Live Edge Dipoles: 2023 Upgrade Advanced Application Notes
The Live Edge Dipoles took 1st Place at the Parts Express Speaker Design Competition August 4-5 in Ohio. 59 contestants. My Open Baffle design on a slab of live edge birch won top score in the Open Unlimited category.
Judge Jerry McNutt said, “The best way to not sound like a box is: Don’t use a box.” Listener comments: The Beryllium tweeters are extremely transparent; bass is rich, expansive and authoritative; thorough top-to-bottom coherence and integration; warm, ambient and enveloping stereo image. Can effortlessly fill a large room to overflowing with palpable output and dynamic range.
I hosted the Chicago Audio Society a few weeks prior. Listeners were consistently impressed not only with rock and jazz but string quartets and large orchestral works. One listener said they combine the open, room-filling sound of Magnepans with the slam and dynamics of a JBL. Sound stage is huge.
The Live Edge Dipoles were featured on the cover of AudioXpress January 2021. The article there provides a thorough description of the system as it was at the time: https://audioxpress.com/article/you-can-diy-perry-marshall-the-live-edge-dipoles or tinyurl.com/LiveEdgeDipolesAudioXpress.
I have since made numerous upgrades. Before I dive into details, a few insights about speaker design; and why, out of thousands of possible choices, I selected the format I used here.
Why ordinary speakers are doomed
When you visit any audio show, 90% of the speakers share the same weary collection of baked-in assumptions:
- They all have boxes
- They all sound boxy
- Wife Appreciation Factor is low or sub-zero
- Simulated woodgrain finish
- On-axis frequency response is obsessed about
- Off-axis response is ignored
- Radiation pattern is ignored
- Bass is omnidirectional
- Drivers interfere with each other
- They have 1” dome tweeters
- Crossovers butcher impulse and phase response
- There is one “center of the will of God” sweet spot
- The room is ignored (and room acoustics are voodoo)
- Inefficient: 1 watt / 1 meter is often mid- to low-80s
- Hard for amps to drive because of extreme impedance variations
- Drivers are small and sound anemic with real (non-audiophile) music
I don’t believe vocals are the toughest thing to get right in a speaker. Percussion is the hardest! Especially with other instruments. Most speakers fail to excel at percussion because they butcher time response; you can easily measure it.
Check the sidebar that reports Step Response in every Stereophile review and witness the butchering with your own eyes. Woofers and tweeters fire with opposite polarities on different dates. Average speaker designers claim you can’t hear the difference. I insist you can. Once you hear the “boxiness” of box speakers, you can’t un-hear it.
With the Live Edge Dipoles, the most delicate of sounds, the wires and shell of a snare drum, are reproduced the startling precision. When the drummer slams the floor tom (like in “Hatesong” by Porcupine Tree), you feel it in your bones.
40 Years of Design
My early teenage years were, shall we say, less than fun. I escaped into Audio projects. I fell in love with a pair of Boston Acoustics A60s but couldn’t afford ‘em. Laid my hands on a McGee catalog and made my own with a Peerless polypropylene woofer and the same Tonegen tweeter as the A60s.
I sold my first pair of speakers to a paying customer when I was 14. I briefly sold my own brand through a local dealer alongside B&W, Denon and PS Audio when I was a senior in high school. Then an Electrical Engineering degree with accent on Control and Communication Systems. Worked as an acoustical engineer for 3 years at Jensen designing OEM drivers for Honda, Ford, Chrysler and Acura.
Since then, audio has been my hobby, with occasional articles in Voice Coil and AudioXpress. I’ve made almost every type including acoustic suspension, reflex, bandpass, transmission line, horn, dipole, ribbon and shaded arrays. I’ve built Motional Feedback subwoofers and used active, passive, analog and digital crossovers. Homes, cars, churches, studios and live bands.
A memorable leap in my journey as a speaker designer was my first Open Baffle design using 12” Faital coaxials. The huge size of soundstage and the lack of boxiness made a striking impression on me, playing quiet classical guitar music at 7 o’clock in the morning. The energy of pro-sound drivers without compression sealed the deal.
Radiation Pattern
Polar pattern is one of the great underappreciated secrets of great speaker design, and it’s a secret to these speakers’ performance. The capacity to deliver low distortion, flat frequency response, transient response, etc. is admirable. But to achieve an excellent distribution pattern at the same time separates the men from the boys. When response is only flat on-axis, your design will sound more like a speaker and less like real music.
When room reflections don’t match direct sound from the speaker, it sounds unnatural. Textbook speaker designs pretend the room is not there. But not only does the bass have room modes and standing waves, but the mid and high frequency reflections add LOTS of clues for your ears. Good clues make your ears happy.
PERFORMANCE SUMMARY
- 30Hz to 25KHz with silky-smooth response in a real room that has real reflections. Not just in an anechoic chamber.
- True Constant Directivity sound pattern 30Hz to 25KHz so imaging is superb anywhere in the room
- Open Baffle Dipole sounds great even behind the speakers
- High Efficiency – 95dB 1 watt/1 meter. Sounds great and plays loud even with “flea watt” single ended vacuum tube amps
- High Power Handling and High Output – 100 watts & 115 dB running full range; 500 watts & 120 dB with subwoofer
- Near-perfect impulse response
- Near-perfect phase response
- Low Distortion (<2% above 60Hz; <1% above 150Hz at 90dB)
- Easy to drive
- Live Edge wood is so beautiful, non-audiophile wives of non-audiophile men gasped when I showed these on a screen in a Zoom meeting.
- 3-way system is bi-amped (not tri-amped) using a MiniDSP 2x4HD.
Above: Polar spectrum of the Live Edge Dipoles, 150Hz-15KHz. Beamwidth is defined as -6dB points relative to 0 degrees, which is the cyan color in the graph. Beamwidth stays between +/-30 and +/-60 degrees across the entire range.
Above: Frequency response in a real room at 0, 15, 30, 45, 60, 75 and 90 degrees. The number of commercially available speakers that come close to these off-axis curves, you can count on one hand. I discuss the polar pattern below. They have a rear-firing “ambience” tweeter which makes the rear radiation very similar to the front. To my ears, dipoles without this don’t sound right. A level pad lets you tailor the ambience to taste.
A true Constant Directivity speaker yields good stereo imaging even if you are standing right next to one of the two speakers. It delivers outstanding sonic images to every seat in the house. When arranged in a triangle pattern with the speakers pointed towards the listener position, every seat sounds great.
Great Stereo Image Everywhere
Below is a crayon drawing of a room with Constant Directivity Dipoles. It shows why Open Baffles with uniform response, toed in about 30 degrees, image better than every other type of speaker.
Stereo imaging is strong anywhere inside the yellow boundary, starting at A.
B, right in front of one speaker, is the only spot where imaging isn’t good.
When you stand at C you hear the right speaker across the room just as well as the left. And yes, you get a decent stereo image there.
Same is true even at D. You can hear the right speaker clearly, even standing just behind the left speaker.
There are no sidewall reflections at the wall near C because it’s in the null plane of the dipole.
