"Inefficient" speakers are "worse" than "efficient" speakers

[Chrispy]

A common complaint by ASR members is that so called "inefficient" speakers, 85db or lower sensitivity for 2.83v at 1M, are inferior to efficient speakers, 90db or higher sensitivity. I find this complaint difficult to understand from an objective and scientific viewpoint, especially with amplifier power being relatively inexpensive these days.
1. Inefficient speakers perform poorly at low volumes. Can someone explain this concept since it makes zero sense to me.
A side question would be more in general, the concept of why certain speakers perform better at lower volumes than other speakers at the same volume. What science is behind that?

2. Inefficient speakers are poorly designed. I can actually understand this comment if there are no benefits to the inefficient design such as poor bass response. Otherwise, isn't speaker design, a series of tradeoffs to obtain a result? Size of cabinet. Bass response. SPL capability. What am I missing?

3. Side question - why do some people convert sensitivity level measured at 2.83v at 1M into different sensitivity level based on impedance? Example - 85db at 2.83v at 1M converted to 82db at 4 ohms. How is this conversion useful?

4. Is this just a preference due to already owning lower power amplifiers? If so, I get that. But that has little to do with better or inferior.

Thank you in advance.

Strictly speaking the answer is "no". You probably have other issues/questions/points but this summary isn't it.

 

[kemmler3D]

If you take a poorly designed inefficient speaker with mediocre drivers, then your concerns are valid

I think @Chris A is actually correct about lower IMD being a characteristic of more efficient speakers, although I think the real impact is from lower excursion and not efficiency per se? Excursion causes IMD no matter what, even in the best drivers.

The rest of the comment I can't say for sure, I'm not good at evaluating electrical distortion, but hysteresis is a thing to consider, it's something Purifi brags about having reduced in their drivers, so it sounds plausible to me.

 

[Chris A]

Yes. The farther the diaphragm moves, the higher the modulation distortion for a given SPL. Higher efficiency/sensitivity means lower modulation distortion (Klipsch's Law). The reason why horn-loaded loudspeakers have higher clarity is due to low modulation distortion sidebands when playing real music (multitone operation).

Also, the thermal compression issues are just as real--and are a function of sensitivity/efficiency--just as is the modulation distortion numbers. You can try to increase the amount of copper cross-section in the voice coil to limit temperature rise, but the limits to driver design don't leave you with much latitude there (moving mass limits, etc.).

So these two subjects are not really about "badly designed" low sensitivity drivers/loudspeakers, but rather all loudspeakers having lower sensitivity. The lower the sensitivity/efficiency, the worse is the distortion due to those two effects. And these two distortion types are rarely discussed on audio forums-but they are quite audible with music dynamics.

One of the reasons why a lot of people like compressed dynamics music is that it tends to suppress the audibility of these two distortion types on their lower sensitivity loudspeakers (and avoid exciting their tinnitus, if they suffer from that malady).

There are a few offsetting effects in the positive column for lower sensitivity/efficiency (such as smaller loudspeaker size and a bit easier time avoiding higher Q resonances, and perhaps lower cost of materials used), but that's pretty much it.

Chris

 

[dlaloum]

I’m not sure where you’re getting this from ASR. Those sound much more like the claims you hear outside ASR from the golden ear crowd.

Is the type of stuff that fuels high sensitivity
“ low crossover count” speaker and low wattage tube systems.

I remember “ back in the day “ lots of the claims about low efficiency speakers or hard to drive speaker sounding lifeless or lacking clarity, and for me, it wasn’t particular Thiel loudspeakers that put that to rest.

They were notoriously complex with crossover parts as well as notoriously, low sensitivity and hard to drive. And yet I found that they sounded particularly clear and detailed as well as dynamically life-like.
So I’ve never worried about such stuff.

I will say that… yes… I have perceived really high sensitivity designs like some horn designs as sounding very dynamic and lifelike. So they might be something to the high sensitivity thing. But ultimately, I think it just comes down to how a speaker is engineered.

