A passive version of this speaker could be built for $1000 per pair for drivers and crossover parts. You can read the thread and see if this kind of performance might meet your need.
https://www.diyaudio.com/community/threads/new-project-tower-3-way-with-twin-8s.378223/post-6816455
I would use a different woofer, probably the SB23NRX, and I would make the cabinet larger, and ported, since a passive design does not have the benefit of a DSP-enhanced bass response.
I would also use a waveguide on the tweeter, and then I would not need to do the extensive beveling around the tweeter. With a waveguide, the directivity goals can be met with a a simple roundover on the easy-to-build rectangular cabinet.
If you (or anyone else) is interested in replicating this project, I will help out all that I can, and I will share everything I know.
j.
https://www.diyaudio.com/community/threads/new-project-tower-3-way-with-twin-8s.378223/post-6816455
I would use a different woofer, probably the SB23NRX, and I would make the cabinet larger, and ported, since a passive design does not have the benefit of a DSP-enhanced bass response.
I would also use a waveguide on the tweeter, and then I would not need to do the extensive beveling around the tweeter. With a waveguide, the directivity goals can be met with a a simple roundover on the easy-to-build rectangular cabinet.
If you (or anyone else) is interested in replicating this project, I will help out all that I can, and I will share everything I know.
j.
In this case though we're talking DIY.
Back loaded horns are a lot of fun and can give impressive performance.
It does look odd, doesn't it ... This post explains it
https://www.diyaudio.com/community/threads/new-project-tower-3-way-with-twin-8s.378223/post-7017084
One of the goals of this project was to try the "current best practice" recommendation of using 1.2xwavelength CTC spacing. In simulation it does well, and my extensive measurements confirmed it. It is a good method of getting a smooth directivity index and power response through the crossover region. I spent almost a year doing simulations and foam-board mockups to prove to myself it would work.
Of course if we could get CTC spacing of 1/4 wavelength we would do it... but the only way, really, is your way, by using a wide band driver crossed to a woofer at ~ 200 - 400 Hz. I prefer a different set of compromises...
Indeed and lumping in all the other thoughts re BLH design is that IME it all boils down to following T/S and choosing the 'close enough' Fs, effective Qts' to get the desired SQ @ 'X' gain BW, i.e. Fs - Fhm where T/S theory peters out:
upper: Fhm = 2*Fs/Qts'
lower: Flc = Fs*Qts'/2 (AFAIK only used for reactance annulled BLHs)
Qts': 2*Fs/Fhm
Qes' = Qts'*Qms/[Qms-Qts']
Rs = [[Qes'/Qes]*Re]-Re
Maybe for some more easily understood from a horn loading POV along with all the extra math (pg. 7):
http://www.xlrtechs.com/dbkeele.com/PDF/Keele (1977-05 AES Preprint) - LF Horn Design Using TS Paras.pdf
(Qts'): (Qts) + any added series resistance (Rs): http://www.mh-audio.nl/Calculators/newqts.html
(Rs) = 0.5 ohm minimum for wiring, so may be higher if a super small gauge is used as a series resistor and/or there's other series resistance.
Actually, looking at current prices, I think it would come in at closer to $1100 per pair. As I said, if I was going to do a passive version of this, I would use an Augerpro waveguide on the SB26CDC tweeter, and I would use the paper cone versions of the SB23 woofers (SB23NRX) in a ported cabinet of about 85 liters. But by making all of those changes, I would have to completely re-measure all the drivers in the finished cabinet... so my existing work would not be much help... it would almost be like starting from scratch.
But for @Marveloudio, the point is this; a decent 3-way can be built for $1000 per pair. If you are looking at a kit, there are a number of good options. I would add the Jeff Bagby Kairos / Continuum to your list.
Surely the usual way (and possibly only when it comes to high fidelity?) is to use a coaxial arrangement of tweeter and midrange rather than a wideband driver?
At 2023 prices, really nice drivers can be purchased for under ~500-ish EU/$ for a set of 4:
10-12" + wide-range option:
Single suspension Alpair 5 (or similar) 2 x 50€
10-12" pro-mid woofers 2 x 150-200€ (FaitalPRO, Lavoce, SB Audience, Eminence...)