1, 2, 3 and 4 illustrate that rear radiation reflects from the wall back into the room. This provides ambience but does not override direct sound.
The “sweet spot” is somewhere between F and G, depending on how “headphone-like” you want your sound.
Place your turntable at T, the null for both speakers. Cuts acoustic feedback 10-15dB.
As you walk from C to E to H, the stereo image remains stable and the volume grows louder. E and H, though much farther away, are louder than C because the directivity pattern is so well-controlled.
Very few speakers do any of this. To my knowledge the number of available commercial designs that achieve constant directivity across the entire spectrum is zero.
Above: Frequency response of the Live Edge Dipoles, on axis at 1.5 meters in my room. No attempt to gate or eliminate room reflections. Room response should taper downward slightly as frequency rises. Some speakers measure more perfectly in an anechoic chamber… but these measure very well in real rooms. Not just in one spot but everywhere in the room! This EQ sounds most neutral to my ears.
In this design I optimized the physical design and shape for radiation pattern, and then I corrected the problems this incurred with Digital Signal Processing.
Upgrades Since the 2021 AudioXpress article:
- I upgraded the aluminum dome in the Radian 5208C to beryllium. The beryllium has very similar frequency response but sounds quite a bit different. Extremely clear and transparent. Sounds “buttery transparent” and not analytical. The tweeter just disappears.
- I changed the tweeter wiring and several EQ settings in the DSP crossover.
- I added Bass EQ options. The stock design is flat down to below 30Hz. This is great for small rooms and listeners who are not “bass heads.” However, if you push them super hard with lots of deep bass, the subs will complain. I added a “Rock and Roll” EQ setting that rolls off below 35Hz. With this setting you can play loud music in a large room and the speakers will handle the full dynamic range without effort.
The standard version with aluminum dome sounds fantastic and I have no criticisms. Though the beryllium diaphragms cost several hundred dollars apiece, they are well worth the dinero. They deliver extreme transparency similar to ribbons and electrostatics, without harshness. The tweeters just disappear and let the music flow through. Finally, they’re not merciless. Many high-def speakers are harsh and unforgiving of ordinary recordings, but the Rolling Stones and AC/DC sound great too.
High SPL Drivers, Amps and DSPs
The Eminence 18 Kappa LF woofers produce 98dB SPL / 1W. The Radian 5208 coax puts out 95dB from the woofer section and ~100dB from the 16-ohm tweeter section.
High efficiency makes them extremely sensitive to electronic hum and noise. Amps with noise and grounding problems, not evident in typical 85-90dB audiophile speakers, are exposed. Furthermore, the MiniDSP 2x4HD has some noise which is not obvious with most speakers, but evident with high efficiency horns.
You can solve this by using an amp with an input control and turning it down a bit (Adcom GFA2535 is a great 4 channel amp and has level controls) or with a padding device like a Harrison Labs 12dB attenuator.
An even better solution is the MiniDSP Flex Eight, which has higher resolution and lower noise floor. I have configuration files for this too, discussed below. I further minimized the noise problem by wiring the rear tweeter (PRV WG175PH) and front tweeter in series, with parallel resistor networks bringing the total impedance down to 11 ohms. This tames the tweeter output to 95dB.
Only the most basic functions are performed by passive components: 200Hz crossover between woofer and midrange; tweeters are wired in series with an L-pad for adjusting level of the rear tweeter; a 20uF capacitor simply protects the tweeters from accidents. All heavy lifting is done by the DSP with 2KHz crossover with FIR phase correction.
Noise sensitivity notwithstanding, the energy, dynamic range and effortless performance you get from efficient pro drivers like Eminence and Radian more than compensate for the noise precautions you have to take. Speakers with vast dynamic range and low distortion sound so effortless and open, you feel compelled to crank the volume up.
Most audiophiles are so conditioned to buckets of intermodulation distortion, they’re stunned when a system doesn’t have it. I call it “Dynamic range to burn.” Once you’ve experienced it, you never go back.
The “Lambda Slab” U-Frame
The triangular sides on the bottom of the cabinet on each side of the 18” woofer pushes the dipole roll off frequency lower. Without them the roll off would start around 100Hz; with them it’s around 60Hz. The wings buy you almost an octave of bass.
It’s easy to get deep bass from an Open Baffle speaker… you just need a Digital Signal Processor and heavy EQ. You pay a price, however, which is that your woofer can easily bottom. Distortion spikes at low frequencies. I have paid close attention to this issue. The bass EQ begins in earnest below 60Hz. This is the curve for the standard “Jazz EQ” DSP file:
Above: EQ reaches 20dB at 29Hz, then rolls off gently below. Yep, 20dB is a ton of EQ, but remember that these have 95-98dB sensitivity, so across 90% of the spectrum they require very little power. Extra power is only demanded by occasional deep bass passages.
I call this the Jazz EQ setting. It’s optimized for lowest group delay and least ringing in the step response. Most speakers have gobs of phase shift at low frequencies which you easily hear on well recorded bass drums. The steeper the filter, the more phase shift. Steep slopes like in actively assisted 6th order reflex designs add many tens of milliseconds of phase delay. This design sidesteps that.
Above: Phase is +/-30 degrees from 40Hz to 20KHz with a max of only 120 degrees around 20-30Hz. Very, very few speakers match this. This cuts group delay and maximizes bass resolution.
Above: Step response. Fast rise time followed by steady descent with no phase reversals. Drivers fire in perfect time with each other. There is little low frequency ringing because the design minimizes phase shift and group delay. Very few speakers match this. This is one reason why clarity and imaging are so precise.
Above: Impulse response. Most speakers shred impulses, which should look like an upside down “T.” The Live Edge Dipoles deliver 80% of their energy in a tight 0.1 millisecond slice. This makes lightning fast transients and exquisite detail.
Bass from most subs, even celebrated designs, sounds more like well-articulated thumps than genuine instruments. Open Baffle speakers bring a welcome exception and sound extremely natural. The Jazz EQ setting achieves flat response to below 30Hz with minimum phase delay. Perfect for listeners with small rooms and / or do not demand the pantleg-flapping bass of car stereo shows and rave clubs.
However if you have a big room and/or want to shake the house with loud rock or electronic music, I added a 2nd EQ configuration file called “Rock & Roll EQ” that adds a dual shelf filter (instead of standard high pass) at 35Hz, with a slight lift between 40-60Hz and above 5KHz:
With this one tweak, the speaker can play 10dB louder than the Jazz setting. So if you like your music loud, even in a large room, this EQ setting will keep your woofers from bottoming or wincing in pain.
Above: Low frequency drive signal from amp. Blue = Jazz EQ. Orange = Rock & Roll EQ. Reduces woofer excursion by 70%, allowing up to 10dB greater output.
This matters in an Open Baffle design. That’s because an acoustic suspension woofer’s max output falls at the rate of 12dB per octave. That is challenging enough if you’re trying to shake the house. But an Open Baffle woofer loses traction at the rate of 18dB per octave (12dB from standard radiation resistance, plus an extra 6 dB due to rear wave cancellation). So every 1/3 octave of bass you ask of an Open Baffle speaker demands +6dB which requires 2X the excursion and 4X the power. One full octave costs you 18dB: 8X excursion and 64X power!
