I do not see this discussed to any extent so I'd like to open this up.
As we all know, different drivers have different sensitivities. As far as I can tell there are two solutions to this issue if you plan to design an accurate loudspeaker:
1. Pad down the higher sensitivity drivers to match your lowest sensitivity driver
2. Increase the number of lower sensitivity drivers until they match the higher sensitivity drivers
I see the pro/cons of these options as thus:
Option 1
Pros
1. Cheaper: Less drivers means lower cost
2. Smaller: More compact for space savings
3. Simplicity: Less drivers means the design and the response is simpler. It is easier to take measurements.
Cons
1. Loss of sensitivity: Your system sensitivity will only be as high as your lowest driver
2. Potential cost of amplifier: You will need so spend more on your amplifier as your system will be harder to drive
3. Less amplifier options: Depending on your final sensitivity, you selection of amplifier may be cut down significantly
4. Less woofer/mid options if you need a higher sensitivity system
Option 2
Pros
1. Higher sensitivity: you would be able to use whatever amplifier you wanted with a high sensitivity system
2. Driver work: The amount of work the lower drivers will have to perform will but significantly cut down. Your woofers will handle much more power before hitting XMAX
3. Power threshold: The system should be able to handle a lot more power before distortion
4. Less amplifier distortion, less woofer distortion
5. With more driver in vertical axis you should be able to significantly reduce vertical cancellations which should better fill a room
6. It will look super impressive: No doubt, a loudspeaker using redundant drivers that goes up to the ceiling looks far more impressive.
7. You can use any woofers you want as sensitivity can be accounted for using multiple drivers.
Cons
1. Much more expensive: You will need multiple drivers, however, your crossover costs should remain the same
2. More complex build: The enclosure will be much more complex to design and build
3. More complex measurements: I have not personally ever measured redundant drivers but I imagine it isn't as cut and dry as measuring just one.
4. It will take up more space: It is going to be much larger. Geometry is geometry.
I'd like to hear all of your thoughts on these two different approaches as I am considering using multiple drivers to up my sensitivity on my next build. I have an example below that is really a hybrid of the two
I will also throw into this mix a discussion of low frequency sensitivity. Most of us are trying to boost the low frequencies using multiple drivers. Tweeters play loudly, woofers do not. I have heard that what matters most for the low end is cone area and power. The driver sensitivity doesn't matter below 150 hz or so. I do not know if this is true. It does not seem like it is true. Then again, I do not know exactly how sensitivity is measured. I hear it is playing a 1000hz tone with 1 watt then measuring the decibels. Due to peaks, cone resonances, and driver frequency response I would think this would not be a perfectly accurate measurement for all drivers. If someone knows, please post that information as I feel it is critical to this discussion.
Please ignore any errors on this XO and design. It is just an example for flare that I tossed together in a hurry.
As we all know, different drivers have different sensitivities. As far as I can tell there are two solutions to this issue if you plan to design an accurate loudspeaker:
1. Pad down the higher sensitivity drivers to match your lowest sensitivity driver
2. Increase the number of lower sensitivity drivers until they match the higher sensitivity drivers
I see the pro/cons of these options as thus:
Option 1
Pros
1. Cheaper: Less drivers means lower cost
2. Smaller: More compact for space savings
3. Simplicity: Less drivers means the design and the response is simpler. It is easier to take measurements.
Cons
1. Loss of sensitivity: Your system sensitivity will only be as high as your lowest driver
2. Potential cost of amplifier: You will need so spend more on your amplifier as your system will be harder to drive
3. Less amplifier options: Depending on your final sensitivity, you selection of amplifier may be cut down significantly
4. Less woofer/mid options if you need a higher sensitivity system
Option 2
Pros
1. Higher sensitivity: you would be able to use whatever amplifier you wanted with a high sensitivity system
2. Driver work: The amount of work the lower drivers will have to perform will but significantly cut down. Your woofers will handle much more power before hitting XMAX
3. Power threshold: The system should be able to handle a lot more power before distortion
4. Less amplifier distortion, less woofer distortion
5. With more driver in vertical axis you should be able to significantly reduce vertical cancellations which should better fill a room
6. It will look super impressive: No doubt, a loudspeaker using redundant drivers that goes up to the ceiling looks far more impressive.
