Assuming equal quailty of drivers in speakers

[3db]

Lets say that the drivers are identical, what is the single biggest factor that make speakers sound different from one another? I'm excluding factors such as room acoustics, different driver types such as omni polar, electrostatics, etc.. What percentage does this factor attribute in differentiating teh sound between speakers?

 

[Joe Schmoe]

Probably the way that the bass is reinforced. Sealed, ported, passive radiator, open baffle, transmission-line, etc. all have different characteristics (and, of course, any given design can be implemented with varying degrees of success.)
Another major factor would be cabinet resonance (or lack thereof.)
A third factor that can have an effect (not sure how significant) is diffraction. Baffles with radiused edges tend to image better.

 

[Swerd]

If all other things are identical, same drivers in the same cabinet, the crossover is clearly the single biggest source of differences in sound. I've heard this demonstrated more than a few times. In a typical 2-way speaker the crossover is in the upper midrange (2 to 3 kHz) where human hearing is very sensitive. This can affect the sound plus or minus one octave, so a crossover affects sound as low as 1 kHz to as high as 6 kHz.

A crossover affects the balance between bass and treble, whether a speaker seems "bright" or "laid back". A properly designed crossover filters out all the breakup noise created by a woofer in its upper range. And a properly designed crossover also prevent tweeter distortion caused when it gets driven at low frequencies. So the crossover affects the amount of noise a speaker makes, or how "fatiguing" it can sound. A crossover shapes the overall frequency response of a speaker.

A distant second is the nature of the bass alignment of the cabinet, i.e. whether the cabinet is sealed, vented, etc. This affects sound only below 200 Hz.

It is debatable how much baffle edge effects can be heard. Those differences can be measured, but it is not clear how many people hear them.

I am not sure how big an effect cabinet resonsances can have. Maybe that is because I have not heard a good comparison of dead vs. noisy cabinets. I would be surprised if it was a big as the major differences caused by crossovers.

 

[Joe Schmoe]

I would also have mentioned crossovers, but I though that that was included under the heading of "assuming equal quality...". (I tend to think of drivers and crossovers as one inseparable entity.)

 

[Nuance AH]

Lets say that the drivers are identical, what is the single biggest factor that make speakers sound different from one another? I'm excluding factors such as room acoustics, different driver types such as omni polar, electrostatics, etc.. What percentage does this factor attribute in differentiating teh sound between speakers?

Without a doubt, the crossover. If you can't get that "right" the quality drivers will be meaningless IMO (not being used to their potential).

Have you ever heard a speaker with average drivers but an amazing crossover/implementation? I have, and the speaker was pretty impressive. In turn, I've also heard speakers with "top notch tier one drivers," but a poor crossover/implementation. These speakers sounded like poo; not worth the price at all.

Definitely the crossover.

Just my opinion.

Edit: DOH - swerd beat me to it. You've won this round my friend.

 

[Nuance AH]

I would also have mentioned crossovers, but I though that that was included under the heading of "assuming equal quality...". (I tend to think of drivers and crossovers as one inseparable entity.)

I used to think that way too, but now I know that the crossover is even more important than the quality of the drivers. Both need not be overlooked; it is critical that they are both handled properly IMO, else the end result will be a sub-par product.

 

[avaserfi]

It is debatable how much baffle edge effects can be heard. Those differences can be measured, but it is not clear how many people hear them.

Diffraction directly effects polar response of a loudspeaker. Such effects can have drastic effects on loudspeaker performance depending on the drivers used, room interaction and resonances within the loudspeaker. For ideal soundstage and imaging as well as timbre information diffraction issues need be eliminated by use of a waveguide (if it allows proper polar response) or baffle design.

I am not sure how big an effect cabinet resonsances can have. Maybe that is because I have not heard a good comparison of dead vs. noisy cabinets. I would be surprised if it was a big as the major differences caused by crossovers.