I have experienced speakers that provide heaps of micro-detail even at low volumes (electrostatics), and others which seemed to skip over the details in the recording unless you "gave them some stick"... had to play loudly! - Then they "came alive"

The stats were ALWAYS "alive".

One example I lived with for a few months, was a set of Klipsch Forte's... which sounded lifeless, and dull at what I consider normal listening levels, but crank then up to "loud", and everything came to life. (note: the "loud" setting didn't require heaps of power, as the speaker is relatively efficient... but it required more SPL than my stats did to achieve "life")

It felt like there was some sort of inertia that had to be overcome... and once you got past a certain level (SPL) - everything was there... but below that level.... sigh.

I've definitely heard this phenomenon - but never really heard a decent explanation of it.

And I'm not convinced that it is a dynamic vs stat speaker issue, as many "standard" speakers don't suffer from this effect in my experience.

It also makes me wonder whether that phenomenon was (is?) sufficiently common in mass market speakers, that it may have generated the common search for "loudness" - lots of people in my experience, had never heard a set of speakers that could reproduce all the micro-detail at low volume levels, hence their focus on achieving the levels of SPL at which they perceived that "quality" was achieved.

 

[Luvchampagne]

Yes. The farther the diaphragm moves, the higher the modulation distortion for a given SPL. Higher efficiency/sensitivity means lower modulation distortion (Klipsch's Law). The reason why horn-loaded loudspeakers have higher clarity is due to low modulation distortion sidebands when playing real music (multitone operation).

Also, the thermal compression issues are just as real--and are a function of sensitivity/efficiency--just as is the modulation distortion numbers. You can try to increase the amount of copper cross-section in the voice coil to limit temperature rise, but the limits to driver design don't leave you with much latitude there (moving mass limits, etc.).

So these two subjects are not really about "badly designed" low sensitivity drivers/loudspeakers, but rather all loudspeakers having lower sensitivity. The lower the sensitivity/efficiency, the worse is the distortion due to those two effects.

That is not what the scientific data shows for high performance drivers. How do you correlate the measurements of these drivers with the lack of distortion?

And these two distortion types are rarely discussed on audio forums-but they are quite audible with music dynamics.

Is this objectively measureable?
Whenever I read something is "audible" in the audiophile world, I don't believe it unless it is measureable.

One of the reasons why a lot of people like compressed dynamics music is that it tends to suppress the audibility of these two distortion types on their lower sensitivity loudspeakers (and avoid exciting their tinnitus, if they suffer from that malady).

There are a few offsetting effects in the positive column for lower sensitivity/efficiency (such as smaller loudspeaker size and a bit easier time avoiding higher Q resonances, and perhaps lower cost of materials used), but that's pretty much it.

Chris

Thank you. Not disagreeing with you on any subjective issue. I'd just like to better understand from an objective and scientifically measureable result.

 

[levimax]

The Ascilab speaker designer mentioned in a post that high Qms drivers are known for better sound at "low volume". High Qms means low mechanical damping so it does make some intuitive sense. I can't say for sure and it could be bias but I just built some large subs with 18" drivers with a high Qms and they do seem to stay balanced and lively at low listening levels. Not sure how one would go about testing this.

 

[Arindal]

The stats were ALWAYS "alive".

One explanation for that might be their tendency towards high and at the same time constant directivity. The level of additional reverb in the room is always low as a consequence of that, so you get this feeling of clarity and ´lively´ impulses even at lower SPL with no midrange-heavy reverb masking the treble details in the direct sound.

One example I lived with for a few months, was a set of Klipsch Forte's... which sounded lifeless, and dull at what I consider normal listening levels, but crank then up to "loud", and everything came to life.

