A Broad Discussion of Speakers with Major Audio Luminaries

[Holmz]

Am I mistaken here...if a speaker (let's say a tweeter) can move quickly enough to adequately respond to a 20 KHz signal, it is "fast" enough, yes?

No!
but maybe it’s @Newman that is mistaken.

 

[audiofooled]

It is an understandable misunderstanding, but a misunderstanding nevertheless.

There is a whole ASR thread discussing this exact question. My best attempt in that thread to explain why the output (SPL) from a driver does not track the velocity or displacement of the diaphragm, no matter how intuitive that seems, is linked here.

Yeah, the auditory perception related to acoustical phenomena and, on the other hand, electromechanical and signal properties seem to be routinely conflated based on intuition. That is without taking into account phase relationships between acceleration, displacement, velocity and pressure.

What may possibly happen in rooms at listening positions, as far as energy transfer between objects/bodies of different acoustic impedance I remark here:

Help me understand Impulse Response, Step Response, and Time Alignment

Please show me a scientific study that shows ... Why?! If there is, an AI will find it today. The problem then is the analysis of the result(s). Without a solid background, or to the contrary, with just the background of long term audio enthusiasm, not too much is gained. Audio is not the right...

www.audiosciencereview.com

 

[gnarly]

This is the "fast bass" audiophile argument. It is a myth. Is that "definitive" enough?

@Fredygump was clearly questioning dynamics from a full-spectrum viewpoint....very appropriately imo.
Calling his post the 'fast bass' argument seems rather disingenuous.

 

[gnarly]

I'd like to make another attempt at explaining why I think SPL is key to the question of, does higher sensitivity correspond with higher dynamics.
First, an attempt at defining/separating the terms transient response from dynamics. Admittedly my own differentiation, but i think it might help.

Transient response equals frequency response. The wider the range of frequency response, the flatter the response....the greater the transient response.
(and the 'faster the bass' lol).
Time and phase alignment are the icing on the cake for additional transient response improvement.
This is all definitional and rooted in the equivalency between a transfer function and impulse impulse response.
Nothing about this definition of transient response requires any degree of specification with regard to level (SPL or electrical).

Dynamics means to me, how well does a speaker produce excellent transient response/frequency response..

It is dependent on every driver section in the speaker maintaining its bandwidth contribution to transient response/frequency response, at the SPL level being driven.
Which means no thermal compression, no amplifier clipping, no excursion limitations...at the grossest level of breakdown. And at a finer level, no excessive rise in distortion or other non-linear responses.

And this is of course very much SPL level dependent,
as every speaker has its transient response/frequency response degrade, once above a certain SPL threshold.
Above that threshold, one or more driver sections will fail to maintain output linearity throughout their passbands. (as @Fredygump was questioning)

Below that SPL threshold, I'd say the speaker has excellent dynamics.
Above it, dynamics begin to suffer, increasingly so as SPL is raised.

Without bringing SPL into the discussion of sensitivity vs dynamics, the discussion is undefined and pointless imo.
(Or efficiency, as sensitivity and efficiency are directly related and the distinction is primarily pedantic at this level).


Without harping further on why I think high sensitivity designs are much more likely to maintain dynamics with increased SPL (seems like common sense to me)

here's a great vid from B&C on driver power handling... (whole thing away from time stamp is worth watching)
It definitely makes me reconsider the power specs we read about drivers...and speakers' power ratings,...... cause let's face it, most all speaker power specs are just the sum of driver manufacturers specs)

 

[RDoc]

<snip>

There is a whole ASR thread discussing this exact question. My best attempt in that thread to explain why the output (SPL) from a driver does not track the velocity or displacement of the diaphragm, no matter how intuitive that seems, is linked here.

cheers

The linked to diagrams attributed to Purify are complete fakes. There is absolutely no way any physical system can respond like that to a square wave.

 

[Holmz]

I'd like to make another attempt at explaining why I think SPL is key to the question of, does higher sensitivity correspond with higher dynamics.
First, an attempt at defining/separating the terms transient response from dynamics. Admittedly my own differentiation, but i think it might help.
…

Help who?