By relieving the 18” woofer of everything below 35Hz, you buy yourself 10dB more dynamic range. The 18” Eminence woofers, bolstered by the U-frame wings and 35Hz roll off on the Rock & Roll EQ, use 14dB of boost at 39Hz and generate high SPLs without effort. You can load both settings into your MiniDSP 2x4 HD and switch between them at will.
Low Distortion
Below: Distortion measured at mid-band output of 90dB SPL/1M in a real room; SPL climbs to 100dB at 40Hz. This is with the Rock & Roll EQ setting. Distortion is under 2% above 60Hz, under 1% above 150Hz. (Data also includes rattles and buzzes of the room itself.) The darker curve is 2nd order, lighter 3rd order:
(continued in next post)
Choice of Wood
There is nothing special about birch, other than I fell in love with the slabs I found at Big Red Sawmill in Palmyra Nebraska. Birch is not heavy. They weigh 57 pounds (26kg) each. Other woods will have a different sound. You may prefer hardwoods like ash, maple, oak or walnut due to their added mass and density.
Polar Pattern Nuances
Below you see a somewhat different representation of the polar response. It calls extra attention to two lobes around 1000Hz where, off axis, the output is +2dB higher than on axis. Near the top of this article you see off-axis frequency response curves and they “bunch” a little bit near 1000Hz.
This is because the Radian 5208 is mounted in a baffle about 20” wide, which means between 500Hz and 1500Hz there is almost no rear cancellation reaching around from the other side. This causes some “bunching” of the off-axis frequency response curves around 1KHz.
The above polar plot accents mild hot spots at +/-30 degrees near 1KHz. This +2dB is the most trivial of problems; but if were the baffle pyramid shaped so as to be narrower at the top, ~12” wide, these regions would go away. It would also require a mild boost in the 120-700Hz range.
The coaxial Radian midrange + compression horn tweeter integrate like a single full range driver, without the beaming or high IM distortion of full range drivers! You can’t get more coherent than a concentric driver where the woofer cone doubles as a Constant Directivity waveguide. Tannoy popularized this format long ago; today, DSP transforms it. When melded with the subwoofer via 6dB in-phase crossover, the transition is invisible to the ear. No time delays, lobes or interference patterns.
Crossover frequencies are 200Hz and 2KHz. 2KHz yields a seamless blend between midrange and tweeter with the DSP’s ability to precision align the signals. The 200Hz keeps most of the low-mid energy coming from the midrange without saddling it with too much bass power.
Even in extreme conditions, the Radian’s excursion is below +/-2mm. If you plan to drive these to super high SPLs, reduce the 200uF capacitors to 100uF.
Why the 18” Woofer & 8” Mid-Woofer Blend so well
I use an identical format using different size drivers (2 x 15” subs + 15” coax with compression horn tweeter) in the Bitches Brew speakers https://tinyurl.com/bitches-brew-diyaudio. This config yields the most coherent top-to-bottom integrity of anything I’ve tested. One audiophile described it as “superb crisp sharp punchy detail, fully integrated like a chorus line all kicking in perfect unison.”
Because woofer and mid are wired in series, and mounted on the same baffle, there is no easy way to separate them. I said they cross at 200Hz, but there’s more to the story.
Below are the electrical drive signals measured at the terminals of 18” woofer, 8” mid and coax tweeter. The point where electrical signals between woofer and mid match is 130Hz. You see the DSP bass boost to the woofer at 30Hz, and a diminished boost to the mid which peaks at 40Hz. This is not a problem because the Radian 5208 has a very tight suspension and a resonance of 90Hz.
#1 priority is: Radian mid must maintain low excursion. It’s rare to see it move, even with heavy bass and punishing volumes. It receives 20dB less signal in the deep bass than the sub. You also see a dip in the pink curve at 22Hz. That is the 22Hz impedance peak of the Eminence (graph appears later in this article) suppressing signals near 22Hz to the Radian. This is the appeal of a series crossover network.
Above: Woofer (green), midrange (orange) and tweeter (purple) drive signals at the speaker terminals. Black is the LF signal coming from the amp which drives woofer and midrange. Please notice that between 30Hz and 130Hz, the voltage at the woofer terminals exceeds the input voltage! This is because of the resonance of the LC network, which is also reflected in the impedance minimum of 4.3 ohms at 90Hz.
The graph below reveals why the 18” Eminence and 8” Radian blend so well. The blue curve is with the Radian midrange in the polarity shown on the schematic. The orange curve is with the Radian’s polarity opposite of the schematic.
With opposite polarity, the 18” and 8” cancel between 100Hz and 1000Hz, with the biggest null at about 250Hz. This means the two drivers produce nearly equal output across 2 octaves, from 150Hz to 600Hz. This is why they blend so seamlessly.
You can tell low frequencies come from the 18, but there is no obvious audible crossover. Both drivers behave well far outside their range. You have none of the disembodied mismatch that so often comes with subs and satellites.
The orange curve also has a null between 1.5KHz and 3KHz. This is where the Radian mid and Radian tweeter overlap. Because the 200Hz crossover is coherent and the mid crosses at 2KHz to a concentric tweeter, the sound is homogenous.
Above: Reversing polarity of the midrange illustrates reinforcement vs. cancellation between woofer, mid and tweeter.
On the schematic you see a 22uF capacitor and 7-ohm resistor. This is a Zobel network which cancels the inductance of the woofer section. If you use a tube amp it minimizes interaction between speaker and amp.
The 18” Eminence subwoofer and 8” Radian midrange are wired in series, with a 200uF capacitor across the sub and a 4.7mH inductor across the midrange. The series wiring causes the impedance peaks of the drivers to work with my design goals. The impedance peak of the 18” woofer at resonance reduces the signal received by the midrange at that frequency. A standard parallel crossover would do the exact opposite.
The inductors (see photo) are low-DCR copper foil units from Madisound. DC Resistance is 0.6 ohm. DCR must be low because the only protection the midrange has from low bass (which has 20dB boost at 30Hz) is low resistance for the inductor in parallel with it. I don’t recommend iron core inductors because they have hysteresis distortion.
This 6dB SERIES (not parallel!) filter is the only crossover between woofer and mid, which is why the blend between the drivers is seamless. 3-way speakers are extremely hard to get right and this is one of the aspects I’m especially proud of.
The midrange is wired in reverse polarity from the woofer. It is in this configuration that the two are most tightly in phase with each other. FIR filter in the DSP corrects global phase and makes the phase response of the speaker flat. A similar filter corrects time errors on the tweeter.
Shelf filters: Less phase shift than Low Pass & High Pass filters
On the left is an 18dB per octave MiniDSP high pass filter at 800Hz, which adds 270 degrees of phase shift below 400Hz. On the right is a 16dB shelf filter, which introduces a maximum of 135 degrees of phase shift at 800Hz, and zero phase shift at low frequencies.
