The effect is there, there can be no doubt of that, even John's analysis shows that it is there. The question is a matter of degree. Is the effect every great enough to be what we call "dynamics"? I don't know, but consider this; at some point, i.e. smaller drivers and ever higher SPLs it has to be a significant factor. Is that level so high that other things overcome its audibility?
But lets not confuse the well studied thermal compression with thermal modulation. Although they stem from the same root, one is short term and the other long term. Different factors affect each of these so the two can be quite disconnected.
In my tests I did frequency response tests every 50 ms (or somewhere in that time frame) on a compression driver system and a smaller 1" dome tweeter system both playing the same SPL level of noise ( the level was fairly high, but not so high that either speaker was damaged.) At each interval the frequency response was measured and it was changing by not insignificant amounts. Some changes - especially at crossover - had as much as 3 dB change. This would clearly be audible if instantaneous, but over what time scale would it become inaudible? Questions not easily answered but not so easily written off either.
But lets not confuse the well studied thermal compression with thermal modulation. Although they stem from the same root, one is short term and the other long term. Different factors affect each of these so the two can be quite disconnected.
In my tests I did frequency response tests every 50 ms (or somewhere in that time frame) on a compression driver system and a smaller 1" dome tweeter system both playing the same SPL level of noise ( the level was fairly high, but not so high that either speaker was damaged.) At each interval the frequency response was measured and it was changing by not insignificant amounts. Some changes - especially at crossover - had as much as 3 dB change. This would clearly be audible if instantaneous, but over what time scale would it become inaudible? Questions not easily answered but not so easily written off either.
Well along the same lines apparently JBL thinks it important enough to use a patented edge-wound wire in the coils of some of their newer drivers. They use wire with a lower TCR helps keep the resistance constant with increasing power levels. This is a side view of the 2216ND used in the new M2 monitor.
Rob
Rob
Attachments
I did sims for a 1" Seas tweeter with copper wire VC. Knowing the DC R, the diameter of the VC and the height of the VC, the number of turns and number of layers it's easy to figure what the wire length and gage are and hence mass. All you have to do is plug in the wire mass, Cv for copper and the DC R and you get the max rate of temperature change, what I called the thermal slew rate. I just went 70 msec because this has to be integrated numerically with sufficiently small time step to follow the signal transient. It took a long time. But the point is that there is no way the thermal response follows the signal. I have no argument that the VC heats up over a longer period of time, and that it results in a drop in sensitivity. But this change is sensitivity does not result in dynamic compression, or reduction of the peak to average SPL radiated by the driver. What it does is reduce the level of play back. I.E. it's like turning down the volume. But everything is reduced in volume, not just peaks.
I'm in total agreement with you that VC heating over time is real and that it results in reduction in output if the applied signal remains constant in level. My argument and analysis are intended to show that the time scale is long relative to musical transients and therefore musical transients are not "compressed" due to rapid changes in VC temperature. Reduction in sensitivity due to VC heating is certainly real but it does not occur over msec (or fractions of msec) time scales.
We know that to get a 3dB drop in sensitivity the Re of the driver has to at least double. I say at least because the driver Z away from resonance(crudely) is Re +j Le and Le is not a function of temperature. We also know that the Q of a simple 2nd order HP passive filter will change by a factor of 2 is Re doubles, again, ignoring the effect of Le. Now, connect a driver to a 2nd order passive filter and start sending MSL bursts through it. Measure the filter transfer function. See how long and at what power levels it takes to make the filter Q change significantly. In my tests it took a long time at levels which would not allow me to remain in the room. Compared to the peaks and valleys of a musical signal the VC temp creeps up over significantly longer time scales at normal play back levels.
I would say that if you look at the power delivered into a driver, the rate of heating is probably better estimated by the time averaged power of some reasonable time interval, say several second. That is really what my plots are saying.
If you don't start with an "objective" definition of "dynamics" there is no question.
After having determined what "dynamics" means we can then examine whether a loudspeaker (in this case ORION) manages to accomplish accurate "dynamics" over its intended range (which in the case of ORION can be taken to be about 15dB to 105dB . . . noise floor to peak SPL for orchestral music in a concert hall environment, as delivered by Redbook CD). That is to say, does it have, in this range and within the limits of audability, a signal invariant linear amplitude transfer function from input signal to acoustic output.
There is no objective evidence that it does not.
