Frequency response graphs

  • Thread starter Dazed_and_confused
  • Start date
D

Dazed_and_confused

Audioholic Intern
Just what all can be understood from a frequency response graph for a speaker? I'm referring to the graphs that plot dB SPL on the y axis along with frequency in Hz along the x axis.

If I'm understanding this right (and I very well may not be!), I know that you can see where the frequency response would tend to be uneven. I also know you can see where the ideal frequency range is. I also understand that these graphs are not necessarily indicative of the response one would get in their own room (due to variations in acoustics and room design).

Is there anything else one can take away from these graphs? Would it be correct to roughly infer the sensitivity of the speaker? For example, if the response curve tends to be a fairly flat line along, say 90 db SPL, does that mean the speaker's sensitivity is 90 db SPL?
 
TLS Guy

TLS Guy

Seriously, I have no life.
Just what all can be understood from a frequency response graph for a speaker? I'm referring to the graphs that plot dB SPL on the y axis along with frequency in Hz along the x axis.

If I'm understanding this right (and I very well may not be!), I know that you can see where the frequency response would tend to be uneven. I also know you can see where the ideal frequency range is. I also understand that these graphs are not necessarily indicative of the response one would get in their own room (due to variations in acoustics and room design).

Is there anything else one can take away from these graphs? Would it be correct to roughly infer the sensitivity of the speaker? For example, if the response curve tends to be a fairly flat line along, say 90 db SPL, does that mean the speaker's sensitivity is 90 db SPL?
It would be helpful, if we knew why you are posting these questions and why you revel in being dazed and confused. You asked a question yesterday, and we are on to the next with no feedback from you as to whether you understood the answer you received to the question. Also if we are in for question of the day we would like to know about.

In answer to today's question, you are basically on track. However, the graph is not much use without knowing the conditions of the measurement. So the frequency response is what it is under the condition of measurement.

As to implying any insight into sensitivity, this again can only be determined under conditions of measurement. For sensitivity the mic has to be on axis at a distance of 1 meter. There are two standards, the response when the driver is driven by a power of 1 watt at one meter.

The other is with a drive of 2.83 volts 1 meter. This later can only be fully interpreted in conjunction with the impedance curve of the speaker. If the impedance of the speaker is 8 ohm then the two conditions of measurement will yield the same result. On the other hand the if the speaker is 4 ohm in the 2.83 volt I meter measurement the 4 ohm speaker will indicate 3db more sensitivity. That occurs however because it will have drawn twice the power of the 8 ohm speaker.

Your question was somewhat vague, but I hope I have answered your question. As I said previously it would be helpful to know you object in view and your background.
 
D

Dazed_and_confused

Audioholic Intern
It would be helpful, if we knew why you are posting these questions and why you revel in being dazed and confused. You asked a question yesterday, and we are on to the next with no feedback from you as to whether you understood the answer you received to the question. Also if we are in for question of the day we would like to know about.

In answer to today's question, you are basically on track. However, the graph is not much use without knowing the conditions of the measurement. So the frequency response is what it is under the condition of measurement.

As to implying any insight into sensitivity, this again can only be determined under conditions of measurement. For sensitivity the mic has to be on axis at a distance of 1 meter. There are two standards, the response when the driver is driven by a power of 1 watt at one meter.

The other is with a drive of 2.83 volts 1 meter. This later can only be fully interpreted in conjunction with the impedance curve of the speaker. If the impedance of the speaker is 8 ohm then the two conditions of measurement will yield the same result. On the other hand the if the speaker is 4 ohm in the 2.83 volt I meter measurement the 4 ohm speaker will indicate 3db more sensitivity. That occurs however because it will have drawn twice the power of the 8 ohm speaker.

Your question was somewhat vague, but I hope I have answered your question. As I said previously it would be helpful to know you object in view and your background.
Thanks for the reply. I'm looking to replace a 25 year old stereo system. Just trying to learn and gather as much information as possible to aid in my decisions. The main decision I'm struggling with is which subwoofer(s) to buy. I was asking about frequency graphs because I'm trying to gauge the sensitivity of the various SVS subs. (I don't see this spec posted). I don't "revel" in being dazed and confused. It was a reference to the movie, of which I'm a fan. I also thought it was a fair screen name since I'm far from an expert in audio matters.

