Loudspeaker Sensitivity

[Michael John]

Does anyone have solid experience in measuring loudspeaker sensitivity?

Manufacturers often provide sensitivity plots, as a function of frequency (1W/1m), and a single sensitivity number, say 98 dB 1w/1m. The frequency plot makes sense but it's not clear to me how the single number is derived. It's appears that some manufacturers pick the peak of the plot whilst others do something else.

I'm currently reading through a copy of IEC60268-5 and it seems to suggest a number of different methods. If someone can jumpstart me, I'd be most appreciative.

 

[Ivan Beaver]

Re: Loudspeaker Sensitivity

It really depends on what you are trying to show-HOWEVER there are also some "catches".

Let's say you measure the freq response a loudspeaker-basically a full range one. And your measurement is calibrated.

One of the things to consider is that the low freq cutoff (typically -3dB) HAS to be tied to the sensitivity-or it means almost nothing.

Some manufacturers like to state these differently-but that is VERY deceiving as to the actual performance.

Let's say that the "average" is 100dB. ie a line drawn through the middle of the response curve. But there is a peak at 1Khz that is 105dB. Some would say the sensitivity is 105dB. BUT if you do that-the point of -3dB would HAVE to be 102dB-which is higher than the average-so there is no "-3dB"

So to find the real -3dB you look for the points on the response graph that are 97dB on each end of the response. (100-3=97dB)

That would be your low freq number that "should" be published. Now one decision to make is-if you want the sensitivity to be "higher"-then you HAVE to accept a higher low cutoff and lower high cutoff.

But if you want to extend the freq response-then you have to state a lower sensitivity-AND maximum ouput (assuming you want the upper and lower freq to be part of the sound output).

One of the "games" that some manufacturers like to play is as follows. Let's keep the same 100dB "middle" of the response. Now state the freq response as +/- 3dB. Now that gives a 6dB window. So if you subtract 6dB from 100dB-then that gives a "flat response that goes down to 94dB (100-6=94dB)

NOW let's give a -3dB number-which would be at 92dB-which is actually NINE dB below the stated sensitivity!!!!!!!!!!!!!!!!

So that freq is MUCH lower in level than you would expect-based on the "simple numbers".

If you don't believe me-just go to some websites and look at the stated -3dB and look at their measured responses and see for your self.

Not everybody does this-but a good number do-you might be surprised of some of the "big names" that do this. I know of some sites that the stated -3dBis actually 15-20dB down from sensitivity. WOW I would love to see that math.

I assume the manufacturer is hoping that people will not look at the graph and just take the stated numbers as fact. NOT.

By looking at the measured data-you can come up with own numbers as to what you feel they should be. But that only applies if the response graph is supplied with a dB scale.

If only a 0 (zero) dB is the reference is supplied-then you have no idea what the "reference" is. Is it the stated sensitivity? How do you know? Did you just "assume" that? What if the sensitivity number is actually referenced to a peak on the response?

Sometimes you have to dig a bit deeper for the REAL answer.

 

[Michael John]

Re: Loudspeaker Sensitivity

Thanks Ivan. Very useful. I'd hoped you would see my post.

IEC60268-5 talks about the effective frequency range being the frequency points 10 dB below the average pressure (at least an octave around the most sensitive part of the spectrum). This all being done with a sine tones or a sweep. Note that the effective range is different to the rated frequency range a manufacturer may choose.

Separately, the standard talks about calculating sensitivity, for a rated or chosen frequency range, using either pink noise (filtered with 24 dB/oct HPF and LPF to the chosen frequency range) OR integrating 1/3 octave noise measurements in the chosen range. In either case the single number is referred back to 1W/1m. Either of these methods seem reasonable in that they avoid picking the highest point on the graph.

From your post, I presume all the numbers you mention are taken from a 1W referred frequency response plot. (e.g.2.83Vrms @ 8ohms.) and this appears different to the IEC60268-5 sensitivity method using pink noise. Is there another standard I should get a copy of? Or does everyone use a 2.83Vrms sine sweep as a defacto standard basis for subsequent calculations?

Thanks,
Michael

 

[Ivan Beaver]

Re: Loudspeaker Sensitivity

Thanks Ivan. Very useful. I'd hoped you would see my post.

IEC60268-5 talks about the effective frequency range being the frequency points 10 dB below the average pressure (at least an octave around the most sensitive part of the spectrum). This all being done with a sine tones or a sweep. Note that the effective range is different to the rated frequency range a manufacturer may choose.

