I guess it depends on where you sit in the concert hall. If the average classical level in my listening room is ~78dB then peaks could hit 100dB at the listening seat. Because distortion is low, it doesn't sound "too loud." Much more than that and the room starts to talk to me (room needs more work). If I were hitting peaks of 100dB or more on less dynamic recordings, it certainly would sound loud.
Is there any good info on average music listening levels? I know I don't listen anywhere near as loud as amplified concerts. Maybe as loud as acoustic, tho I'm not sure.
There's an important and subtle point here. AMT drivers are close cousins to compression drivers, not ribbons. Why?
If Oskar Heil's AMT patent has any meaning, it's in the acceleration that the pleated diaphragm gives ... supposedly in the 5x range. This isn't all that far from the 5x compression ratio we see in the more moderate-ratio compression drivers.
There are consequences for using an acceleration multiplier: coupling to the air-load is improved, and efficiency goes up. I'm probably wrong here, but it seems that the virtual diaphragm has its mass reduced by these techniques.
As for loudspeaker distortion measurements, I am not sure they are meaningful, except to detect drivers with design or manufacturing faults (VC rubbing, off-center spider, etc.). Correlation of loudspeaker THD with audibility is fairly low, in my experience. Sudden jumps in THD or IM distortion at certain frequencies can point to driver resonances, though, and choice of crossover slope affects out-of-band IM distortion of tweeters.
In a similar way, THD distortion measurements of a complete amplifier are not a reliable measure of quality, except to detect outright faults; a more meaningful index is closely examining the transfer curve of the open-loop circuit, which tells you about crossover distortion, thermal tails, and signal-dependent time delays which affect loop stability. In amplifiers, we're looking for internal overshoots that saturate active devices, and how long these devices take to recover from the overshoot condition. This can be very difficult to detect if you're not allowed to open up the box and start poking around.
We don't have the ability to "open up" a working driver in the same way; attaching probes alters the behavior of the moving system, so the techniques have to use indirect methods. You'd need to measure VC heating dynamically, and use laser holography to examine the diaphragm at various G-loads.
The only thing I can think that might correlate with subjective impressions of dynamic range would some sort of pulse measurement that would quickly heat the voice coil and then measure distortion as it cools down. Maybe a burst of pink-noise at 100 dB SPL lasting a one to five seconds (which would heat the VC), then quickly measuring the distortion spectrum as it cools down.
Last edited:
And there's the rub, Pano. It is not distortion at all that causes you to think it's not "too loud". I used to think so too.
I measured the JBL 2445 large format CD on a 300 Hz tractrix horn and compared it to a B&W 602 at 2m at 100db (thats 106 db at 1m) outdoors. The distortion was about the same between the two. Even higher order distortion was about the same. And those are very high levels. Why then does the horn/CD combo sound so much more dynamic inside?
I'm guessing that this is the secret behind the fidelity of Tom's horns as well. By raising the level of direct sound compared to the reflected sound, and delaying the reflections, you give the listener consistent and clear cues about the sound, creating a sense of cognitive ease and the perception of dynamics, smoothness and so on.
Lynn, the measurements I made were of both THD and individual orders of distortion (upto 12th). Research from Geddes suggests that it is the higher order distortion that is audible. Lower orders are masked by the fundamental. But even higher orders were similar between the B&W and the horn/CD combo, at quite high levels.
Last edited:
Hi Jmmlc,
Can you show an example of such an impulse? I want to see it and then compare it to that from the few horns, CDs, ring radiators I have.
Thanks!
Would frequency response affect the way we perceive dynamics? Would peaks in for instance the presence band increase perceived dynamics? If anyone have seen any similarities between the frequency responses (or power responses) of speakers that sound dynamic versus those that do not, it would be interesting.
I don't know if this has any effect, I'm just throwing out an idea. But I read an article some time ago, where it was mentioned that very small changes in frequency response (of amplifiers in this case) produced subjective changes in parameters like dynamics and clarity. This were tiny differences, in the order of 0.1dB. The differences are of course much larger in loudspeakers. But the observation is interesting.
Is there any research on what factors that influence our perception of dynamics? It is probably much more complicated than just level differences.
-Bjørn
I don't know if this has any effect, I'm just throwing out an idea. But I read an article some time ago, where it was mentioned that very small changes in frequency response (of amplifiers in this case) produced subjective changes in parameters like dynamics and clarity. This were tiny differences, in the order of 0.1dB. The differences are of course much larger in loudspeakers. But the observation is interesting.
Is there any research on what factors that influence our perception of dynamics? It is probably much more complicated than just level differences.
