Speaker dynamic range at the listening position

[sceptical1]

How can level matched AB listening tests be valid for speakers if the inverse square law rule is sidestepped as in the case of a point source vs line source or any variation where the power response (just a guess?) differs considerably at distance.
Even measuring at the listening position gives an unfair advantage to one or the other, if so which one is favored? The speaker with superior dynamics will on average sound louder playing music and therefore be preferred because the assumption is that louder is better, the very reason for making an attempt to equalize volume during comparisons.

I doubt that trying to normalize a subjective evaluation to the extent that it could be considered scientific proof is frought with insurmountable problems. That is, it can't be done.

 

[kyuu]

You're guessing there's a difference in "dynamics" (please define what you mean by this) at the listening position between speakers of different types, but you don't know. And this guess makes blind testing invalid.

Ok.

Even if we assume your premise is true for the sake of argument, that would seem to be a legitimate reason to prefer one speaker versus another.

 

[levimax]

You're guessing there's a difference in "dynamics" (please define what you mean by this) at the listening position between speakers of different types, but you don't know. And this guess makes blind testing invalid.

Ok.

Even if we assume your premise is true for the sake of argument, that would seem to be a legitimate reason to prefer one speaker versus another.

I think I get his question which is if 2 speakers are level matched at the LP with pink noise but if the speakers differ markedly in directivity would this change the perceived loudness with actual music playing and make one speaker perceived to be "louder" and therefore preferred when not really better? In essence is directivity a confounding variable? I don't know the answer but I would think this is discussed in Dr. Tooles book?

 

[kyuu]

I think I get his question which is if 2 speakers are level matched at the LP with pink noise but if the speakers differ markedly in directivity would this change the perceived loudness with actual music playing and make one speaker perceived to be "louder" and therefore preferred when not really better? In essence is directivity a confounding variable? I don't know the answer but I would think this is discussed in Dr. Tooles book?

Considering that speakers with wider directivity are generally preferred in studies of preference the premise seems unlikely. as narrow directivity would necessarily result in more of the energy going directly, more or less, to the listener. But since most of the off-axis energy will likely reach the listener's ears as reflections anyway, I don't think it actually makes much of a difference. An exception might be outdoor settings, I suppose. But it's already common knowledge and practice to use speakers with controlled directivity for outdoor settings I believe.

I don't see why the "dynamics" would be any different between two speakers whose average SPL is level matched at the listening position in any circumstance, however, at least due directly to different directivities.

 

[boxerfan88]

Just sharing my opinion ...

How can level matched AB listening tests be valid for speakers if the inverse square law rule is sidestepped as in the case of a point source vs line source or any variation where the power response (just a guess?) differs considerably at distance.

1. I think it can be valid if the pink noise average SPL (be it dBA or dBC) levels can be matched to be very close.
2. Once the above is done, different speaker type will interact with the room differently and the direct/reflected sounds may combine differently at the listening position, however the SPL will be almost the same because of #1.

Even measuring at the listening position gives an unfair advantage to one or the other, if so which one is favored? The speaker with superior dynamics will on average sound louder playing music and therefore be preferred because the assumption is that louder is better, the very reason for making an attempt to equalize volume during comparisons.

Because of #1, they are "level matched" therefore good enough for AB listening test.
Now, if we turn our attention to dynamics, again because of #1 the "average SPL" should be almost the same, however, the "peak/instantaneous SPL" may differ because of different speaker's "dynamic capability".

I doubt that trying to normalize a subjective evaluation to the extent that it could be considered scientific proof is frought with insurmountable problems. That is, it can't be done.

Maybe so, yet I still think AB listening test is still worth trying, for the experience or learning or even for fun.

And as said by someone somewhere before, doesn't matter what speaker is in play, the room will be the OverLord

/

 

[Francis Vaughan]

One needs to define dynamic capability. As discussed there doesn’t seem to be any sort of objective measure that this means.
Actual dynamic range of a loudspeaker can be affected by a range of transient phenomena. The main one is voice coil heating. Better drivers are generally less affected. You see this tested. Smaller loudspeakers generally are affected more, as it is harder to get rid of the heat. Higher efficiency speakers can be less affected as they don’t need as much power and thus dissipate less energy as heat for the same loudness.
Any other level based changes to performance are likely evidence of serious design flaws. Directivity changes with loudness might result from cone breakup. Not impossible, but not something one would expect in any modern speaker costing more than a few dollars.

Eventually any loudspeaker ceases to be able to play above physically possible limits. Before they get there, most will start to run into significant non-linear behaviour. We don’t tend to listen to them at these levels.

Loudspeakers of sensible design and construction, used within their design limits, won’t have any “dynamic capability” variations.

The question of how to normalise loudspeakers of fundamentally different radiation patterns, with point versus line at an extreme, is a different matter. Neither speaker should be expected to have changes in pattern dynamically. Not unless something is badly wrong. They should be expected to deliver energy into the space they are in linearly with input level. At any given listening position, the relative loudness between speakers of different designs should be expected not to vary, no matter what the absolute levels are. The actual sound field and how it works in the space is another matter. But unless something is broken, dynamic changes should be one factor that doesn’t matter.

 

[gnarly]

One needs to define dynamic capability. As discussed there doesn’t seem to be any sort of objective measure that this means.
Actual dynamic range of a loudspeaker can be affected by a range of transient phenomena. The main one is voice coil heating. Better drivers are generally less affected. You see this tested. Smaller loudspeakers generally are affected more, as it is harder to get rid of the heat. Higher efficiency speakers can be less affected as they don’t need as much power and thus dissipate less energy as heat for the same loudness.

