siddharthdas
Active Member
Box speaker suffer from a problem that's really difficult to fix without clever analog/mechanical engineering and/or DSP - that's dispersion. Since we use mechanical drivers to reproduce sound, we are limited by their physics. Bass frequencies radiate in a sphere - i.e. we will hear bass at almost the same loudness at the same distance from from the driver anywhere in the room (let's ignore floor bounce/ baffle step etc for the moment). Midrange and highs radiate in a hemisphere in the front of the driver (until such point as the wavelength of sound is close to driver diameter). When driver diameter is comparable to or larger than the wavelength of the sound, then sound beams in a narrow cone. Now the problem here is energy density, also known as power response. For bass, you are radiating double the energy into your room for the same on-axis SPL, as compared to higher frequencies. Simply put power radiation throughout the sonic spectrum varies with frequency-range and driver diameter. That makes the sound emanating from most speakers inherently unnatural.
Now, it is unclear if the "power response vs. perception" studies give us a definitive answer for which parts of the frequency range benefit most from constant directivity (i.e. uniform dispersion). Technically however, it is possible to do that somewhat easily for mids and highs (using waveguides for tweeters etc). For bass it takes a lot of serious engineering - Kii audio (http://www.kiiaudio.com) uses DSP, and Dutch & Dutch (https://dutchdutch.com) used some ingenious mechanical engineering. I think both provide an option for linear phase FIR filtering. Some insane horn speakers (https://www.magico.net/product/ultimate.php) probably do that as well. All dipole speakers do controlled dispersion by design (https://steinwaylyngdorf.com/freestanding/). And finally omnipolar speakers also do that by design ((http://www.mbl.de/?lang=en) albeit with hugely increased room interaction. With each kind of design there are sonic and engineering trade-offs.
That said, the thing therefore to look for is whether a speaker has controlled directivity to begin with. If not, they may still sound nice, but those speakers are starting off on the back-foot
- no matter whether they cost $100, or $1000,000.
Will be delighted to hear everyone's experiences with constant directivity speakers, and side-by-side controlled (blind, ideally) comparisons with other types of speakers.
Now, it is unclear if the "power response vs. perception" studies give us a definitive answer for which parts of the frequency range benefit most from constant directivity (i.e. uniform dispersion). Technically however, it is possible to do that somewhat easily for mids and highs (using waveguides for tweeters etc). For bass it takes a lot of serious engineering - Kii audio (http://www.kiiaudio.com) uses DSP, and Dutch & Dutch (https://dutchdutch.com) used some ingenious mechanical engineering. I think both provide an option for linear phase FIR filtering. Some insane horn speakers (https://www.magico.net/product/ultimate.php) probably do that as well. All dipole speakers do controlled dispersion by design (https://steinwaylyngdorf.com/freestanding/). And finally omnipolar speakers also do that by design ((http://www.mbl.de/?lang=en) albeit with hugely increased room interaction. With each kind of design there are sonic and engineering trade-offs.
That said, the thing therefore to look for is whether a speaker has controlled directivity to begin with. If not, they may still sound nice, but those speakers are starting off on the back-foot
- no matter whether they cost $100, or $1000,000. Will be delighted to hear everyone's experiences with constant directivity speakers, and side-by-side controlled (blind, ideally) comparisons with other types of speakers.
) those designs, here is a little bit of context on the my systems' building blocks, from the speaker point of view. A 2 x Satori WO24P-8 driver cabinet with sealed bass alignment lies at the heart of the following designs. The range that these drivers handle vary from around 200Hz to 1200Hz, depending upon the configuration. All systems are fully active DSP crossover implemented with 64000 tap FIR filters conditioning the signal that goes to each driver. For now I use a 6 channel ESI audio interface form D/A conversion and the analog signal is sent to separate amps driving each driver. All the polar responses that I will show are measured using a semi-dual channel measurement system with a "turn-table" allowing me to capture the angular frequency responses. I use Baffle-step corrected nearfield + gated far-field measurements to estimate the full range polar response.
