15” Open Baffle 2-way with Live Edge Wood, Constant Directivity, Passive Xover and Punchy Bass
What would happen if a Magnepan MG1.7 and a Klipsch Heresy had a baby?
View attachment 1221614
View attachment 1221615
Many builders here have seen my
Bitches Brew Open Baffle Speakers, the
Flanagangsters, and
Live Edge Dipoles. I love using slabs of live edge wood to make Dipoles. For this system I had a beautiful piece of Walnut in the garage. I wanted to create a passive design that would come as close as possible to these larger, active DSP systems while using a passive crossover, thus maintaining the simplicity of a traditional stand-alone speaker.
At only 35 pounds (16kg), these register at 1/2 to 1/3 the weight of the other designs listed above. Dimensions 28”H x 19”W x 18”D (71cm x 48cm x 46cm).
Sensitivity 93dB
Impedance 4 ohms
Power rating 100w/ch
Frequency Response 40Hz-20KHz
I wanted the expansive, room-filling sound of Open Baffle; and wanted to achieve bass reflex punch without sounding “boxy.” I also wanted Constant Directivity, meaning that when you move off axis, the level changes but the tonal balance does not. This provides excellent imaging for every seat in the house and not just the one “sweet spot.”
View attachment 1221616
Above: Frequency response in real room, 1/6th octave. No attempt to exclude room reflections. Average of 5 positions at a distance between 1 and 2 meters.
View attachment 1221618
Above: 1/6th octave frequency response at 0, 15, 30, 45, 60, 75 and 90 degrees.
View attachment 1221619
Above: Polar plot from -90 degrees to +90 degrees on front plane of the speaker. Largely maintains constant directivity across the entire band, with cardioid pattern between 150Hz and 400Hz. (This is inevitable because of the U-frame enclosure and is the tradeoff incurred by extending the bass an octave downward to 60Hz, as compared to a flat baffle.) There's slight widening of the pattern around 1700Hz, just above the crossover frequency. Radiation is very even across a +/-30 degree listening window.
I chose the
Eminence Alpha 15A woofer with these parameters:
SPL 94dB
Fs 41
Qt 1.26
Xmax 3.8mm
it works very well in an Open Baffle. The high Q means it already has +2dB built in bass boost to counteract the rolloff of the baffle, which starts at around 65Hz. Further down I’ll explain the Bass EQ circuit I designed to extract an extra 5.5dB between 40-50Hz.
After
much deliberation I chose the
SB Acoustics Satori TW29R tweeter. I mounted it in a
Visaton WG148R waveguide.
The Satori is one of the best tweeters I’ve used. It has very high resolution and great imaging, while still retaining the gentle edge of a soft dome. Transparent and revealing without being excessively analytical.
I chose this tweeter because I didn’t want the sound of a traditional compression driver. The kind of horn one would typically mate this with might easily sound nasal or “shouty” despite whatever EQ magic one might throw at it. I wanted a tweeter that, when mounted in a waveguide, could keep up with the 94dB sensitivity of the woofer and even give an extra +1 or +2dB at the high end.
With its standard faceplate, this tweeter has about 94dB SPL in the 10-20KHz range and a waveguide raises SPL further below 10KHz.
“Lambda” U-Frame Open Baffle
I like using triangular “wings” on the sides of OB speakers to extend bass response. Without wings, these woofers will start rolling off at 120Hz and you’ll get little deep bass. With wings, or U-Frame dipole that is lambda shaped in the side view, the cutoff begins around 65Hz and drops 6dB per octave. The triangle shape keeps the distance from woofer to edge spread across a wide range rather.
I discovered a way to add 4-7dB in the 40-60Hz range so the sound is a little “plump” in the deep bass range. I describe this in the next section.
Passive Bass Boost
This design adds a 5.5dB boost at 45Hz using an unusual passive high pass filter. I can’t be the first person in history to do this, but I’ve never seen this explained or modeled anywhere else. These components are in section “1” of the crossover schematic: a 500uF capacitor and 20mH inductor.
View attachment 1221620
I explain the circuit in section 1 in detail in my post “
Open Baffle Bass Boost: +4 to +7dB w/ Passive Xover. No DSP.” I call it the "Marshall Bass Boost." Here you’ll find guidelines and examples for designing your own. It replaces the 25-ohm impedance peak at 41Hz with a 3-ohm dip, drawing almost 4X more power from the amplifier around 45Hz.
This also serves as a subsonic filter, protecting the woofer from signals below 40Hz. Signal falls to -10dB at 30Hz and -17dB at 20Hz. It provides substantial protection from out-of-band signals. Not only do dipoles suffer from rear wave cancellation, they are easily overdriven to large excursions by frequencies you can’t hear. This filter boosts the signal where it matters and cuts it where it doesn’t. The components required for this cost about $40 per channel and are well worth the investment.
View attachment 1221623
Above: Large 20mH inductor and 500uF capacitor boost response at 45Hz.
Drive Signals for Each Driver
In the red curve below you can see the +5.5dB boost at 45Hz as well as the sharp filter.
View attachment 1221624
Above: The signal reaching the Eminence Alpha 15A woofer is in RED. The signal to the SB TW29R tweeter (front) is in GREEN. The signal to the
PRV WG230Ti tweeter (rear) is in BLUE.
Woofer crossover (Section 2 in the schematic)
The 1mH inductor, 10-ohm resistor plus LCR notch network in Section 2 of the schematic form a shelf filter with a deep null at 1900Hz. This suppresses the +12dB breakup mode of the Eminence woofer. However, I don't roll off the woofer above that frequency. This is because that would add even more phase shift. I avoid steep crossovers except when absolutely necessary. The woofer is already set back 3-4 inches (7-10cm) behind the plane of the tweeter, so a steep crossover would only make that problem worse.