The SVS website gives frequency response graphs noting "2-meter ground plane away from reflective structures (quasi-anechoic)". How would I interpret that in context of the 1 watt at one meter standard?

I'm not trying to post a "question of the day", just asking things as I think of them.
 
TLS Guy

TLS Guy

Seriously, I have no life.
Thanks for the reply. I'm looking to replace a 25 year old stereo system. Just trying to learn and gather as much information as possible to aid in my decisions. The main decision I'm struggling with is which subwoofer(s) to buy. I was asking about frequency graphs because I'm trying to gauge the sensitivity of the various SVS subs. (I don't see this spec posted). I don't "revel" in being dazed and confused. It was a reference to the movie, of which I'm a fan. I also thought it was a fair screen name since I'm far from an expert in audio matters.

The SVS website gives frequency response graphs noting "2-meter ground plane away from reflective structures (quasi-anechoic)". How would I interpret that in context of the 1 watt at one meter standard?

I'm not trying to post a "question of the day", just asking things as I think of them.
Now I understand. I was starting to think that you were an idle university student, and I was doing your term paper!

You have a big fundamental misunderstanding here. Subs are powered so there is no sensitivity spec. They are powered from a line voltage and the subs contain an amp.

The smaller the sub the more inefficient it will be. If it is a sealed sub it will be very inefficient in the last octave and need significant Eq requiring gobs of power.

The size of a sub is the major determinant of its overall performance. There is truly no substitute for size when it comes to bass reproducers.

Now of course manufacturers want to put their best foot forward so they parse the data as a rule.

Third party reviews with objective measurements are your best guide.

What you want to know about a sub are primarily these factors.

The F3 point. That is the frequency the sub is 3db down from its reference level.

You need to know its maximum acoustic output and the distortion at that output.

You need to know the change in frequency response in relation to power drive.

You need to look for undesirable ripple in the response.

A very important factor is Q and not often quoted. The Q is a measure of the degree to which the reproduction is resonant. A lot of people like a Q of around 0.7. I think that is plenty high. A Q above 0.7 definitely sounds bad and results in a boomy gooey bass spreading all over the place and exciting room resonances. I prefer a Q around 0.5 and believe that to be optimal.

Last but not least, the reliability of the amp needs to be taken into consideration and customer service. Sub amps are I think the most unreliable devices on the audio scene by and large. Most are generally unserviceable. So failure generally requires a new amp.

One other issue you need to be aware of. The commercial world plays the numbers game. This is especially true of bass extension and specifically F3. It is common place for commercial subs to design for lowest F3 to the detriment of the overall design.

You might want to take a look at this thread, which really drives home that size is the major determinant of overall performance.

If you have practical skills you would do well to consider going the DIY route. On the whole the best subs are DIY, and in terms of value per dollar it is no contest.
 
Swerd

Swerd

Audioholic Warlord
TLS Guy mentioned Q in regard to woofers. It is an important feature of speakers, but it can be a difficult topic to understand. So I thought I would add to what he said. For anything that vibrates or resonates, I think of Q as a shape or slope of a response curve.

An electrical filter will oscillate or ring, to some extent, after the signal stops. The steeper the slope of this filter, the longer it will ring. Similarly, the higher the Q value of this filter, the more it will ring. Mechanical filters work the same way.

Speaker cabinets act as mechanical high-pass filters for woofers or subwoofers mounted in them. It passes frequencies above the cut-off or low frequency limit of the design, and the signal rolls off below this point at a rate determined by the design. The cabinet dimensions affect the characteristics of the high-pass filter, letting a designer tune the woofer’s bass response. A sealed cabinet acts essentially as a 2nd order high-pass filter, and a vented enclosure typically resembles a 4th order high-pass filter.

This figure shows the low end of the frequency response curves of a woofer in various sized boxes with different system QTC.

Ideally, a woofer will respond to a single bass tone by starting and stopping soon afterwards. A vented design will continue to respond (ringing or oscillating) about twice as long after the signal stops, as a sealed design. This transient response can be visualized in an impulse response curve, where a single pulse of sound is introduced, and the response is graphed vs. time.