But if 105dB is "10" you can't turn it up to "11". No one disputes that . . .
Here, let me try to make it very simple. Hopefully you all understand the difference between DC power, AC RMS power and instantaneous power. What I am saying is the when subject to an AC signal of sufficiently high frequency (higher audio frequencies) the VC temp varies in time in a manner that is substantially the same as it would when subjected to a DC power of the same magnitude as the RMS AC power. The temperature does not follow the instantaneous power.
I also think there may be a miss conception of what I mean by long and short time scales. Talking about transients I thinking fractions of msec. In that sense a second is a long time. There is no misapplication of physics here, there appears to be a lack of understanding of the physics.
I also think there may be a miss conception of what I mean by long and short time scales. Talking about transients I thinking fractions of msec. In that sense a second is a long time. There is no misapplication of physics here, there appears to be a lack of understanding of the physics.
John
The thermal response does not have to "follow the signal" to be a factor as you suggest - this requirement is unnecessary and not one that I would ever claim to happen or expect.
But there IS compression, there has to be and you admit that. But neither you nor I has shown over what time scale this ceases to be audible as a "dynamic compresssion" and becomes just a static level shift. 70 ms is certainly too short to come to the conclusions that you have. And how do you explain my measurement results? (Admittedly this was a long time ago and I don't have them readily available so I won't press this issue, but had they been as insignificant as you suggest I would certainly not have pursued the issue. I would simply have let it drop as yet another audio factor that is irrelavent. But it was significant - at least as I recal it. I have not strongly pursued it because if it is a factor then the solution is trivial - just use more efficeient speakers, which I already do. The fact is that the subject interests me and I believe that it is real - it is real, is it audible? I also believe that the whole idea could be wrong, just as you say, I don't know. Dynamics is something that seems to avoid quantification and yet it does not seem to go away as a subjective issue. What makes me believe that it may be artificial is how unreliable the perceptions are - often 180 degree opposite opinions. Not a good basis for a significant audible factor - but just what one would expect from an insignificant one.)
Hi
Dynamics, a subject of interest but to discuss it, one must acknowledge that how our hearing experience is not how microphones and loudspeakers work are not the same things.
Power compression in the first order is a change in the driver’s frequency response and sensitivity which is a result of the VC heating. As the temperature rises, so does the VC wire’s Rdc and has doubled at about 230 C and some modern VC’s can tolerate operation into the 300+C range.
At an ancient AES conference Doug Button (JBL) presented a paper on power compression showing that with a high enough VC temperature, that increasing the power can actually cause a reduction in sound output. That had followed one I gave on the elimination of power compression in the Servodrive (servomotor driven) subwoofers using a cooling system which ran off of a small amount of input power. It was one of those old AES presentations I met Earl too.
Anyway a rough rule of thumb is pc begins around 1/8 to 1/10 the rated power and “how long” it takes to reach X, depends on the VC thermal time constant and rate of input.
In a large woofer motor, that might be 15-45 seconds, a tiny tweeter VC heats up “faster” but has much less effect on the frequency response part compared to a tuned woofer like a vented box (which depends strongly on driver Qe).
The VC is connected to a radiator and that to suspension elements.
These are also variables. A little appreciated fact is that loudspeaker cones may not be the simple piston we wish they were, this may or may not matter depending what you do with one but in the case of a low frequency horn like the lab-sub driver or Tapped horns at work, the amount the cone flexes can cause a significant error between what you get and what you wanted.
Radiator suspensions usually have some amount of hysteresis as well, in other words, how they behave depends partly on the very recent history. The more a driver depends on mechanical damping (think gooey dome tweeter etc), the more likely it is that how it behaves also depends on what it had been doing just before, the ambient temperature and age.
I like horns, they are fun to design especially the modern ones at work where the ranges tie together into one source but one of the things that makes them work at high power is that they depend on the acoustic load for damping and that is pretty well behaved (comparatively speaking), only having the gas law as a complication .
While I love hifi, In a different area of loudspeakers where I must spend a lot more time, there are a couple things which to me seem like they may be useful clues or a direction to look for those with time and curiosity..
As the new building Fire codes are being applied, the concern for voice intelligibility is developing legal teeth. Who cares how loud an emergency announcement is if you can’t hear what it says, well, now they do and a lot of line arrays are coming down / not going up.