Audible resonance, both of the driver and cabinet, have been shown to be correlated with loudspeaker preference. All other things equal (and baring specific anomalies) a resonant loudspeaker will almost always be perceived as less pleasing than an inert one.

It seems many people seem to be unfamiliar with resonance as an issue due to lack of an easily accessible inert reference.

 

[gus6464]

Take a look at this driver: Usher 8945P

Now assuming you use a tweeter that is capable of being crossed over at the following three frequencies with no problems how do you think each speaker will sound different?

1.5KHz
2KHz
3KHz

I would guess that at 1.5KHz the speaker would sound a bit laid back in the midrange. At 2KHz the speaker will still sound laid back between 1-1.5KHz but then start to have a bit of a foward characteristic. Then at 3KHz I would definitely find the speaker to be forward.

 

[avaserfi]

Take a look at this driver: Usher 8945P

Now assuming you use a tweeter that is capable of being crossed over at the following three frequencies with no problems how do you think each speaker will sound different?

1.5KHz
2KHz
3KHz

I would guess that at 1.5KHz the speaker would sound a bit laid back in the midrange. At 2KHz the speaker will still sound laid back between 1-1.5KHz but then start to have a bit of a foward characteristic. Then at 3KHz I would definitely find the speaker to be forward.

The data provided by a simple on-axis and phase response plot of this tweeter is not sufficient to make such judgments of perceived tonality. Also, most crossover implementations do not just crossover with no effect on driver response. So, for this driver proper crossover implementation could actually tame peaks/dips etc... It seems common after all, for speaker designers to include their preferred tonality via crossover implementation.

In general, crossover point has no direct effect on a speakers sound presuming the drivers being integrated are sufficiently capable of this high/low a crossover point*. This also presumes that the crossover is being designed to be linear rather than color sound.

Rather, what causes described perception would be the lower Q changes in on/off axis frequency response as well as possible cabinet and driver resonance [these are just examples, not conclusive].

*This refers to driver center to center distance, distortion, polar response, driver resonances etc...

 

[gus6464]

The data provided by a simple on-axis and phase response plot of this tweeter is not sufficient to make such judgments of perceived tonality. Also, most crossover implementations do not just crossover with no effect on driver response. So, for this driver proper crossover implementation could actually tame peaks/dips etc... It seems common after all, for speaker designers to include their preferred tonality via crossover implementation.

In general, crossover point has no direct effect on a speakers sound presuming the drivers being integrated are sufficiently capable of this high/low a crossover point*. This also presumes that the crossover is being designed to be linear rather than color sound.

Rather, what causes described perception would be the lower Q changes in on/off axis frequency response as well as possible cabinet and driver resonance [these are just examples, not conclusive].

*This refers to driver center to center distance, distortion, polar response, driver resonances etc...

To assume that all manufacturers will use a driver capable of performing within it's crossover limits nowadays is kind of wishful thinking. Companies have to watch out for costs in relation to price after all.

 

[avaserfi]

To assume that all manufacturers will use a driver capable of performing within it's crossover limits nowadays is kind of wishful thinking. Companies have to watch out for costs in relation to price after all.

Such a presumption was never made, the explanation above was for illustrative purposes alone. It is clear, based on many loudspeaker's performance, that performance (in many areas) is often sacrificed for cost savings. Even so, it is impossible to make predictions of sound quality based on knowing the crossover point and on-axis response alone. Other data, as previously stated, is required to correlate measurement to human perception.

 

[gus6464]

Such a presumption was never made, the explanation above was for illustrative purposes alone. It is clear, based on many loudspeaker's performance, that performance (in many areas) is often sacrificed for cost savings. Even so, it is impossible to make predictions of sound quality based on knowing the crossover point alone. Other data, as previously stated, is required to correlate measurement to human perception.