Had the very same impression, and can offer a potential explanation other way ´round: although they offer a narrow directivity, it is pretty uneven in the midrange. If you listen at lower SPL, the reverb in the room is dominated by the frequency band 1-2K for which our ear is pretty sensitive, so this midrange/presence-heavy reverb is capable of masking treble details, creating a muddled tonality. Once you crank it up, the treble gets decreasingly masked and sounds more detailed and dynamic again.

And I'm not convinced that it is a dynamic vs stat speaker issue, as many "standard" speakers don't suffer from this effect in my experience.

It IMHO has nothing to do with dynamic vs. horn vs. planar principle, but with dominating frequency bands in the reverb due to bands offering overly low directivity index.

A neutral reverb with no dominant midrange, is what some mixing engineers working at low SPL are preferring, and many of them are enthusiastic about Kii Audio offering similar properties of excellent clarity at low levels (which use very conventional dynamic drivers but achieve the aforementioned properties by cardioid design).

 

[Mnyb]

I think the general design of the speakers dominates completely. High or low efficiency.

It seems to me be literally 1000’s of compromises involved designing speakers and drivers .

Many high efficiency speakers suffer from a very ragged frequency response resonances and abrupt changes in directivity ?

The good ones might need greater listening distance to make the drivers blend properly, hence a big room migth also be handy.

My 2c small listening spaces short listening distances and cheap amplifier power makes sacrifice some efficiency to get other goal seems like a tempting road forward for many speaker designers.

 

[Mike123]

1. Inefficient speakers perform poorly at low volumes. Can someone explain this concept since it makes zero sense to me.

Sensitivity at 2.83V is just a point on the graph. It doesn't make much sense if we're interested in low volumes. We need a graph of sensitivity versus power. If it's linear, then point 1 is incorrect. If stiff speakers don't have a linear sensitivity versus power, then at 0.283 or 0.0283V they'll play worse than sensitive speakers. I can assume that some stiff speakers behave exactly like that. That is, to move a stiff moving system, a certain minimum power level is needed. And below this limit, the speaker behaves very quietly, plays different volume levels nonlinearly, and thus introduces linear distortions at low volumes.

 

[Mnyb]

It mostly seems to gravitate around how big of a bass driver and subsequent big box you can tolerate.

Treble drivers and many midranges seems very efficient already, they are preceded by resistor networks in passive speakers and in active speakers smaller amps and attenuated levels

 

[Arindal]

Treble drivers and many midranges seems very efficient already, they are preceded by resistor networks in passive speakers

It appears to be a tendency, but not a general rule. While with tweeters there is little necessity to design really low-sensitivity models, existing midrange cones can show pretty low sensitivity. Particularly the small ones designed to be combined with a subwoofer at low x-over freq like 100 or 90Hz.

That is, to move a stiff moving system, a certain minimum power level is needed.

Which would mean that such speaker would fail miserably at any SPL with any music showing dynamic variations. I don't really think this is a common problem.

If excursion is not proportional to input level at lower SPL, it would show up as significant dynamic expansion plus distortion.

Many high efficiency speakers suffer from a very ragged frequency response resonances and abrupt changes in directivity ?

Same can be said about certain low-efficiency models, do not really see a connection here. Maybe except the fact that a speaker designer can exchange sensitivity for improving the frequency response, but on the other hand it is easier to control directivity with certain high-efficiency concepts.

 

[Looneybomber]

I think @Chris A is actually correct about lower IMD being a characteristic of more efficient speakers, although I think the real impact is from lower excursion and not efficiency per se? Excursion causes IMD no matter what, even in the best drivers.

IMD (amplitude and frequency modulation) is directly impacted by excursion, yes, but it’s independent of sensitivity or input power. In fact, higher efficiency could bring about more IMD per unit of power.

Think of a 6” woofer. X liters of displacement at Y Hz equals Z decibels. Simple acoustics. A certain amount of IMD will be produced with X because X will have a specific amount of excursion associated with it. Let’s call that, IMD(X). If it takes 1W to achieve X, we will also have IMD(X). If it takes 100W to achieve X, then we also have IMD(X). The system with higher sensitivity has higher IMD per watt but an equal amount of IMD per decibel since output and IMD is directly related to excursion and not power.