The total system Q of the woofer combined with cabinet is a ratio of values (with no units) that defines both the shape of the frequency response curve roll-off. It also determines the amount of damping to oscillation or ringing after the signal stops. This is also spoken of as “transient response”. QTC is the term for sealed cabinets, and the similar QTS term is for ported reflex cabinets.

A sealed enclosure with a QTC of 0.5 is considered a “critically damped alignment”. An impulse response curve shows that it responds to the initial pulse and has little or no overshoot. For a given driver, a QTC of 0.5 requires the largest box. This low Q alignment has a downward-sloping response curve, but offers the best possible transient performance and the lowest frequency extension at -10dB.

A system QTC of 0.577 (also known as a Bessel alignment) has the most linear phase response and offers slightly less damping than Q = 0.5.

When QTC = 0.707 (a Butterworth alignment) there will be the flattest frequency vs. amplitude response. This is a common alignment for woofers because it offers a fuller sound, while still being reasonably well damped.

System QTC near 1.0 delivers a peaked response, but allows the smallest box size still considered by some to be acceptable. A woofer with a QTC larger than 1.0 is a boom box with a peaked response curve, and an impulse response that rings and rings. Guess where most less expensive home theater subwoofers fall? Woofers and subwoofers that play on after the signal has stopped (due to a high Q and the resulting ringing) sound slow and muddy.

This next figure (sorry about the size, it was the best I could find) shows the amount of ringing in the impulse response curves where QTC ranges from 0.5 to 2.0.


Looking at the first figure, it is easy to say, “Go for that big response where Q is 2 and get the most bass response for your money” High Q drivers usually do have less expensive small magnets, so this certainly will cost less. But along with the illusion of great bass at low cost is the hidden cost of lack of control over the movement of the woofer cone. It allows a big response around the woofer’s resonance frequency, but furnishes poor control over transients, causing muddy sounding or ringing bass. The second figure shows the response of a impulse response graphs of speaker several different QTC values. QTC of 2 hangs on the longest after the pulse has ended, much like the ringing of a bell. As QTC gets lower, the ringing ends sooner.
 
TLS Guy

TLS Guy

Seriously, I have no life.
Swerd has given you a good explanation of this important topic, not usually mentioned on the glossies or by salesmen.

For reference this is what a critically damped impulse looks. This comes from my TL system.

 
D

Dazed_and_confused

Audioholic Intern
Subs are powered so there is no sensitivity spec. They are powered from a line voltage and the subs contain an amp.
I still don't follow. Why can't you still measure the spl given 1 watt at a distance of 1 meter from the speaker?

The size of a sub is the major determinant of its overall performance. There is truly no substitute for size when it comes to bass reproducers.
The size of the sub itself, the enclosure, or the ratio between the two?

If you have practical skills you would do well to consider going the DIY route. On the whole the best subs are DIY, and in terms of value per dollar it is no contest.
I'm not too good in the DIY department. Thank you for your reply. Very informative. Still trying to understand all of this but it's starting to come together (I think) :)
 
D

Dazed_and_confused

Audioholic Intern
TLS Guy mentioned Q in regard to woofers. It is an important feature of speakers, but it can be a difficult topic to understand. So I thought I would add to what he said. For anything that vibrates or resonates, I think of Q as a shape or slope of a response curve.

An electrical filter will oscillate or ring, to some extent, after the signal stops. The steeper the slope of this filter, the longer it will ring. Similarly, the higher the Q value of this filter, the more it will ring. Mechanical filters work the same way.

Speaker cabinets act as mechanical high-pass filters for woofers or subwoofers mounted in them. It passes frequencies above the cut-off or low frequency limit of the design, and the signal rolls off below this point at a rate determined by the design. The cabinet dimensions affect the characteristics of the high-pass filter, letting a designer tune the woofer’s bass response. A sealed cabinet acts essentially as a 2nd order high-pass filter, and a vented enclosure typically resembles a 4th order high-pass filter.

This figure shows the low end of the frequency response curves of a woofer in various sized boxes with different system QTC.