Most of the time one can tell how a loudspeaker is going to sound just on the FR graph alone. Assuming a speaker is designed to give a neutral response, how does a +10dB bump from 2KHz-3KHz not matter. I know most people find neutral sounding speakers to be somewhat dull and that is why the house sound is implemented.

 

[3db]

So from the posts received thus far, its fair to say that crossover is the the number #1 component in making speakers sound different from one another.

How much affect would cabinet resonance have towards making speakers sound different? It gets mentioend alot around here but I wonder how much this really plays into the sound.

 

[avaserfi]

Most of the time one can tell how a loudspeaker is going to sound just on the FR graph alone. Assuming a speaker is designed to give a neutral response, how does a +10dB bump from 2KHz-3KHz not matter.

At most simply knowing on-axis response will give one a extremely general idea of tonality and nothing more. Three speakers could relatively easily be designed that had near identical on-axis responses while all sounding vastly different.

Some of the citations below also deal directly with loudspeaker measurements as they relate to perception. On-axis response is an important piece of the puzzle, but is by no means definitive. Also there is anecdotal evidence for this, in the early 1970s Ian Paisely (Mirage/NRC) designed a loudspeaker that was 30Hz to 20kHz +/- ¼ db this loudspeaker while a feat was founded to sound terrible subjectively. Later, Paisely discovered this was primarily due to the units terrible off-axis response. This lead to research whose findings mimic Toole's in discovered off-axis response importance.

Now, take the following three speakers all three of which are extremely linear devices. While there on-axis responses are very similar, is it very likely they will sound the same?

How much affect would cabinet resonance have towards making speakers sound different? It gets mentioend alot around here but I wonder how much this really plays into the sound.

In Toole and Olive's work it was found that audible resonances greatly effects loudspeaker preference and rating [1][2]. Later, the two researchers worked together to review prior work and finally expand on the subject of resonance audibility [3][4]. These works have stood up to peer review and scrutiny over the years and within the past few years have been used to create a successful mathematical model that can near perfectly (r = 0.995) loudspeaker preference of a listener [5].

[1] Toole, E. Floyd. Loudspeaker Measurements and Their Relationship to Listener Preference - Part 1. J. Audio Engineering Soc., Vol. 34, No. 4. 227 - 235. April 1986.
[2] Toole, E. Floyd. Loudspeaker Measurements and Their Relationship to Listener Preference - Part 2. J. Audio Engineering Soc., Vol. 34, No. 5. 323 - 348. May 1986.
[3] Toole, E. Floyd. The Modification of Timbre by Resonances: Perception and Measurement. J. Audio Engineering Soc., Vol. 36, No. 3. 122 - 142. March 1988.
[4] Olive, E. Sean et al. The Detection Thresholds of Resonances at Low Frequencies. J. Audio Engineering Soc., Vol. 45, No. 3 116 - 128. March 1997.
[5] Olive, E. Sean. Multiple Regression Model For Predicting Loudspeaker Preference - Part 2. Harman International. Convention Paper 6190. October 2004.

 

[Swerd]

Take a look at this driver: Usher 8945P

Now assuming you use a tweeter that is capable of being crossed over at the following three frequencies with no problems how do you think each speaker will sound different?

1.5KHz
2KHz
3KHz

I would guess that at 1.5KHz the speaker would sound a bit laid back in the midrange. At 2KHz the speaker will still sound laid back between 1-1.5KHz but then start to have a bit of a foward characteristic. Then at 3KHz I would definitely find the speaker to be forward.

When I say "crossover" I don't mean just the circuits that select the roll-off frequencies for the drivers, but all the various additional circuits that smooth the response or improve the overall performance of the speaker. There are plenty of commercially available speakers with crossovers that skip one or several of these, but no good speaker omits them.