Separately, the standard talks about calculating sensitivity, for a rated or chosen frequency range, using either pink noise (filtered with 24 dB/oct HPF and LPF to the chosen frequency range) OR integrating 1/3 octave noise measurements in the chosen range. In either case the single number is referred back to 1W/1m. Either of these methods seem reasonable in that it avoids picking the highest point on the graph.

From your post, I presume all the numbers you mention are taken from a 1W referred frequency response plot. (e.g.2.83Vrms @ 8ohms.) and this appears different to the IEC60268-5 sensitivity method using pink noise. Is there another standard I should get a copy of? Or does everyone use a 2.83Vrms sine sweep as a defacto standard basis for subsequent calculations?

Thanks,
Michael

I am not aware of ANY manufacturer that uses the 1 watt method.

NOW BEFORE anybody gets excited-many STATE that it is based on 1 watt. But it is NOT. It is based on the voltage at the rated impedance (2.83V for 8 ohm 2 volt for 4 ohm etc) that it was tested at.

If they REALLY put 1 watt into the loudspeaker-then the applied voltage would be all over the place (due to the impedance not being constant)-so the freq response would be even more ragged than it is now.

And besides we DON'T use POWER amplifiers to drive speakers. YES they are called that-but in reality they are just VOLTAGE amplifiers. They take a given input voltage and apply a certain amount of gain to it. The power is the result of the applied voltage into the loudspeaker load.

As long as you stay below the maximum output-then the voltage coming out of the amp will be the same-no matter what impedance loudspeaker is hooked to it.

If it were a true "power amp", then the voltage would change with different impedance loads.

Yes I may be getting "picky", but it is important to realize the differences.

For what it is worth-Danley uses the applied voltage (2.83V) for sensitivity. It doesn't matter the impedance. Because that is what you are applying-voltage NOT power.

And just to throw a little more fuel on the fire. The rated impedance is not always the actual impedance-or an average.

For example the Danley TH115. It is rated at 4 ohms-but the ACTUAL impedance that it should be "rated" at is 5 ohms.

HOWEVER it is amazing how many people are confused by anything other than 2-4-8-16 ohms. So if we calculated the voltage to be applied for a 4 ohm load-and then applied it to a 5 ohms load-the measured output would be lower than if it was calculated for 5 ohms.

Origionally the Sh50 was rated at 6 ohms-because that was what it should be rated at. I was getting tons of calls asking what kind of amp to use-because people could not find a spec for a 6 ohm amp.

So we changed the rated impedance to 4 ohms and the phone calls stopped.

Many people simply don't understand the voltage-power-impedance relationships. And that is part of audio 101. As is being able to read a basic spec sheet and understand what it means.

I am not talking about Heyser spirals here-but a basic response curve and what it really means.

 

[Michael John]

Re: Loudspeaker Sensitivity

Thanks again Ivan.

Yes, I'm aware of the amps being essentially a voltage device, the use of fixed voltage 2.83Vrms (not power) and the use of nominal impedance, rather than average or min. (IEC60268-5 specifically calls this out saying that for a rated nominal impedance, the measured impedance shouldn't drop below 80% of the nominal. In other words if the measured is lower, the rated nominal should be changed.)

Considering the fixed applied voltage of 2.83Vrms; it could refer to the RMS of noise, tones or a tone sweep. From your posts and others sources I read, people often seem to be using a tone sweep and calculating everything from that. However IEC60268-5 specifically talks about using pink noise (albiet band limited to the frequency range of choice) to calculate a single sensitivity number. The two methods mathematically will lead to different results. How different, I don't know.

So I'm wondering is there a IEC or AES standard that specifically describes measuring sensitivity close to the methods manufacturers are using? Or is everyone using a similar and reasonable defacto standard?

Yep, I hear you re the 6 ohms. In some of our cinema software tools, I added the ability to interpolate between an amp's 2/4/8/16 wattage numbers for speaker arrays where the combined parallel/series impedance would be non-standard. Little things like this really help.

I am not aware of ANY manufacturer that uses the 1 watt method.

NOW BEFORE anybody gets excited-many STATE that it is based on 1 watt. But it is NOT. It is based on the voltage at the rated impedance (2.83V for 8 ohm 2 volt for 4 ohm etc) that it was tested at.

If they REALLY put 1 watt into the loudspeaker-then the applied voltage would be all over the place (due to the impedance not being constant)-so the freq response would be even more ragged than it is now.