7. You can use any woofers you want as sensitivity can be accounted for using multiple drivers.
Cons
1. Much more expensive: You will need multiple drivers, however, your crossover costs should remain the same
2. More complex build: The enclosure will be much more complex to design and build
3. More complex measurements: I have not personally ever measured redundant drivers but I imagine it isn't as cut and dry as measuring just one.
4. It will take up more space: It is going to be much larger. Geometry is geometry.
I'd like to hear all of your thoughts on these two different approaches as I am considering using multiple drivers to up my sensitivity on my next build. I have an example below that is really a hybrid of the two
I will also throw into this mix a discussion of low frequency sensitivity. Most of us are trying to boost the low frequencies using multiple drivers. Tweeters play loudly, woofers do not. I have heard that what matters most for the low end is cone area and power. The driver sensitivity doesn't matter below 150 hz or so. I do not know if this is true. It does not seem like it is true. Then again, I do not know exactly how sensitivity is measured. I hear it is playing a 1000hz tone with 1 watt then measuring the decibels. Due to peaks, cone resonances, and driver frequency response I would think this would not be a perfectly accurate measurement for all drivers. If someone knows, please post that information as I feel it is critical to this discussion.
Please ignore any errors on this XO and design. It is just an example for flare that I tossed together in a hurry.
As a standalone statement in a passive speaker, this is incorrect. If you have an active sub/woofer with a separate amp and are OK with using much more power there, it's more reasonable.
Mostly true, but if the driver obviously isn't reproducing that frequency at a reasonable level, it's not going to be spec'd at that frequency. Imagine a tweeter with a 3 kHz resonance. No way its sensitivity is going to be measured at 1 kHz.
I never give speaker sensitivity a second thought. Well almost never. I think speaker sensitivity is mostly a marketing gimmick.
I always want the tweeter to be padded down.
High sensitivity woofers, are great, if they go deep enough, don't cost too much, are low distortion, and don't need a box that is too big.
It's not uncommon to get higher overall performance from a much more expensive speaker. That higher performance may only be realized at high SPL in a big room.
A high sensitivity speaker might have a lower impedance, and draw more power than one that is higher impedance. So, although you could risk clipping an amp, I think an 83dB speaker is sufficient. I don't see a need for a better amp. I have a sub, so the speakers draw much less power, and there's not significant driver distortion at low frequency
A while back, I thought I wanted higher sensitivity for home theater. I talked myself out of it.
I would never add drivers to match a tweeter. It might be an advantage to not need to pad a mid-range, as resistors may get hot. I've not had that issue personally, although I do check for any signs of heat in them.
So I go with option one. I like the pros listed, and the cons listed are not relevant. In fact, con #4 should be listed under option two, not one. Con #2 is saying that it's a con for your sensitivity to be what it is. It's a number. It's not a con that it's a lower number.
I always want the tweeter to be padded down.
High sensitivity woofers, are great, if they go deep enough, don't cost too much, are low distortion, and don't need a box that is too big.
It's not uncommon to get higher overall performance from a much more expensive speaker. That higher performance may only be realized at high SPL in a big room.
A high sensitivity speaker might have a lower impedance, and draw more power than one that is higher impedance. So, although you could risk clipping an amp, I think an 83dB speaker is sufficient. I don't see a need for a better amp. I have a sub, so the speakers draw much less power, and there's not significant driver distortion at low frequency
A while back, I thought I wanted higher sensitivity for home theater. I talked myself out of it.
I would never add drivers to match a tweeter. It might be an advantage to not need to pad a mid-range, as resistors may get hot. I've not had that issue personally, although I do check for any signs of heat in them.
So I go with option one. I like the pros listed, and the cons listed are not relevant. In fact, con #4 should be listed under option two, not one. Con #2 is saying that it's a con for your sensitivity to be what it is. It's a number. It's not a con that it's a lower number.
Last edited:
Hi, datasheet sensitivity (voltage sensitivity) alone doesn’t tell the full story. What matters in practice is system efficiency, particularly at low frequencies, because that’s where most of the power is consumed. In a typical passive speaker with a flat frequency response, a 40W amplifier will mostly be working to reproduce low frequencies, while higher frequencies require much less power.
