2026-04-07
my project
The ripole subwoofers are done! Time for a sweep. Also painted the HF Peavy horns yellow, thinking of getting the peavy logo as a tramp stamp.
Speaker Building and Research Log
2026-03-26
Review of HiFi Pursuit Listening Room No.3
I arrived at Listening Room No. 3 at the Cooper Hewitt, Smithsonian Design Museum prepared, frankly, to be underwhelmed. Too often, systems that borrow the visual language of classic horn loaded hi-fi promise romance and deliver nostalgia at the expense of precision. The system by Devon Ojas avoids that trap entirely. It is a rare example of a system that embraces character without losing control.
Horn loaded systems, when properly executed, still outperform most modern PA architectures in terms of transient articulation and midband intelligibility. This installation makes that case strongly. The system is highly coherent across the room, with very few modal hotspots. Stereo width is limited, but that is an architectural constraint rather than a system failure. The payoff is spatial consistency, with energy distributed evenly into all corners without collapse or overhang.
The low frequency section is particularly revealing. Direct radiating fostex 31.5in subwoofers are used instead of folded horn or tapped horn alignments. This choice produces significantly higher group delay control in the upper bass region, and the subs feel notably faster than typical large format horn-loaded low frequency systems. Crucially, they are positioned and delayed in a way that achieves acceptable time alignment with the midbass horn section, avoiding the usual LF smear that occurs when physical path length and electrical delay are not tightly controlled.
The high frequency architecture is where the system becomes most interesting. The super tweeters, similar to the JBL 2405, are not an extension layer but a structural necessity. They offload extreme HF energy from the Altec 1505 style multicell horns, allowing the compression drivers to remain within a lower distortion operating band. The multicell dispersion pattern produces a highly diffuse upper mid and lower high frequency field, avoiding the axial beaming common in radial horn designs. The result is a high frequency image that is not point sourced but volumetric, with a noticeable reduction in spatial collapse at off axis positions.
Tonally, this is not a flat reference system. There is an intentional spectral contouring, with elevated sub bass and a slightly lifted air band. It reads closer to a controlled smile curve than a linear target. That choice is not incidental. It produces a perceptual increase in scale and physicality, particularly in the transition from low bass to lower midrange, where many horn systems tend to thin out if not compensated.
The crossover implementation feels functionally musical rather than purely technical. Each band is isolated into discrete horn or driver systems, effectively segmenting the spectrum into instrument-like domains. The result is a reduction in intermodulation within each band and a noticeable increase in transient separation. Electronic material, especially fast synth sequences and arpeggiated content, remains clean under load, with minimal harmonic intermodulation between bass and midband content.
Room treatment is appropriately aggressive without becoming overdamped. The space suppresses early reflections enough to maintain clarity while preserving sufficient decay to avoid anechoic dryness. This balance is essential in a system of this scale and dispersion geometry.
Where this system becomes more interesting is when considered against contemporary club sound design trajectories.
Venues such as Nowadays demonstrate a fairly successful approach to high frequency management. The elevation of HF elements at the top of the stacks, combined with distributed high frequency dispersion units suspended from the ceiling, follows a design logic consistent with the legendary Paradise Garage sound system topology. This vertical HF distribution does a strong job of reducing harsh axial beaming and creating a more even HF field across the dancefloor.
However, the tradeoff is in the low frequency domain. The multi-bin low frequency arrays, particularly the cluster of 2 bass horns or kick bins positioned on one side of the floor, and the 4 bass bins at the bottoms of the quad stacks introduce asymmetry in LF propagation and increase temporal smear in the 40–120 Hz region. The system becomes perceptually slow and muddy in the bass due to enclosure loading, path length variance, and room coupling that is not fully controlled at those wavelengths. The system design at Public Records, also designed by OJAS, can exhibit excessive low frequency energy in an acoustically under-treated environment. The use of multiple large format low frequency W bins introduces significant stored energy in the room and slow bass due to the long horn path combined with the already large room. In that context, the system tends to overpopulate the modal region, making fast tempo material, particularly techno or high BPM electronic music, feel less temporally resolved than intended. The issue is not output capability but decay time and room interaction in the sub and low bass region.