And besides we DON'T use POWER amplifiers to drive speakers. YES they are called that-but in reality they are just VOLTAGE amplifiers. They take a given input voltage and apply a certain amount of gain to it. The power is the result of the applied voltage into the loudspeaker load.

As long as you stay below the maximum output-then the voltage coming out of the amp will be the same-no matter what impedance loudspeaker is hooked to it.

If it were a true "power amp", then the voltage would change with different impedance loads.

Yes I may be getting "picky", but it is important to realize the differences.

For what it is worth-Danley uses the applied voltage (2.83V) for sensitivity. It doesn't matter the impedance. Because that is what you are applying-voltage NOT power.

And just to throw a little more fuel on the fire. The rated impedance is not always the actual impedance-or an average.

For example the Danley TH115. It is rated at 4 ohms-but the ACTUAL impedance that it should be "rated" at is 5 ohms.

HOWEVER it is amazing how many people are confused by anything other than 2-4-8-16 ohms. So if we calculated the voltage to be applied for a 4 ohm load-and then applied it to a 5 ohms load-the measured output would be lower than if it was calculated for 5 ohms.

Origionally the Sh50 was rated at 6 ohms-because that was what it should be rated at. I was getting tons of calls asking what kind of amp to use-because people could not find a spec for a 6 ohm amp.

So we changed the rated impedance to 4 ohms and the phone calls stopped.

Many people simply don't understand the voltage-power-impedance relationships. And that is part of audio 101. As is being able to read a basic spec sheet and understand what it means.

I am not talking about Heyser spirals here-but a basic response curve and what it really means.

 

[Ivan Beaver]

Re: Loudspeaker Sensitivity

Thanks again Ivan.

Yes, I'm aware of the amps being essentially a voltage device, the use of fixed voltage 2.83Vrms (not power) and the use of nominal impedance, rather than average or min. (IEC60268-5 specifically calls this out saying that for a rated nominal impedance, the measured impedance shouldn't drop below 80% of the nominal. In other words if the measured is lower, the rated nominal should be changed.)

Considering the fixed applied voltage of 2.83Vrms; it could refer to the RMS of noise, tones or a tone sweep. From your posts and others sources I read, people often seem to be using a tone sweep and calculating everything from that. However IEC60268-5 specifically talks about using pink noise (albiet band limited to the frequency range of choice) to calculate a single sensitivity number. The two methods mathematically will lead to different results. How different, I don't know.

So I'm wondering is there a IEC or AES standard that specifically describes measuring sensitivity close to the methods manufacturers are using? Or is everyone using a similar and reasonable defacto standard?

Yep, I hear you re the 6 ohms. In some of our cinema software tools, I added the ability to interpolate between an amp's 2/4/8/16 wattage numbers for speaker arrays where the combined parallel/series impedance would be non-standard. Little things like this really help.

I am not the one to argue with standards-but the nice thing about standards is that there are so many to choose from.

However-regarding using noise-have you ever tried to use noise and set it to 2.83V? With it jumping all over the place-the best you could do would be some sort of visual average-on both the output level and the measuring device.

In my opinion-this could lead to all sorts of "errors" and we know which way this can go.

And how would you relate freq response (-3dB. -10 points) to band limited noise-unless you use a WHOLE LOT of bands? To me (I have not read the "standard") that would mean a different test-which means they (sensitivity and freq response) are no longer tied together-and just opens up another can or worms for incorrect specs.

Regarding impedance-to ME (not necessarily the opinions of others) the impedance rating should be used to give an idea of the load that "overall" is presented to an amp.

So what if it dips down a little bit at one freq? What if the rest of the curve is well above the rated impedance. I feel it should be more of an "average" across the intended freq range of usage-unless it get really low for a wide are of response. No I do not want to begin to put any "numbers" to that statement.

Let's look at example you give. Let's say a loudspeaker is called 8 ohms and there is a little dip 1/3 oct wide (how much resolution is used in the measurement will affect what the graphs looks like also) that is 6 ohms (less than 80% of the rated impedance) but the average impedance is more like 9 ohms.

Are they saying it should be rated at 4 ohms? That number would not be anywhere near close to the average load presented to the amp or give the user an idea of how many speakers they could run off of an amp channel.

It is MY opinion (again not that of others) that specs should not be a "contest" but rather give the user a REAL WORLD idea of what to expect from the particular product.

And this goes beyond speakers-look at some amplifiers (particularly in powered speakers). Some rate the power as a 50ms test burst. That is ONE CYCLE at 20Hz or TWO cycles at 40Hz!!!!!

Is that real? When was the last time you had any type of material that had 1 or 2 cycles in it?