…
Transient response equals frequency response. The wider the range of frequency response, the flatter the response....the greater the transient response.
(and the 'faster the bass' lol).
Time and phase alignment are the icing on the cake for additional transient response improvement.
This is all definitional and rooted in the equivalency between a transfer function and impulse impulse response.
Nothing about this definition of transient response requires any degree of specification with regard to level (SPL or electrical).

Dynamics means to me, how well does a speaker produce excellent transient response/frequency response..

…

Transient Response (TR) maybe mathematically has some overlap with frequemncy response.
But I am imagining you are thinking a flat magnitude FR and somehow hoping that that will have a stunning TR or Impulse response (IR).

You can get a nice flat FR and have poor TR and poor IR.

…
It is dependent on every driver section in the speaker maintaining its bandwidth contribution to transient response/frequency response, at the SPL level being driven.
Which means no thermal compression, no amplifier clipping, no excursion limitations...at the grossest level of breakdown. And at a finer level, no excessive rise in distortion or other non-linear responses.
…

For some instantaneous thing, thermal heating is not going to happen.
I am not sure I agree at all with the low distortion part, and I’d almost bet that “dynamic systems” can have a lot of distortion with increasing SPL.

…
And this is of course very much SPL level dependent,
as every speaker has its transient response/frequency response degrade, once above a certain SPL threshold.
Above that threshold, one or more driver sections will fail to maintain output linearity throughout their passbands. (as @Fredygump was questioning)
…

Is that what they were questioning ?

…
Below that SPL threshold, I'd say the speaker has excellent dynamics.
Above it, dynamics begin to suffer, increasingly so as SPL is raised.
…

This seems like hope and common sense, versus factual

…
Without bringing SPL into the discussion of sensitivity vs dynamics, the discussion is undefined and pointless imo.
(Or efficiency, as sensitivity and efficiency are directly related and the distinction is primarily pedantic at this level).


Without harping further on why I think high sensitivity designs are much more likely to maintain dynamics with increased SPL (seems like common sense to me)
…

More hope and common sense?

 

[DonH56]

I'm unclear what the difference is between sensitivity and efficiency in this context.

Sensitivity is how much SPL is produced at 1 m with 1 W input. Reducing self-heating in the voice coil means less wasted power (energy) to heating the voice coil wires resulting in higher output for a given input, thus higher efficiency (greater output power for less input power). It could be argued that more sensitive speakers are more efficient but it does not always work out that way.

 

[levimax]

It could be argued that more sensitive speakers are more efficient but it does not always work out that way.

I am having a hard time understanding how a more sensitive speaker would also not be more efficient. Can you give an example to help me understand.

 

[IPunchCholla]

You can get a nice flat FR and have poor TR and poor IR.

I find myself having a hard time grasping this. Can you point to some reading or examples?

 

[gnarly]

I am having a hard time understanding how a more sensitive speaker would also not be more efficient. Can you give an example to help me understand.

I'd like to see an example too.

They are different for sure, .....but as we also know to be so directly related, that I think the belaboring the distinction is pointless..

Efficiency and sensitivity conversion - loudspeaker percent and dB per watt and meter loudspeaker efficiency versus sensitivity vs speaker sensitivity 1 watt = 2,83 volt box chart - sengpielaudio Sengpiel Berlin

Efficiency and sensitivity conversion - loudspeaker percent and dB per watt and meter dB/W/m loudspeaker efficiency versus sensitivity vs speaker sensitivity 1 watt = 2,83 volt box chart - sengpielaudio Eberhard Sengpiel

sengpielaudio.com

 

[DonH56]

I am having a hard time understanding how a more sensitive speaker would also not be more efficient. Can you give an example to help me understand.

