On-axis vs. Power Response

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I note that many designers strive for maximally flat on-axis response. I am assuming that to achieve this the frequency response curve for the total amount of acoustic power is often made worse.

Specifically baffle-step compensation has me worried. Makes sense if we listen in an anechoic chamber, in a living room I would think that boost would cause a bump in the power response, will we hear this emphasis?

So, when I listen to my speakers what do I hear? The on-axis response mostly affects what aspects of the music? What aspects are hugely affected by the power response?

Spacing the loudspeaker a couple of feet from the back wall will measure huge peaks and dips in the response due to reflective cancellation and reinforcement. How does this affect what we hear?

Allison addressed these problems by placing the speakers against the wall, his speakers sounded pretty good to me, but this practice certainly does not garner much favor.

Has anyone had excellent results with speaker designs similar to the Allison?
 
There is no definitive answer to too many of your questions. Reproducing recorded music is a complicated endeavor. Psychoacoustics makes thing untidy. Many theories have been put forward to correct problems we do not fully understand.

Let us look at placing loudspeakers close to a rear wall. If the loudspeaker transducer is flush mounted with the rear wall, then most wall interactions are minimized. Indeed, this is one way that some manufacturers test their transducers.

If the transducer is spaced away from the rear wall, then the interactions will not be minimized. Here comes the psychoacoustics. The human brain does not weight all sounds at the same value. We hear early reflections differently than we hear late reflections. We also key on higher frequencies more than on lower frequencies.

If you have a loudspeaker that has depth to the cabinet, then you can only place the transducer so close to the rear wall. When the cabinet is spaced several feet from a rear wall the interaction frequencies are lower where our brains don't care about it as much. Move the cabinet closer to the wall and while the magnitude may decrease, the frequencies of the interactions increase and our brains care about them more.

Also, while the Allison's may have done well at low frequencies, they (like all loudspeaker transducers) innately altered the sound at higher frequencies. These higher frequency inaccuracies were inherent in the designs of the transducers and had nothing to do with placement. Some people liked the sound and some did not. To a large extent, whether you liked the sound or not had little to do with how they were designed to interact with low frequency room interactions.

Mark
 
One situation that is definitely not preferable is when the on-axis response is flat and the off-axis response is looking like a roller-coaster ride.
This happens often with multiway systems that use drivers up to a too high frequency (where they beam) before the response is handed over to the next one with broad radiation.

Regards

Charles
 
I'd be interested to know something from those of you with more horn experience than I...

With a constant directivity horn, the response within the horns designed radiating angle is pretty damn even, but what happens when you move outside of the horns designed radiating angle? Does the response just drop off very sharply but stay quite even, or does the response become very uneven all of a sudden?

I'm just wondering because at least with a direct radiator, while the response drops at high frequencies even a little bit off-axis, the roll-off is smooth and predictable.
 
Hi

You hear at least two things depending where you are.
On axis, you hear direct sound, which has the spectrum shape you measure on axis in an anechoic environment.
Added to that, you have all the reflected sound, the “reverberant field”.
The spectrum of the reverberant field closely resembles the total acoustic power vs frequency that the speaker radiates.
On axis response is a function of total acoustic power and the directionality* at that frequency. *ability to concentrate the radiate the energy in one preferred direction over an omni directional source.

Floyd Toole’s testing showed people prefer that the reverberant field have roughly the same spectrum as the on axis response.

An extreme example where this is not the case, take an A-7 Voice of the theater.
On axis, they can be pretty flat but the horn produces a pronounced narrowing of its radiation pattern as the frequency climbs (a property of nearly all curved wall horns).
As a result, the power response has “no” high frequency energy.
On axis, they can sound fine but anywhere outside the pattern and one has a very dark murky sounding speaker.

In a perfect world one has a single speaker with more or less flat response and constant directivity which means the radiated power spectrum is the same as the on axis response.