From a systems perspective, there is an interesting design question here around scale efficiency. It is not always acoustically or economically optimal to deploy extremely large diaphragm low frequency drivers in club environments where transient speed and articulation are as important as SPL. There is a case to be made for hybridizing paradigms.
A synthesis of approaches could be more effective. The spatial HF distribution principles of Paradise Garage sound system era and systems like Floating Points' Sunflower Sound System, combined with the modular midbass horn logic seen in contemporary designs associated with designers like Shorty and newer club system builders, could be integrated with faster low frequency approaches such as dipole configurations or high quality open horn systems in the HOQS lineage. This combination would likely improve temporal coherence in house, experimental, and beat-driven music contexts without sacrificing scale.
Likewise, replacing conventional dome or point-source HF systems with multi-entry horn topologies, fully separated into independently amplified and DSP controlled bands, would allow far more precise spatial tuning of venues and larger sound stages. With proper FIR-based correction and per-band delay alignment, the entire system could be optimized not just for SPL or coverage, but for groove integrity and transient readability.
What becomes clear across all of these comparisons is that the Listening Room No. 3 system succeeds precisely because it avoids the excesses found in larger scale club deployments. Smaller volume, tighter treatment, and more disciplined crossover partitioning reduce phase instability. At the same time, the inherent dispersion characteristics of the multicell horns suggest that in a larger architectural volume, the system could scale into something significantly more immersive, potentially achieving a far wider lateral image without sacrificing coherence.
2026-03-26
my project
The Ripole subs me and Jackson are making with the old black widow 15in peavy drivers I harvested from a (much bigger and heavier) PA are coming along! Cant wait to do a sweep of these, hope they go down to at least 40hz, it's a waste of time if they dont go under 100hz.
2026-02-10
my project
Setup
2025-12-29
my project
I cant believe the year is almost over. Here are some renderings and models I have been making of tractrix and multicell tractrix horns.
2025-12-09
my project
The clouds have been SEVERED. Added 15 in woofers and replaced the tweeter with an 8ohm fostex tweeter, now all the drivers are 8ohm and the soundstage has much more bass ;). Very happy. Because of the matching impedance I can now use them with tube amplifiers, and the crossovers are back where they are supposed to be. These will pair nicely with my ripole subwoofers.
(Feb 2nd 2026 update) I finally measured these speakers and they are pretty much what I expected, not too hifi lol. They're lacking highs dramatically and mids by quite a bit. Here's the frequency response before and after eq. Setup in my room I just have the record player going through the eq, and into the amp. The room I am in is pretty flat with little nodes and resonances.
2025-10-21
I still have not done a freq sweep on the cloud speakers, and I still haven't heard them outside of the weird acoustics of my insane carpeted huge office space of a studio but one thing i learned for sure is that I need to do something about the bass response and the tweeters. I'm learning about dipole speaker design and how an efficient and small RiPole design for the subwoofer is a good use of space and allows the bass to be more directional, in a figure 8, where the air back of the driver is pushing air backwards out of phase. I don't think the enclosed and now ported bass drivers were a good idea. I think I also stuffed them way too much with fiberglass and lost most of the back volume due to that. In my next design I want to try a stereo pair of these RiPole subs with an open baffle full range driver on top. I have these coaxial drivers I found in an old organ that I would love to put to use as well, but the drivers are different left to right. Efficient open baffle and dipole designs also I think I would enjoy more because I am not designing for a club or big spaces, mostly just small rooms and my bedroom. I also want to use them with hifi tube amps which require efficient speaker systems to get the most loudness.
A RiPol (Ridtahler dipole) subwoofer is a specific type of compact, folded-baffle dipole design that uses two woofers in an opposing push/push configuration to produce a unique cardioid-shaped radiation pattern. This design lowers the woofer's resonant frequency, which allows for very deep bass with a relatively small cabinet and reduces room standing waves by canceling out low-frequency energy in certain directions. This results in clearer, more musical bass that is less localized and has less impact on nearby electronics.