Ideally, a woofer will respond to a single bass tone by starting and stopping soon afterwards. A vented design will continue to respond (ringing or oscillating) about twice as long after the signal stops, as a sealed design. This transient response can be visualized in an impulse response curve, where a single pulse of sound is introduced, and the response is graphed vs. time.

The total system Q of the woofer combined with cabinet is a ratio of values (with no units) that defines both the shape of the frequency response curve roll-off. It also determines the amount of damping to oscillation or ringing after the signal stops. This is also spoken of as “transient response”. QTC is the term for sealed cabinets, and the similar QTS term is for ported reflex cabinets.

A sealed enclosure with a QTC of 0.5 is considered a “critically damped alignment”. An impulse response curve shows that it responds to the initial pulse and has little or no overshoot. For a given driver, a QTC of 0.5 requires the largest box. This low Q alignment has a downward-sloping response curve, but offers the best possible transient performance and the lowest frequency extension at -10dB.

A system QTC of 0.577 (also known as a Bessel alignment) has the most linear phase response and offers slightly less damping than Q = 0.5.

When QTC = 0.707 (a Butterworth alignment) there will be the flattest frequency vs. amplitude response. This is a common alignment for woofers because it offers a fuller sound, while still being reasonably well damped.

System QTC near 1.0 delivers a peaked response, but allows the smallest box size still considered by some to be acceptable. A woofer with a QTC larger than 1.0 is a boom box with a peaked response curve, and an impulse response that rings and rings. Guess where most less expensive home theater subwoofers fall? Woofers and subwoofers that play on after the signal has stopped (due to a high Q and the resulting ringing) sound slow and muddy.

This next figure (sorry about the size, it was the best I could find) shows the amount of ringing in the impulse response curves where QTC ranges from 0.5 to 2.0.


Looking at the first figure, it is easy to say, “Go for that big response where Q is 2 and get the most bass response for your money” High Q drivers usually do have less expensive small magnets, so this certainly will cost less. But along with the illusion of great bass at low cost is the hidden cost of lack of control over the movement of the woofer cone. It allows a big response around the woofer’s resonance frequency, but furnishes poor control over transients, causing muddy sounding or ringing bass. The second figure shows the response of a impulse response graphs of speaker several different QTC values. QTC of 2 hangs on the longest after the pulse has ended, much like the ringing of a bell. As QTC gets lower, the ringing ends sooner.
 
D

Dazed_and_confused

Audioholic Intern
Is there a way to calculate these Q values for subs? I don't see Q values listed for the SVS subs I'm looking at. Can Q be derived from knowing the subwoofer's diameter and cabinet dimensions?
 
TLS Guy

TLS Guy

Seriously, I have no life.
I still don't follow. Why can't you still measure the spl given 1 watt at a distance of 1 meter from the speaker?



The size of the sub itself, the enclosure, or the ratio between the two?



I'm not too good in the DIY department. Thank you for your reply. Very informative. Still trying to understand all of this but it's starting to come together (I think) :)
You could measure the drive of the internal amp, but as a point of comparison you would need a lot more information. It would not be a point of reference and for sealed subs would vary with frequency. Sealed subs have a high F3 without equalization, so the sensitivity measurement would decline with decreased frequency. It is much more helpful to know that actual power required to produce a given spl at a given frequency.

It is the size of the enclosure that is the major determinant of overall performance. Driver size per se only relates to performance indirectly. A larger cone is heavier, and therefore tends to a lower Fs. That is the free air resonance of the driver and to all intense and purposes determines the low frequency cut off of the system. In addition output will be greater for a given cone travel, so you won't hit the mechanical limits as quickly. JL audio have, or did and 8" sub driver that was a very capable driver.

It is reduced enclosure that really reduces efficiency in the last octave and makes small subs require enormous power and leads to a lot of expense and difficulties. You can see that in the power requirements for the various designs. The alignment ported and then the isobarik alignment using four of the same drivers really gives the game away. Although four of the same drivers are used rather than two, the cabinet volume is halved and the power demands are doubled. Whereas if I had done an alignment with four drivers loaded conventionally, the enclosure size would have doubled and the power demands would have been halved.

