Flint: High Dynamics, Low Distortion, State of the Art Hi-Fi Custom Speakers (2018)

[Flint]

Introduction​

Many of you recall my amazing speakers built back in 2008/2009 in which I put into practice design principles not common in commercial speakers but which address issues related to imaging, distortion and dynamic performance. I made them taller and wider than most, off-set the mid and high frequency drivers, focused on the baffle and placement more than aesthetics, and aggressively isolated the drivers from vibration. They were amazing, but one design aspect related to vibration isolating the drivers resulted in the speakers becoming ruined.

I had cut ¼ inch thick sheets of very soft Sorbothane into rings and used those as gaskets behind the mid-woofer, suspending them and absorbing all vibration from the driver and thus reducing distortion and tightening the impulse response both measurably and audibly. It was a brilliant design, but alas the Sorbothane failed and turned into a black tar-like goop that seeped into the rear of the mid-woofer frames and eventually out the front of the baffle and destroying things. Yuck. I cried.

But, I love listening to music far too much to give up altogether on having a killer stereo. So, I got back into designing and building speakers and built a new speaker rig, replacing everything in my old system, other than the IB subs, and starting over with the latest and greatest designs in a cost is no object system.

Design Decisions
Choosing the Transducers (woofers, midranges, and tweeters)
​

When considering a new system, I always want to replace the old with something significantly better than the previous system. Sometimes “better” means more convenience or greater efficiency, but with speakers I always want better audible performance – like VASTLY superior performance. Since speakers are always the weakest link in any audio system, going for the best I can afford always makes sense, even if it means not getting the amps, cables, power conditioners, or other parts I would like. So, I started my design process by researching the raw transducers on the market. Given the advances in speaker design since 2008, I knew there were some amazing options out there. Dome tweeters were improving, as were midrange drivers and even mainstream woofers. Materials were getting more and more refined and less and less “sexy” – always a good thing. Gimmicks are getting less common as well, which makes life much easier.

However, as I started my search a new driver design which initially appeared about as gimmicky as they come was announced by Scan-Speak, an incredibly respectable transducer maker out of Denmark. What did they launch? Well, it was a set of tweeters and midrange drivers with, and wait for it, elliptical voice coils. Not round voice coils like we’ve had for 100+ years, but elliptical. In the tweeter the elliptical VC meant the dome was now also elliptical, a strange look. In the woofer the cone was still round, but the dust cap at the center was elliptical making it appear more like an optical illusion than a new design. The claims Scan-Speak made was that the design reduced distortion by a huge amount. In fact, the distortion was so reduced that the designers didn’t have to apply dampening techniques in the cone or dome which reduce resonances at the cost of efficiency. So, these new speakers had an order of magnitude lower distortion AND significantly higher efficiency. The rated sensitivity of the tweeter was 95dB 1W 1M, which was 4dB higher than the previous top of the line tweeter from the company. 4dB is more than double the power – so a 100 watt amp could be replaced with a 50 watt amp using the new tweeter and the in room loudness peaks would remain the same. Crazy! But unbelievable.

So I continued contemplating designs with other transducers while waiting for real world tests of the new Scan-Speak elliptical VC drivers to show up. To my surprise, a famous kit designer out of Denmark presented his bookshelf speaker based on these new drivers, and low and behold he said all the claims from Scan-Speak were more than true and even provided standard measurements to demonstrate the improved performance. Hmmmm…. But that’s just one dude in a socialist nation who depends the great USA for their defense. Cannot trust those pinkos, so I waited for more tests from real people and continued working on designs with some new drivers from Morel that looked really good – one drawback is that the top of the line Morel tweeter is the same one I was replacing in my old dead system and I wanted significantly better performance. There were the new Beryllium tweeters from Scan-Speak and SB Acoustics I was considering, but I’ve never liked metal domes. It was a long frustrating wait.