-Bjørn
Yes Bjørn - couple of years back someone on the forum mentioned "false dynamics" caused by peaky response. Peaks in the response would certainly add to the dynamics. 
Shouty horns could do that.
ra7, I have to disagree. At least part of what makes it sound too loud is distortion. It's not the only thing, for sure. I do thing the room plays a large part. Maybe box talk above a certain level?
Shouty horns could do that.ra7, I have to disagree. At least part of what makes it sound too loud is distortion. It's not the only thing, for sure. I do thing the room plays a large part. Maybe box talk above a certain level?
Yes, distortion in general makes it sound louder, or unpleasant. But if you compare the relative levels of distortion in the horn/CD combo and a decent performing cone/dome, you will find very little difference. When I measured it myself, I could not believe it. But that's what it is.
I don't know if it's possible for you, but maybe you can measure the little Vifa TC9 and your Altec horns outside at 100 db above 1 kHz and compare the distortion. Just be sure to give the Vifa a widish baffle
STEPS will plot distortion of individual harmonics as a percentage. Very easy to compare.I just want to note that I appreciate this digression for the regular programming here. There is no doubt that Lynn's design will be good sounding. I have lived with a very similar speaker the last couple of years, and it is indeed very good. I'm interested in knowing why it is good.
Last edited:
Hmm, something near middle-row say 20 or more meters away - 78 db average sounds about right with that type of music.
As for compelling data on an average.. I've not seen anything. It's mostly all focused on hearing loss and damage.
Hmm.. perhaps a new thread with a poll is in order?Back when I was listening to other peoples systems and the occasional audio dealer, I'd take along the analog rat-shack meter. Even at "demo" levels the average usually wasn't more than 93 db for contemporary music, and almost 10 db less for symphonic works. (..again, demo levels, not normal listening levels.) This was taken at the listening position, though in most cases the distance was rarely greater than 3 meters. I'd note though that the noise floor in all instances (though certainly variable), was *much* lower than at something like an expo or "show".
The key point to note was that even with loudspeakers that weren't really designed for significant spl (Spica in particular), the increase in "volume" for these speakers didn't *detract* from a sense of dynamics. That subset of "dynamics" that we might describe as "ease" or "effortlessness" - yes. Compression on the TC-50 in the lower mid's and bass in particular was pretty brutal with even modest peaks, but the sound was still "dynamic". Plenty of "snap" and "drive", and perhaps even more tactile sounding at louder volumes. (..it's been a long time though, so my memory could be faulty.)
Last edited:
Hi guys
What are dynamics, I don’t mean what it “sounds like” but what is it in the signal?
Lynn is right in his thinking in that if one recorded the amplitude vs time and drive the speaker with a signal, one could plot the loss over time resulting from power compression or voice coil heating.
If one used noise, one could plot the change in frequency response the added resistance causes,
I am reminded of an experience in the old days.
A manufacture was promoting it’s amplifiers with an A/B/X switcher and wanted to do it at the speaker company I worked at.
They brought the stuff, we had gathered a number of pro- amps, a Chevin, a couple Crowns and others and I brought my Threshold stasis from home as well as the music.
There were some small audible differences which fell into two sort of behaviors. That was until we raised the level a little bit more and after a couple tracks, my threshold sounded a little less dynamic than the pro amp I was comparing to.
The front panel power indicator said the peaks were -20dB and it wasn’t that loud so I grabbed an oscilloscope and looked a the speaker drive.
What I saw was unexpected, there were very short even single peaks which were Voltage clipped with my Threshold.
Now, most of the time the signal was more like the level indicator said but on the taxing music I had chosen, it was those short clipped peaks that made it less dynamic.
Sure you have heard clipping etc, but this was totally inaudible as a flaw, the only difference were these few very very short missing peaks and you only heard that when compared A/B to one without that limit.
The difference it made was only in the dynamic impression, it was way too short to hear as a flaw.
If you want to see if this exists, an oscilloscope is your friend.
Part of what we hear as dynamics IS related to very fast stuff I am sure.
What If one wanted greater time resolution than the envelope amplitude vs time?
One could make a test signal that was one tone but gated on and off, doing it slowing and ending with a “fast on off”.
Lets say you had 1KHz and you turned it on and off from 1 time per second up to 20 or 30 times per second.
One can examine the difference between the on and off period, the ideal being that ideally the speaker also stars and stops according to the signal.
You can examine the result of the real thing looking at the difference between the on and off, the less difference there is, in this kind of measurement, the lower the Modulation Transfer Function is.
