[music_man]

A loudspeaker is rated 30-25,000 +/- 6db. The same speaker is rated at 45-20,000 +/- -3db.
I am half asleep but the math does not seem to add up. 6 down is not that much difference.
Other than I do not trust anyone that provides specs only of more than 3db.
They do, so good there but it does not seem to be making sense.
On the other hand I can barely see right now. Going to bed check your answers in the morning.
Thank's

 

[castleofargh]

not sure there is a problem here. if both specs come from the same conditions using the same target, then the underlying message is that the deviations grow bigger once outside of the 45hz-20khz range, but are still contained within +/-6dB(which means up to 12 on the entire curve) from 30hz to 25khz. so the 30-45hz and the 20-25khz areas are where those extra +/-3dB of deviation are occurring.
just the typical low end roll off in speakers could be enough to explain this.

 

[music_man]

Thank you, I am sure when I rest up I will get it. Obviously that is at it's limit.

 

[bigshot]

-6dB is a considerable amount. I think most specs are +/-3dB because that is a typical manufacturing tolerance.

 

[amirm]

Other than I do not trust anyone that provides specs only of more than 3db.
They do, so good there but it does not seem to be making sense.

It is a marketing attempt to relax the +-3 dB and show better numbers in low frequencies mostly.

Regardless, they are all junk specs. You need to see the actual graphs and see where those variations are. And those measurements need to cover different axis relative to speaker.

 

[71 dB]

I made an example of a fictional loudspeaker with the given specs:

±3 dB means the response varies 6 dB (making it a plus minus thing gives "smaller" number). ±6 dB means 12 dB variation, but the zero level can and almost always is different. In this example the response varies between -9 dB at 30 Hz to +3 dB at 6.3 kHz. Hopefully this makes sense.

 

[music_man]

yes! thank you. that was really nice to take the time to do that. I feel companies use -6db or even -10db to mask the sad truth. this was said already sort of. anything serious is not rated more than +/- 3db and really zero imo. getting to -10db is such a huge difference in volume that those are just bs companies imo. that is exactly why I look at a graph. if they do not provide one, not a serious company imo.

 

[gregorio]

I feel companies use -6db or even -10db to mask the sad truth. this was said already sort of. anything serious is not rated more than +/- 3db and really zero imo.

Yes, but the reality (or "sad truth" if you want to call it that) is that it's not possible or desirable to create a speaker with an entirely linear response. You cannot make a speaker which is linear down to 0Hz or linear up to infinity kHz and so they all have to start rolling-off somewhere in the frequency spectrum (two "somewheres" in fact, one in the low freqs and one in the high freqs), have -3dB somewhere else, reach -10dB somewhere else and at some point reach minus-infinity (no output). So, the question isn't if a speaker has +/- 3dB or +/-10dB because all speakers must have both (and at both ends of a frequency range), the question is where those variations occur.

G

 

[71 dB]

yes! thank you. that was really nice to take the time to do that. I feel companies use -6db or even -10db to mask the sad truth. this was said already sort of. anything serious is not rated more than +/- 3db and really zero imo. getting to -10db is such a huge difference in volume that those are just bs companies imo. that is exactly why I look at a graph. if they do not provide one, not a serious company imo.

No problem! The graphs we see are actually smoothed versions of the real measured data, because the smoothed curves describe better how our ears hear the sound. The raw data looks like this:

It's pretty wild at higher frequencies, but it's not as bad as it looks, because that's the sonic radiation to one direction only, typically measured at the radiation axis of the treble tweeter. Just 1° off-axis the green "data" is very different. The wild ride of the response curve at high frequencies is due to diffractions (from e.g. the edges of the speaker cabined) so it's sound energy scattering to different angles. At low frequencies diffractions don't happen, because the wavelength is large compared to the speaker cabinet and so the green raw data "calms down."