So, while I waited and played with new design ideas, I got to help Heeman with his stunning 13” woofer cabinets. That was a good time. Then I built my huge line array speakers and master bathroom speakers. Along the way the trade rag, Voice Coil Magazine, published a test bench review of the new Scan-Speak elliptical speakers, called the Ellipticor line, and the extremely trustworthy and often brutally honest Vance Dickason (author of the Loudspeaker Design Cookbook) gave them not only his highest compliments, he was almost gushing in the article. This thrilled me as I desperately wanted the claims by Scan-Speak to be true. I immediately emailed Vance to confirm the tone of his article and he couldn’t have been more positive about a set of speakers. Damn, it appears these are the greatest thing in the high fidelity speaker market in a LONG LONG time.

With that I started pricing a system and given the insanely high prices for these new hand-made in Denmark speakers, I was struggling to justify using them. At first, I considered abandoning my proven MTM D’Appolito array which has proven audible in-room benefits and going with a mere MT design which saved me the price of two of the midrange drivers, but I simply could not go backwards. So, I settled on finding a way to pay for the MTM solution. Then, I considered building a woofer enclosure similar to Heeman’s since I knew how good it sounded, but as much as I love Heeman, I had to be better. So I chose the little brother to his 13” woofer and went with two 10” woofers resulting in 18% more cone area, 52% greater motor strength, 84% more power handling, and 110% higher sensitivity (per speaker channel) which should result in higher peak output, lower distortion, and more powerful dynamics.
So, my drivers were chosen, all from Scan-Speaker, all hand-made in Denmark, and all the absolute state of the art for the classes:

ScanSpeak Revelator 26W/8861T
https://www.madisoundspeakerstore.com/approx-10-woofers/scanspeak-revelator-26w/8861t-10-woofer/

Scanspeak Ellipticor 18WE/4542T-00
https://www.madisoundspeakerstore.c...k-ellipticor-18we/4542t-00-7-midwoofer-4-ohm/
https://www.audioxpress.com/article/test-bench-scan-speak-ellipticor-18we-4542t00-midbass-woofer

ScanSpeak Ellipticor D3404/5520-00
https://www.madisoundspeakerstore.c...r-d3404/5520-00-38mm-elliptical-dome-tweeter/
https://www.audioxpress.com/article...lipticor-d3404-552000-tweeter-from-scan-speak

Ellipticor release announcement:
https://www.audioxpress.com/news/sc...t-models-from-its-new-ellipticor-driver-range
​

​

 

[Flint]

Enclosures & Baffles Design​

I proved with my previous speakers that a larger baffle with offset tweeters and midrange drivers results in vastly superior imaging and sound field as well as an improved in room performance. I knew I would stick to that overall concept. But I also have this idea in my head that I could form a speaker company that sells modular speakers where each component in a speaker is designed as a module which is stacked on top of other modules to get a finished speaker. This would allow for audiophiles to get the type of tweeter they love, like ribbon or metal dome, rather than my preferred soft dome tweeter, without having to build an entirely different full speaker enclosure. I could stock a bunch of different tweeter modules, mid-woofer modules, bass modules, and midrange modules and let the buyer piece together what they can afford, fit into their room, and prefer. So, I decided to test that concept with these speakers.

I came up with a finished look and feel I like and then modularized the components and then fine tuned those blocks into the final designs. Below is the drawing I build my speakers from:

Drawing 1: Initial design for Flint’s Statement Speakers, 2018

The idea of the design is to get the most from the drivers within the limitations of my skills as a cabinet maker, the size of my listening room, and the materials I have access to. Let’s look at each of the design elements as they affect the performance.

Width 1st Benefit: The baffles are a total of 19.5 inches wide. The midwoofer is centered 7.8 inches from one side and 11.7 inches from the other. This means the baffle will reinforce the forward radiation of all sound coming from the speakers at all frequencies above 217 Hz (the ¼ wavelength of 15.6 inches – or double the distance of the speaker centers to the nearer side of the baffle). The roll off below 217Hz will be 3dB per octave down to 145 Hz (the ¼ wavelength of 23.4 inches – or double the distance from the speaker centers to the farther side of the baffle), where the roll off will be 6dB per octave. So, if I can pick a crossover frequency between the woofers and the midrange drivers which is above 217 Hz, the output from the midrange drivers will not require EQ.