So far as room effects, anything you do that alters the signal, potentially reduces your ability to extract information from the recording. Reflections, especially close ones are VERY destructive so far as the stereo image is concerned.
This is why Omni speakers, while constant directivity, make everything sound the same, the structure of room reflections is added to everything.
The more one can reduce these reflections (usually through directivity) compared to the direct sound, the more the system sounds like the individual recordings and not the room.
Room reflections in some cases can be “nice” but that should not be confused with transferring the most information from the recording to your ears.
Best,

Tom Danley
 
  • Like
Reactions: 1 user
You can find supporters of both omnidirectional loudspeakers and highly directional loudspeakers. As I stated earlier, involving the human brain makes things very untidy.

By setting as my standard of reproduction the accurate transformation of an electical impluse into an acoustical impulse, I favor one over the other. I favor a system that maximizes the reproduction of the sound engineered into the recording. Even with this as my reference, however, room effects (and power response) are still important. You do not want directional lobing or large peaks in the off-axis or power response of a system. You do not want large peaks in the on-axis and or anachoic response of a system.

To complicate matters even more, there are numerous examples of omnidirectional, directional, and everything in between that produce large peaks both on and off-axis. There is a path that needs to be traveled to maximize performance of any system. While I favor one theory over the other, few loudspeakers come close enough to maximum performance to test or to prove which theory may produce the most accurate sound reproduction.

Some people, however, will wish to simplify the discussion to argue for just one theory regardless of the details. It might be best to avoid loudspeaker designs from those authors.

Instead, I would look for loudspeakers are really good in one aspect and not bad in the other. In other words, always avoid the bad.

Lastly, always be carefull about descriptions of sound. While the A 7s may have sounded muddy off-axis, simply being directional will not produce muddy sound. The loss of high frequencies will make the sound dull. That alone will not produce muddy sound. I suspect something else was also going on.

Mark
 
Hi

“By setting as my standard of reproduction the accurate transformation of an electical impluse into an acoustical impulse, I favor one over the other. I favor a system that maximizes the reproduction of the sound engineered into the recording. Even with this as my reference, however, room effects (and power response) are still important. You do not want directional lobing or large peaks in the off-axis or power response of a system. You do not want large peaks in the on-axis and or anachoic response of a system.”

Yes, yes yes.
In a larger room, larger audience all the problems are more severe making all these issues even more critical if one is to understand words , voices music etc.
Here is my solution for those problems used in larger scale reproduction, no lobes, constant directivity, impulse response of one source, reproduces a square wave over a decade in frequency (220Hz – 2600Hz).

http://www.danleysoundlabs.com/synergy_horn.asp?model=SH 50

http://www.danleysoundlabs.com/pdf/danley_tapped.pdf

http://www.proaudioreview.com/pages/s.0037/t.8805.html

creeping into homes

http://www.zerogain.com/forum/showthread.php?t=13956&page=33

http://www.audioasylum.com/forums/hug/messages/12/128285.html

“Lastly, always be carefull about descriptions of sound. While the A 7s may have sounded muddy off-axis, simply being directional will not produce muddy sound. The loss of high frequencies will make the sound dull. That alone will not produce muddy sound. I suspect something else was also going on.”

Yes, the A-7 has a very tilted radiated power spectrum while the on axis response is flat.
If you are off axis, you hear the falling acoustic power as the frequency increases.
In the A-7, exponential, tractrix and other traditional curved horns, this narrowing was set to off set the falling power response from the driver. On a Cd horn, that same driver needs “CD compensation” to make up for that falling power spread over a constant angle.
Best,
Tom Danley
 
  • Like
Reactions: 1 user
The on axis response (usually near Tweeter height) is a single discrete frequency response, a singular discrete measurement. Once we go off axis we are measuring somewhat more than that. As most speakers are symmetrical laterally (drivers are vertically aligned), then a single off axis measurement is in reality two measurements. In terms of import the off axis should weighed as double that of the singular on axis. For this reason it is not unreasonable to say the the on axis measurement is the frequency response and the off axis response represents the power response. I would think 25-30 degrees as reasonable.

A consistent power response is undeniably a major consideration.