Thanks for including that FR graph. It's a useful illustration. That Usher woofer has a typical FR curve of a woofer with a Kevlar or carbon fiber cone (after I looked it up, it's a paper/carbon fiber composite cone). The raised response at the high end of its FR at 3-4 kHz is typical. This bump in response may be due to a reflected resonance coming from where the stiff cone meets the flexible rubber surround, or it may be due to breakup, when the cone stops moving back and forth like a rigid piston and begins to ripple and flex. Carbon fiber and Kevlar woofers do this worse than paper woofers, but not as bad as metal drivers.

What ever causes these upper midrange peaks, they are noise that you don't want to hear. The crossover should suppress those frequencies to the point where they are inaudible. Otherwise it might produce an irritating edge to the speaker’s sound. Speakers that at first listen seem to add detail over and above what is in the recording generally suffer woofers with these kind of upper midrange peaks. Not surprisingly, many people mistakenly believe this fatiguing sound comes from the tweeter and not the woofer.

So to go back to your question that follows the FR graph, I would begin rolling off that woofer at 1.5 kHz or lower, to avoid hearing any of those resonance or breakup peaks. The slight dip just below 1.5 kHz can easily be smoothed out so it won't be noticed. Remember, this is a region where the tweeter also contributes about half the sound and that will help minimize the dip.

My other point about choosing the crossover frequency reflects avaserfi's earlier post that the FR curve you posted only shows on-axis response. To get the maximum dispersion and imaging from a speaker, a crossover designer should select a woofer roll-off frequency that is below the point where the woofer's off-axis response drops significantly below the on-axis response. That's why a manufacturer should provide a FR curves that show on-axis and off-axis curves. Here is an example where on-axis and 30° and 60° off-axis FR curves are shown in one graph.

Diffraction directly effects polar response of a loudspeaker. Such effects can have drastic effects on loudspeaker performance depending on the drivers used, room interaction and resonances within the loudspeaker. For ideal soundstage and imaging as well as timbre information diffraction issues need be eliminated by use of a waveguide (if it allows proper polar response) or baffle design.

Your points are correct, but I was limiting my earlier comments about baffle diffraction to what happens when two identical speakers are compared. One speaker is in a cabinet with squared edges, and the other has rounded edges with a ¾" radius. Here is an example.

The differences a small but measurable. It is debateable how audible those differences are.

So from the posts received thus far, its fair to say that crossover is the the number #1 component in making speakers sound different from one another.

How much affect would cabinet resonance have towards making speakers sound different? It gets mentioend alot around here but I wonder how much this really plays into the sound.

That’s something that I've also wondered about, but until we see a quantitative comparison we won't know. The standard FR curves for speakers, and the curves showing cabinet resonances are like comparing apples and oranges. I don't understand how to directly compare them.

 

[avaserfi]

Your points are correct, but I was limiting my earlier comments about baffle diffraction to what happens when two identical speakers are compared. One speaker is in a cabinet with squared edges, and the other has rounded edges with a ¾" radius. Here is an example.

The differences a small but measurable. It is debateable how audible those differences are.

This is the same situation I was referring too as well. The example you linked measured radii that were not sufficiently large for optimal performance and still came to results that would be audible*. If a more ideal situation was measured such as square corners compared to a loudspeaker with 2" edge radii a larger, more audible, difference would be seen.

*There has been a significant amount of research with regard to audibility of peaks and dips in frequency response as they relate to Q, frequency and even source material. The only time the audibility of such differences are debated is when there is a lack of understanding (or knowledge of) this research.

That’s something that I've also wondered about, but until we see a quantitative comparison we won't know. The standard FR curves for speakers, and the curves showing cabinet resonances are like comparing apples and oranges. I don't understand how to directly compare them.

This specific complaint is fully discussed in the article I previously mentioned in this thread. The audibility of resonance is fully quantified in a variety of acoustic situations as it relates to Q and playback level.

Toole, E. Floyd. The Modification of Timbre by Resonances: Perception and Measurement. J. Audio Engineering Soc., Vol. 36, No. 3. 122 - 142. March 1988.