Width 2nd Benefit: The offset centers for the drivers results in a significant way the impact on the in-room frequency response be addressing the issue of diffraction from the baffle edges. I wrote lengthy articles about diffraction back in the day on the S&V forum and later posted about diffraction and this offset design on the precursor to this forum when I built my previous speakers. But, if the speakers are centered on the baffle, there will be a comb filter effect which is doubled in amplitude versus this odd-order (2/5ths) off-set approach where the comb filtering will be contradicted between each side of the baffle. Needless to say, there are thousands of papers written about the pitfalls of diffraction. I have personally measured the impact on amplitude response, impulse response, distortion, and off-axis response. And, I have personally heard the difference in my previous system where the center speaker was using exactly the same drivers in a narrow, centered driver speaker compared to my left or right speakers which had those drivers in wide, off-set drivers. The difference was real, and the benefits are clear. This is why companies like B&W have spherical baffles/enclosures, and others have off-set baffles.

Heavily Rounded Baffle Edges: Again, to help address edge diffraction I used a massive 7/8 inch radius round-over router bit. Diffraction is created when a sound wave is moving along a flat surface and encounters a sharp edge. At the point where the wave hits the edge it diffracts into a new sound source at the edge with sound radiating away at all possible directions on that plane. So, rounding the edge cuts back on the diffraction. The radius of the edge determines what frequencies will diffract versus those which bend around the edge. My 7/8 inch radius round-over ensures frequencies above 1,933Hz will not diffract at all. The designed crossover frequency for the tweeter is 1,970Hz at 24 to 48 dB per octave. Thus the side radiating sound off the tweeter should have effectively zero diffraction off the edge of the baffle. The midrange is already turning directional at those frequencies and is not sending any output to the side above 1,000Hz (at 1,000Hz, the 90 degree off axis output is 12dB below the forward 0 degree axis output). So, nearly all diffraction is effectively eliminated, then I offset the drivers to reduce the complimentary impact of what little diffraction is still present. This is overkill, but it doesn’t cause any other negative performance issues and I personally think it looks damn cool.

Driver Alignment 1st Benefit: As you can see from the drawing, the speakers are stacked from the bottom to the top in a woofer – woofer – midrange – tweeter – midrange vertical line. The top three drivers are arranged to comply with the D’Appolito array where the output from the midrange drivers will beam in the upper range and thus reduce lower midrange reflections off the floor and ceiling. The distance from the center of the tweeter to the center of each midrange results in an ideal crossover frequency of 1,970Hz. Using a Linkwitz-Riley aligned crossover algorithm will make things perfect in this range, especially if I can perfectly align the distance of the acoustic center from the listener’s ears – something I can easily do with my digital crossover. The end result will be perfect beaming from just above 500Hz to the crossover frequency near 2,000Hz – thus reducing my need to put acoustic treatment on the floor which are effective in that range. The tweeter, however, will still disperse in a wide pattern which may benefit from floor treatment. I am still contemplating that product.

Driver Alignment 2nd Benefit: With any MTM alignment, you get a better sense of a point source for the sound. It appears to your ears that all the sound coming from the midrange drivers and the tweeter are all coming exactly from the center of the tweeter, thus creating better imaging and a more precise soundstage. If you can also accomplish everything I mentioned above (time alignment, perfect D’Appolito tuning, etc.), the perception will be that a single driver centered on the tweeter is producing all the sound in the critical midrange (500Hz to 5,000Hz). I can verify this effect from my previous speakers. With the early tests of these new speakers it is shocking how perfect the perception of a center speaker is presented to the listener. It is almost scary.

Driver Alignment 3rd Benefit: The woofers sitting below the lower midrange places the midrange at a point in my listening space where I will get a floor bounce null at about 295 Hz. I can address that by setting the woofer crossover at 300Hz where the null will be at a much higher frequency because the woofers are closer to the floor (shorter distance equals higher frequency). Also, placing the two woofers close to the floor reinforces their output by placing them close to a large reflective surface which will offset the baffle step null from the rear wall (the rear wall null will be approximately 104Hz with a peak at about 145Hz). With a solid +6dB from the floor reinforcement, I can apply EQ without worrying about power compression or clipping.