It can also be argued that a speaker has many frequency responses. It changes every time you change the microphone.

Joe R.
 
Part Two: Making a Good Loudspeaker – Imaging, Space and Great Sound in Rooms
By Dr. Floyd E. Toole, Vice President Acoustical Engineering,
Harman International Industries, Inc.
This section begins with an explanation of the "circle of confusion" that pervades the audio industry, showing how important loudspeakers are in the preservation of audio artistry. However, loudspeakers are technical devices, and engineers need the appropriate science to design them properly. The situation is complicated by the fact that we listen in diverse rooms, all of which influence what we hear from the loudspeakers. There is a discussion of some of the important measured parameters of loudspeakers, and how it is possible to design loudspeakers in ways that makes them inherently more "room friendly" – having the potential of sounding good in a variety of different room settings.



ALL of the most preferred loudspeakers are ones that exhibit the
flattest, smoothest families of curves.
They exhibit the fewest, and the lowest level, resonances. They have
the flattest, smoothest, widest bandwidth frequency responses when
measured from all angles.
They have similar shapes in all of the curves – i.e. they have quite
constant, or at least smoothly changing, directivity as a function of
frequency.
Can we measure what we can hear? No, but we sure have made a
good start.
Loudspeakersandroomspt2.pdf





http://www.harman.com/about_harman/technology_leadership.aspx


10.24.2004
Audio, Science in the Service of Art: Read >>


10.24.2004
Direction and Space – the Final Frontiers: Read >>


09.07.2003
A New Laboratory for Evaluating Multichannel Audio Components and Systems: Read >>


10.30.2002
Subwoofers: Optimum Number and Locations: Read >>


01.31.2002
Part One: How Many Loudspeakers? What Kind? Where?: Read >>


01.31.2002
Part Two: Making a Good Loudspeaker – Imaging, Space and Great Sound in Rooms: Read >>


01.31.2002
Part Three: Getting the Bass Right: Read >>
 
I fall into the "reverberant field matters" camp, and would agree with the comments made here by Tom Danley and mike.e. I consider the on-axis response to be relatively unimportant, unless you listen exactly on-axis in an anechoic chamber or outdoors.

To address Tenson's question, in my experience a constant-directivity horn or waveguide is much more likely to have smooth off-axis performance than is a direct radiator driver. In my opinion one way to make good use of this is to match up the radiation pattern of woofer and horn at the crossover frequency. This often calls for good upper midrange performance from a fairly large-diameter woofer.

In my opinion, two main reasons the reverberant field matters are these: 1) The reverberant field is a significant contributor to perceived tonal balance, spaciousness or ambience, richness, and texture; and 2) when the reverberant energy has a significantly different spectral balance from the first-arrival sound, the ear/brain system's localization mechanism literally has to work harder to correctly classify the reberberant energy as reflections instead of new sounds. This increased "cpu usage" can result in listening fatigue - literally, a headache.

There are many different ways to "get the reverberant field right", some of which are more room-friendly than others (like speakers with well-controlled forward-facing radation patterns), and some of which can give a richer presentation in a large room (omnis, dipoles, bipoles, and other polydirectional types when designed for good power response).

Getting back to bunzinthesun's original question about the baffle step, in my opinion baffle step compensation is appropriate in some specific applications (nearfield listening, lack of boundary reinforcement). However, I think that in most applications you're better off without it.

Duke.
 
audiokinesis said:

Getting back to bunzinthesun's original question about the baffle step, in my opinion baffle step compensation is appropriate in some specific applications (nearfield listening, lack of boundary reinforcement). However, I think that in most applications you're better off without it.

Duke.

I can't quite agree with that. Boundary reinforcement only starts somewhere below 200Hz as the system gradually becomes spherical in radiation pattern. As the diffraction loss has usually reached a max by that frequency, you will end up with a large lack of energy near 300Hz which is arguably the power centre of most genres, especially orchestral. I can only agree if you had a wall flush mounted speaker system, no step then needs to be compensated for.