"Sensitivity" is also affected by the environment—placing a speaker in a corner versus in an open field can change its efficiency because total radiated power is distributed differently.
In general, high-sensitivity systems have large cone areas and big enclosures, or they use horn loading. Small speakers inherently have lower sensitivity, and speakers of similar size and type tend to have similar efficiency if they cover a comparable bandwidth. Basically just due to sound wavelength being huge at lows.
Higher sensitivity generally means lower non-linear distortion at a given SPL, but the key is having "enough sensitivity" for your needs—going beyond that doesn’t always offer practical benefits.
In this sense you'd always want as big of a system as practical. Yeah I mean physically big, big cones, big magnets, everything big. If you need more output and bandwidth from some limited physical size you'd use more advanced transducers, highly optimized box alignments, more power with optimized cooling schemes, and eventually start narrowing the bandwidth. This is public address territory though, sound attenuates with distance means it's got to be very loud at close to reach everyone in a big crowd. Not really a concern at home, because we are always close and there is no crowd.
It is huge subject involving both acoustics and how transducers and amplifiers work. I hope you'll get enough info to be able to get alogn in practice 🙂
"Sensitivity" is also affected by the environment—placing a speaker in a corner versus in an open field can change its efficiency because total radiated power is distributed differently.
In general, high-sensitivity systems have large cone areas and big enclosures, or they use horn loading. Small speakers inherently have lower sensitivity, and speakers of similar size and type tend to have similar efficiency if they cover a comparable bandwidth. Basically just due to sound wavelength being huge at lows.
Higher sensitivity generally means lower non-linear distortion at a given SPL, but the key is having "enough sensitivity" for your needs—going beyond that doesn’t always offer practical benefits.
In this sense you'd always want as big of a system as practical. Yeah I mean physically big, big cones, big magnets, everything big. If you need more output and bandwidth from some limited physical size you'd use more advanced transducers, highly optimized box alignments, more power with optimized cooling schemes, and eventually start narrowing the bandwidth. This is public address territory though, sound attenuates with distance means it's got to be very loud at close to reach everyone in a big crowd. Not really a concern at home, because we are always close and there is no crowd.
It is huge subject involving both acoustics and how transducers and amplifiers work. I hope you'll get enough info to be able to get alogn in practice 🙂
Last edited:
You could use logic to come up with a system, lets make sone requirements first, then you'd go and figure out what kind of a system could manage it reasonably.
A proper playback system should be capable of 20-20kHz bandwidth, right? Perhaps 40Hz - 15kHz is ok, you decide. You'd want it be capable of roughly 80db average SPL at your listening spot, because ear is most linear there, it's comfortably loud and fun level to listen, and about all commercial sound is done at that level so should sound great.
Music you mostly listen might have 15db peaks above average. In addition your listening distance robs some, 6db every doubling of distance. You might like some bass boost as well, a house curve. Factor these in and your bass peaks might be somewhere 110db @ 1m 30Hz or so in a typical livingroom listening scenario, 80db average at the spot. Could be higher, could be lower, you can calculate for your own application.
Now, if you have 2W amp you need a system that can do 110db 1m 30Hz with 2W. This would be huge system, ~105db/1W/1m sensitivity at 30Hz territory. If you have a 1000W amp, you only need ~75db/1W/1m sensitivity at 30Hz, it's very different system although fills your target. Or, you could just change the requirements, the requirements are important here, perhaps little distortion doesn't matter 🙂 Likely small bookshelf speaker doesn't withstand 1000Watts, so perhaps some sensible middle ground, as big as practical like ~100W amp, 15" bass, compression driver, perhaps it's enough for fun system 😉
A proper playback system should be capable of 20-20kHz bandwidth, right? Perhaps 40Hz - 15kHz is ok, you decide. You'd want it be capable of roughly 80db average SPL at your listening spot, because ear is most linear there, it's comfortably loud and fun level to listen, and about all commercial sound is done at that level so should sound great.
Music you mostly listen might have 15db peaks above average. In addition your listening distance robs some, 6db every doubling of distance. You might like some bass boost as well, a house curve. Factor these in and your bass peaks might be somewhere 110db @ 1m 30Hz or so in a typical livingroom listening scenario, 80db average at the spot. Could be higher, could be lower, you can calculate for your own application.