Not offhand and I am generally staying away from ASR lately so not going to dig. We had to calculate an example in one of the college acoustics classes I took a millennia or so ago. Basically you can make a speaker that has high sensitivity and yet wastes a lot of power, I think related to the impedance curve and crossover/driver choices that lead to a good sensitivity number but poor overall efficiency (or vice-versa). Part of it is due to how sensitivity is often defined, as a function of efficiency, which makes for an easy calculation but is not necessarily true to the physics. Sensitivity is also a single-frequency measurement whilst efficiency is generally calculated for the overall (broadband) transfer function. By limiting the definitions you can make the argument they are related as I mentioned, and that is how most audiophiles treat it. It often goes the other way, with midrange and tweeter drivers attenuated to provide more extended bass relative to the upper drivers' levels, reducing sensitivity though woofer efficiency could be higher. In that case the upper drivers are not necessarily less efficient, but power is dissipated in the attenuating resistor, reducing overall efficiency. Note speakers are electromechanical devices with a number of things affecting sensitivity and efficiency, such as magnet strength, spider and surround resistance, voice coil impedance, pressure coupling (e.g. cone vs. horn radiators, ports, passive radiators, etc.), and so forth that complicate the calculations. In the end sensitivity is how SPL a watt produces at 1 kHz 1 m away whilst efficiency depends on how much power is dissipated by anything that does not directly produce sound. Small single-driver speakers tend to be fairly sensitive but inefficient.

The difference is far easier to show in electronics than speakers. You can design a highly sensitive amplifier with appropriate gain staging, but the end result could be very inefficient, such as a class A vs. B vs. D output stage.

This is a different metric than "dynamics", with audio commentary often claiming falsely that compression drivers are more "dynamic" due to their higher sensitivity. It takes less power to provide a wide dynamic range, but compression drivers are not inherently more "dynamic" than conventional or planar drivers. IME experience compression drivers and their associated horns often emphasize the upper treble by design and due to beaming in the horns, making them "brighter" and thus sound more "dynamic" to folk.

 

[Everett T]

Here is a nice summation of sensitivity

 

[gnarly]

Not offhand and I am generally staying away from ASR lately so not going to dig. We had to calculate an example in one of the college acoustics classes I took a millennia or so ago. Basically you can make a speaker that has high sensitivity and yet wastes a lot of power, I think related to the impedance curve and crossover/driver choices that lead to a good sensitivity number but poor overall efficiency (or vice-versa). Part of it is due to how sensitivity is often defined, as a function of efficiency, which makes for an easy calculation but is not necessarily true to the physics. Sensitivity is also a single-frequency measurement whilst efficiency is generally calculated for the overall (broadband) transfer function. By limiting the definitions you can make the argument they are related as I mentioned, and that is how most audiophiles treat it.

Thanks for that....all makes sense.
Especially the part about staying away from ASR ! lol.... keep telling myself it's a bad habit...

Anyway, I can see (and have known) that measuring efficiency is a different measurement than a simple formula conversion from sensitivity.
Often I've seen where using manufactures specs, the sensitivity and efficiency don't tie together using the general conversion formula.
I think because most of the time sensitivity is just eyeballed from the response curve, to a number the manufacturer feels they can justify.

Personally, I like to measure sensitivity as an integration across the driver's pass band. I use pink with all driver processing in place, and measure average RMS voltage at driver terminals, while simultaneously with measuring acoustic LEQ SPL. Gives real world voltage sensitivity imo. Add in measuring average current over same time interval, and real world power sensitivity too. Todays true RMS, averaging meters make it pretty easy to do i think.

Anyway again, I think the subject gets way too much attention.....generally from just wanting to prove points...lol again

 

[MattHooper]

Just buy some Quad 57s, problem solved. People, we've had electrostatic speakers and over-engineered TLS from people like Bill Perkins for years now. Building a great speaker is just honest engineering. Most of what we have now is just marketing bs from greedy people selling overpriced crap.

The 57s are one of my favourite speakers and I have owned the 63s.

However, I would not be satisfied with either of those speakers. I’d love to on the 57s in addition to my main speakers just as a place to visit, but I moved on from electrostatics to dynamic speakers and never regretted it.

 

[Floyd Toole]

Sensitivity is how much SPL is produced at 1 m with 1 W input.

dB SPL @ 1 watt @ 1 m iis the "traditional" statement of sensitivity, but it needs elaboration, and it has been replaced.