Woofer Boxes:
Basic: 6-sided box, 18mm ply + 6-8mm dowels (these are easy to learn and fun. You can go for a cool riveted look, or shave them down.)
Ambition: angled cuts + a bit more maths for a stronger box.
TBH, I never really cared much for multiway baffles with drivers locked in position. Stacked boxes allow for way more tuning and experimentation.
Wave-guided wide range
I took the plunge carving shallow horns / wave guides. post #10 for visuals. A bit crazy I know, but so is 3d printing in its own way.
My XO design process:
I usually start with frequency sweeps (apart from casual listening), looking at the raw wave height in software like Audacity, AND listening carefully to assess whether I actually like the sound at a certain frequency. Previously, I had a situation where the "break even" point between a 10" subwoofer (in a non-ideal cube box) and a 3.5" wide-range was above 500 Hz -- much higher than I'd expected. The old w.r. drivers had really poor mid-bass and low mids in their ~1L boxes with resistive venting, but I didn't know that before I built them. So the XO design really has to come after some initial building and listening.
-Air-core inductor for the woofers. (For my low power requirements I was able to use high gauge enamel wire).
-Series padding resistor + parallel inductor for the W.R.
Further refinement:
Add parallel notch filters placed in series to EQ any problem areas. I plan to do them one by one, until diminishing returns kick in, so it's important to trust your ears to prioritize.
It would be a similar thing with DSP or active filters: if you only have a few biquads to play with, or knowing that every filter stage adds noise and distortion, you need to somehow choose what needs fixing first. I'm a passive XO convert, but only plan to use 1st order filters and notches passively.
I still plan to do multi-amping and active filters that focus on other aspects that passive filters are not so good at, like a Linkwitz transform to extend the bass. One idea that intrigues me is to use a small inductor on the wide range + pre-emphasis to boost the treble actively.
10-12" + wide-range option:
Single suspension Alpair 5 (or similar) 2 x 50€
10-12" pro-mid woofers 2 x 150-200€ (FaitalPRO, Lavoce, SB Audience, Eminence...)
Woofer Boxes:
Basic: 6-sided box, 18mm ply + 6-8mm dowels (these are easy to learn and fun. You can go for a cool riveted look, or shave them down.)
Ambition: angled cuts + a bit more maths for a stronger box.
TBH, I never really cared much for multiway baffles with drivers locked in position. Stacked boxes allow for way more tuning and experimentation.
Wave-guided wide range
I took the plunge carving shallow horns / wave guides. post #10 for visuals. A bit crazy I know, but so is 3d printing in its own way.
My XO design process:
I usually start with frequency sweeps (apart from casual listening), looking at the raw wave height in software like Audacity, AND listening carefully to assess whether I actually like the sound at a certain frequency. Previously, I had a situation where the "break even" point between a 10" subwoofer (in a non-ideal cube box) and a 3.5" wide-range was above 500 Hz -- much higher than I'd expected. The old w.r. drivers had really poor mid-bass and low mids in their ~1L boxes with resistive venting, but I didn't know that before I built them. So the XO design really has to come after some initial building and listening.
-Air-core inductor for the woofers. (For my low power requirements I was able to use high gauge enamel wire).
-Series padding resistor + parallel inductor for the W.R.
Further refinement:
Add parallel notch filters placed in series to EQ any problem areas. I plan to do them one by one, until diminishing returns kick in, so it's important to trust your ears to prioritize.
It would be a similar thing with DSP or active filters: if you only have a few biquads to play with, or knowing that every filter stage adds noise and distortion, you need to somehow choose what needs fixing first. I'm a passive XO convert, but only plan to use 1st order filters and notches passively.
I still plan to do multi-amping and active filters that focus on other aspects that passive filters are not so good at, like a Linkwitz transform to extend the bass. One idea that intrigues me is to use a small inductor on the wide range + pre-emphasis to boost the treble actively.
The only coax chassis designs I know without on axis chancellation from reflections and wide tweeter dispersion are from KEF or Genelec Ones. Both not good sources for DIY ...
I like this comment. It's what I feel like saying when someone makes the claim that beaming drills a hole in your head, which of course is not the same issue.
For noise reduction?