 

[Flint]

Module 3-Point Stacking Design: While the modular design was initially chosen to test the idea of selling customizable modular speakers, I realized I could isolate each module with soft silicon rubber sheets. This would aid in lowering distortions and improving the impulse response. After I first stacked up the modules in my house, I immediately realized I could angle a module downward or upward by putting thicker isolating material between the rear points or front points. Since these speakers place the tweeter a few inches higher than my ears when sitting in the listening position, I can angle the tweeter downward, so it is aimed directly at my ears, and then angle the midrange above it even more, so it is also aimed the same as the lower midrange. This added benefit wasn’t considered when I drew up the plans, but in the current setup in my listening room I am taking advantage of this capability and I can hear and even measure the improvements.

Drawing 2: Angling the tweeter and upper midrange driver to aim at listener’s ear.

Upper Blank Baffle Module: When I chose to go with a modular design, I didn’t consider what the top of the stacked system might look like. After drawing it up without anything above the upper midrange driver, I couldn’t ignore the abrupt and strange truncated top, so I drew the upper midrange module with a taller baffle. However, that would eliminate the module approach by making two different modules for upper and lower midrange units in MTM designs. So, I came up with a blank module which provided some baffle reinforcement above the drivers, and added a more appealing look to the total system. To simplify my build work, I just made a copy of the tweeter module and instead of cutting a speaker hole in it, I rounded over the top side to look pleasant. Quite simple, really. When I first stacked up my speakers, I enjoyed playing with multiple configurations ranging from all the way from a basic MT to all the modules in one stack with a WWWWMMTTMM I was impressed they all stacked up so easily.


Photo 1: Silly optional modular stack consisting of a W-W-M-M-T-M-M-B alignment.



Summary: Every aspect of the size, shape, driver positioning, and alignment is about improving the performance of the speaker by reducing or eliminating in-room acoustical detriments. The result should be high efficiency, better imaging and soundstage, greater dynamic punch, improved impulse response, and lower distortions not related to the drivers themselves.​

 

[Flint]

Construction​

Over the years I have learned that simple off the shelf MDF is by far the best material to build speaker cabinets from. Yes, there are other materials which could be better. But, the better materials are either far too complicated to work with, like aluminum or concrete, or too expensive to acquire, like aluminum or magnesium honeycomb multilayer sheets. So, I am happy to stick with MDF for all the panels needed in my speakers. One material I would love to work with is granite. I believe it would make an excellent front baffle material in the designs I use. However, I need to learn more about how to affix wood to it and get an airtight seal AND ensure it will last the test of time. Maybe if I go into this business I can find a materials expert and mechanical engineer to help me figure it out.

So, I went to Lowe’s and stocked up on a couple of sheets of MDF. In total, I believe I used just over 2.5 sheets of MDF for the speakers. I also relied on simple Tightbond I standard wood glue to hold everything together. As I have mentioned many times before, I have learned to avoid all forms of metal construction affixers, like nails, brads, screws, or whatever. I find they are only useful at the point of assembling because once the glue dries that joint is stronger than the MDF itself. So, a screw or brad actually damages the MDF where it is affixed and reduces the strength at that point. Metal parts also make scraping, sanding, routing, and sealing the MDF much more difficult and potentially dangerous. So, my speakers are made of MDF and wood glue and not much more.

Here’s a short photo essay of the building process:


Photo 2: All the wood panels before any gluing of parts for a stereo pair of speakers.


Photo 3: Wood parts after rounding over and all other cutting.



Photo 4: Silicone Rubber Gaskets for Midrange Drivers and Tweeters.

 

[Flint]

Photo 5: Tweeter Countersink on Baffle.


Photo 6: Midrange Driver Countersink on Baffle.


Photo 7: Baffles Laid Out for Viewing.


Photo 8: Bass Cabinets Clamped without front or rear baffles.


Photo 9: Lacquered Baffles.

 

[Flint]

Photo 10: Gluing the Enclosure to the Midrange Baffle (Using Dumb-bells as claiming weights).


Photo 11: Gluing the Bass Enclosure to the Baffle (using kettle balls as clamps/weights).