Today the standard method of measuring loudspeaker sensitivity is to use an input of 2.83 volts and measure the sound pressure level in an anechoic space in the far field of the source. 2.83 volts generates 1 watt into 8 ohms. a number that was selected because 8 ohms is the nominal load for defining the power output of amplifiers. For consumer and monitor loudspeaker systems the far field is typically 2 m or more, depending on the size of the loudspeaker, and the measured SPL is adjusted by calculation to what it would have been at 1 m. So, the measured SPL level at 2 m is lifted by 6 dB (a factor of 2 in distance in this example) to give us the sensitivity at the standard distance of 1 m. Only single transducers and small loudspeaker systems yield valid measurements at 1 m. Sensitivity should be specified as the average SPL over a frequency range, such as 300 Hz to 3 kHz (there are other possible ranges) to avoid manufacturers picking a peak in the frequency response to get a favorable specification. Ignore any specifications that do not specify how sensitivity was measured, and absolutely ignore in-room measurements.

But this statement alone is not enough information. We need to know the impedance characteristics.

Loudspeaker systems and transducers do not have constant impedance, even though manufacturers specify a "nominal" impedance. A lot of "8 ohm" loudspeaker systems have impedances that more correctly would be described as 4 ohms, and they may have minimum impedances even lower at some frequencies, and maximum impedances much higher. That means that the power into the loudspeaker varies with frequency, so there is no meaningful power sensitivity except at specific frequencies, which is a useless situation, but the charade goes on. Loudspeakers with real impedances lower than the specified number extract more power from the amplifier which is a problem for customers. Many inexpensive power amplifiers cannot drive lower impedances, leading to clipping and protective circuit activation - distortions we want to avoid. To be safe, choose power amplifiers that are designed to safely drive 4 ohms, and even better, 2 ohms. The best ones will double power into halved impedances, but many do not, but are still stable. In multichannel amplifiers be careful to note how many channels are driven for the power specification, and over what frequency range the specification holds.

The current standard is a voltage sensitivity, which for solid state amplifiers (constant voltage sources) is appropriate. It works so long as the impedance minimum does not result in current limiting of the output stages, so a full impedance specification needs to include a minimum impedance. Some incompetent system designs have had impedances that drop to under 1 ohm. Ignorant reviewers thought that such loudspeakers were able to "reveal" differences between power amplifiers, when in fact they were the problem. The result has been generations of massive monoblock "arc welder" power amps that can drive very low impedances and remain stable - expensive solutions for problems that should not exist.

Efficiency is a measure of acoustical power out compared to the electrical power in, and the percentages are low. Measuring the acoustical power out is complicated so it is almost an academic concept. In addition, a percentage is not very helpful in practical applications. The word "efficiency" is widely misused in common parlance.

 

[DonH56]

dB SPL @ 1 watt @ 1 m iis the "traditional" statement of sensitivity, but it needs elaboration, and it has been replaced.

Today the standard method of measuring loudspeaker sensitivity is to use an input of 2.83 volts and measure the sound pressure level in an anechoic space in the far field of the source. 2.83 volts generates 1 watt into 8 ohms. a number that was selected because 8 ohms is the nominal load for defining the power output of amplifiers. For consumer and monitor loudspeaker systems the far field is typically 2 m or more, depending on the size of the loudspeaker, and the measured SPL is adjusted by calculation to what it would have been at 1 m. So, the measured SPL level at 2 m is lifted by 6 dB (a factor of 2 in distance in this example) to give us the sensitivity at the standard distance of 1 m. Only single transducers and small loudspeaker systems yield valid measurements at 1 m. Sensitivity should be specified as the average SPL over a frequency range, such as 300 Hz to 3 kHz (there are other possible ranges) to avoid manufacturers picking a peak in the frequency response to get a favorable specification. Ignore any specifications that do not specify how sensitivity was measured, and absolutely ignore in-room measurements.

But this statement alone is not enough information. We need to know the impedance characteristics.

Loudspeaker systems and transducers do not have constant impedance, even though manufacturers specify a "nominal" impedance. A lot of "8 ohm" loudspeaker systems have impedances that more correctly would be described as 4 ohms, and they may have minimum impedances even lower at some frequencies, and maximum impedances much higher. That means that the power into the loudspeaker varies with frequency, so there is no meaningful power sensitivity except at specific frequencies, which is a useless situation, but the charade goes on. Loudspeakers with real impedances lower than the specified number extract more power from the amplifier which is a problem for customers. Many inexpensive power amplifiers cannot drive lower impedances, leading to clipping and protective circuit activation - distortions we want to avoid. To be safe, choose power amplifiers that are designed to safely drive 4 ohms, and even better, 2 ohms. The best ones will double power into halved impedances, but many do not, but are still stable. In multichannel amplifiers be careful to note how many channels are driven for the power specification, and over what frequency range the specification holds.