The ripole sub will give me deep bass, tuning the port gets you about -7 to -10 hz from the resonant frequency of the driver. So with the peavy black widow drivers I should be able to get down from like 90 hz to 30 or 40.
2025-09-22
my project
The cloud speakers are done! This was a wild learning experience. I was super excited when the crossover came in to set it up. I did a lot of testing at 4ohms with the crossover around 290 for the bass and 2400 for the mid to tweeter, and at 8 ohms with the crossover for the bass around 300 and the tops around 3400. I decided to stick with running the speakers at 4 ohms. One of the immediate things i noticed was that there wasn't enough bass at all. This could be because the bass drivers I was using were 4 ohm, and the mid ones were 8 ohms and covering a wider range, but also because the mid drivers were open back and the subs were enclosed, so you're getting twice as loud of a sound coming out of the mid range drivers. I added two power resistors, whatever I had lying around in the studio to the back of the mid range drivers to make them about 10db quieter which helped but they still feel loud and I don't think ill get much out of these cheap 10in bass drivers, so I'm leaving it where it is for now, and maybe ill add another resistor in series to the back if its too mid heavy. That said, there is a lot of clarity in those mids, not so much in the highs, I wish i had used compression drivers or added a piezo or super tweeter to get up to 20khz, but this is the gonzo build. It just needed to happen and now its done. They sound great! I have a lot of ear fatigue from lots of listening so I'm going to take a break from them, listen to them in new spaces and ask for peoples opinions on how they sound so I don't feel crazy.
2025-09-09
my project
The cloud speakers are almost complete! Today I ordered the crossover. When me and Matt did a sweep on the low, mid and high frequency drivers, we found that the mid range drivers roll off around 2400 hz. When doing the crossover manually with my modular synth I found it works well for the woofer driver around 300hz. I didn't really want to design the crossover from scratch so i went looking for one for similar specs for cheap, but could only find one with a crossover of 300hz and 3000hz. This means most likely there will be a drop between 2400/2800 hz and 3000hz, that I'm hoping isn't too drastic. I think it also rolls off on the bottom end around 100hz, which isn't ideal but its what i got for these. Tested them all today with separate amplifiers and no crossover and it sounds pretty good! Won't know how flat it is until doing another full sweep after the crossover is wired up. The crossover I'm using is a cheaper copy of the Dayton Audio XO3W-375/3K 3-Way Crossover. The one I'm using is the Audiopipe CRX-3003K. I have a feeling the super high frequencies are not really there yet too, maybe I will have to add some sort of super tweeter or piezo tweeter at the top, maybe with some simple tractrix horn. And the crossover matches the speakers in blue!
After doing some calculations and looking at the schematic of the crossover, it turns out my crossover points are actually shifting because of the ohm of my drivers. The tweeter driver is 6ohm, the midrange 8ohm and the woofer 4ohm. If I use the crossover in 8ohm mode, the low pass cutoff for the woofer becomes 425hz and the tweeter high pass becomes 3465hz, which is way too high for where my midrange driver rolls off. Instead I'll have to use it at 4ohms: the tweeter rolloff becomes 2400hz which is actually better for my midrange driver, and the woofer low pass becomes 212hz which is pretty low, but I believe the midrange driver can go that low or at least with only a little rolloff so I'm okay with that.
2025-08-29
The Speaker Stack
The speaker stack is a representation, a totem, a shrine, a ritual object in translating sound.
Each layer of the speaker stack directly correlates to its own architecture and program.
The bottom of the speaker stack, the bass, beneath a layer of grass, represents the ground and the dirt and the earth beneath our feet. We dance on this land, we kick our feet on this dirt together. The indigenous practice of drum circles and dance give thanks and homage to the ground beneath our feet, we celebrate by moving on the ground. The bass frequencies, with their long wavelengths, take seconds to fill a room, they have the longest history. They tell the stories of the elders, as the high frequencies tell the stories of the future, moving past our ears faster than we can perceive. We can hear down to 60hz, but no more than 20,000.