Photo 12: Final Baffle Finishing


Photo 13: First Time Stacking Modules from Rear



Photo 14: First time Stacking Modules from Front

 

[Flint]

Photo 15: Tweeter Module & Silicone Strips for Front Edges


Photo 16: Midrange Modules


Photo 17: Stacked Modules in initial Tests


Photo 18: Example of Angled Baffles, Note the Pegs and Bands to hold modules in place


Photo 19: Final Installation and Placement



To come later… Tuning, Measuring, and Listening

 

[Flint]

One aspect of the design process I forgot to add in my initial posts was my predicted mid and woofer bass extension. The midrange is in a sealed enclosure which, according the Vance Dickason, could allow the driver to accept over 58V from an amp without exceeding the excursion limits and thermal limits of the driver. 58V into 4 ohms is a whopping 841 Watts!!!! Think about that!!! With two of these drivers in my rig, I could theoretically put a 1,600 watt amp (into 4ohms) on them!!!

Anyway, the bass enclosures are a modified B4 alignment similar to what I designed for Heeman and the midrange enclosures are sealed and stuffed. Here's the predicted responses of each:

Diagram 1: Predicted bass extension of the Midrange and Woofer speakers in system.

The woofer extends to nearly 30Hz at -3dB and 21Hz at -10dB.

The midrange is clean to about 115Hz at -3dB and has a limit of about 70Hz at -10dB.

While I didn't get a chance to test them outdoors to fully confirm, the in room close mic measurements perfectly match the predictions.

 

[team_tsp]

Still absorbing everything, but amazing job! In many ways it's an elegant design, and based on everything you've mentioned they probably sound amazing. Can't wait to hear them! I'm sure I'll have plenty of questions soon to come lol.

 

[The DirtMerchant]

Awesome.

 

[Huey]

Agreed! Pretty cool!

 

[Haywood]

Did you end up staying with the angled baffles? It is hard to tell from the final picture. How did angling the modules impact the sound?

 

[Flint]

Did you end up staying with the angled baffles? It is hard to tell from the final picture. How did angling the modules impact the sound?

Yes, I did angle the baffles downward by using multiple layers of the 1/8 inch Silicone Rubber.


Photo 20: Side view of Rear showing the Silicone used to angle the uppermost modules downward

As you can see, I placed four small pieces of 1/8 inch silicone between the rearmost point of the support that sticks out the rear of the modules. This puts the rear 3/8 inch farther apart than the front which has one 1/8" sheet of Silicone between the modules. The result is a slight downward tilt to the tweeter and an equally increased tilt from the upper midrange driver. The top blank baffle module is inline with the top midrange module.

The result is the tweeter being aimed directly at my ear when sitting upright in my "critical listening pose" on the sofa. The lower midrange unit is aimed slightly below the ear and the upper midrange is aimed an equal angle above my ear.

Why do I do this?

The characteristic of any driver varies depending on the how far off axis you are from the driver. While it isn't all the important to be exactly on axis for any specific driver, it is important that if two drivers are producing the same sound they are both projecting at the same angle to the listener's ear. So, if the bottom midrange is aimed directly four inches below my ear canal, then the upper midrange should also be aimed directly four inches above my ear. They should also both be exactly the same distance from my ear canal.

Now, if I can place the tweeter at exactly ear level and it is firing perfectly horizontally forward, then a midrange drive below and above the tweeter will be the same off-axis angle from the ear and the same distance from the ear. However, in my rig the tweeter is about 4 inches higher than my ear, so I need to have the upper midrange angled down to be symmetrically off-axis to the lower midrange, and by raising the rear that tips the baffle forward, and thus moves the driver closer to my ear.

With this physical alignment, the tweeter is acoustically centered perfectly and the two midrange drivers are perfectly symmetrical and thus will present a more perfectly in-phase response in the operating range.

Here's a photo of the baffle angles to show how the tweeter and upper midrange lean forward:

Photo 21: Baffle Angles for Tweeter and Upper Midrange Modules

(note: I hung a white towel over a music stand to get a white background so you could see the baffle with better contrast)

As you can see from the baffle angles, the tweeter center is moved a few mm forward and the upper midrange acoustic center is moved exactly double the tweeter. So, if the tweeter is moved forward 2mm, the upper midrange is moved 4mm. I haven't measured it though, as the math is pretty easy to do with angles alone and actual distances just fall in line.

So, that's the benefit of angling the modules forward on the tweeter and upper midrange.