The two filters are obviously very different and serve different purposes. But any time I can get away with using a shelf or notch instead of high- or low-pass filter, I’ll take it every time. If you study my original Birch Dipole writeup from 2020 at
http://www.tinyurl.com/liveedgedipoles
…you can see this preference in the crossover design, which I explain in detail. The place I use this the most is woofers. Most of the time you have to tame peaks at the top end of the woofer’s range. If you do it with notch or shelf filters instead of low pass filters, you cut the phase shift in half, impulse and step response improve, there is less group delay and the system sounds more natural. I do this not only with DSP but also in traditional passive crossovers.
The only traditional crossover slope in the DSP is an 18dB/octave high pass filter at 800Hz protecting the tweeters from low frequencies. I kept it well away from the 2KHz crossover frequency to minimize phase shift. The rest of the EQ was done with shelf filters. This left fewer problems for the Finite Impulse Response filters to solve.
Judges’ comments from the 2023 Speaker Design Competition
Judges were Vance Dickason from Voice Coil Magazine, Jerry McNutt of Eminence, and Peter Noerbaek of PBN Audio. Judging forms, not necessarily in that order:
Performance in every category was 8, 9 or 10 points. Net 9.0 out of 10.
Add a Sub?
Hardly anyone would suggest these speakers lack bass. However, the clutch does slip below 35Hz with rapidly diminishing returns. If you are a “bass head” and want room shaking 15Hz-30Hz vibes, cross over the sub low: 35-50Hz. Open baffle speakers possess window-like transparency that box speakers don’t match. If you crave Open Baffle sound, I also recommend instead of Reflex you choose OB, Transmission Line or Acoustic Suspension for your sub.
Once I showed off the Live Edge Dipoles at a “Bourbon Night” listening party at the Drake hotel ballroom. I crossed them at 50Hz to an 18” subwoofer. The dipoles had no trouble filling the room with music using a 60-watt amp. I arranged chairs in a semicircle as you see, and every person in the room enjoyed an excellent stereo image.
Few were audiophiles. None had ever experienced anything like this before. They made playlists and heard their favorite songs in a brand new way.
When you toe in the speakers as seen here, imaging is superb everywhere in the room, including center listener position. You also don’t want the wide birch panels exactly parallel to the back wall, as it can create slap back. Toeing them in helps that too.
No Steep Filters
I am conservative applying filters. I only use brute force when absolutely necessary. I dislike steep slopes because they add time delay. The MiniDSP has both IIR (standard) filters, plus a FIR DSP option for each channel. I do as much EQ and crossover work as I can with peak and shelf filters. At the end, I use Eclipse FIR software to generate “maximum phase” correction curves. I also use FIR to apply fine-grained 1/48th octave EQ on the tweeters.
Easy Impedance
Above: Impedance curve of Eminence 18” Woofer + Radian 8” Midrange section. Resonance 22Hz, Qt = 0.5. Average 6 ohms, minimum 4.3 ohms at 60Hz. Load is largely resistive. Phase angle is in red. Easy to drive.
Above: Tweeter section impedance curve. 11 ohms, super easy to drive, especially for small tube amps. Phase angle is at the top.
Construction
I commissioned Seth Cothron in Chicago to build the cabinets (instagram.com/studio38designs and he’s very skilled and easy to work with). All I gave him was some sketch drawings and circles drawn on the wood:
DSP File Download for MiniDSP 2x4HD
These are configuration files for the MiniDSP 2x4HD and include FIR filters. These are optimized for my particular speakers and the idiosyncrasies of my individual drivers. The “Jazz EQ” (bass down to 25 Hz & no steep filters for linear phase) and “Rock & Roll EQ” files (flat down to 35Hz with steep attenuation below) are available in a ZIP file here:
https://tinyurl.com/birchdipole2023
You’ll want to measure and tailor your EQ for optimum performance.
(Continued in next post)
There is nothing special about birch, other than I fell in love with the slabs I found at Big Red Sawmill in Palmyra Nebraska. Birch is not heavy. They weigh 57 pounds (26kg) each. Other woods will have a different sound. You may prefer hardwoods like ash, maple, oak or walnut due to their added mass and density.
Polar Pattern Nuances
Below you see a somewhat different representation of the polar response. It calls extra attention to two lobes around 1000Hz where, off axis, the output is +2dB higher than on axis. Near the top of this article you see off-axis frequency response curves and they “bunch” a little bit near 1000Hz.
This is because the Radian 5208 is mounted in a baffle about 20” wide, which means between 500Hz and 1500Hz there is almost no rear cancellation reaching around from the other side. This causes some “bunching” of the off-axis frequency response curves around 1KHz.
The above polar plot accents mild hot spots at +/-30 degrees near 1KHz. This +2dB is the most trivial of problems; but if were the baffle pyramid shaped so as to be narrower at the top, ~12” wide, these regions would go away. It would also require a mild boost in the 120-700Hz range.
The coaxial Radian midrange + compression horn tweeter integrate like a single full range driver, without the beaming or high IM distortion of full range drivers! You can’t get more coherent than a concentric driver where the woofer cone doubles as a Constant Directivity waveguide. Tannoy popularized this format long ago; today, DSP transforms it. When melded with the subwoofer via 6dB in-phase crossover, the transition is invisible to the ear. No time delays, lobes or interference patterns.
Crossover frequencies are 200Hz and 2KHz. 2KHz yields a seamless blend between midrange and tweeter with the DSP’s ability to precision align the signals. The 200Hz keeps most of the low-mid energy coming from the midrange without saddling it with too much bass power.
Even in extreme conditions, the Radian’s excursion is below +/-2mm. If you plan to drive these to super high SPLs, reduce the 200uF capacitors to 100uF.
Why the 18” Woofer & 8” Mid-Woofer Blend so well
I use an identical format using different size drivers (2 x 15” subs + 15” coax with compression horn tweeter) in the Bitches Brew speakers https://tinyurl.com/bitches-brew-diyaudio. This config yields the most coherent top-to-bottom integrity of anything I’ve tested. One audiophile described it as “superb crisp sharp punchy detail, fully integrated like a chorus line all kicking in perfect unison.”
Because woofer and mid are wired in series, and mounted on the same baffle, there is no easy way to separate them. I said they cross at 200Hz, but there’s more to the story.
Below are the electrical drive signals measured at the terminals of 18” woofer, 8” mid and coax tweeter. The point where electrical signals between woofer and mid match is 130Hz. You see the DSP bass boost to the woofer at 30Hz, and a diminished boost to the mid which peaks at 40Hz. This is not a problem because the Radian 5208 has a very tight suspension and a resonance of 90Hz.
#1 priority is: Radian mid must maintain low excursion. It’s rare to see it move, even with heavy bass and punishing volumes. It receives 20dB less signal in the deep bass than the sub. You also see a dip in the pink curve at 22Hz. That is the 22Hz impedance peak of the Eminence (graph appears later in this article) suppressing signals near 22Hz to the Radian. This is the appeal of a series crossover network.