From Toole's paper "Audio - Science in the Service of Art" (available online):

"Why is it that resonances are so important? Because they are the fundamental building blocks of almost all of the sounds we are interested
in hearing. High-Q resonances define the pitches. Medium- and low-Q resonances define the timbres, allowing us to distinguish between different voices and instruments. It is subtle differences in the resonant structure of sounds that are responsible for the nuances and shading of tone in musical sounds. Our ears are very highly attuned to the detection and evaluation of resonances, and it is therefore no surprise that listeners zero in on them as unwanted “editorializing” when they appear in loudspeakers."

 

[Swerd]

This is the same situation I was referring too as well. The example you linked measured radii that were not sufficiently large for optimal performance and still came to results that would be audible*. If a more ideal situation was measured such as square corners compared to a loudspeaker with 2" edge radii a larger, more audible, difference would be seen.

*There has been a significant amount of research with regard to audibility of peaks and dips in frequency response as they relate to Q, frequency and even source material. The only time the audibility of such differences are debated is when there is a lack of understanding (or knowledge of) this research.

The example shows that a ¾" radius roundover was sufficient to essentially eliminate the diffraction peaks. A 2" radius would not produce a different result - at least not with the speakers used in that link.

This specific complaint is fully discussed in the article I previously mentioned in this thread. The audibility of resonance is fully quantified in a variety of acoustic situations as it relates to Q and playback level.

Toole, E. Floyd. The Modification of Timbre by Resonances: Perception and Measurement. J. Audio Engineering Soc., Vol. 36, No. 3. 122 - 142. March 1988.

From Toole's paper "Audio - Science in the Service of Art" (available online):

"Why is it that resonances are so important? Because they are the fundamental building blocks of almost all of the sounds we are interested
in hearing. High-Q resonances define the pitches. Medium- and low-Q resonances define the timbres, allowing us to distinguish between different voices and instruments. It is subtle differences in the resonant structure of sounds that are responsible for the nuances and shading of tone in musical sounds. Our ears are very highly attuned to the detection and evaluation of resonances, and it is therefore no surprise that listeners zero in on them as unwanted “editorializing” when they appear in loudspeakers."

Thanks for that reference. I recently got Toole's book, and I hope I'll find a similar section about this subject in it. I've only begun reading it.

 

[avaserfi]

The example shows that a ¾" radius roundover was sufficient to essentially eliminate the diffraction peaks. A 2" radius would not produce a different result - at least not with the speakers used in that link

This is primarily because only on-axis and 15 degree measurements were shown. Other diffraction issues would be more visible as the measurements moved further off-axis presuming a waveguide was not used. The other alternative (aside from a waveguide causing this) is that the drivers used do not have sufficient polar response to fully excite these diffraction issues at more extreme angles.

 

[AudioDude]

Lets say that the drivers are identical, what is the single biggest factor that make speakers sound different from one another? I'm excluding factors such as room acoustics, different driver types such as omni polar, electrostatics, etc.. What percentage does this factor attribute in differentiating teh sound between speakers?

There are several factors that play roles in speakers. The material of the actual driver will determine if the sound is free or muffled. The cross-over components can either help you or hurt the sound greatly. The cabinet in which the speakers are mounted in can either give good resonance or fair resonance. The cubic space needs to be pre determined and measure in order to get great speaker sound and response. Example; we purchase our outdoor speakers from outdoorspeakerstore.com and learned that the omni speakers we purchased was less expensive than the Bose omni type speakers. We also know that the speakers we purchased have an 8" driver & the bose only has a 4.5" driver within their cabinet. Which do you think delivered the better quality sound? Not the Bose! In the real world, you cannot get more sound out of a smaller speaker when compared to a speaker of greater mass. The smaller drivers can only produce so much bass response regardless of the cabinet size. Also, the cross over networks within need to have a good variation of say 20Hz to 20,000Hz. Keeping that in mind will help anyone purchasing speakers for either indoors or outdoors.

AudioDude