The speaker cabinet is a temple for the speaker driver, an architecture to acoustically amplifying its voice. Of the driver is the person in program, the cabinet is the architecture to make sure that voice is heard loud and clear. There's a reason the wound copper, the heart of what makes a speaker driver work, is called the voice coil. If your skin, like a drum, is the paper cone attached to the voice coil that's excited with voltage as purpose to make sound, what would you say?
The architecture of the horn lets your voice free, extending your voice through the people.
This mechanical object, the speaker driver, so technical in nature, explodes with sound when attached to the proper enclosure. The design of the box, makes a huge difference in the sound of the speaker, as the design of a space make a huge impact on the voices that pass through it. It's this balance that makes speakers so beautiful. It's not Ying and Yang, it's Kaisen: continuous improvement or an uphill battle. There's no perfect balance, and that's what makes the experience of spaces so unique.
The wood against the speaker diaphragm extends the frequencies through the enclosure and amplifies the sound. Coupling the body to the architecture in the same way creates this deep connection to the space you're interacting with.
The Yari Kanna Copy Paper, the hand carving wood tool in direct connection to the manufactured, machined, clean printer paper. It's what makes these ritual objects. The use, over and over. The sound system playing songs, hymns, archival recordings, live performances, hours and hours and miles and miles of wavelengths that makes this architecture tell its own story.
There is this ephemeral nature to an object that emits sound: you hear a song and then it's over, you may never hear it again, there's no evidence it was ever played, you may even forget it the next day, but you can always go back to this object and experiment it new, like a sunrise the next morning.
There is real power in the speaker stack. The bass frequencies are so long and big that you can feel them physically, the high frequencies if not tamed can sound shrill like a baby crying or nails on a chalkboard. Years and years of club design and acoustics for the discotheque have led to the design of systems with such personalities and are a testament to our desire to feel sound on a physical comfortable level. This amplifier sounds "warm". "The highs are easy on the ears". A kick box you can "feel in your chest". It's the same as the car audio people that put 12 speakers in the back of their car, pull up to an empty lot where all of their friends also put 12 speakers in the backs of their cars, and see which systems make your hair fly into the air the most when you push the bass.
This is the reason why in loudspeaker design, these tuned systems, not for the room or the venue, but for the people and their ears, are so highly praised, weather it be the smooth deep sound of house music at the paradise garage in heart of NYC the 90s, or the cold harsh punchy kicks reverberating through warehouses at tresor and berghain in Berlin. The most popular and praised speakers designers being Altec Lansing and JBL, speakers originally designed for large scale theatre use, and home audio listening. These designs have been remixed and adapted for small and large scale speaker systems since the 70s and modified versions of their designs are still used in massive venues around the world.
When sound is architecture, these speakers appealed to many programs. There's the designer; that wants them for the living room, the audiophile, that wants the best most "realistic" reproduction of sound for his record collection. "I want to feel like the Beetles are right in front of me on stage". The average consumer that knows the name is good and wants to listen to anything. The musician that wants personalized sound for amplifying their sound. The monk that treats the sound system as a device for deep listening, where electronically amplified sound is needed, like the ritual of dance music.
It is clear, the speaker stack is the temple for the voice.
2025-08-27
Some things to think about for myself when designing these horns. In Hornresp I need to specify that I'm folding the horn, if I'm going to go about designing my own W-Bin for pro Dayton audio drivers. For driver specs, 500w and above is fine, I want high sensitivity for efficiency and loudness. Of course, the resonant frequency — if I want the driver to go low I need to pick a driver with a low resonant frequency, maybe around 40 or 30hz. The Vas (Equivalent Compliance Volume) should be bigger rather than smaller. The larger volume pushes more air. Qts (and its components, Qms, Qes) — Higher mechanical damping (Qms) and controlled total Qts (often between 0.3 and 0.5) are ideal for ported/W-bin designs, ensuring flatter response and better transient behavior. Bl — Force factor. Higher Bl means stronger motor, giving better control at high SPLs. Xmax — Maximum linear cone excursion. A driver with high Xmax supports higher output before distortion—critical for club-level SPL. Sd — Effective cone area. Larger Sd increases displacement volume (Vd), improving low-end output.