Above: Woofer (green), midrange (orange) and tweeter (purple) drive signals at the speaker terminals. Black is the LF signal coming from the amp which drives woofer and midrange. Please notice that between 30Hz and 130Hz, the voltage at the woofer terminals exceeds the input voltage! This is because of the resonance of the LC network, which is also reflected in the impedance minimum of 4.3 ohms at 90Hz.
The graph below reveals why the 18” Eminence and 8” Radian blend so well. The blue curve is with the Radian midrange in the polarity shown on the schematic. The orange curve is with the Radian’s polarity opposite of the schematic.
With opposite polarity, the 18” and 8” cancel between 100Hz and 1000Hz, with the biggest null at about 250Hz. This means the two drivers produce nearly equal output across 2 octaves, from 150Hz to 600Hz. This is why they blend so seamlessly.
You can tell low frequencies come from the 18, but there is no obvious audible crossover. Both drivers behave well far outside their range. You have none of the disembodied mismatch that so often comes with subs and satellites.
The orange curve also has a null between 1.5KHz and 3KHz. This is where the Radian mid and Radian tweeter overlap. Because the 200Hz crossover is coherent and the mid crosses at 2KHz to a concentric tweeter, the sound is homogenous.
Above: Reversing polarity of the midrange illustrates reinforcement vs. cancellation between woofer, mid and tweeter.
On the schematic you see a 22uF capacitor and 7-ohm resistor. This is a Zobel network which cancels the inductance of the woofer section. If you use a tube amp it minimizes interaction between speaker and amp.
The 18” Eminence subwoofer and 8” Radian midrange are wired in series, with a 200uF capacitor across the sub and a 4.7mH inductor across the midrange. The series wiring causes the impedance peaks of the drivers to work with my design goals. The impedance peak of the 18” woofer at resonance reduces the signal received by the midrange at that frequency. A standard parallel crossover would do the exact opposite.
The inductors (see photo) are low-DCR copper foil units from Madisound. DC Resistance is 0.6 ohm. DCR must be low because the only protection the midrange has from low bass (which has 20dB boost at 30Hz) is low resistance for the inductor in parallel with it. I don’t recommend iron core inductors because they have hysteresis distortion.
This 6dB SERIES (not parallel!) filter is the only crossover between woofer and mid, which is why the blend between the drivers is seamless. 3-way speakers are extremely hard to get right and this is one of the aspects I’m especially proud of.
The midrange is wired in reverse polarity from the woofer. It is in this configuration that the two are most tightly in phase with each other. FIR filter in the DSP corrects global phase and makes the phase response of the speaker flat. A similar filter corrects time errors on the tweeter.
Shelf filters: Less phase shift than Low Pass & High Pass filters
|
|
|
On the left is an 18dB per octave MiniDSP high pass filter at 800Hz, which adds 270 degrees of phase shift below 400Hz. On the right is a 16dB shelf filter, which introduces a maximum of 135 degrees of phase shift at 800Hz, and zero phase shift at low frequencies.
The two filters are obviously very different and serve different purposes. But any time I can get away with using a shelf or notch instead of high- or low-pass filter, I’ll take it every time. If you study my original Birch Dipole writeup from 2020 at
http://www.tinyurl.com/liveedgedipoles
…you can see this preference in the crossover design, which I explain in detail. The place I use this the most is woofers. Most of the time you have to tame peaks at the top end of the woofer’s range. If you do it with notch or shelf filters instead of low pass filters, you cut the phase shift in half, impulse and step response improve, there is less group delay and the system sounds more natural. I do this not only with DSP but also in traditional passive crossovers.
The only traditional crossover slope in the DSP is an 18dB/octave high pass filter at 800Hz protecting the tweeters from low frequencies. I kept it well away from the 2KHz crossover frequency to minimize phase shift. The rest of the EQ was done with shelf filters. This left fewer problems for the Finite Impulse Response filters to solve.
Judges’ comments from the 2023 Speaker Design Competition
Judges were Vance Dickason from Voice Coil Magazine, Jerry McNutt of Eminence, and Peter Noerbaek of PBN Audio. Judging forms, not necessarily in that order:
Performance in every category was 8, 9 or 10 points. Net 9.0 out of 10.
Add a Sub?
Hardly anyone would suggest these speakers lack bass. However, the clutch does slip below 35Hz with rapidly diminishing returns. If you are a “bass head” and want room shaking 15Hz-30Hz vibes, cross over the sub low: 35-50Hz. Open baffle speakers possess window-like transparency that box speakers don’t match. If you crave Open Baffle sound, I also recommend instead of Reflex you choose OB, Transmission Line or Acoustic Suspension for your sub.
Once I showed off the Live Edge Dipoles at a “Bourbon Night” listening party at the Drake hotel ballroom. I crossed them at 50Hz to an 18” subwoofer. The dipoles had no trouble filling the room with music using a 60-watt amp. I arranged chairs in a semicircle as you see, and every person in the room enjoyed an excellent stereo image.
Few were audiophiles. None had ever experienced anything like this before. They made playlists and heard their favorite songs in a brand new way.
When you toe in the speakers as seen here, imaging is superb everywhere in the room, including center listener position. You also don’t want the wide birch panels exactly parallel to the back wall, as it can create slap back. Toeing them in helps that too.
No Steep Filters
I am conservative applying filters. I only use brute force when absolutely necessary. I dislike steep slopes because they add time delay. The MiniDSP has both IIR (standard) filters, plus a FIR DSP option for each channel. I do as much EQ and crossover work as I can with peak and shelf filters. At the end, I use Eclipse FIR software to generate “maximum phase” correction curves. I also use FIR to apply fine-grained 1/48th octave EQ on the tweeters.
Easy Impedance
Above: Impedance curve of Eminence 18” Woofer + Radian 8” Midrange section. Resonance 22Hz, Qt = 0.5. Average 6 ohms, minimum 4.3 ohms at 60Hz. Load is largely resistive. Phase angle is in red. Easy to drive.
Above: Tweeter section impedance curve. 11 ohms, super easy to drive, especially for small tube amps. Phase angle is at the top.
Construction
I commissioned Seth Cothron in Chicago to build the cabinets (instagram.com/studio38designs and he’s very skilled and easy to work with). All I gave him was some sketch drawings and circles drawn on the wood:
DSP File Download for MiniDSP 2x4HD
These are configuration files for the MiniDSP 2x4HD and include FIR filters. These are optimized for my particular speakers and the idiosyncrasies of my individual drivers. The “Jazz EQ” (bass down to 25 Hz & no steep filters for linear phase) and “Rock & Roll EQ” files (flat down to 35Hz with steep attenuation below) are available in a ZIP file here:
https://tinyurl.com/birchdipole2023
You’ll want to measure and tailor your EQ for optimum performance.
(Continued in next post)
Last edited:
MiniDSP Flex Eight
The MiniDSP Flex Eight offers 8 channels instead of 4 with significantly lower noise, with some other features compared to the 2x4HD. The 2x4HD has separate FIR taps on each output channel, which I take full advantage of in the Birch Dipole design. But the Flex Eight has only two FIR taps – only on the main input channels. This requires a different strategy.
