@andy19191 "dirty data" (free field)
Let's assume placing a microphone at 2m distance on-axis (red dot) of this left speaker.
Then we go to the right in 10cm steps (max 2m to the right), taking a frequency
response at each step.
In this way, we do not increase the angle by constant steps (the increment in angles
getting smaller of course) and we also move farther away from the speaker.
But with the curves falling in level now - even at low frequencies - we get some
visual separation of the curves, which i think makes the picture more instructive.
Horizontal steps (10cm):
Now we move the microphone vertically (max 2m upwards) from the on-axis point.
Vertical steps (10cm):
I think this diagrams point to how the averaged sound power curves (shown above) of such an arrangement
tend to be very smooth (sound power falling with frequency, DI rising with frequency), especially when
taking more points in space into account.
Let's assume placing a microphone at 2m distance on-axis (red dot) of this left speaker.
Then we go to the right in 10cm steps (max 2m to the right), taking a frequency
response at each step.
In this way, we do not increase the angle by constant steps (the increment in angles
getting smaller of course) and we also move farther away from the speaker.
But with the curves falling in level now - even at low frequencies - we get some
visual separation of the curves, which i think makes the picture more instructive.
Horizontal steps (10cm):
Now we move the microphone vertically (max 2m upwards) from the on-axis point.
Vertical steps (10cm):
I think this diagrams point to how the averaged sound power curves (shown above) of such an arrangement
tend to be very smooth (sound power falling with frequency, DI rising with frequency), especially when
taking more points in space into account.
Last edited:
The frequency response varies smoothly and deeply (i.e. audible) and since it also varies significantly in space will not be correctable with EQ. It would need to be designed out by, for example, configuring the drivers more along the lines of existing multiple driver speakers that work reasonably(ish). However, what is missing, if we set aside imaging quality, is to what extent reflections may restore some perceived tonal balance and, if implemented reasonably well, enhance spaciousness which tends to be the advantage of such speakers to offset their disadvantages. Do you have simulation software to hand to provide this relevant but missing information?
@andy19191
Some thoughts on your concerns:
The on axis response is smooth and "dead flat". There is no need for EQ (except for the baffle step, which will
be of concern for the woofer mostly, depending on crossover frequency chosen).
When listening (especially to stereo) i will stay in the preferred listening position/seat, with the speakers "toed in".
I will (usually) not be "walking around" and even if doing so, the power response of (both) speakers will be the
prominent attribute, because it is hard (impossible) for a human to perceptively separate e.g. direct sound from
the contribution of early reflections especially when leaving the stereo listening zone (which is inherently small
due to stereo itself not due to the single speaker and his directivity).
To attribute this speaker concept "poor imageing" (phantom source localization in stereo?) ist a pure speculation.
The stereo inherent phenomenon of "mid phantom disappearence" at frequencies >3Khz (*) is not made worse
with this concept, on the contrary it is more likely to be mitigated (and of course these mid- and high frequency
arrangements are particularly good at making up "believable" mid phantom images).
Also combing due to (low order and close) mirror sources in the room is mitigated due do direct sound decorrelating
progressively towards larger off-axis angles and higher frequencies starting from the lower midrange.
A loudspeaker which is "totally coherent" in radiation places a problem in usual living rooms of home listeners,
that usually are "reserved" when it comes to extensive acoustical room treatment with diffusers and absorbers.
Most HiFi enthusiasts think, that such "usual" untreated rooms are "the problem":
They are not IMO. The "common way" of making loudspeakers is the biggest problem of course, when seen in
conjunction with the room (i.e. loudspeaker/room interaction).
___________________________
(* see referenced paper from Bennet et al. )
Some thoughts on your concerns:
The on axis response is smooth and "dead flat". There is no need for EQ (except for the baffle step, which will
be of concern for the woofer mostly, depending on crossover frequency chosen).
When listening (especially to stereo) i will stay in the preferred listening position/seat, with the speakers "toed in".
I will (usually) not be "walking around" and even if doing so, the power response of (both) speakers will be the
prominent attribute, because it is hard (impossible) for a human to perceptively separate e.g. direct sound from
the contribution of early reflections especially when leaving the stereo listening zone (which is inherently small
due to stereo itself not due to the single speaker and his directivity).
To attribute this speaker concept "poor imageing" (phantom source localization in stereo?) ist a pure speculation.
The stereo inherent phenomenon of "mid phantom disappearence" at frequencies >3Khz (*) is not made worse
with this concept, on the contrary it is more likely to be mitigated (and of course these mid- and high frequency
arrangements are particularly good at making up "believable" mid phantom images).
Also combing due to (low order and close) mirror sources in the room is mitigated due do direct sound decorrelating
progressively towards larger off-axis angles and higher frequencies starting from the lower midrange.
A loudspeaker which is "totally coherent" in radiation places a problem in usual living rooms of home listeners,
that usually are "reserved" when it comes to extensive acoustical room treatment with diffusers and absorbers.
Most HiFi enthusiasts think, that such "usual" untreated rooms are "the problem":
They are not IMO. The "common way" of making loudspeakers is the biggest problem of course, when seen in
conjunction with the room (i.e. loudspeaker/room interaction).
