflectors are reinforced ¾” plywood. The reflectors are press fit and secured against the back of the driver frames as well as the upper and lower WG walls. So far, the soft duct walls are 1” OC 703 sprayed with 50/50 thinned flat black latex. Duct walls are attached to ¼ round rails on both the reflectors and main WG walls.
The three LCRs were built as nine modules, plus eighteen subwoofer panels/WG mouths. To speed construction, the strategy was to build the center HF waveguide and lower woofer first. The center is unique, and all of the construction improvements identified there would be directly implemented in the slightly deeper L&R matched pairs. For example, the large flat walls of the woofer WGs needed to be stiffer…easier to build than retrofit.
To reduce theater down-time, and so they would fit through the 3’ theater door, all of the modules were built in a nearby recreation room. To prove the final product before the original baffle wall was disturbed, the WGs were loaded with drivers and tested before installation and final paint.
Over 50 pounds of screws and several cases of PL Premium were used during construction of the original baffle wall, so it took a month to remove the original wall and bracing. During that time, I became a surgeon with various saws and really great friends with a 4’ pry bar! The guys who haul my trash, not so much!!!
Each upper woofer WG was hoisted into position using three come-alongs with wedges and plumb bobs for adjustment. Then the lower WGs were moved into position. Testing acoustic impulse arrival at the MLP determined precise fore and aft location of each upper WG relative to the lower. Once exact positions were confirmed, the upper WGs were lag-bolted and foamed to the multilayer OSB ceiling.
Next, horizontal framing and lips were added across the entire wall. Tweeter WGs were positioned and all of the modules fastened together.
Framing for the subs was added, along with 2x6 bracing directly to the exterior concrete wall to ensure subwoofers would not shake the baffle wall. Frames for the central upper and lower subs are triangulated for modest vector cancellation and attached to the immobilized upper and lower woofer WGs. Finally, 1.5” plywood subwoofer mounting plates are added, along with corner fillets and final paint.
With this refresh, one of the things up for improvement was the “stereo triangle”. The Octagon was originally designed with the L&R speakers forming the equilateral triangle so prized by audiophiles. However, with truly outstanding imaging -even in simple stereo- the speakers seemed a little too confined laterally. This time, the extra real estate outside the L&R speakers was used to increase the spread to about 70* relative to the MLP.
Aligning upper and lower woofer acoustic centers tilts the upper part of the new wall toward the listening area. This tilt shifts the 120” AT screen closer to viewers for a larger apparent presentation…which dovetailed nicely with a new projector. Moving the screen required the minor inconvenience of moving the projector box further back…oh well!
ELECTRONICS AND SOFTWARE:
Audio is very direct with no DA/AD/DA signal conversion in the playback chain. All source material is Intel PC based, playing via JRiver. In addition to playing all sources, JRiver also hosts the crossover and impulse correction files generated by Acourate.
Acourate is a very sophisticated software “toolbox”, including measurement and psychoacoustic correction using a PC as the processing engine. The price for this sophistication is (for me) an extremely steep learning curve, complicated by somewhat sketchy documentation. Fortunately, Uli Brugemann, the software author, is extremely supportive and responded to my numerous noob questions quickly.
For those new to Acourate, or considering purchase, I highly recommend Mitch Barnett’s book “Accurate Sound Reproduction Using DSP”. That book saved me many hours, and even more questions to Uli. “Mitchco’s” book is available on Amazon via Kindle and thoroughly documents his implementation of Acourate.
Tri-amping the LCRs plus the subs and eight surrounds requires 18 Acourate filters. Here is a representative configuration file for the convolver, along with universal JRiver channel mapping.
The software outputs to a Lynx Aurora 16 DAC through an AES-16e digital controller. The 16 channel DAC drives all of the power amps directly, maintaining a high sample rate source as close to the speakers as practical.
Power amps are the three original Parasound A-21’s for the three LCRs, two Emotiva XPA-5’s for the surrounds, and a bridged XPA-2 for the subs. Movies and concerts are stored on 8x4Tb Western Digital hard drives.
Over the years I’ve tried “flat”, “smiley curves”, the B&K curve, as well as various down-tilts on the top end. None of them sounded completely natural and I always eventually wanted something different.
Yet another “curve” emerged from JBL/Harmon/Toole/Olive studies of listener preference.
In the most preferred response, you can visualize a down-tilted straight line from DC to light. While observing the straight line, you realize that any deviation in the path between A and B increases change per octave. Theoretically, minimizing the rate of change can result in lower masking of adjacent frequencies and therefore facilitate detail retrieval, so the straight line becomes a very intriguing target.