Are there other variables that change in real life? Yes. But, as a thought exercise, one must look at at the AM and FM portions of IMD and sensitivity together in a vacuum and ignore the other variables. Once done, it’s easy to see how the two have nothing to do with one another directly, only indirectly through excursion.

 

[Heinrich]

Regarding power dissipation, there's something to be considered that again centers on Hoffmann's Iron Law. The latter says, that the efficiency watts per boom in bass is determined by the internal box volume. And nearly nothing else, regardless of driver type, not even the size of it within reasonable bounds. Don't beleive me, or Mr. Hoffmann? Try a simulation program for speaker / box alignment, and you'll see. It is in the formulas. (Don't be fooled by the so called 'transfer function', which is 'normalised', but look at the real output.)

If you start from there, things become pretty easy. Define which sound pressure level at which frequency (in deep bass), given a certain input voltage has to be achieved, and chose the volume accordingly.
As a second step chose by what degree you are willing to attenuate the mids. This decision will determine the relation mids to bass, hence the nominal bass extension.

Attenuation can be done in different ways. Digital/analog equaliser, passive crossover ... all these will keep the power demand of the speaker driver's motor the same. The extra 'power' is disspated elsewhere, or is not generated (eq) to begin with.

Only rarely the attenuation of the mids is made by using a particularly heavy cone. It simplyy doesn't work well. But that wouuld be the only case, that a low sensitivity would heat the driver's motor beyond need.

Reiterated, if you take some internal box volume, all is settled, no further decision to be made. Any bass from that enclosure will cost you so much power. Only the relation to the mids is a matter of choice, but does regularly not affect power dissipation in the driver.

ps: conversely, you cannot determine the speaker's real power dissipation, given some odd playback level, from the sensitivity numbers in any way!

 

[dlaloum]

One explanation for that might be their tendency towards high and at the same time constant directivity. The level of additional reverb in the room is always low as a consequence of that, so you get this feeling of clarity and ´lively´ impulses even at lower SPL with no midrange-heavy reverb masking the treble details in the direct sound.



Had the very same impression, and can offer a potential explanation other way ´round: although they offer a narrow directivity, it is pretty uneven in the midrange. If you listen at lower SPL, the reverb in the room is dominated by the frequency band 1-2K for which our ear is pretty sensitive, so this midrange/presence-heavy reverb is capable of masking treble details, creating a muddled tonality. Once you crank it up, the treble gets decreasingly masked and sounds more detailed and dynamic again.



It IMHO has nothing to do with dynamic vs. horn vs. planar principle, but with dominating frequency bands in the reverb due to bands offering overly low directivity index.

A neutral reverb with no dominant midrange, is what some mixing engineers working at low SPL are preferring, and many of them are enthusiastic about Kii Audio offering similar properties of excellent clarity at low levels (which use very conventional dynamic drivers but achieve the aforementioned properties by cardioid design).

I will say that over the long term (40 years) I have tended to strongly prefer wide dispersion designs... and whether this affects (how this affects?) the "aliveness" at low volumes, is an interesting question.

Thank you for positing an interesting reasoning.

 

[007Shorty]

1. Inefficient speakers perform poorly at low volumes. Can someone explain this concept since it makes zero sense to me.

There are many reasons why a loudspeaker can have poor efficiency (crossover, driver, design, etc.). A general statement like this is most likely wrong.

My hypothesis for good audio reproduction at low levels is the use of drivers with the lowest possible RMS values (Thiele/Small parameter for mechanical resistance/friction of the driver's suspension (i.e., 'lossiness'), in N s/m).

 

[Arindal]

I have tended to strongly prefer wide dispersion designs... and whether this affects (how this affects?) the "aliveness" at low volumes, is an interesting question.