NO-REAL music is longer-sometimes VERY much longer in duration. So that particular amp cannot produce its rated output and the level drops down quite a bit-as much as 6dB or more. So the use of the very short tone bursts is nothing but a "game" that some manufacturers use to get higher numbers-that the end user expects to be able to get-yet cannot. So what that means is that the maximum output of the loudsepaker woudl have to be 6dB lower-now who would buy that?

But it the loudspeaker can produce a 1 or 2 cycle tone at a particular SPL, then that is "considered" to be the "peak" output. YEAH-that is real USABLE-NOT! Just try putting a normal SPL meter up to it and drive it as hard as you want-and see if you can measure anywhere near the ratings. To me-you should be able to do that-but I am just weird in that way.

Yes I do agree that modern amps are not the continuous output types I grew up with-that is fine-but the short term peaks have to be long enough to reproduce a "normal" (that needs to be defined and everybody has different opinions on that) audio signal.

There is at least one loudspeaker manufacturer who has a stupid high number that is basically a "Maximum physical output" and as far as I can tell they give an SPL number that the loudspeaker will produce one a one cycle tone right at the limit of destruction.

Is that real? NO-but are they lying-I would assume not-but they are HIGHLY misleading the user as to how loud of a SPL they can get out of the loudspeaker in a real world situation.

I try to stay in the real world and not live in fantasy land-but that is probably because I am to honest and was a user for decades before I got into install and manufacturring.

So I want to know the REAL WORLD results I can expect out of a particular product-NOT some special test procedure that gives some special number-What about ME! and MY gigs-can I get the type of numbers the manufacturers provide-THAT is what is important to me.

And that is the type of numbers I try to present on the spec sheets I write.

And I will stand behind any of the numbers I put on a spec sheet-even though they are as "good" as some other products-at least you (the average Joe) can reproduce them in your own back yard. And that is what is important to me.

Getting off soap box now.

 

[Phil Graham]

Re: Loudspeaker Sensitivity

Does anyone have solid experience in measuring loudspeaker sensitivity?

Manufacturers often provide sensitivity plots, as a function of frequency (1W/1m), and a single sensitivity number, say 98 dB 1w/1m. The frequency plot makes sense but it's not clear to me how the single number is derived. It's appears that some manufacturers pick the peak of the plot whilst others do something else.

I'm currently reading through a copy of IEC60268-5 and it seems to suggest a number of different methods. If someone can jumpstart me, I'd be most appreciative.

The proper way to do this is integrating over the RMS behavior in the bandpass of interest with the stated stimulus of record. I've got a resource on this that I can share if you send me a PM.

Even then it's not that simple, bc the axial sensitivity is not representative of the overall power response. I personally feel that Ivan's way of doing things, while not to a "standard" is a very reasonable course of action.

I would ask, what is your specific design intent? The sensitivity number is a pretty weak one to use from an engineering standpoint (which is where I know you're coming from).

 

[Michael John]

Re: Loudspeaker Sensitivity

Hi Phil,

Thanks.

The proper way to do this is integrating over the RMS behavior in the bandpass of interest with the stated stimulus of record.

Yep.

Even then it's not that simple, bc the axial sensitivity is not representative of the overall power response. I personally feel that Ivan's way of doing things, while not to a "standard" is a very reasonable course of action.

I completely agree.

I would ask, what is your specific design intent? The sensitivity number is a pretty weak one to use from an engineering standpoint (which is where I know you're coming from).

Good point. I should have mentioned this up front...

When evaluating cinema system designs, we're using a combination of calculations, using data from spec sheets, and judgement calls based on past experience with different types/brands of gear. It's clear there are differences between manufacturers as to how they derive their numbers and I'm trying to get an idea of the exact (and varied) methods in use in the industry so that we're in a better position to argue for or against a piece of gear that is borderline. IEC60268-5 is the only standard I'm aware of, but it appears people aren't quite following it's methods.

 

[Michael John]

Re: Loudspeaker Sensitivity

Hi Ivan,

Thanks again....

..... regarding using noise-have you ever tried to use noise and set it to 2.83V? With it jumping all over the place-the best you could do would be some sort of visual average-on both the output level and the measuring device.

That's right, and the standard describes time averaging.

And how would you relate freq response (-3dB. -10 points) to band limited noise-unless you use a WHOLE LOT of bands? To me (I have not read the "standard") that would mean a different test-which means they (sensitivity and freq response) are no longer tied together-and just opens up another can or worms for incorrect specs.