And interesting it is…so far, the down tilted straight line is my favorite listening “curve”. To me, the sound is more natural than other approaches.
If you want to try the straight line tilt with a single target, the 80/20 rule works well…adjust the top end so that 20% of your source material sounds a bit harsh and the other 80% provides full detail retrieval. To make it a worthwhile project, be sure to achieve very smooth response…narrow notches are not bad, but avoid FR peaks like the plague.
After you’ve satisfied the 80/20 rule, listen for a few days before deciding if you like what you hear. If you want a little more boom or sizzle, simply change the tilt to satisfy your ears, your speakers, and your listening environment.
The elephant in the room is the lack of standards in the recording industry. Like any other curve, one tilt simply does not suit all dynamic and/or digital compression schemes. Blu-ray movies seem to be the most consistent, some BR concerts suffer from the loudness wars, and CDs are all over the map. Management at Netflix must think everyone listens to 2” TV speakers.
The short version: more dynamic and/or data compression = more tilt.
I have implemented about a dozen JRiver preset “zones” with different tilts for different quality sources. Except for a 30Hz HPF on the surrounds, all channels have the same frequency response within each zone. “Generic” zones are implemented in both linear and minimum phase with -6.5db tilt from 10Hz to 20kHz (or about 0.6db per octave). The generic tilt sounds a bit too soft for most high quality material and a little harsh with most compressed sources. For higher quality lossless material, linear phase filters with less tilt are implemented. Minimum phase filters with greater tilt are used for more highly compressed sources requiring lower latency with JRiver’s WDM driver.
One of the benefits of implementing Acourate is that only one set of high quality measurements is required. After quality measurements are stored, system output can be adjusted to any specific target response, and it can be done in a matter of minutes. Over the coming months, additional tilts will be added as “perfect” settings emerge for different recording styles. “Perfect” for my speakers, my room, and my ears. Long term, I’d like to zero-in on just a few presets since constant switching interferes with the joy of listening.
Here is frequency response of all three LCRs from the MLP vs target response in red. +/-1db from 10Hz to 20kHz except -2db just below 100 and 200 Hz. (upper frequency scale is 24kHz)
If you’ve read this far, you may be thinking “finally, the bottom line!”
Clearly, measurements can’t tell the whole story of any system. To wrap some words around the results, the system is quite natural and very immersive; “the sound” is quickly forgotten in favor of program material. A wide soundstage is centered on the screen so acoustic images are where they belong, with well defined depth and lateral position of individual voices and instruments.
Improvement above crossover comes, predominately, from the beryllium domes with, perhaps, a little help from the larger waveguides. A life punctuated by top fuel, machine guns, and rock concerts has left me extremely sensitive to any misbehavior in the upper registers…and Beryllium clears that up by sounding very much like silk domes.
If beryllium top end is silk, the midrange is “butter”…I was surprised to hear smoother upper mids. Some of the improvement may be from the waveguides…or it may be something not revealed in our standard test routines. Whatever the reason, if anybody with large format drivers is on the fence about Be diaphragms, jump in…they are great!
Another pleasant surprise is woofer “impact”. Old hat to horn lovers, but clean LF is delivered at high level with little cone movement. This is directly attributable to the LF waveguides.
LF clarity is improved, probably via better in-room behavior of the “W/W” configuration. With the same membrane rear wall, and virtually the same planar SBA, subwoofer results are unchanged…smooth, powerful, crisp, and clean.
Of course Acourate software ties it all together with clean impulse response provided via time-correct linear-phase crossovers.
WOULDA, COULDA, SHOULDA:
Though it may not have made an appreciable difference in results, I now see more opportunity in the system.
First, I should have done more work on flat angled lips for the tweeter WGs. If the WGs had been a full 7’ wide, vertical directivity would have improved and floor bounce reduced for less reliance on software correction. Don’t know if it would have ever been successful…and finding a solution now would be very irritating!!! The WGs are mounted so they can “easily” be replaced. (Didn’t I say I wasn’t doing this again?)
It would have been easier if the room were deep enough to accommodate standard front-loaded woofer WGs. In retrospect, I should have more fully explored “flat” 2x3 stacks in the throats, or a dozen 10” rather than 36x7”. That might have prevented a cracked rib experienced during very tedious woofer installation!
More an ongoing effort than a do-over, but my understanding of Acourate is still very much in its infancy. There is no question in my mind that it can provide even greater performance. Results are already significantly better than I can achieve with JRiver crossovers and PEQ, but the learning curve is very steep for an old analog guy like me.
Would I do it again? My biggest regret is not doing it sooner!
“It isn’t what you buy, it’s what you build.”