I could also accomplish the same thing by sitting higher so my ear is the same height as the tweeter's acoustic center. Likewise, I could accomplish the same thing by tilting the entire system forward so the entire baffle front is exactly perpendicular to my ear. Neither of those options were available to me. If, however, I replace my listening room sofa, which is a possibility since the current one is 16 years old and in need of some serious help or a trip to the curb, and the new sofa is taller or shorter, I would adjust the tilt of the tweeter and upper midrange to realign to my ear.

 

[Flint]

By the way, the two photos in the post above clearly show the pegs and bands I used to help secure the modules to one another.

What I wanted to accomplish was to stabilize the stack so the modules didn't slip around or move from vibrations while playing music. I felt they were relatively secure when I first stacked them up, but the silicone rubber sheet material I was placing between each module to reduce vibration between modules (which I am convinced makes a big difference) is slippery - that's why they use Silicone to make cooking molds and such. The stretchy bands would also help absorb vibration and potentially improve the vibration absorbing qualities of the silicone by putting more pressure on the thin sheets.

So I decided I would use rubber bands to pull the modules together and increase the pressure on the silicone and hopefully increase the stability of the stacked modules. I press-fit dowel pieces into holes I had cut into the back sides of the modules and went shopping for rubber bands of the right length, but I soon remembered that rubber bands are unreliable and tend to age poorly. I considered tiny bungee cords, but I couldn't find any small enough for this more delicate task. I then looked up scientific gear for rubber seal rings, even silicone rings and was struggling to know what to buy.

Then one day I was walking through Walmart with like 10 cans of lacquer in my cart when I saw a HUGE package of black hair ties. That was my eureka moment. Elastic hair ties are similar to bungee cords as they have multiple rubber fibers banded together by a fabric wrap and would work extremely well for my task. So I purchased the largest package they had with the thinking that I'll replace all of them every 6 months, or so, to ensure they remain tight and not give out without my seeing it.

So, in those photos above you see unpainted dowel segments with hair ties pulling them together.

 

[Zing]

they probably sound amazing

There ain't no "probably" about it.

 

[Flint]

There ain't no "probably" about it.

Golly, thanks.

I think the bass is a little strong, but I'll be refining the crossover for awhile longer.

 

[Zing]

I think the bass is a little strong

Only in presets 2 and 3. 1 seemed awfully well balanced to me.

 

[Flint]

Only in presets 2 and 3. 1 seemed awfully well balanced to me.

You have great taste, sir.

 

[Flint]

In a separate thread I posted the wiring diagram for these speakers and my rig, but it has changed since then. Instead of large amps, I moved to smaller amps which reduces the noise floor, takes up less space, generates less wasted heat, and looks cleaner. I also added a ground loop isolator and decoupled the Blu-Ray player by converting the coax digital to optical and adding a remote controlled optical switch. I also added a basic mechanical RCA Audio switch so I could quickly switch from the Echo Dot to a long analog cable running to the rear of the room for my test & measurement PC, or I can plug in a phone or portable player (sometimes visitors bring their music in that format rather than discs). I also show the 12V power supply I use to feed the trigger power control for all the amps. The power supply is plugged into an Alexa controlled smart AC switch, so I can just tell Alexa to turn on or off the "Stereo" and that switch goes on or off and the amps turn on or off. I love that feature. The Smart AC switch has a button in case the smart features are not working as well as a dedicated phone app in case Alexa is down.



Right now I have separate remote controls for the Blu-Ray player, TV, Amazon Fire TV, Toslink switch, and MiniDSP. I will program my Philips Pronto to control everything with scripts later, but 90% of the time I am only using the MiniDSP control for volume and the Amazon Fire TV for video viewing or the Blu-Ray remote for music. So, generally, once everything is setup, I only need one remote for navigation and one for volume.

 

[Flint]

Holy moly!

Tonight, after a day of working in my yard and cleaning up my porch, I was relaxing while watching the 2nd episode of "The Rookie" on Hulu when I noticed the percussion/drum track at the end section was coming from directly to my left side. Not from the left speaker or some ambient "around the room" sound, but directly from my left. It was as if there was a surround speaker at the left of my sofa.

Insane!!!!