An enclosure that tries to force the quart out of the proverbial pint pot requires a huge power drive.

When it comes to subs and all speakers for that matter you have to take most manufacturers claims with a grain of salt. That is where third party measurements are so important.
 
AcuDefTechGuy

AcuDefTechGuy

Audioholic Jedi
Swerd

Swerd

Audioholic Warlord
I still don't follow. Why can't you still measure the spl given 1 watt at a distance of 1 meter from the speaker?
The main reason why a speaker's sensitivity is measured, is to let a potential buyer know what size amplifier would be big enough to drive the speaker. If the sub is sold with it's own internal amp, sensitivity measurements become meaningless.

That was the short answer, TLS Guy already provided a longer & more detailed answer.
 
Swerd

Swerd

Audioholic Warlord
Is there a way to calculate these Q values for subs? I don't see Q values listed for the SVS subs I'm looking at. Can Q be derived from knowing the subwoofer's diameter and cabinet dimensions?
The total system Q of a speaker can be estimated, but it requires several electro-mechanical parameters of a speaker known as Theile/Small (or TS) parameters. It also requires knowing the internal volume of a cabinet, and if the speaker is ported, it also requires knowing the port length and diameter.

It is also possible to recognize if a speaker has an undesirably high Q simply by looking for a low frequency hump in the frequency response curve. This, of course, requires honest reporting by a manufacturer or a third party.

As an example, I'd guess that the elevated response between 60 and 130 Hz suggests (but does not prove) that this speaker was designed with a high system Q, and suffers from muddy sounding bass.

 
D

Dazed_and_confused

Audioholic Intern
It is also possible to recognize if a speaker has an undesirably high Q simply by looking for a low frequency hump in the frequency response curve. This, of course, requires honest reporting by a manufacturer or a third party.
This is very helpful. I don't necessarily need an exact Q value (although it would be nice). Thanks!
 
AcuDefTechGuy

AcuDefTechGuy

Audioholic Jedi
It is also possible to recognize if a speaker has an undesirably high Q simply by looking for a low frequency hump in the frequency response curve. This, of course, requires honest reporting by a manufacturer or a third party.

As an example, I'd guess that the elevated response between 60 and 130 Hz suggests (but does not prove) that this speaker was designed with a high system Q, and suffers from muddy sounding bass.
Well the bass region is significantly affected by the room. So the bass "hump" appears to be very common. You can see this bass hump (around 130Hz) even with the $22,000 Revel Salon2 and probably most speakers.

http://www.stereophile.com/content/revel-ultima-salon2-loudspeaker-measurements

The equally excellent $20,000 KEF Reference 207/2 has the "bass hump" at ~ 60Hz.

http://www.stereophile.com/content/kef-reference-2072-loudspeaker-measurements
 
Swerd

Swerd

Audioholic Warlord
Yes, bass response of speakers is significantly affected by the room boundaries, the walls, ceiling and floor. But measuring speaker responses can be done to eliminate or minimize the room boundary effect from the bass range. They can be considered good enough to allow anyone to predict how a speaker might perform in most rooms.

The frequency response graphs published at Soundstage Network are done by the Canadian National Research Council. They have a genuine anechoic chamber where they can accurately measure bass response without those pesky room boundary reflections.

It is also possible to use computer software and measurement hardware to minimize these problems. Careful microphone placement (near or far depending on the size of the drivers), and keeping the measurement time gate short enough (roughly 4 milliseconds) can minimize measuring reflected sounds from room boundaries. This is what Stereophile does. While not done in a true anechoic chamber, they are done carefully enough to be considered accurate and reliable. They clearly state details in the accompanying article that describe how they approximate an anechoic response with their measuring tools:

Fig.4 shows how these individual drive-unit outputs add up on the tweeter axis in the farfield. Looking through this graph's small ups and downs, the Salon2's response is extraordinarily flat, from the upper bass all the way through its 30kHz upper limit. In fact, the tweeter's output starts to rise just below 30kHz, suggesting that the beryllium dome's primary resonance lies above this frequency. (My measurement microphone is calibrated to only 30kHz.) At lower frequencies, the broad rise in output in the upper bass will be mainly due to the nearfield measurement technique. The speaker's low-frequency response extends almost down to 20Hz. This is a true full-range loudspeaker.