Right again. The standard does optionally describe integrating 1/3 octave banded noise measurements, if you have them already, to create the single sensitivity number. As to the single noise method causing the sensitivity and freq response to be unrelated; I completely agree with you. I thought it odd when I read it..

Best,
Michael

 

[Ivan Beaver]

Re: Loudspeaker Sensitivity

And it no wonder there is so much confusion in audio.

One thing I find interesting is that even with a graph-people have no idea what it means.

On a different forum there was a guy who was arguing that a particular loudspeaker would easily reproduce 20Hz. His argument was that since 20Hz was shown on the graph-and the freq response line went down (yeah 40dB down) to 20Hz-that was "good".

Why is a simple graph so hard to read? Apparently it is-based on the number of people I have meet that have no idea how read it.

 

[Michael John]

Re: Loudspeaker Sensitivity

For anyone who hasn't left due to boredom...

I just read through AES2-2012. (It's referenced widely for loudspeaker power measurements - mainly the 2 hours without a significant change in parameters.)

It only has 3 small paragraphs on sensitivity and describes calculating broad-band sensitivity using a test signal of 2-octave wide filtered pink noise centered at 1 kHz, or 2-octave wide filtered noise centered around the mean of the upper and lower frequency limits. (The filtering is not specified but from other parts of this standard and IEC60268-5, it's probably 24dB/oct Butterworth HP and LP filters.)

So at least AES2 and IEC60268-5 are consistent.

 

[Phil Graham]

Re: Loudspeaker Sensitivity

Hi Phil,

When evaluating cinema system designs, we're using a combination of calculations, using data from spec sheets, and judgement calls based on past experience with different types/brands of gear. It's clear there are differences between manufacturers as to how they derive their numbers and I'm trying to get an idea of the exact (and varied) methods in use in the industry so that we're in a better position to argue for or against a piece of gear that is borderline. IEC60268-5 is the only standard I'm aware of, but it appears people aren't quite following it's methods.

Michael,

Is this for providing feedback to integrators that are doing design/builds of multiplex cinema-type spaces? Like an application support group for your cinema processor customers?

You might be able to incent your loudspeaker suppliers to subject themselves to some particular battery of tests on their loudspeaker in exchange for some sort of "Dolby approved" licensure. Heck you may even be able to drive revenue from that by charging a maintenance fee.

Just a thought.

 

[Michael John]

Re: Loudspeaker Sensitivity

Thanks very much. Yep, we've had a film services organisation supporting film production around the world for many, many years and I'm not privy to their business models. I'm currently supporting them from the Research side...

Michael,

Is this for providing feedback to integrators that are doing design/builds of multiplex cinema-type spaces? Like an application support group for your cinema processor customers?

You might be able to incent your loudspeaker suppliers to subject themselves to some particular battery of tests on their loudspeaker in exchange for some sort of "Dolby approved" licensure. Heck you may even be able to drive revenue from that by charging a maintenance fee.

Just a thought.

 

[Don Boomer]

Re: Loudspeaker Sensitivity

Hey Michael

I'm not clear ... are you wanting to measure a "driver" or "speaker system"? If the latter then mic position can make a big difference. Manufacturers don't often specify the mic placement. Unless specified the mic is usually placed on-axis and in front of the HF driver. But sometimes it's taken on the "design axis", typically half way between the woofer and tweeter. In this case the distance of 1 meter is likely a lot closer than anyone is actually gonna listen to the box and there can be big differences (peaks and dips) with only a small change in the vertical position of the mic.

 

[Michael John]

Re: Loudspeaker Sensitivity

Thanks Don. Yes, the latter. From what I've gathered, it's common to measure beyond 1 metre (to minimise these effects) and refer the measurements back to 1 metre.

Hey Michael

I'm not clear ... are you wanting to measure a "driver" or "speaker system"? If the latter then mic position can make a big difference. Manufacturers don't often specify the mic placement. Unless specified the mic is usually placed on-axis and in front of the HF driver. But sometimes it's taken on the "design axis", typically half way between the woofer and tweeter. In this case the distance of 1 meter is likely a lot closer than anyone is actually gonna listen to the box and there can be big differences (peaks and dips) with only a small change in the vertical position of the mic.

 

[Ivan Beaver]

Re: Loudspeaker Sensitivity

Thanks Don. Yes, the latter. From what I've gathered, it's common to measure beyond 1 metre (to minimise these effects) and refer the measurements back to 1 metre.