Take any loudspeaker in a room, use ARTA to take the MTF’s up close and then far away, as you move out of the near field, the MTF’s at the higher rates degrade, measure two speakers eq’d the same, one with a lot of directivity and one with little and look at the MTF’s, compare the stereo image and you will see why i think this is an exciting way to look at loudspeakers.
With speech recognition, there is no doubt what so ever that the STIpa measurement (based on MTF’s in 7 bands) is a language independent indicator of being able to understand random words based on the dynaimcs of the sound and so (I think) has to be based on a fundamental hearing process.
The MTF's are the root of that.
Hi Jean-michel
I came to exactly the same conclusions, see the little puffer fish here haha;
http://www.diyaudio.com/forums/mult...ns-beaten-behringer-what-157.html#post3403094
Best,
Tom Danley
What are dynamics, I don’t mean what it “sounds like” but what is it in the signal?
Lynn is right in his thinking in that if one recorded the amplitude vs time and drive the speaker with a signal, one could plot the loss over time resulting from power compression or voice coil heating.
If one used noise, one could plot the change in frequency response the added resistance causes,
I am reminded of an experience in the old days.
A manufacture was promoting it’s amplifiers with an A/B/X switcher and wanted to do it at the speaker company I worked at.
They brought the stuff, we had gathered a number of pro- amps, a Chevin, a couple Crowns and others and I brought my Threshold stasis from home as well as the music.
There were some small audible differences which fell into two sort of behaviors. That was until we raised the level a little bit more and after a couple tracks, my threshold sounded a little less dynamic than the pro amp I was comparing to.
The front panel power indicator said the peaks were -20dB and it wasn’t that loud so I grabbed an oscilloscope and looked a the speaker drive.
What I saw was unexpected, there were very short even single peaks which were Voltage clipped with my Threshold.
Now, most of the time the signal was more like the level indicator said but on the taxing music I had chosen, it was those short clipped peaks that made it less dynamic.
Sure you have heard clipping etc, but this was totally inaudible as a flaw, the only difference were these few very very short missing peaks and you only heard that when compared A/B to one without that limit.
The difference it made was only in the dynamic impression, it was way too short to hear as a flaw.
If you want to see if this exists, an oscilloscope is your friend.
Part of what we hear as dynamics IS related to very fast stuff I am sure.
What If one wanted greater time resolution than the envelope amplitude vs time?
One could make a test signal that was one tone but gated on and off, doing it slowing and ending with a “fast on off”.
Lets say you had 1KHz and you turned it on and off from 1 time per second up to 20 or 30 times per second.
One can examine the difference between the on and off period, the ideal being that ideally the speaker also stars and stops according to the signal.
You can examine the result of the real thing looking at the difference between the on and off, the less difference there is, in this kind of measurement, the lower the Modulation Transfer Function is.
Take any loudspeaker in a room, use ARTA to take the MTF’s up close and then far away, as you move out of the near field, the MTF’s at the higher rates degrade, measure two speakers eq’d the same, one with a lot of directivity and one with little and look at the MTF’s, compare the stereo image and you will see why i think this is an exciting way to look at loudspeakers.
With speech recognition, there is no doubt what so ever that the STIpa measurement (based on MTF’s in 7 bands) is a language independent indicator of being able to understand random words based on the dynaimcs of the sound and so (I think) has to be based on a fundamental hearing process.
The MTF's are the root of that.
Hi Jean-michel
I came to exactly the same conclusions, see the little puffer fish here haha;
http://www.diyaudio.com/forums/mult...ns-beaten-behringer-what-157.html#post3403094
Best,
Tom Danley
Yes Gary, I've noticed that and been thinking about it. Why would that be?
My system is presently set so that -20dB pink noise equals about 78dB at the listening seat. That doesn't really put the peaks very high, but it seems plenty loud.
I'm pretty sure that most would be in agreement here when I say that a large group of listeners enjoy some 2nd order distortion products for reasons unknown. The extreme subjectivity of this hobby makes it almost impossible to develop any kind of true reference.
For me, once a conventional dome becomes stressed, the faults are immediately apparent....something that doesn't happen with a horn loaded CD in the home environment. But at the same time regardless of the listening fatigue from the stressed dome, or the increased perception of detail from the CD, I still slightly prefer the stereo image of the conventional hifi/dome speaker? Maybe I haven't yet experienced that ethereal quality of a well designed waveguide speaker in the right acoustic space. The quest, like Lynn is what keeps me coming back to forums!