View attachment 1221625
Above: nearfield measurements of woofer and tweeter and the 1250Hz crossover frequency. The woofer rolls off naturally at 12dB/octave above 3KHz. (See the drive signal to the woofer a few paragraphs above, it heads towards 0dB at high frequencies.) The lack of additional filtering reduces group delay. In my opinion the minimal filtering just sounds better.
Matching Woofer to Tweeter
I’m stretching both drivers to the limits of their range. The 15” woofer start to beam above 1KHz, and the WG148R waveguide stops being directional below 2KHz. So there is a little discontinuity in the radiation pattern between 1-2KHz.
View attachment 1221627
Above: Frequency response at 0, 15, 30, 45, 60, 75 and 90 degrees. Very nearly constant directivity across the band. At about 1500Hz you can see some “bunching” of the radiation pattern, where 30-45 degrees off axis, the output is slightly stronger than on-axis. This would go away if the waveguide was 8” (20cm) diameter instead of 6” (15cm). I chose an off the shelf Visaton waveguide, but a larger 3D printed waveguide would solve this. The bunching at 300Hz is because at this frequency, the radiation pattern is more cardioid than dipole.
Tweeter Crossover
The crossover (Schematic Section 3) for the TW29R is a 24dB/octave filter at 1250Hz. There is no getting around using a steep filter for a 29mm dome tweeter being asked to go so low.
Section 4 of the schematic is an RLC notch filter. It suppresses a minor peak at 3KHz. Some builders might not feel it’s necessary, but this detail erased a last bit of objectionable coloration and leaves the upper midrange silky smooth and just slightly recessed.
The slightly rising response on axis compensates for falling response off axis. When you average the 0, 15, 30, 45 and 60 degree curves together, you get a slightly downward-tilting response, which is my goal for voicing the speakers:
View attachment 1221628
Above: Average of 0, 15, 30, 45 and 60-degree response on the front side of the speakers. Plot also includes 2nd and 3rd harmonic distortion.
Mounting the Satori Tweeter to the Visaton Waveguide
The Satori TW29R tweeter has a not-quite-flat faceplate which doesn’t match the WG148R waveguide. I had to add a bead of modeling clay about 3mm thick in order to fill the gap. You can see the modeling clay in the picture:
View attachment 1221629
Rear Tweeter
The rear tweeter is a
PRV WG230Ti. I chose it because it has smooth response and 105db sensitivity so you can easily pad it down to match the front tweeter. I find dipoles sound best when the rear output at high frequencies is a few dB higher than the front output. This adds spaciousness without interfering with detail.
Section 5 of the schematic shows a 7-ohm resistor in series with the PRV tweeter. You can choose any value resistor you wish. This value sounds best to my ears. Below is the rear output:
View attachment 1221630
Above: Rear measurements at 0, 15, 30, 45, 60, 75 and 90 degrees off axis. The system sounds best to my ears when the rear tweeter has a rising response. If you build this system you can play with that resistor and find the value that sounds best to you.
I’m normally a fan of simple crossovers, but due to the multiple conflicting priorities in getting a 2-way passive Dipole speaker to achieve: punchy bass, smooth response, Constant Directivity, high efficiency, high power handling and ultra-transparent high end simultaneously, a price had to be paid. The price was a rather complex crossover. Fortunately it’s the parts count and perfectionistic details that are complex, not the essential concept.
VituixCad files for this design are available here:
https://tinyurl.com/walnutdipolesvituix
Areas for Possible Improvement?
A similar woofer with a curvilinear cone instead of straight-sided cone would likely not have the 1.9KHz breakup mode. It would also have wider dispersion above 1KHz and blend with the tweeter better. An 8” / 20cm waveguide would likewise blend better with the woofer. A more pricey version of this system could be executed with the SB Acoustics 15OB350 woofer and the SB Acoustics 29mm Beryllium Waveguide tweeter. (The crossover would have to be completely redesigned. It would still lend itself to the
Marshall Bass Boost circuit.)
So… how do they sound?
These speakers throw a HUGE soundstage. They image very well everywhere in the room and have a very transparent and detailed yet silky and not-too-analytical high end. The soft dome tweeters are easy on the ears.
The bass is just slightly on the rich side, and lower midrange is warm and buttery. The overall tone is warm, spacious and detailed with a bit of extra sparkle on the top end. They are equally at home with jazz or rock. ZZ Top sounds terrific. “Burning Down the House” by the Talking Heads is an acid test for bass overload, because the bass drum and percussion parts have a lot of very low frequency energy that easily overloads many woofers, especially in OB and Reflex. The Marshall Bass Boost circuit protects these woofers from bottoming, and they can play quite loud. Bass drums and guitars are authoritative.
There is also a lot to be said for their high efficiency. They are easily 5-6dB more efficient than most audiophile speakers. They’re sensitive enough to use with “flea watt” Single Ended Tube amps.
Generally I like solid state amps better, but I do enjoy playing with tubes. Some speakers just sound better with tubes. These sound outstanding with my Yaqin MC-13L 40W amp, which has much more than enough power to drive these.
They excel with massed voices and percussion. Imaging is both spacious and precise with considerable depth. Instruments have a "bloom" and roundness that you almost never hear from standard box speakers.
My non-audiophile friends who walk in the room have an immediate positive reaction towards their appearance. The Walnut cabinets (made by Seth Cothron @studio38designs) look fantastic.
As Jerry McNutt, president of Eminence Speaker says, “The best way to not sound boxy is to not use a box.”
View attachment 1221631
This is the drawing I gave my carpenter, which he used to build the speakers:
View attachment 1221696
