Hi all,
just a quick - and long overdue - post to briefly illustrate the latest incarnation of my own home system.
As the title of the thread suggests, this is essentially a "classic" two way "large format monitor" speaker, comprising a 15" Woofer and a large-format compression driver mounted to a horn.
A few key points:
- THE WOOFER is NOT a typical modern PA-style 15-incher (high Mms, ultra-high BL, very low Qes, lowish Qms, highish Fs), NOR is it a "vintage" component (Altec, etc.).
Instead, it is a Fostex FW405N, which is a RARE example of a modern, recently designed (albeit now discontinued) 15" woofer with some key "old school" characteristics (medium Mms, medium BL, medium Qes, high Qms, low Fs).
This, IMHO, stikes the "goldilocks" compromise between efficiency, low-level-detail retreival, "musicality" and linearity.
- THE COMPRESSION DRIVER is NOT a typical modern PA-style unit (rapid-flare exit, high Fs, often Titanium, or sometimes polymer diaphragm).
Instead, it is a 1990's JBL 2450 (2" exit version), retrofitted with a Truextent Beryllium diaphragm.
This, IMHO, strikes the "goldilocks" compromise between headroom (100mm diaphragm), low-level retreival and lack of resonances (Be), and ability to "load" down to mid-low frequencies (slow-flare exit)
- THE HORN is NOT a typical modern PA-style "constant directivity" waveguide (rapid-flare expansion, poor low-frequency loading, sometimes nasty diffraction slots in the older designs...), NOR is it a "vintage" component (e.g., Altec multi-cells or classic "sectoral" horns a-la 511, etc.).
Instead, it is a Yamamoto SS-300, which is a sophisticated evolution of a classically-derived hypex radial horn, but with a smooth profile devoid of any sharp corners, any discontinuities or diffraction-inducing steps along its expansion, and a smooth round-elliptical throat adaptor whose expansion parameters are perfectly matched to those of the JBL 2450 driver used. It is also carved out of a solid block of Japanese cherry wood, thereby killing virtually all internal resonances. Its "raw" frequency response is very smooth and extended from approx. 300Hz to over 15kHz (-6dB), with useful output to 20kHz. A mild "hump" centred around 2kHz is taken care of by means of a passive parallel-RLC trap.
- THE CROSSOVER is ACTIVE, but NOT DSP-based, NOR any standard off-the-shelf "pro" unit.
Instead, it is a bespoke Marchand unit, with asymmetrical low-pass and high-pass slopes individually tailored to the Woofer and Driver+Horn responses, and aiming to obtain the best possible phase-match between the two unit's outputs over a wide frequency range centred on the crossover frequency, which - incidentally - is a LOW 500Hz. For those who are interested, the topology is similar to that adopted by TAD in their glorious 2401, 2402 and 2404 monitors (i.e., 6th order low pass and 2nd order high-pass electrical slopes).
The whole system is driven by a mid-2000s Esoteric multi-bit DAC/Preamp > Marchand active crossover > Yamaha class AB amp (woofers) / Accuphase class A amp (drivers).
A few pictures to illustrate the finished product below.
Cheers!
Marco
just a quick - and long overdue - post to briefly illustrate the latest incarnation of my own home system.
As the title of the thread suggests, this is essentially a "classic" two way "large format monitor" speaker, comprising a 15" Woofer and a large-format compression driver mounted to a horn.
A few key points:
- THE WOOFER is NOT a typical modern PA-style 15-incher (high Mms, ultra-high BL, very low Qes, lowish Qms, highish Fs), NOR is it a "vintage" component (Altec, etc.).
Instead, it is a Fostex FW405N, which is a RARE example of a modern, recently designed (albeit now discontinued) 15" woofer with some key "old school" characteristics (medium Mms, medium BL, medium Qes, high Qms, low Fs).
This, IMHO, stikes the "goldilocks" compromise between efficiency, low-level-detail retreival, "musicality" and linearity.
- THE COMPRESSION DRIVER is NOT a typical modern PA-style unit (rapid-flare exit, high Fs, often Titanium, or sometimes polymer diaphragm).
Instead, it is a 1990's JBL 2450 (2" exit version), retrofitted with a Truextent Beryllium diaphragm.
This, IMHO, strikes the "goldilocks" compromise between headroom (100mm diaphragm), low-level retreival and lack of resonances (Be), and ability to "load" down to mid-low frequencies (slow-flare exit)
- THE HORN is NOT a typical modern PA-style "constant directivity" waveguide (rapid-flare expansion, poor low-frequency loading, sometimes nasty diffraction slots in the older designs...), NOR is it a "vintage" component (e.g., Altec multi-cells or classic "sectoral" horns a-la 511, etc.).
Instead, it is a Yamamoto SS-300, which is a sophisticated evolution of a classically-derived hypex radial horn, but with a smooth profile devoid of any sharp corners, any discontinuities or diffraction-inducing steps along its expansion, and a smooth round-elliptical throat adaptor whose expansion parameters are perfectly matched to those of the JBL 2450 driver used. It is also carved out of a solid block of Japanese cherry wood, thereby killing virtually all internal resonances. Its "raw" frequency response is very smooth and extended from approx. 300Hz to over 15kHz (-6dB), with useful output to 20kHz. A mild "hump" centred around 2kHz is taken care of by means of a passive parallel-RLC trap.
- THE CROSSOVER is ACTIVE, but NOT DSP-based, NOR any standard off-the-shelf "pro" unit.