Now, if you have 2W amp you need a system that can do 110db 1m 30Hz with 2W. This would be huge system, ~105db/1W/1m sensitivity at 30Hz territory. If you have a 1000W amp, you only need ~75db/1W/1m sensitivity at 30Hz, it's very different system although fills your target. Or, you could just change the requirements, the requirements are important here, perhaps little distortion doesn't matter 🙂 Likely small bookshelf speaker doesn't withstand 1000Watts, so perhaps some sensible middle ground, as big as practical like ~100W amp, 15" bass, compression driver, perhaps it's enough for fun system 😉
(a bit similar to previous post, but I've already have written)
It all depends on the expected SPL, distance from speakers, room size and minimum low frequency response you need.
With these requirements, then you start to choose the speaker options and engineer them to a reasonable performance (distortion, frequency response, cost, size, impedance etc).
I don't like to start engineering a system by fixing one variable, such as sensitivity or any other.
I prefer to establish the goal and, then, engineer the multiple variables to achieve it.
My goal example: SPL average 80dB and 95dB peak in a 5x4m room, at 2m distance and flat response to 26Hz.
Let's design.
Considering a 3-way speaker, I have to use what is commercially classified as "subwoofer" in order to achieve 26Hz @ -3dB.
This first choice implies in low senstitivy. Let's consider 83dB SPL.
Mid-bass and dome tweeter will have much higher sensitivity, so I go for padding them.
At 1m, disregarding acoustic +3dB coupling, I need 0.5W to get 80dB SPL.
At 2m, I loose 6dB, so I need 2W (+6dB or 4 times the power) to get 80dB SPL.
If I want 15dB headroom for peaks above 80dB SPL, I need 15dB more or 31 times the power, which is 62W.
Good class AB with 31+31W - very easy to build or cheap to buy.
In this case, sensitivity was not a problem and it is what I more or less have at home.
It all depends on the expected SPL, distance from speakers, room size and minimum low frequency response you need.
With these requirements, then you start to choose the speaker options and engineer them to a reasonable performance (distortion, frequency response, cost, size, impedance etc).
I don't like to start engineering a system by fixing one variable, such as sensitivity or any other.
I prefer to establish the goal and, then, engineer the multiple variables to achieve it.
My goal example: SPL average 80dB and 95dB peak in a 5x4m room, at 2m distance and flat response to 26Hz.
Let's design.
Considering a 3-way speaker, I have to use what is commercially classified as "subwoofer" in order to achieve 26Hz @ -3dB.
This first choice implies in low senstitivy. Let's consider 83dB SPL.
Mid-bass and dome tweeter will have much higher sensitivity, so I go for padding them.
At 1m, disregarding acoustic +3dB coupling, I need 0.5W to get 80dB SPL.
At 2m, I loose 6dB, so I need 2W (+6dB or 4 times the power) to get 80dB SPL.
If I want 15dB headroom for peaks above 80dB SPL, I need 15dB more or 31 times the power, which is 62W.
Good class AB with 31+31W - very easy to build or cheap to buy.
In this case, sensitivity was not a problem and it is what I more or less have at home.
Something I probably should have said earlier is that it can be helpful to estimate the usable sensitivity in your desired band by looking at the frequency response plot. The sensitivity spec itself can be a little squirrely due to non-flat response, the fact that the test can be run multiple ways, that it's sometimes stated as 1 watt and other times 2.83 V, could be based on different standards or no standard, doesn't tell you anything about enclosure effects, etc. Sensitivity is a good place to start for gross estimates, but it doesn't always have the specificity required. But this is relevant for the full audio band, not something specific to the bass range.
I don't have the full standard, but regarding an update to IEC Standard 60268-5, in 2015 Klippel said:
https://www.klippel.de/fileadmin/klippel/Bilder/Know-How/Literature/Papers/TUTORIAL_W.KlippeL-IEC Standard Project 60268-5 A and B AES 2015.pdf
"Sensitivity
Calculated from the frequency response and effective frequency range, as the sound pressure level produced at 1 m on the reference axis by an applied voltage of 2,83 V.