Fig.4 Revel Ultima Salon2, anechoic response without grille on listening axis at 50", averaged across 30° horizontal window and corrected for microphone response, with the complex sum of the nearfield responses plotted below 300Hz.​

 
AcuDefTechGuy

AcuDefTechGuy

Audioholic Jedi
I think there is a good reason why we don't see too many measurement of the Philharmonic or Salk speakers below 200Hz. I bet you would see a bass hump between 60-130Hz too.

I just looked at the Sierra Tower's measurement. There is a bass hump around 80-90Hz also.

http://www.ascendacoustics.com/pages/products/speakers/SRT/srtmeas.html

I don't think these bass humps (Salon2, KEF 207/2, Sierra, etc.) mean that the Q-values are high or the bass is muddy in any way.
 
TLS Guy

TLS Guy

Seriously, I have no life.
I think there is a good reason why we don't see too many measurement of the Philharmonic or Salk speakers below 200Hz. I bet you would see a bass hump between 60-130Hz too.

I just looked at the Sierra Tower's measurement. There is a bass hump around 80-90Hz also.

http://www.ascendacoustics.com/pages/products/speakers/SRT/srtmeas.html

I don't think these bass humps (Salon2, KEF 207/2, Sierra, etc.) mean that the Q-values are high or the bass is muddy in any way.
I'm sorry but I respectfully disagree. I think it is a design issue.

The KEF frequency response.


Kef impulse response, off the graph!


Revel Salon 2 Frequency response.



Revel Salon 2 impulse response. Again off the graph, but about 7 to 8 msec



The ascend impulse extends to 4 msec, so better.

My reference dual TLs.

Frequency response.


The small dip hump from 35 to 65 Hz is the acoustic transition between the lines. This is obviously a very difficult design challenge and I was lucky, I think, to get it this good.

Impulse response 0.5 msec



Now I will let you judge which of those speakers has a bass delivery that really does sound like live.

All of the other speakers are ported reflex. This in of itself is a resonant system. When you add a passive crossover with a low crossover point you add insult to injury as you see in the first two.

The other problem is commercialism. I don't have to worry about that. My point is that you can actually make a ported cabinet sound acceptable tight, is you sacrifice some bass extension, and sacrifice a little output, by slightly damping beyond the accepted norm.

However you might loose sales if the competitors speaker has a lower F3 than yours!

I personally think you are all forgiving of this because it is so much the norm in speakers you are used to it.

However I have a very strong hunch indeed that if you lived for a while with a speaker that did not exhibit that behavior, you would never live again with the former.
 
Swerd

Swerd

Audioholic Warlord
I don't think these bass humps (Salon2, KEF 207/2, Sierra, etc.) mean that the Q-values are high or the bass is muddy in any way.
You're generalizing too much, and forcing broad conclusions that really can't be made. I think you do know better. I only said peaks in this bass range suggests high Q. It doesn't prove it.

There clearly are other room-related reasons why a frequency response curve can have the appearance of bass peaks. It depends on the details of how the frequency response curve was measured.
I think there is a good reason why we don't see too many measurement of the Philharmonic or Salk speakers below 200Hz. I bet you would see a bass hump between 60-130Hz too.
Are you really suggesting that Philharmonic or Salk speakers are designed with high Q bass peaks? Or are you playing Devil's Advocate?

Dennis doesn't show frequency response measurements below 200 Hz because he doesn't have an anechoic room, or the capacity that Stereophile has to simulate one. He readily admits that, and I think that's being honest.
 
TLS Guy

TLS Guy

Seriously, I have no life.
I actually am putting almost all the weight on the impulse responses, which suggest that the frequency responses are likely correct.

Of those speakers mine is the only low Q speaker of the bunch. The others are high Q, which from the design is what you would expect.
 
newsletter

  • RBHsound.com
  • BlueJeansCable.com
  • SVS Sound Subwoofers
  • Experience the Martin Logan Montis
Top