Below you see the final acoustic output of the drivers using the Flex Eight DSP. I attempted to get as close as I could to a low-Q, well-behaved 12dB/octave acoustic slope from both woofer and tweeter, using almost entirely peak filters and shelf filters. Acoustically, it behaves approximately like a three-way system with 12dB/octave acoustic slopes on every driver.
Above: Woofer+Bass mid acoustic output (blue); tweeter output (red).
Above: IIR Digital EQ curves, as programmed into the Flex Eight. Also has a 6dB per octave high pass filter on the tweeter at 600Hz not shown here. Most of the heavy lifting is done by shelf filters, which have less phase shift. A 0.26msec delay aligns midrange with tweeter.
Before FIR correction, the system has 360 degrees total phase shift between 50Hz and 15KHz. Two very simple 1st order maxphase FIR slopes, below, made with Eclipse FIR Designer, reverse this:
Step response (below left) is excellent. Phase response is +/-30 degrees across 9 octaves, below right:
You can download the MiniDSP Flex Eight files at https://tinyurl.com/birchdipoleflex8
Driver Substitutions?
I used the Radian 5208C-16 Beryllium for 200Hz-2KHz and 2KHz-20KHz. Outstanding. You can save money with the standard aluminum diaphragm which is quite good. The beryllium tweeter is superb, the bass midrange has a very neutral character, and the radiation pattern is great. The only substitute I might endorse is its bigger brother, the 10” Radian 5210, also available with both aluminum and beryllium 1.75” domes. You’d need to measure and adjust the DSP.
Rear tweeters are the affordable PRV WG175PH with an L-pad for volume adjustment. Substitute other compression horn tweeters as you desire. Working out the fine details is up to you.
I chose the Eminence Kappa Pro 18LF-8 18” woofers for 25Hz-200Hz. They have Sd of 1180 cm^3 and 8mm Xmax, for a robust Volume Displacement of 930 cm^3. Reasonably priced and sturdily built.
You should select woofers with an absolute minimum sensitivity of 93dB SPL. I do not recommend conventional low efficiency high-Xmax subwoofers with high mass and thick heavy rubber surrounds. I do not expect they will perform well in this design. They will not integrate well with the Radian midrange. Stick with pro format woofers with lighter cones and accordion suspensions.
Other subs that are good candidates:
The SB Acoustics 15OBO350 15” woofer will also work well here. I use it in the Bitches Brew Live Edge Dipoles. See www.perrymarshall.com/speakers and full details at https://tinyurl.com/bitches-brew-diyaudio. 11.5mm Xmax means its Volume Displacement is comparable to the Eminence.
The Faital 18XL1800 is an exotic 18” woofer with 20mm Xmax. I have not tried it, but if you want much more Xmax while retaining high efficiency, FaitalPro makes excellent drivers and it looks to be a superb choice.
The Acoustic Elegance Dipole18, 16ohm with dual coils wired in parallel, is likely an excellent option. Any change to the woofer will require attendant changes in overall EQ and you are on your own for that.
Remember: These could have been box speakers
It would be easy enough to ditch the dipole configuration, lose the rear tweeter and put these same parts in a conventional 4-6 cubic foot / 100-150 liter ported box. They would still deliver Constant Directivity above 1000Hz. They’d play as low and loud as any home listener could ever ask for: a state-of-the-art rendition of the Klipsch Cornwall.
But they wouldn’t be gorgeous anymore. Wife Appreciation Factor would drop like a rock. Sonically they would lose their magical edge, because this particular driver combo in a dipole configuration hits that elusive sweet spot. Literally the only thing they don’t dish out with aplomb is room-shaking, sub-35Hz bass. In every other respect they’ll hold their own with any commercial design costing less than a brand-new Mercedes.
A Pair of Your Own
I recommend you visit to a lumber store that carries live edge wood and pick out a couple of slabs that call out your name. You’ll be the only person on earth with speakers that look like that.
Order the parts and thus has your adventure begun.
The Live Edge Dipoles hold their own against the most prestigious brands in the world - B&W, KEF, Estelon, YG Acoustics, Wilson Audio, Focal, TAD, Genelec and Dynaudio. In fact they do a number of things better.
I don’t know of any commercially available speaker that combines Open Baffle, Constant Directivity, Linear Phase, High Efficiency and High Power like these do. A growing number of DIYers on the forums have now built versions of these, and thrilled with the results. It’s your turn.
Carpe Diem – Seize the Day,
Perry Marshall
The MiniDSP Flex Eight offers 8 channels instead of 4 with significantly lower noise, with some other features compared to the 2x4HD. The 2x4HD has separate FIR taps on each output channel, which I take full advantage of in the Birch Dipole design. But the Flex Eight has only two FIR taps – only on the main input channels. This requires a different strategy.
Below you see the final acoustic output of the drivers using the Flex Eight DSP. I attempted to get as close as I could to a low-Q, well-behaved 12dB/octave acoustic slope from both woofer and tweeter, using almost entirely peak filters and shelf filters. Acoustically, it behaves approximately like a three-way system with 12dB/octave acoustic slopes on every driver.
Above: Woofer+Bass mid acoustic output (blue); tweeter output (red).
|
|
|
Above: IIR Digital EQ curves, as programmed into the Flex Eight. Also has a 6dB per octave high pass filter on the tweeter at 600Hz not shown here. Most of the heavy lifting is done by shelf filters, which have less phase shift. A 0.26msec delay aligns midrange with tweeter.
Before FIR correction, the system has 360 degrees total phase shift between 50Hz and 15KHz. Two very simple 1st order maxphase FIR slopes, below, made with Eclipse FIR Designer, reverse this:
Step response (below left) is excellent. Phase response is +/-30 degrees across 9 octaves, below right:
|
|
|
You can download the MiniDSP Flex Eight files at https://tinyurl.com/birchdipoleflex8
Driver Substitutions?
I used the Radian 5208C-16 Beryllium for 200Hz-2KHz and 2KHz-20KHz. Outstanding. You can save money with the standard aluminum diaphragm which is quite good. The beryllium tweeter is superb, the bass midrange has a very neutral character, and the radiation pattern is great. The only substitute I might endorse is its bigger brother, the 10” Radian 5210, also available with both aluminum and beryllium 1.75” domes. You’d need to measure and adjust the DSP.
Rear tweeters are the affordable PRV WG175PH with an L-pad for volume adjustment. Substitute other compression horn tweeters as you desire. Working out the fine details is up to you.
I chose the Eminence Kappa Pro 18LF-8 18” woofers for 25Hz-200Hz. They have Sd of 1180 cm^3 and 8mm Xmax, for a robust Volume Displacement of 930 cm^3. Reasonably priced and sturdily built.
You should select woofers with an absolute minimum sensitivity of 93dB SPL. I do not recommend conventional low efficiency high-Xmax subwoofers with high mass and thick heavy rubber surrounds. I do not expect they will perform well in this design. They will not integrate well with the Radian midrange. Stick with pro format woofers with lighter cones and accordion suspensions.