The reference is 1M-but if you actually measure at that distance-there are all kinds errors that can come into the measurement. And usually in a "good way" for the spec and a "bad way" for the designer trying to use that spec in a prediction.

When you measure to close-the low freq is going to measure higher in level and lower in extension than it would if you measure further away.

This is due to physical size of the cabinet acting as a boundary-kind of like the proximity effect of a mic. Use it close to your mouth and the bass is louder.

So the manufacturer will like the results-but if you are trying to use the spec for some useful information-such as predicting what the SPL at a particular freq will be at a particular distance.

Then there is also the discussion of "what is 1M"? Sure we know the distance-but measured from where?

I have heard (but cannot prove) that at least one manufacturer used to measure from the throat of the HF driver.

This can be a real problem-like with our J4. The HF drivers are so far back in the horn-that they are about a meter before they get to the front of the cabinet.

So at 1M-the drivers are actually 2M away-or 6 dB is already lost due to distance.

If you measure at a further distance and then "back calculate", you can minimize the effect of the depth of the cabinet.

So where you measure-and how you measure can really affect the result-and what the manufacturer "wants" out of the spec. Data that designers can use or "better specs".

The only way to be sure is to measure the cabinets in question yourself-under the SAME conditions.-but even then that can cause some errors.

It is one thing to measure-and quite another to do it right and get real usable accurate results.

 

[Brian Oppegaard]

Re: Loudspeaker Sensitivity

And this goes beyond speakers-look at some amplifiers (particularly in powered speakers). Some rate the power as a 50ms test burst. That is ONE CYCLE at 20Hz or TWO cycles at 40Hz!!!!!

Is that real? When was the last time you had any type of material that had 1 or 2 cycles in it?

NO-REAL music is longer-sometimes VERY much longer in duration. So that particular amp cannot produce its rated output and the level drops down quite a bit-as much as 6dB or more. So the use of the very short tone bursts is nothing but a "game" that some manufacturers use to get higher numbers-that the end user expects to be able to get-yet cannot. So what that means is that the maximum output of the loudspeaker would have to be 6dB lower-now who would buy that?.

Testify! It seems like the misleading specs have gotten worse lately with the trend toward smaller power supplies. 2 cycles of 40Hz tone burst do sound like thump thump thump, but disco is dead and modern electronic music as you note has many long passages of high level low frequency almost sine waves. Anything shorter than 40ms is a bogus spec IMHO and if you really want to shake em til you break em you have to have at least several seconds of full power on tap. At AES a few years ago I suggested three power specs for an amplifier instead of just one, corresponding to three time intervals. It was called 4 4 4 in homage to a (briefly) leading political candidates slogan. It stands for the power delivered for 40 milliseconds (the minimum significant hold time), 4 seconds (a crescendo), and 4 ever (long term continuous power until thermal). I have this spec on my web site for my amps and few competitors. Comparative Performance - SpeakerPower

 

[Don Boomer]

Re: Loudspeaker Sensitivity

From what I've gathered, it's common to measure beyond 1 metre (to minimise these effects) and refer the measurements back to 1 metre.

Yep if you have a really big anechoic chamber. I think the point in all of this is that without the manufacturer specifying how measurements are made you can never assume apples to apples.

 

[Phil Graham]

Re: Loudspeaker Sensitivity

This is due to physical size of the cabinet acting as a boundary-kind of like the proximity effect of a mic. Use it close to your mouth and the bass is louder.

Pedantic technical note here. The proximity effect of a pressure gradient microphone is due to the variance in levels applied across the gradient. In the far-field, where the pressures on both sides are basically the same, there is a natural roll off of the microphone. When the manufacture compensates for this rolloff, and then the performer gets dramatically closer to one side of transducer, the pressure is no longer nearly the same. It is in this context that the compensation for the mic's inherent rolloff with a farfield source reveals itself as the proximity effect.

 

[Ivan Beaver]

Re: Loudspeaker Sensitivity

Pedantic technical note here. The proximity effect of a pressure gradient microphone is due to the variance in levels applied across the gradient. In the far-field, where the pressures on both sides are basically the same, there is a natural roll off of the microphone. When the manufacture compensates for this rolloff, and then the performer gets dramatically closer to one side of transducer, the pressure is no longer nearly the same. It is in this context that the compensation for the mic's inherent rolloff with a farfield source reveals itself as the proximity effect.

Exactly the reason I said "Kind of like"-meaning the "effect" is similar-but not "achieved" the same way-meaning that up close the bass is louder.

As usual-once you dig a bit deeper-it gets more complicated.