Loudspeakers have also directivity. They "shoot" the sound mostly forward, toward the listener in the listening spot. At low frequencies this directivity is very small with normal boxed speakers so bass sounds are radiated equally to all directions. When the frequency raises, the wavelength gets shorter and around 1 kHz the wavelength is about the same size as a typical boxed 2-way "bookshelf" speaker. For floor standing 3-way -speakers this happens at a lower frequency. That's way on axis-response should raise up with frequency a few decibels, because off-axis responses are "treble shy." What we hear is a combination of the direct sound, early reflections and reverberation. With typical acoustics the direct sound is just something like 10 % of all the sound energy we hear. That's why listening room acoustics is so important.

Frequency curves don't tell much about sound quality to a layman, but an acoustic engineer immersed to speaker design can spot design defects and predict sonic characteristics. The +/- X dB frequency ranges alone are pretty useless without other knowledge about the speaker and as such are "marketing data" rather than meaningful specs of the products concerning the sound quality. It's like evaluating a house based on it's volume. A small recently renovated house is certainly better to live in than a large empty hall!

Yes, but the reality (or "sad truth" if you want to call it that) is that it's not possible or desirable to create a speaker with an entirely linear response. You cannot make a speaker which is linear down to 0Hz or linear up to infinity kHz and so they all have to start rolling-off somewhere in the frequency spectrum (two "somewheres" in fact, one in the low freqs and one in the high freqs), have -3dB somewhere else, reach -10dB somewhere else and at some point reach minus-infinity (no output). So, the question isn't if a speaker has +/- 3dB or +/-10dB because all speakers must have both (and at both ends of a frequency range), the question is where those variations occur.

G

Because of the directivity issues we want the on axis response to raise a few decibels from bass to treble so that the sound is balanced together with the effect of power response (what the speaker radiates to all directions).

English is not my first language, but I believe "flat" is more preferred term to describe frequency response curves. The term "linear" has association to non-linear distortion. So, linear response is free of non-linear distortion even if the frequency response is curvy as hell in which case the system suffers from linear distortion.

The bold part is on point!

 

[music_man]

I was an audio engineer for 42 years. I should have known my question but I was exhausted. I still am quite frankly. I am getting up there.

I am using unique very high end bookshelf speakers that by design have mostly overcome reflections, proximity, directionality, etc. they literally have the perception of throwing sound 30' behind them. I was not asking about those. I was just concerned companies are trying to show lower frequencies than are actually usable. mainly this came from the sub world. that is wrong too imo. a cubic foot box going to 18hz since they measured it at -10db. it is basically an outright lie. interestingly these speakers are measured at +/- 1db. Go figure.

 

[castleofargh]

I was an audio engineer for 42 years. I should have known my question but I was exhausted. I still am quite frankly. I am getting up there.

I am using unique very high end bookshelf speakers that by design have mostly overcome reflections, proximity, directionality, etc. they literally have the perception of throwing sound 30' behind them. I was not asking about those. I was just concerned companies are trying to show lower frequencies than are actually usable. mainly this came from the sub world. that is wrong too imo. a cubic foot box going to 18hz since they measured it at -10db. it is basically an outright lie. interestingly these speakers are measured at +/- 1db. Go figure.

as Greg explained, the relevance of those specs relies on both the margin and the frequency range covered. we could imagine the crappiest speaker publish specs saying it's accurate withing +/-0.1dB, between 70 and 72hz. ^_^. what's misleading is when a manufacturer pick the range that's convenient for him, because then we can't check with another speaker using specs in a different range of frequencies. they're not comparable data anymore if they don't define the same range.
otherwise, it's a spec like others. informative, to a point.

 

[music_man]

Both were 20-20. I did not at the time realize an extra 3db was that much difference. of course it is +/- 6db is a 12db difference in volume. that is a lot. the companies that rate subs at -10db through their usable frequency range are dishonest. or at least the measurement is not useful. they want their 8" sub to look like it goes to 12hz. impossible at proper volume and just trying to trick people. you will notice the legit companies refrain from this.

 

[bigshot]

Just using an audio editor will quickly allow you to do easy experiments to figure out what numbers relate to in real world sound.

 

[music_man]

Unfortunately I am on my way to return my Engineering sample. It makes sense though. Simple math. Now that I am awake. 12db is a lot. Of course there is a perceived difference in volume there.