Passive filtering can be good at linkwitz transform given the box/driver is in the 'good' range of qtc:
Here is the kind of response you can have from passive eq ( still need a makeup gain cell to compensate for circuit's losses though(unity gain):
https://www.uaudio.com/webzine/2004/july/graphics/lowboostcut.jpg
With basic coaxials the listening position is usually designed to be 15-30 degrees off-axis and not on-axis. Given the smallness of the on-axis area this is not much of an issue in practise.
Personally I had no issues with a KEF tweeter+midrange coaxials in a recent(ish) 3 way DIY design. Opted for them because they were both cheaper and better performing than the equivalents from the companies directly serving the DIY market. Obviously pre-sales DIY support is absent but KEF provide well above average post-sales support for their hardware. At the 2 way budget end the coaxials on the DIY market look more competitive. Never tried to work with Genelec drivers but they appear well designed though more integrated with the chassis. Anyone have experience using them for DIY?
No experience with Genelec's coax but if someone want to play with something approaching the principle, second hand Cabasse's loudspeaker using their BC/TC range of drivers is an option. If you like wide dispersion they are very nice sounding ( experienced them in a Goelette 500).
Easier to find in continental EU though.
Bc ( coaxials):
https://www.forumcabasse.org/wiki/BC12
https://www.forumcabasse.org/wiki/BC13
Tc ( triaxials):
https://www.forumcabasse.org/wiki/TC21
https://www.forumcabasse.org/wiki/TC22
https://www.forumcabasse.org/wiki/TC23
I prefer controled directivity so gone Kef too.
Easier to find in continental EU though.
Bc ( coaxials):
https://www.forumcabasse.org/wiki/BC12
https://www.forumcabasse.org/wiki/BC13
Tc ( triaxials):
https://www.forumcabasse.org/wiki/TC21
https://www.forumcabasse.org/wiki/TC22
https://www.forumcabasse.org/wiki/TC23
I prefer controled directivity so gone Kef too.
Last edited:
That vaguely rings a bell... It might be something else, but I'm eyeing "constant directivity" as a possible culprit. Flat coverage with a frequency response that stays even inside a 60° x 90° rectangle seems like an admirable technical goal, but maybe it's not quite right for musical enjoyment?
To reduce voice coil distortion by raising the output impedance by adding a linear inductance, as I would with a woofer. But because I don't want to low-pass a tweeter, the active filter EQs it back with a high-pass shelf.
I think it is. The cleaner direct sound is relaxing, not because it's reduced.. you can fix that, but because it's clean and doesn't make you think about it. It isn't fatiguing.
There are those that say anechoic is boring.. It isn't like that because with control, you don't need to treat your room too much and you have late reverb, which is good. However if I were asked to pick between the two in an emergency, I'd prefer anechoic.
I'll just outright say that I don't have enough experience with horns to be definitive. One thing I suspect is that a CD horn may produce brighter echoes if there's a reflective wall behind the audience. It could be that if I reduce the peaks in my (approx.) exponential horns, they will be too dull.
Other factors could be at play as well, like the amplification or the amount of gain, as a 105dB/W horn with phase plug will have different compromises from a wave guide with a mere 95dB/W sensitivity at best.
Other factors could be at play as well, like the amplification or the amount of gain, as a 105dB/W horn with phase plug will have different compromises from a wave guide with a mere 95dB/W sensitivity at best.
I would think best practice is a sphere with smooth response on any given axis. This way any reflections going to the listening position will all have a similar response, and therefore a similar sound and timbre. Working backwards from that premise, if we identify the primary reflections (which CTA2034 does) then we can optimize for those critical angles. When you do that you will see that 1/4 wavelength ctc isn't the only solution.
When other criteria are met, getting the horizontal offaxis to be smooth and gently is sloping is pretty easy to do. But vertical relfections are harder - therefore the 1/4 conventional wisdom. But once you design for the primary reflections and see the other solutions available, you get can get pretty good vertical response too. Here is the ceiling reflections from my design using a wider ctc:
You won't do that with many other ctc ratios. Or fullranges for that matter. Coincident coxials have the potential, but they almost always bring other compromises, besides not being widely available to DIYers. I used the Kef Q100 driver with an optimized crossover versus some 2-way designs of my own. I really liked some of the things the Q100 offered - coherent, large midrange. But the treble was unrefined, and overall my own Scanspeak Discovery design sounded better, better-behaved, and more pleasing to listen to.