Good post, thx.
Fully concur we need to define dynamic capability. The definition and design goal I use for my DIY speaker hobby, is having unclipped and uncompressed headroom for all peak transients above average SPL.....across the entire spectrum. So pretty much an engineering definition than I'm strongly convinced pays off audibly.
So I've been studying, and testing/measuring, my speakers ability to have such headroom.

Voice coil heating, how quickly it occurs as a driver's average SPL is increased, has been a bit of a surprise. So has the resulting rise in impedance, which reduces SPL.
I think the relatively short duration sine sweep compression tests like seen here on ASR and Erin's, are a good test for this......especially if multiple runs are observed.
The driver's impedance rise also requires that the amp powering it have the increased voltage capability needed, to keep from clipping. Funny how it's not so much a matter of the amp running out of power in terms of current), but being a matter of insufficient rail voltage.

When I look at most home audio speakers and the amps driving them, it becomes apparent how often they have to be compressing and clipping rather majorly, in frequency regions across the spectrum.....when being used at the higher average SPL they are supposedly capable of. Particularly so for passives.
I think it's a considerably bigger issue than commonly thought.

 

[JeremyFife]

As I understand the question: if you level match signals at the speaker binding posts (by voltage) then the measured levels at the listening position may still be different - and the louder speaker has an advantage.

Surely that's common, if speakers being compared have different sensitivities? and commonly controlled for ... have I misunderstood?

 

[levimax]

As I understand the question: if you level match signals at the speaker binding posts (by voltage) then the measured levels at the listening position may still be different - and the louder speaker has an advantage.

Surely that's common, if speakers being compared have different sensitivities? and commonly controlled for ... have I misunderstood?

For speakers you can't "match voltage" due to different efficiencies as you mentioned so you have to match SPL at the LP usually with pink noise. Not as neat and clean as measuring a DAC's output voltage.

 

[sceptical1]

I may have meant sound power instead of power response, here is an explantion:
https://community.sw.siemens.com/s/article/what-is-sound-power

I want to know if some speakers can output a higher peak level than what is assumed to be normal according to the inverse square law or is a small 2 way maybe attenuating by more than 6db per doubling of distance. Something that explains why some speakers sound lively. I use a line array as an example simply because there can be close to no attenuation.
Dynamic range in this case has nothing to do with maximum undistorted output, compression or sensitivity. I mean something easily heard at low levels like 60 or 70db.

 

[Francis Vaughan]

As you note, a line array does not fall away with inverse square. Where you get into trouble is that in the real world with the huge range of wavelengths involved, the actual radiation pattern of any loudspeaker is frequency dependent. So in the end you need to design the loudspeaker to work for you at the listening position. Moreover it must work to deliver energy into the room so that reflected sound also has a sensible frequency response. It isn’t just a matter of the direct sound.

This is a problem with loudspeakers in general. Not just a line array, but any speaker with multiple drivers for the same frequency range will have higher on axis response if they are further apart than say a quarter of a wavelength. Line arrays are an extreme form.

In the end all that is happening is that the on axis efficiency is higher, and the directivity different. So even if the direct frequency response is flat, there can be issues with the reflected energy and possibly the overall apparent frequency response is not flat. Which is why the spinorama is such a useful tool.

The manner in which sound is radiated won’t change with loudness. So even if there is some perceived difference at low levels it isn’t directly due to the radiation patterns.

Any apparent liveliness in the sound is likely due to frequency response anomalies. An enormous amount of subjective character ascribed to loudspeakers comes down to nothing more than this. At low levels of listening our ears intrinsic variation in response with loudness comes into play, which confounds a lot of subjective impressions.

 

[levimax]

As you note, a line array does not fall away with inverse square. Where you get into trouble is that in the real world with the huge range of wavelengths involved, the actual radiation pattern of any loudspeaker is frequency dependent. So in the end you need to design the loudspeaker to work for you at the listening position. Moreover it must work to deliver energy into the room so that reflected sound also has a sensible frequency response. It isn’t just a matter of the direct sound.

This is a problem with loudspeakers in general. Not just a line array, but any speaker with multiple drivers for the same frequency range will have higher on axis response if they are further apart than say a quarter of a wavelength. Line arrays are an extreme form.

In the end all that is happening is that the on axis efficiency is higher, and the directivity different. So even if the direct frequency response is flat, there can be issues with the reflected energy and possibly the overall apparent frequency response is not flat. Which is why the spinorama is such a useful tool.

The manner in which sound is radiated won’t change with loudness. So even if there is some perceived difference at low levels it isn’t directly due to the radiation patterns.

Any apparent liveliness in the sound is likely due to frequency response anomalies. An enormous amount of subjective character ascribed to loudspeakers comes down to nothing more than this. At low levels of listening our ears intrinsic variation in response with loudness comes into play, which confounds a lot of subjective impressions.

What do you think of the theory that drivers with high Qms are more "lively" at low levels? This was mentioned by the Ascilabs designer.

 

[Francis Vaughan]

Impossible to say. What are the measurable parameters that control apparent liveliness? What it is a listener hears as liveliness?

A higher Qms would be expected to dictate a different Qes and/or enclosure to yield a good overall response. Maybe a higher Qms is a proxy for a set of choices made in the driver design. It suggests less mechanical damping. Such choices might show up with additional stored energy and concomitant frequency response anomalies. But maybe if they are benign enough they just add a little character to the speaker.

But why only at low levels? For a small loudspeaker higher levels might mean that there is some effect from voice coil heating. Does the alignment change enough at different power levels that a higher Qms ameliorates something when heating is not a factor? Lots of possible competing second order effects.

This is possibly one of the rare cases where the spectral decay is more revealing of behaviour. Whilst the information is in the basic frequency response, sometimes a different presentation makes something more obvious.

I really doubt there is a useful blanket rule about Qms.