For me, once a conventional dome becomes stressed, the faults are immediately apparent....something that doesn't happen with a horn loaded CD in the home environment. But at the same time regardless of the listening fatigue from the stressed dome, or the increased perception of detail from the CD, I still slightly prefer the stereo image of the conventional hifi/dome speaker? Maybe I haven't yet experienced that ethereal quality of a well designed waveguide speaker in the right acoustic space. The quest, like Lynn is what keeps me coming back to forums!
Interesting article, I had always been doubtful of the effects of thermal compression in response to very short transients (eg dynamic compression occurring during a fractional second period) since the thermal time constant is simply too long even for small drivers.
An important point not addressed by your article though is changes in frequency response occuring over time due to the change in Re with VC heating. The small changes in sensitivity with Re you demonstrate would not be important if they were flat across the board, but in any practical speaker the result will actually be a change in frequency response as the sensitivity loss will be frequency dependent.
I'm of the opinion that its these sometimes subtle longer term shifts in frequency response with VC/crossover heating that can cause one speaker to sound less or more dynamic than another in high SPL sections, in other words its a perceptual thing centred around the frequency response, (we perceive certain frequency response errors as more or less dynamic sounding than a flat baseline) not a result of actual compression occurring moment to moment in under a second.
These frequency response changes take two forms - one is changes to the response of an individual driver due to the effects of its modified impedance curve, particularly at crossover frequencies on passive networks since the damping of the crossover is often critically affected by small impedance changes of the driver.
The other is the different thermal characteristics of different drivers in a multiway system, and I think this is probably the more important and audible one.
Imagine a 3 way with a perfectly flat response when cold, when you start playing it loud for just a minute or so there could potentially be a 0.5dB loss in sensitivity of the tweeter as the tweeter reaches its "meandering average" as you call it. The much larger midrange driver might experience only 0.05dB sensitivity loss in the same conditions due to its very different thermal mass properties.
0.5dB doesn't sound like a large frequency response error but if is a shelf such that everything above 3Khz (2.5 octaves) has dropped approximately the same amount it is VERY audible indeed.
I guess my point is that while we can sometimes say that changes in sensitivity of individual drivers with heating is small enough that its not a big deal and perhaps not even audible in many cases, as soon as you put together two different drivers with vastly different thermal characteristics and cross them over, the different sensitivity losses with heating completely messes up our carefully matched and balanced crossover.
Worst case scenario is a midrange to tweeter crossover point as the tweeter is usually far more delicate than the other drivers, has a much shorter thermal time constant, and we're right in the most sensitive part of the ears response so small errors through the crossover region are most likely to be noticed.
Low efficiency cone and dome systems seem to be most likely to fall foul of this issue. A full range driver eliminates this sudden shelf like change in thermal characteristics between midrange and treble (as its the same driver) while a highly efficient compression tweeter avoids it practically speaking by simply having a lot greater dynamic margin than a dome and less thermal dissipation.
Perhaps a greater stability of frequency response during high SPL passages is what compression driver / high efficiency driver fans notice when they say it sounds more dynamic ?
Last edited:
Consider that the woofer gets the largest part of energy from the spectrum and low frequencies require more energy to move the more moving mass of the woofer.
Power compression is the term for what happens when the voice coil gets warm and impedance rises. This usually happens to the woofer first, especially compression drivers due to theyr high sensitivity are attenuated by about 6-8 dB and will not get heated up so fast. Changes in impedance and sensitivity will make x-over drift and let the system loose balance to sound thin and agressive.
Power compression is the term for what happens when the voice coil gets warm and impedance rises. This usually happens to the woofer first, especially compression drivers due to theyr high sensitivity are attenuated by about 6-8 dB and will not get heated up so fast. Changes in impedance and sensitivity will make x-over drift and let the system loose balance to sound thin and agressive.
2. order distortions sound pleasant because they are even harmonics that are present in the sound of an instrument and can enhance the clearness of the sound up to a certain level.
It may be support for the idea that directivity improves clarity. The improvements would be constant even though the signal level changes.
Next time Arta is up an running I'll have to spend some time learning about the MTF and SPIpa measurements that Tom has been using. These measurements sound interesting.
With the changes made over the last few months the system is quite a bit more dynamic and open sounding.
At this point the subs are mounted on swings, the Beta-8a drivers are in 16L boxes tuned to 68Hz with ~75% fill of Bonded Logic Ultratouch cotton insulation, and the high end is handled by wood 1Khz .707 Cléac'h profile horns driven by Radian 475pb drivers.
At this point the subs are mounted on swings, the Beta-8a drivers are in 16L boxes tuned to 68Hz with ~75% fill of Bonded Logic Ultratouch cotton insulation, and the high end is handled by wood 1Khz .707 Cléac'h profile horns driven by Radian 475pb drivers.