Other subs that are good candidates:
The SB Acoustics 15OBO350 15” woofer will also work well here. I use it in the Bitches Brew Live Edge Dipoles. See www.perrymarshall.com/speakers and full details at https://tinyurl.com/bitches-brew-diyaudio. 11.5mm Xmax means its Volume Displacement is comparable to the Eminence.
The Faital 18XL1800 is an exotic 18” woofer with 20mm Xmax. I have not tried it, but if you want much more Xmax while retaining high efficiency, FaitalPro makes excellent drivers and it looks to be a superb choice.
The Acoustic Elegance Dipole18, 16ohm with dual coils wired in parallel, is likely an excellent option. Any change to the woofer will require attendant changes in overall EQ and you are on your own for that.
Remember: These could have been box speakers
It would be easy enough to ditch the dipole configuration, lose the rear tweeter and put these same parts in a conventional 4-6 cubic foot / 100-150 liter ported box. They would still deliver Constant Directivity above 1000Hz. They’d play as low and loud as any home listener could ever ask for: a state-of-the-art rendition of the Klipsch Cornwall.
But they wouldn’t be gorgeous anymore. Wife Appreciation Factor would drop like a rock. Sonically they would lose their magical edge, because this particular driver combo in a dipole configuration hits that elusive sweet spot. Literally the only thing they don’t dish out with aplomb is room-shaking, sub-35Hz bass. In every other respect they’ll hold their own with any commercial design costing less than a brand-new Mercedes.
A Pair of Your Own
I recommend you visit to a lumber store that carries live edge wood and pick out a couple of slabs that call out your name. You’ll be the only person on earth with speakers that look like that.
Order the parts and thus has your adventure begun.
The Live Edge Dipoles hold their own against the most prestigious brands in the world - B&W, KEF, Estelon, YG Acoustics, Wilson Audio, Focal, TAD, Genelec and Dynaudio. In fact they do a number of things better.
I don’t know of any commercially available speaker that combines Open Baffle, Constant Directivity, Linear Phase, High Efficiency and High Power like these do. A growing number of DIYers on the forums have now built versions of these, and thrilled with the results. It’s your turn.
Carpe Diem – Seize the Day,
Perry Marshall
Hi Perry (moved to this thread)
I read your notes (which are in fact studies) with great pleasure.
After finishing my first ever OB project with many trials and errors and (to my ears) very good result at the end (with passive serial xo), I will start the Bitches Brew project.
It will take time, of course, and DSP will be a completely new land for me, but why not...
Right at the beginning: how critical is the wood material and even more the shape of the baffle? Parts are easier to collect, but the baffle seems to play a big role. And to the DSP: my preamps and power amps are balanced - do you have a recommendation for a balanced DSP?
Many thanks in advance!
I read your notes (which are in fact studies) with great pleasure.
After finishing my first ever OB project with many trials and errors and (to my ears) very good result at the end (with passive serial xo), I will start the Bitches Brew project.
It will take time, of course, and DSP will be a completely new land for me, but why not...
Right at the beginning: how critical is the wood material and even more the shape of the baffle? Parts are easier to collect, but the baffle seems to play a big role. And to the DSP: my preamps and power amps are balanced - do you have a recommendation for a balanced DSP?
Many thanks in advance!
There is a reason they make violins out of one material and not another. The wood makes a difference but I can't tell you exactly how. In general with a speaker, the more dense, hard and heavy the wood, the better. However you will notice that I did not choose dense woods for my builds. I picked out what was beautiful and available at the lumber store. Oak and Ash are dense woods.
I would love to do one of these speakers using marble or granite instead of wood. Maybe someone on the forum will try that.
The shape of the baffle does make a difference. If you study the Live Edge Dipoles above you will see that there is some "bunching" in the polar pattern around 1KHz (see below), and in the heat map you see two yellow hot spots at +/-40 degrees. That's because the baffle is about 20" wide and the midrange is only 8" wide.
@CharlieLaub has written about this extensively. He calls it a "minimally baffled dipole" in recent articles in AudioXpress discussed here:
https://www.diyaudio.com/community/threads/charlielaub-ob-article-in-audioxpress-sept-2023.402166/
The way you sidestep that problem in Live Edge Dipoles is you cut the cabinet so it tapers from 22" wide at the bottom down to 10-12" wide at the top. Then that bunching will disappear. You will also get less obstruction of reflections from the rear wall. (You will also have to trim down a piece of live edge wood and it may not look as natural.) That will also require you to adjust the DSP and boost frequencies between 120Hz and 700Hz a few extra decibels, because you'll get some baffle loss.
The Bitches Brews do not have that problem because the midwoofer is 15" and the baffle is 20", only a little wider. That's why if you look closely the Bitches Brew polar pattern does not have any of the "bunching" that the Live Edge Dipoles have. Then by having U-frame "lambda wings" to extend the bass, along with a 100Hz crossover, I manage to make the speaker dipole across the entire range while still having a minimally baffled 15" coaxial bass mid.
Think of it as a pyramid shaped baffle where the bottom of the pyramid wraps around the side. That pushes the dipole loss cancellation down to below 50Hz, which buys you about +6dB to +8dB more bass.
Here are the graphs for comparison.
Live Edge Dipoles:
Bitches Brew:
Re: Balanced inputs and outputs - it appears that the MiniDSP "Flex" (not Flex Eight) has an option for balanced outputs, see https://www.minidsp.com/products/minidsp-in-a-box/flex and this unit will work just fine with either design. You'd have to keep it biamp as I've done, and not triamp since the Flex only has 4 channels. I can't promise that you can use the exact same configuration files I provided, but you can certainly use the same curves and FIR filters.
I would love to do one of these speakers using marble or granite instead of wood. Maybe someone on the forum will try that.
The shape of the baffle does make a difference. If you study the Live Edge Dipoles above you will see that there is some "bunching" in the polar pattern around 1KHz (see below), and in the heat map you see two yellow hot spots at +/-40 degrees. That's because the baffle is about 20" wide and the midrange is only 8" wide.
@CharlieLaub has written about this extensively. He calls it a "minimally baffled dipole" in recent articles in AudioXpress discussed here:
https://www.diyaudio.com/community/threads/charlielaub-ob-article-in-audioxpress-sept-2023.402166/
The way you sidestep that problem in Live Edge Dipoles is you cut the cabinet so it tapers from 22" wide at the bottom down to 10-12" wide at the top. Then that bunching will disappear. You will also get less obstruction of reflections from the rear wall. (You will also have to trim down a piece of live edge wood and it may not look as natural.) That will also require you to adjust the DSP and boost frequencies between 120Hz and 700Hz a few extra decibels, because you'll get some baffle loss.
The Bitches Brews do not have that problem because the midwoofer is 15" and the baffle is 20", only a little wider. That's why if you look closely the Bitches Brew polar pattern does not have any of the "bunching" that the Live Edge Dipoles have. Then by having U-frame "lambda wings" to extend the bass, along with a 100Hz crossover, I manage to make the speaker dipole across the entire range while still having a minimally baffled 15" coaxial bass mid.