The current standard is a voltage sensitivity, which for solid state amplifiers (constant voltage sources) is appropriate. It works so long as the impedance minimum does not result in current limiting of the output stages, so a full impedance specification needs to include a minimum impedance. Some incompetent system designs have had impedances that drop to under 1 ohm. Ignorant reviewers thought that such loudspeakers were able to "reveal" differences between power amplifiers, when in fact they were the problem. The result has been generations of massive monoblock "arc welder" power amps that can drive very low impedances and remain stable - expensive solutions for problems that should not exist.

Efficiency is a measure of acoustical power out compared to the electrical power in, and the percentages are low. Measuring the acoustical power out is complicated so it is almost an academic concept. In addition, a percentage is not very helpful in practical applications. The word "efficiency" is widely misused in common parlance.

Thanks for the clarification and update, Floyd. Glad to see you're still carrying the torch!

I actually had a line about 2.828... Vrms but deleted it. Should've listened to my gut. Decades ago the FTC debated how sensitivity should be defined (something you were likely part of). I remember a huge debate about what bandwidth and signal type (swept sine, noise source, what color noise, etc.) should be used, but manufacturers pushed for a single point measurement at 1 kHz. Honestly don't remember the actual standard; I had moved on to higher frequency stuff.

One of the few quotes I remember from my grad acoustics class was when I asked the professor how loud 1 acoustic watt was. He just said "It'd blow your ears off!"

 

[Floyd Toole]

You can get a nice flat FR and have poor TR and poor IR.

I find myself having a hard time grasping this. Can you point to some reading or examples?

It may be understood by recognizing that transducers, within their operating frequency ranges, are minimum-phase devices. That means that the impulse response can be calculated from the amplitude vs. frequency response: the frequency response alone. When the frequency response is flat and smooth there are no resonances and no time-domain misbehaviour.

However in multi-way loudspeaker systems there are multiple transducers operating over different frequency ranges, combined using crossover networks. The positional differences in acoustic centres of the tranducers and their placement relative to each other add time delays and the crossover networks add phase shift. These are non-minimum-phase characteristics and time-domain behaviour is altered. All is well if the final system yields a flat smooth frequency response on and smooth behaviour off axis because humans are not sensitive to phase shifts and group delays within the ranges encountered in conventional loudspeaker systems.

As much as we like to see perfect square waves and step responses, our ears are not very fussy.

 

[bmc0]

Measuring the acoustical power out is complicated so it is almost an academic concept.

Could it not be approximated from the electrical impedance along with the same set of measurements needed to compute the spinorama?

 

[Floyd Toole]

Just buy some Quad 57s, problem solved. People, we've had electrostatic speakers and over-engineered TLS from people like Bill Perkins for years now. Building a great speaker is just honest engineering. Most of what we have now is just marketing bs from greedy people selling overpriced crap.

I debated about commenting, but this statement is simply so far from reality that I feel I must.

I am very familiar with the Quad 57 and 63, and show anechoic measurements and subjective evaluation results in all editions of my book. My real response is to the statement describing the present state of loudspeakers - it is far from "crap" and getting better by the day. Excellent sound at affordable prices now exists. A few loudspeakers on the market push the limits of what is possible, and increasing numbers are well past the point of diminishing returns. It is because of "audio science" the name of this forum. There is plenty of BS and unsubstantiated opinion floating around, but for those who are serious, there is a growing base of solid knowledge, and superb products as a result.

 

[Floyd Toole]

Could it not be approximated from the electrical impedance along with the same set of measurements needed to compute the spinorama?

Probably true, but I don't see a thermodynamic efficiency percentage as an answer to an existing problem, and of no direct use in the design of audio systems - that I can think of. It is an interesting statistic in comparing cones, domes and horns, but that has been known for decades. Any suggestions?