2025-08-26
Two new things. The Levan Horn and the Bertha horn are different. The one below with the extension are part of both designs, but the Bertha horn is more recent, and uses 2 15in drivers or 2 18in drivers, the original Levan Horn had 1 18in driver. The mid bass/top used in the original Paradise garage design is also below. I think I will build my original found W bin design and modify it for modern Dayton Audio drivers, with a goal of flat (as possible) 30-80hz. Hornresp is annoying to use, but I think it will give me good simulations for this when I start. I want to use basic angles and normal wood lengths to make my life easier. I also plan to curve all of the folds in the horn and put insulation where needed, to smooth out freq response. Depending how high the superscooper goes and how low my tractrix compression driver horn goes I may not need a conical horn on top, I want to make the system 3 way.
This is my horn response for the W bin with the 15in drivers, according to my bad calculations from my bad rhino model of it, and the TS I'm looking for in the Imperial for the Dayton audio drivers, something similar to this:
2025-08-24
Shorty's mid bass cabinets that he's used at Nowadays are modified Altec 817 for 18in drivers with larger ports. It's too complicated at my level to figure out that horn design custom for 18in woofers and ports so for now I'm going to stick to the 18in scooped J horn for the mid bass design and a conical horn for mid tops before the tweeters.
Found some interesting woodworking techniques and tools that will be useful for speaker and horn building. The first being layering wood to a rough curve and glueing and sanding them into the final shape shown in here. There is also a weird putty used on the back of these speakers mixed with beads and a binder. I think its a mixture of dampening compound, tar, paint, binder, structural putty and/or enameled or textured speaker cabinet paint. I don't think it would be wise to use resin or plaster. In the video, it seems like the putty can be thick, and applied on top of the back instead of sanding, if the walls allow for it.
2025-08-22
The other way to do it would be to do the bass cabinet down to 20hz with the Levan horn Richard Long and "Shorty" use for Paradise Garage and this new venue Shorty is opening. The Levan horn is huge, uses 18in drivers. The horn I'm thinking of using uses two 15in drivers. I'm also thinking about portability and where these speakers are going/moving them around. Probably will put casters on the back but don't want them to rattle.
And finally, a link to a couple of horn making techniques: first, the paper mache horn!!
And the fiberglass horn
2025-08-20
I don’t know much about hi-fi audio just yet, but I want to start documenting all of the awesome projects I’m inspired by at the moment so I know where they are, how to reference them, and to keep images of these wild builds. I’ll probably do that on this log. Here are some I’ve been interested in recently. Most of these are from forums and internet archive hi-fi audio sites. There’s a real sense of craft, construction, and sculptural presence to them.
The first is this concrete horn with a phase plug for mid-bass frequencies. The goal of the phase plug is to equalize path length—in a cone driver, sound from the outer edge has farther to travel to the throat than sound from the center. A phase plug helps compensate for this difference so the wavefront entering the horn is more coherent. It also improves coupling efficiency by narrowing and shaping the exit path so the driver’s output better matches the throat size of the horn, avoiding losses. Additionally, it reduces high-frequency cancellation (beaming or lobing). Even in the mid-bass, frequencies are short enough that interference can occur, and a phase plug helps smooth the response and extend usable bandwidth upward. It also helps control directivity and wavefront shape, forming a more ideal wavefront (spherical or planar, depending on design) to match the horn’s flare. At lower frequencies (below ~150 Hz), the wavelength is long enough that phase plug effects are less critical, so many designs skip it. For higher mid-bass or low midrange (~150–500 Hz), a phase plug can make a big difference in clarity, efficiency, and reducing comb-filtering effects.
The other DIY tractrix horn (without a phase plug) I’m interested in is mentioned in my last post below.