Think of it as a pyramid shaped baffle where the bottom of the pyramid wraps around the side. That pushes the dipole loss cancellation down to below 50Hz, which buys you about +6dB to +8dB more bass.
Here are the graphs for comparison.
Live Edge Dipoles:
Bitches Brew:
Re: Balanced inputs and outputs - it appears that the MiniDSP "Flex" (not Flex Eight) has an option for balanced outputs, see https://www.minidsp.com/products/minidsp-in-a-box/flex and this unit will work just fine with either design. You'd have to keep it biamp as I've done, and not triamp since the Flex only has 4 channels. I can't promise that you can use the exact same configuration files I provided, but you can certainly use the same curves and FIR filters.
Last edited:
My project is similar, but very slow going. I’ve chosen dual 15’s for the bass/midbass region as well using the Faital 15pr400 with a .5 inductor for the bottom driver. Midrange driver isn’t worked out yet but I’m using the Aurum Cantus Aero Striction AMT tweeter in a dipole configuration. Been just listening to these with different baffle shapes and trying to determine by ear where the need to hand off to a mid driver. My initial plan (and it may still go that way) would be no mid driver and cross the AC down at 1.2khz………I’m just not sure how I feel about the FP driver that high up yet
Post in thread 'Ultimate Open Baffle Gallery' https://www.diyaudio.com/community/threads/ultimate-open-baffle-gallery.123512/post-7213962
@perrymarshall so you would change the tilt of the Bitches Brew to 5 degree. Change anything else like the wings (bunching around 100Hz) [shape, parallel] or stay the same with 5 degree? Or change anything else?
Very nice speaker designs!
@perrymarshall so you would change the tilt of the Bitches Brew to 5 degree. Change anything else like the wings (bunching around 100Hz) [shape, parallel] or stay the same with 5 degree? Or change anything else?
Very nice speaker designs!
I would change the tilt from 10 degrees backwards to 5 degrees backwards.
I would not change anything else about the cabinet or the structure.
MiniDSP 2x4HD files: https://tinyurl.com/bitchesbrewdsp
MiniDSP Flex Eight files: https://tinyurl.com/bitches-brew-dsp-flex8
This is the current schematic:
Regarding "bunching" at 100Hz: I'm not confident in the measurements that show it, mostly because it's in a 12' x 14' room and I have a suspicion that a different microphone position would make those go away. It may not be real.
What I can say is that if you're concerned about it, build it as a 3 way triamped active system, because if you play with the crossover points and slopes you will get rid of it. Taken at face value, the problem might disappear if you moved the crossover from 100Hz down to 70Hz. It would stress the bass mid a bit more but I don't think that's a problem.
One solution to the baffle shape dilemma with the Live Edge would be to saw the log at an angle to give you a natural tapper from bottom to top. This would weaken the baffle some, but with the thickness and back bracing I doubt it would be an issue. It would be wasteful and quite expensive but I think it would look very interesting.
Funny you ask, I was just comparing last night. The two are in adjacent rooms.
Overall they sound surprisingly similar. Definitely a strong family resemblance in both appearance and sound character. They are both very detailed with a huge soundstage and a very 3 dimensional presentation.
The Walnut Dipoles are warmer sounding with a somewhat stronger upper bass around 100-200Hz. They don't have any obvious flaws. The tweeter is great it's a little softer sounding than the Bitches Brew B&C Horn. They are more forgiving of equipment deficiencies.
The Bitches Brews have a full octave more bass and quite a bit more high end resolution. Not to mention 15dB more dynamic range. They project a larger sound, and because of active DSP you can remove any colorations at will. In the end with careful listening they are hands down better than the 15As at doing almost everything. They are also not forgiving of flaws in the signal chain. If your amplifier or other components have any "grain" or harshness, they will reveal that mercilessly. As you may know Jakob said they compare very favorably to the $300K Avantgardes.
So considering the Walnut Dipoles w/ 15As are FAR less expensive, smaller and simpler, and play nicer with a wide variety of "mid-fi" components, I'm quite proud of them. They will compete easily with any commercially available speaker under $15K.
Last edited:
Great info, many thanks Perry!
In the meantime - and with this further info - I decided to build the Walnut Dipoles first, just because I have the Alphas and many of the required parts - and ordered the missing ones. And of course because I want to learn more and also want to make a comparison with my OBs, before I go for DSP.
Sadly, the PRV tweeter is not available here in Europe, so the postage costs from the US are much higher than the price of the two tweeters, crazy... Maybe I could replace it with an SB Audience Bianco 34CD-PK in a H225 horn, but this horn is much bigger.
For the baffle I would use 40mm (1,57") oak, I hope it works.
In the meantime - and with this further info - I decided to build the Walnut Dipoles first, just because I have the Alphas and many of the required parts - and ordered the missing ones. And of course because I want to learn more and also want to make a comparison with my OBs, before I go for DSP.
Sadly, the PRV tweeter is not available here in Europe, so the postage costs from the US are much higher than the price of the two tweeters, crazy... Maybe I could replace it with an SB Audience Bianco 34CD-PK in a H225 horn, but this horn is much bigger.
For the baffle I would use 40mm (1,57") oak, I hope it works.
One issue with the H225 could be that it's too deep. You need to check. Definitely you don't want it deeper than about 10" / 25cm. Any good horn that will go down to 1200Hz will work.
As I said to another guy last week don't be afraid of DSP. It's easier than you think and far better than passive. The only reason I did the Walnut Dipoles passive xover was for the challenge. I enjoy working against a constraint, it's a game I like to play with myself. (This was not an easy crossover to design BTW.) But if it was just to have a great sounding set of speakers I would choose DSP over passive in a heartbeat.
As I said to another guy last week don't be afraid of DSP. It's easier than you think and far better than passive. The only reason I did the Walnut Dipoles passive xover was for the challenge. I enjoy working against a constraint, it's a game I like to play with myself. (This was not an easy crossover to design BTW.) But if it was just to have a great sounding set of speakers I would choose DSP over passive in a heartbeat.
Nice job. I was wondering what design software was used. I just did a quick read of the posts and may have missed it, but it would be great to see this design modeled in VituixCAD2 or any of the available freeware speaker design software. In the past year I have been modifying my speaker designs to produce the widest possible uniform dispersion and have found the results to be very satisfying. I was able to improve an older speaker build by just lowering the crossover points so the drivers were not beaming. The desire for narrow dispersion speakers to exclude the room acoustics for control room monitoring and mixing is much different from the desire to produce a performance in your listening room which requires a very wide dispersion speaker to fully illuminate and include the room acoustics. With non-resonant sealed box designs I build I started to target the full forward, +/- 90 degree dispersion or a full 360 degree omni directional pattern. The miniDSP products work well. I have recently used the Hypex multi-channel plate amps with build in DSP and found them to produce a very compact, cost effective solution as well.
Last edited:
If you have a microphone and speaker test software, the MiniDSP configuration software has less learning curve than any speaker modeling program like VituixCad or HornResp.