Narrow-band sensitivity: the test signal is 1/3-octave filtered noise centered at 1 kHz, or at the geometric mean of the limit frequencies of the effective frequency range if different from 1 kHz. The
frequency shall be stated.
Broad-band sensitivity: the test signal is 2-octave filtered noise centered at 1 kHz, or at the geometric mean of the limit frequencies of the effective frequency range if different from 1 kHz. The frequency shall be stated"
. . .
"Effective Frequency Range
Method:
a) The frequency response SPL(f,r) shall be measured in the rated frequency range according at resolution corresponding to narrow band filter with a stated bandwidth (typically B = 1/9).
b) The mean sound pressure level SPLmean is calculated in the stated frequency
c) The limits of the effective frequency range shall be determined where the smoothed frequency response is not more than 10dB below the mean sound pressure level SPLmean
Last edited:
you could add that depending on driver configuration, distance, and reproduced wavelength there will be some lobing/directivity influence, which can be beneficial or ruin the sound.
"Usable sensitivity", very interesting concept. Look at the curve, rather than a number.
We see woofers or subwoofers with high sensitivity, but how high is it at 30Hz?
Not a surprise when we discover that is not that high for most of them, unless they have a huge diameter and will be installed in giant boxes.
I very much like this second method and I am going to use this going forward. Makes total sense to me to use the sensitivity based on the chart and the usable bandwidth that I plan to use.
It gets a little more complicated at very low frequencies, since the standard frequency response plot could be infinite baffle, IEC test baffle, small box, or optimal box. You have to pay attention to what's really being shown.
And so this again calls back the tidbit I added about sensitivity under 150hz being a little bit hard to define. When you're talking about 30 hz you're really talking about how sensitive is the port tuning, TL tuning, or horn tuning. I definitely do not have enough knowledge or experience to answer this question.
Sure! In this case the comparison must be with the woofer in the planned enclosure, not comparing woofers by their infinity baffle standard curves.
What I do is to select woofers, simulate a box with them and, then, compare how much SPL I get in the whole range, specially at around 30Hz using similar box volumes.
Yes, it's difficult to define, since for bass response we need the box.
Simulation is key here to make things easy.
Take the target woofers T/S parameters, take the volume you plan to have, simulate the woofers and observe the SPL at the frequency range you want, with the same signal applied and with similar volumes.
Back in the day when everything was expensive it made sense to really prioritize efficiency. Now amps are dirt cheap and it's not hard to build a 9x dB speaker. If cost is a constraint, get a bigger amp and pad it down. Or even better, get a DSP and amps for each driver.
Is option 2 even possible?
I've never been fortunately enough in a design where some manner of level adjustment was not necessary.
Whether or not to attempt to add more drivers is not really the issue. It's how big and how much output do you want to attempt?
There's a big difference between a 2-way with a 6" driver and a 3-way with a single 10" driver or multiple 10" drivers. Your choices, if you are lucky are to use a 4 or 8 Ohm mid-woofer or woofer, and then to consider whether to use 1 or 2.
The 4 to 8 Ohm difference gives you an extra 3 dB. Paralllel drivers add 6 dB, but regardless of those choices I'm going to be extremely lucky to find my woofer and mid or woofer, mid and tweeter are somehow not requiring level adjustments.
I've never been fortunately enough in a design where some manner of level adjustment was not necessary.
Whether or not to attempt to add more drivers is not really the issue. It's how big and how much output do you want to attempt?
There's a big difference between a 2-way with a 6" driver and a 3-way with a single 10" driver or multiple 10" drivers. Your choices, if you are lucky are to use a 4 or 8 Ohm mid-woofer or woofer, and then to consider whether to use 1 or 2.
The 4 to 8 Ohm difference gives you an extra 3 dB. Paralllel drivers add 6 dB, but regardless of those choices I'm going to be extremely lucky to find my woofer and mid or woofer, mid and tweeter are somehow not requiring level adjustments.
Oh most definitely. A bunch of commercial high end hifi brand has a loudspeaker like this as their flagship model. Watch a hifi convention review on YouTube. You'll see all sorts of them.
I haven't seen too many people build them here. Likely because of cost and complexity. Probably also not needed as I think they are meant to fill a very large room. I have my shop which is pretty big and has 16 foot high ceilings so I am thinking about building one for there.