Instead, it is a bespoke Marchand unit, with asymmetrical low-pass and high-pass slopes individually tailored to the Woofer and Driver+Horn responses, and aiming to obtain the best possible phase-match between the two unit's outputs over a wide frequency range centred on the crossover frequency, which - incidentally - is a LOW 500Hz. For those who are interested, the topology is similar to that adopted by TAD in their glorious 2401, 2402 and 2404 monitors (i.e., 6th order low pass and 2nd order high-pass electrical slopes).
The whole system is driven by a mid-2000s Esoteric multi-bit DAC/Preamp > Marchand active crossover > Yamaha class AB amp (woofers) / Accuphase class A amp (drivers).
A few pictures to illustrate the finished product below.
Cheers!
Marco
It works exactly as intended. I.e., the asymmetrical crossover slopes work in tandem with the unfiltered "raw" driver responses AND their relative physical offset to arrive at the desired phase match over a frequency range centred on the crossover frequency.
See phase measurements below:
I agree. I find them gorgeous, too, and I'd wanted a pair for so long.
Alas, they're also almost 4 grand! Each.
Ouch.
The only things that I purchased second hand directly from Japan were the Yamamoto horns (took me years to find a pair available for sale), and the Esoteric DAC.
Hello Marco
Is this (text below) what you finally chose for the system you describe here? See: https://www.diyaudio.com/community/...e-delay-offset-for-horns-particularly.345388/
Hi!
I was the originator of that thread.
The reason I later "lost interest" was the realisation that none of those "quasi-optimal" crossovers had a chance of being correctly implementable in real life, when also duly taking into account the fact that all real exponential or hypex horns already exhibit a "natural" 4th-to-6th order high-pass behaviour at Fc. Therefore, all those nice asymmetrical crossovers that call for a 2nd order high-pass on the tweeter leg are a physical impossibility.
Additionally, what I have come to realise is that what really matters most of all is not minimizing phase rotation of the summed response per se, but rather achieving the best possible phase matching between the two filtered drivers, in a wide frequency range around the acoustical crossover frequency, and up/down to where each driver's response is -12dB.
All of this has led me to re-evaluate the goodness of Pioneer's original passive crossover designs for their Exclusive studio monitors of the '80s and '90s, namely:
This combination seems to work very well indeed for the typical: 15" woofer + ~300Fc Hypex horn, ~600-700Hz crossover, and ~20-25cm offset.
But, the final, real acoustic high-pass transfer function of the tweeter is a combination of the intrinsic acoustic high-pass of the unfiltered tweeter and 2nd order electrical high-pass, and it looks nothing like a textbook high-pass filter (and is certainly NOT 2nd order!).
Cheers,
Marco
Is this (text below) what you finally chose for the system you describe here? See: https://www.diyaudio.com/community/...e-delay-offset-for-horns-particularly.345388/
Hi!
I was the originator of that thread.
The reason I later "lost interest" was the realisation that none of those "quasi-optimal" crossovers had a chance of being correctly implementable in real life, when also duly taking into account the fact that all real exponential or hypex horns already exhibit a "natural" 4th-to-6th order high-pass behaviour at Fc. Therefore, all those nice asymmetrical crossovers that call for a 2nd order high-pass on the tweeter leg are a physical impossibility.
Additionally, what I have come to realise is that what really matters most of all is not minimizing phase rotation of the summed response per se, but rather achieving the best possible phase matching between the two filtered drivers, in a wide frequency range around the acoustical crossover frequency, and up/down to where each driver's response is -12dB.
All of this has led me to re-evaluate the goodness of Pioneer's original passive crossover designs for their Exclusive studio monitors of the '80s and '90s, namely:
- a 6th order electrical LP on the woofer (essentially Linkwitz-Riley)
- a 2nd order electrical HP on the tweeter/horn (quasi-Butterworth)
This combination seems to work very well indeed for the typical: 15" woofer + ~300Fc Hypex horn, ~600-700Hz crossover, and ~20-25cm offset.
But, the final, real acoustic high-pass transfer function of the tweeter is a combination of the intrinsic acoustic high-pass of the unfiltered tweeter and 2nd order electrical high-pass, and it looks nothing like a textbook high-pass filter (and is certainly NOT 2nd order!).
Cheers,
Marco
Yes, exactly.
However, instead of using a passive crossover, I did it using a bespoke active analogue crossover, with individual HP and LP filter slopes and "shapes" of my own design.
Both.
Simulation (using individual gated measurements for the two unfiltered drivers) shows "cleaner" phase traces down to lower frequencies, while filtered measurements from the listening spot are limited by in-room reflections. But both are in agreement where it matters, i.e., around the crossover frequency.
Thank you!
I admit I like it very much - sonically, first and foremost. The good looks don't hurt either ;-)
Sorry - deleted my reply post because I messed up the quick simulations. Will repost the correct results asap.
I simulated what happens, and it does NOT look pretty.
Essentially, you can still get a phase match, but ONLY exactly at the crossover frequency. The two phase traces then DIVERGE wildly as soon as you move even slightly away from that frequency. (This causes ripples in the Frequency response, too, by the way.)
The Step Response is also negatively affected BIG TIME.
1) original crossover (+/+ polarity)
2) opposite polarity crossover
Overall, I'd say it's a MUCH WORSE result, and not worth pursuing for just the sake of being able to push the horn back.