___________________________
(* see referenced paper from Bennet et al. )
Supplement to the paper mentioned above:
"A New Approach to the Assessment of Stereophonic Sound System Performance"
Bennet, Barker, Edeko (1985)
Figure 5 (Page 318):
My take on this
The belief of "strictly coherent" radiating tweeters (or even any kinds of tweeters) contributing
to "pin point imageing" - which often is said/believed to be an attribute of high quality stereo
reproduction - is IMO just a myth:
The geometry of a stereo setup is very limited in virtually mimicking sound coming from the
front (0 degree angle) of the listener at high frequencies >3Khz.
The best what a loudspeaker can do in that frequency range is IMO deliver smooth and flat
on-axis direct sound and an "unobtrusive" sound power curve, together making up an inroom
response that is close to the Bruel&Kjaer preferred inroom response (1974).
"A New Approach to the Assessment of Stereophonic Sound System Performance"
Bennet, Barker, Edeko (1985)
Figure 5 (Page 318):
My take on this
The belief of "strictly coherent" radiating tweeters (or even any kinds of tweeters) contributing
to "pin point imageing" - which often is said/believed to be an attribute of high quality stereo
reproduction - is IMO just a myth:
The geometry of a stereo setup is very limited in virtually mimicking sound coming from the
front (0 degree angle) of the listener at high frequencies >3Khz.
The best what a loudspeaker can do in that frequency range is IMO deliver smooth and flat
on-axis direct sound and an "unobtrusive" sound power curve, together making up an inroom
response that is close to the Bruel&Kjaer preferred inroom response (1974).
Last edited:
If I may add my two cents: comb-filtering is worst along the line through/past two drivers (e.g. MTM), but here it is mitigated by the third (non-collinear) driver. They are of course symmetrically equi-distant to the (tweeter) on-axis listening seat, and mid-to-far-field not too much angular-distance apart with respect to the on-axis ear, provided the XO LPF isn't set too high.
My reflector-point-source experiment dealt with up to three tweeters near-field, half-space/hemispherically omni-directional to very high frequency.
(Re: stereo imaging, please see my separate thread via the above post.)
My reflector-point-source experiment dealt with up to three tweeters near-field, half-space/hemispherically omni-directional to very high frequency.
(Re: stereo imaging, please see my separate thread via the above post.)
Last edited:
@wchang (you wrote)
"If I may add my two cents: comb-filtering is worst along the line through/past two drivers (e.g. MTM), but here it is mitigated by the third (non-collinear) driver. They are of course symmetrically equi-distant to the (tweeter) on-axis listening seat, and mid-to-far-field not too much angular-distance apart with respect to the on-axis ear, provided the XO LPF isn't set too high."
I can follow, what you say.
Crossover frequency in the loudspeaker proposed has to be low enough, that is true. Displacement
vectors between drivers are occuring only once (uniquely) in the above array of 3 small wideband units.
The distances between elements are designed to get the desired "soft rollof" in power response that
is needed to approach an acknowledged/preferred inroom response.
To let the array do what it was designed for, XO has to be low enough, which is also beneficial for the
integration with the woofer.
"If I may add my two cents: comb-filtering is worst along the line through/past two drivers (e.g. MTM), but here it is mitigated by the third (non-collinear) driver. They are of course symmetrically equi-distant to the (tweeter) on-axis listening seat, and mid-to-far-field not too much angular-distance apart with respect to the on-axis ear, provided the XO LPF isn't set too high."
I can follow, what you say.
Crossover frequency in the loudspeaker proposed has to be low enough, that is true. Displacement
vectors between drivers are occuring only once (uniquely) in the above array of 3 small wideband units.
The distances between elements are designed to get the desired "soft rollof" in power response that
is needed to approach an acknowledged/preferred inroom response.
To let the array do what it was designed for, XO has to be low enough, which is also beneficial for the
integration with the woofer.
Last edited:
I missed the part about the founder of Larsen speakers saddly passing away. Sorry for that and hope I have not upset anyone.
What may be the reason behind these curves found so far ...
Bruel&Kjaer:
Toole/Olive:
Bruel&Kjaer:
Toole/Olive:
Last edited:
Maybe it would be interesting in this context to talk about the receiver too, instead of discussing
the sender's characteristics only?
(And of course, there is a "channel" in between.)
https://en.wikipedia.org/wiki/Hermann_von_Helmholtz
https://en.wikipedia.org/wiki/Helmholtz_reciprocity
the sender's characteristics only?
(And of course, there is a "channel" in between.)
https://en.wikipedia.org/wiki/Hermann_von_Helmholtz
https://en.wikipedia.org/wiki/Helmholtz_reciprocity
I think the general consensus is that average listening rooms absorb more at higher frequencies and most speakers have increasing directivity with increasing frequency. The Toole paper also points out that steady-state in-room measurements at low frequencies show elevated bass levels. All of this contributes to the preferred in-room frequency response curves shown, which exhibit a gentle downward trend with increasing frequency.
From Toole's paper:
"Whatever the shape of the spectrum of the direct sound
from musical instruments, or loudspeakers reproducing
recordings of those musical instruments, the steady-state
sound field in a normal room will exhibit a version of that
spectrum that rises at lower frequencies."
. . .
"Sound reproduction spaces are much less reflective:
. . . . Loudspeakers designed with flat on-axis
frequency responses, so as to accurately reproduce the
initial timbral signature of the recorded sounds, will therefore
exhibit a rising sound power output at lower frequencies.
The only exceptions would be arrays designed to maintain
high DI at low frequencies."