As our loudspeakers are used in large spaces because of the intelligibility and there is now a legal mandated measurement for voice intelligibility (STIPA), it made sense to attend a recent seminar.
I kept thinking and wondering what part of hearing information here would apply to music or stereo imaging?
It might well apply since the voice range is where our greatest acuity is..
http://dl.dropbox.com/u/2787059/ECSSIW_Brochure.pdf
Speech transmission index - Wikipedia, the free encyclopedia
What is interesting first of all is that one can have a music reproduction system which has little intelligibility.
A Choir re-enforcement system or organ reproducer in a Church or artificial reverberation system in a large venue are examples of very low stipa systems and stipa has NO impact on the pleasing nature or not because in that use, nothing is happening fast, there is no information being transmitted..
Ok, so I lean the other way, I want to be able to reproduce fireworks which are instantaneous events, that requires that the low mid and high frequency components arrive as they would have live, not re-arranged in time (which spreads out the energy envelope). I want to see every way possible what the mic senses as pressure is not what I fed into the speaker, where it is not a straight wire to sound pressure conversion.
To me, it appears what we should be looking at more carefully are the envelopes the energy comes in, not just instantaneous event like “impulse response”. For example, the presence of clipping of a short string of instantaneous peaks in dynamic music are not audible as clipping or even a fault until you compare to the same case not clipped, then the unclipped sounds somehow more “dynamic” not louder. What we hear / interpret as “loudness” as something more like a short averaged C weighted SPL measurement and NOT as the instantaneous Voltage waveform from a microphones measured as the actual pressure.
Audio information like word intelligibility??, where is that hidden? In part,contained dynamically.
The root of the stipa measurement is a series of Modulation Transfer Function measurements.
Here is a nice explanation of MTF from the optical world where the idea came from (and is then adapted to audio).
Modulation Transfer Function - YouTube
At the seminar, I asked the co-developer of the stipa measurements how high they had investigated the MTF’s. He said it appeared we could hear modulations much higher than used in STIpa (for voice).
I pressed him a bit farther and he said around 30Hz rate (above the stipa ranges) was as high as he thought was about as high hearing acuity extended (as his concern was only voice intelligibility and not corruption of a more complex signal like music, I suspect the audibility extends higher yet).
Many of you guys have ARTA which has a simple MTF measurement .
When I got home, for fun, at the listening position, I have measured a couple consumer grade (craptastic)2 ways in the same place as an SH-50, they deliver VERY different looking MTF’s with the little 2 ways falling apart at higher MTF’s. In this case, the much greater level room reflections would be part of what effects the 2 ways and comparatively destroys their stereo image at the LP (compared to near field listing ).
Since corruption of the MTF at lower rates is what lowers / destroys stipa, I think that is also logical / probable that the same kind of corruption is also tied into preserving / destroying the recorded information at the listening position we use to determine the phantom images from the recording.
Anyway, I don’t have time to pursue this avenue other than casually but I know some of you guys are keen to figure out what maximizes the recorded information that reaches your ears at the listening position.
Have a look at that, do some comparative measurements, see if it doesn’t look like conditions / loudspeakers which result in better / higher MTF’s sort of correlate to better stereo imaging (when you have a pair of the speakers).
By stereo image, I don’t mean a pleasing wall of sound but more like a large window to another space, one where the location of a voice etc between the speakers is much stronger than the awareness of the R and L speakers being the source of sound producing the phantom image.
Best
Tom Danley
Danley Sound Labs
Dynamics, a subject of interest but to discuss it, one must acknowledge that how our hearing experience is not how microphones and loudspeakers work are not the same things.
Power compression in the first order is a change in the driver’s frequency response and sensitivity which is a result of the VC heating. As the temperature rises, so does the VC wire’s Rdc and has doubled at about 230 C and some modern VC’s can tolerate operation into the 300+C range.
At an ancient AES conference Doug Button (JBL) presented a paper on power compression showing that with a high enough VC temperature, that increasing the power can actually cause a reduction in sound output. That had followed one I gave on the elimination of power compression in the Servodrive (servomotor driven) subwoofers using a cooling system which ran off of a small amount of input power. It was one of those old AES presentations I met Earl too.
Anyway a rough rule of thumb is pc begins around 1/8 to 1/10 the rated power and “how long” it takes to reach X, depends on the VC thermal time constant and rate of input.