Another wild speaker I came across recently is a set of backwards open-baffle speakers in Ricardo Villalobos’s studio. They use four 18-inch drivers tilted around 15–25 degrees to create a beam of bass from the rear wave of the drivers, with a compression driver in the center. I want to DIY something like this—they’re kind of insane. With my projects, I think it will be relatively straightforward to achieve a wide frequency response, but things like room acoustics, phase alignment, and amplification will be the real challenge. I keep wondering how these speakers avoid severe phase issues when the woofers are projecting sound backward—it’s fascinating.
2025-08-18
I'm thinking of building a 3- or 4-way system with 12–15-inch woofers in a W-bin design, similar to the Altec W-bin subwoofers used in theater applications from the 1940s–1970s, with these dimensions:
For the ultra-low frequencies (anywhere between 20–60 Hz), I want to tune some transmission line tapped horns using large sub drivers, and pair them together to match the dimensions of the W-bin. I haven’t chosen or calculated them yet, but I plan to improvise a bit due to my specific dimensional constraints, focusing mainly on calculating the horn length and selecting an appropriate driver. Ideally, all speakers used in the project will be 8 ohms—likely from Dayton Audio. This is my reference for the math behind transmission line horns:
"In theory, to achieve maximum efficiency, horn dimensions should be one wavelength long and one wavelength in mouth circumference at the lowest frequency. At three octaves higher, the wavelengths are one-tenth of the horn size and too small for the horn to direct them. These small wavelengths bounce around inside the horn chaotically. It is essential to rescale a smaller horn for the next three octaves. Long horns are for low frequencies. Short horns are for high frequencies. Saxophones, trumpets, French horns, tubas, etc., obey this rule. There is no such thing in physics as long-throw or short-throw horns. These irresponsible marketing terms loosely refer to horn directivity. Low-frequency horns using cone speakers can be one-half, but not less than one-quarter wavelength at the lowest frequency, with reduced efficiency. A 40 Hz wavelength is 8 m / 24 ft. It is not practical to make a bass horn this size, so the horn is shortened (truncated) to one-quarter wavelength: 2 m / 6 ft. Bass bin horns can be folded to control size, made in sections, and grouped to form a single bin."
For the midrange horns, I want to load a 12–15-inch driver in the style of Altec A7 VOT or A1 speakers, aiming for tractrix-style horns with two or four sides extending from the driver. This is my build reference for them.
For the high frequencies, I want to design exponential horns for one or two compression drivers, either 3D printed or lathe-cut. Alternatively, I might try building a DIY version of the classic Altec acoustic lens, or experiment with 3D printing/Rhino designs for a multicellular or sectoral horn like the Altec 1505, 1005, or 805 (A and B versions)—most likely the B version, since it’s modular and easier to print that way.
2025-07-30
my project
Right now I'm building these hi-fi open-baffle hybrid speakers using sub drivers that Hoolihan dropped off at the studio, midrange Page drivers that were used as PA speakers in the ceilings at my studio—probably in the offices that occupied the space in the ’90s and early 2000s—and tweeters Eli harvested from some studio monitors. This is my first hi-fi build, and I'm doing it right: trying to design a proper crossover and everything. For a while, I was quad-amping, using my modular system to create crossover filters before the amp. The Page drivers are open baffle, and the bass (bad-ass serious sub) drivers are in a closed box with no port. Matt Azevedo is helping me with the calculations. I need a Windows computer to run Hornresp and Room EQ Wizard, and I also need to order some more parts for the crossover. That said, I've harvested some power caps and inductors from old speaker cabinets in the basement of the Firehouse. I'm learning a lot about crossover design and speaker impedance. The Page drivers had transformers on the back, which made me think they would have a super high impedance, but they turned out to be just 8 ohms. I've also been looking at a lot of Japanese hi-fi listening rooms with wild compression drivers running full-range systems, connected to insane sculptural horns that spiral through the house into the attic to achieve the length needed for bass response—and I want to build something like that for my architecture thesis.