My guess would be: if the wide dispersion is relatively linear over several frequency bands (like with a wide-baffle design) and early reflections from the sides are not dominant, chances are high this will contribute to excellent behavior at lower SPL. It is easier to take such evenly wide-dispersing speakers, adjust listening distance and wall treatment, than finding more narrow ones which keep their directivity properly constant in midrange and presence.

I encourage everyone to verify my hypothesis, which is naturally supported by anecdotal evidence mostly. I found it interesting that vastly different loudspeaker concepts are listed as examples of excellent dynamics and treble transparency at low SPL, varying from electrostatic planars over dipoles, fullrange horns, hybrid horns with very big midwoofers, cardioids, line sources, broad baffles. They have very little in common except the unusually high midrange directivity. There are a few exceptions from my theory, though, but that might be explained by broader brillance radiation pattern (some recording engineers name certain B&W and PMC models as examples).

the use of drivers with the lowest possible RMS values (Thiele/Small parameter for mechanical resistance of the driver's suspension (i.e., 'lossiness'), in N s/m)

And what do you consider to being ideally low Rms?

Just asking because I stumbled across some drivers with pretty high Rms being used for cardioids and compact subwoofer designs which are not really known for bad behavior at low SPL.

 

[007Shorty]

And what do you consider to being ideally low Rms?

Just asking because I stumbled across some drivers with pretty high Rms being used for cardioids and compact subwoofer designs which are not really known for bad behavior at low SPL.

Large woofers typically have higher Rms values than small ones. Therefore, Rms values can only be compared within a single size class. Rms is calculated from Qms and various other parameters. Qms must be high to achieve low Rms values, and Qms is independent of the size class. I would say a Qms value of >5 is very good, and >10 is excellent.

I'm not sure to what extent this principle can be applied to tweeters.

 

[Chris A]

IMD (amplitude and frequency modulation) is directly impacted by excursion, yes, but it’s independent of sensitivity or input power.

This is an interesting assertion.

1) Could you explain how cone/dome diaphragm excursion is not a function of input power, given a motor design?

2) How can you achieve higher sensitivity besides increasing the driver's physical diaphragm area...or the number of windings in its voice coil?

That [i.e., IMD and thermal compression distortion] is not what the scientific data shows for high performance drivers.

Really? Can you point me to a specific case that you're asserting here (i.e., modulation distortion and thermal compression distortion)?

Chris

 

[Heinrich]

I encourage everyone to verify my hypothesis, ...



And what do you consider to being ideally low Rms?

Just asking because I stumbled across some drivers with pretty high Rms being used for cardioids and compact subwoofer designs which are not really known for bad behavior at low SPL.

I like your approach to consider Fletcher/Munson and masking for evaluating the merits of certain dispersion characteristics. That might need further, if you will clinical investigation. For now it is speculative, and that is utterly o/k as a scientific tool to spark progression.

But, do we see a problem, where does it come from? We may consider that the hypothetical problem statement originates in an urban legend circulated for ages in certain groops. We interprete quite freely, with high effort though, what the word 'performs poorly, if ...' might mean to these people. In parts you take over the associative, in contrast to definitive, wording. I dismiss this approach. We've seen so many fails when trying to apply rigorous logic to such sloppy blanket statements. From my experience it is a waste of time and effort.

 

[Heinrich]

My hypothesis for good audio reproduction at low levels is the use of drivers with the lowest possible RMS values (Thiele/Small parameter for mechanical resistance/friction of the driver's suspension (i.e., 'lossiness'), in N s/m).

With courtesy of Sir Karl Popper, you've got a hypothesis, please state a method to falsify it. Why do we circulate wild assumptions, that sometimes feel a little bit naive, and not prove them? We do that for ages, always asking others to think and do. The RMS is a DIY thing, right. So, do it yourself? (Not meant to be rude, maybe strict? Sorry.)