But I agree that the degree of compensation has a huge effect on the final result and you can easily overcook it. But I cannot stomach an anemic balance. I'm sure that I have a majority agreement on that.

Having said that, I'm no great fan of narrow baffles - and I like the sound of open baffles, which by nature are not narrow.

Joe R.
 
Well thought-out point, Joe Rasumssen. You are absolutely correct that in many if not most cases, the baffle step occurs well above the region where boundary reinforcement kicks in.

I painted with too broad a brush there, and from the wrong angle. My excuse is that I don't really deal much with the type of speaker likely to have a baffle step issue (narrow speakers whose woofers cross over well above the baffle step frequency). But such speakers are probably much more common than the types I normally play with.

If we're talking about a stand-mount speaker, we can raise the frequency where boundary reinforcement starts to help out by placing the speaker on a short stand so the woofer is closer to the floor. But if it's a floorstander, there's not much we can do to get the woofer closer to the floor.

Of course if we're talking about a DIY project, the builder can either make the speaker wide enough to push the baffle step down close to the boundary reinforcement region, or cross over the woofer at the upper end of the baffle step region. With the latter approach, the baffle step is accounted for when the levels are set in the crossover, so no additional dedicated baffle-step circuitry is required. In many cases this leads to a three-way system, which imho is often a good idea - but that's another topic.

I don't like anemic speakers either, and try to minimize or sidestep the baffle step issue altogether in my designs.

Duke
 
Very well put by Joe Rasmussen and Audiokinesis! Thanks.

Those points are perfectly valid for all passive speakers. I am wondering with active dipole speakers the situation could be a bit different. The dipole peak with a proper baffle width may very well compensate the the area between boundary reinforcement and the baffle step rolloff. Just a thought.

Regards,
Bill
 
"good" power response

Hi

Just some thoughts and questions to those of you with more experience.

Clearly, a consistent power response is desirable, but when it comes to exactly what this means, things aren't very clear - especially for lower directivity systems (monopoles, dipoles and their combinations).

A great source of information and theory is here:
http://www.musicanddesign.com/PowerMatching.html

Mostly, a flat power response is suggested as the design goal. However, a real room will have an absorbtion characteristic that is frequency dependent and will influence the developed sound field. Consequently, it appears to me that one should achieve a power response that is roughly complementary to the room absorbtion characteristic. It looks like most rooms have higher absorbtion on mids and highs and lower on bass, which would mean that the power response should actually decrease towards the bottom end (of course, the on axis response still needs to be reasonably flat !)

Am I wrong ? Is there any method or rule of thumb to evaluate a room's reverberant characteristic vs. frequency ?
 
Member
Joined 2008
Paid Member
This topic caught my interest. I got a UMIK-1 recently and did some REW measurements. I compared Sonido SFR200A in open baffle and Lowther DX-2 in a Hedlund horn. There is a 15" OB bass, active crossover at 160 - 350 Hz.

Lowther wins with the sound and by a large margin. I wondered why. First, I did some nearfield measurements. The Lowther has a bump in the high frequencies, which looks ugly. The Sonido is nice and flat and maybe even extends slightly higher. I just could not get over the bad looking FR of the Lowther.

Then, listening position measurement comes - the Lowthers are suddenly nice and flat (on axis, but far field!) and the Sonidos are rolled off with a constant slope. And that suddenly made sense to me. The room influence evened out the peak and unfortunately rolled off the nice and smooth FR of the Sonidos.

So for a full range driver, the bump can actually be a good thing. And the Sonido OBs call for a rear firing ambience tweeter.
 
To address Tenson's question, in my experience a constant-directivity horn or waveguide is much more likely to have smooth off-axis performance than is a direct radiator driver. In my opinion one way to make good use of this is to match up the radiation pattern of woofer and horn at the crossover frequency.

Hello Duke

I agree but you can also do it with direct radiators if you cross them over low enough and use a waveguide on the tweeter. Revel does this as does JBL case in point the LSR32 way back in 1997

1997 LSR32
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.