In a large woofer motor, that might be 15-45 seconds, a tiny tweeter VC heats up “faster” but has much less effect on the frequency response part compared to a tuned woofer like a vented box (which depends strongly on driver Qe).
The VC is connected to a radiator and that to suspension elements.
These are also variables. A little appreciated fact is that loudspeaker cones may not be the simple piston we wish they were, this may or may not matter depending what you do with one but in the case of a low frequency horn like the lab-sub driver or Tapped horns at work, the amount the cone flexes can cause a significant error between what you get and what you wanted.
Radiator suspensions usually have some amount of hysteresis as well, in other words, how they behave depends partly on the very recent history. The more a driver depends on mechanical damping (think gooey dome tweeter etc), the more likely it is that how it behaves also depends on what it had been doing just before, the ambient temperature and age.
I like horns, they are fun to design especially the modern ones at work where the ranges tie together into one source but one of the things that makes them work at high power is that they depend on the acoustic load for damping and that is pretty well behaved (comparatively speaking), only having the gas law as a complication .
While I love hifi, In a different area of loudspeakers where I must spend a lot more time, there are a couple things which to me seem like they may be useful clues or a direction to look for those with time and curiosity..
As the new building Fire codes are being applied, the concern for voice intelligibility is developing legal teeth. Who cares how loud an emergency announcement is if you can’t hear what it says, well, now they do and a lot of line arrays are coming down / not going up.
As our loudspeakers are used in large spaces because of the intelligibility and there is now a legal mandated measurement for voice intelligibility (STIPA), it made sense to attend a recent seminar.
I kept thinking and wondering what part of hearing information here would apply to music or stereo imaging?
It might well apply since the voice range is where our greatest acuity is..
http://dl.dropbox.com/u/2787059/ECSSIW_Brochure.pdf
Speech transmission index - Wikipedia, the free encyclopedia
What is interesting first of all is that one can have a music reproduction system which has little intelligibility.
A Choir re-enforcement system or organ reproducer in a Church or artificial reverberation system in a large venue are examples of very low stipa systems and stipa has NO impact on the pleasing nature or not because in that use, nothing is happening fast, there is no information being transmitted..
Ok, so I lean the other way, I want to be able to reproduce fireworks which are instantaneous events, that requires that the low mid and high frequency components arrive as they would have live, not re-arranged in time (which spreads out the energy envelope). I want to see every way possible what the mic senses as pressure is not what I fed into the speaker, where it is not a straight wire to sound pressure conversion.
To me, it appears what we should be looking at more carefully are the envelopes the energy comes in, not just instantaneous event like “impulse response”. For example, the presence of clipping of a short string of instantaneous peaks in dynamic music are not audible as clipping or even a fault until you compare to the same case not clipped, then the unclipped sounds somehow more “dynamic” not louder. What we hear / interpret as “loudness” as something more like a short averaged C weighted SPL measurement and NOT as the instantaneous Voltage waveform from a microphones measured as the actual pressure.
Audio information like word intelligibility??, where is that hidden? In part,contained dynamically.
The root of the stipa measurement is a series of Modulation Transfer Function measurements.
Here is a nice explanation of MTF from the optical world where the idea came from (and is then adapted to audio).
Modulation Transfer Function - YouTube
At the seminar, I asked the co-developer of the stipa measurements how high they had investigated the MTF’s. He said it appeared we could hear modulations much higher than used in STIpa (for voice).
I pressed him a bit farther and he said around 30Hz rate (above the stipa ranges) was as high as he thought was about as high hearing acuity extended (as his concern was only voice intelligibility and not corruption of a more complex signal like music, I suspect the audibility extends higher yet).
Many of you guys have ARTA which has a simple MTF measurement .
When I got home, for fun, at the listening position, I have measured a couple consumer grade (craptastic)2 ways in the same place as an SH-50, they deliver VERY different looking MTF’s with the little 2 ways falling apart at higher MTF’s. In this case, the much greater level room reflections would be part of what effects the 2 ways and comparatively destroys their stereo image at the LP (compared to near field listing ).
Since corruption of the MTF at lower rates is what lowers / destroys stipa, I think that is also logical / probable that the same kind of corruption is also tied into preserving / destroying the recorded information at the listening position we use to determine the phantom images from the recording.
Anyway, I don’t have time to pursue this avenue other than casually but I know some of you guys are keen to figure out what maximizes the recorded information that reaches your ears at the listening position.
Have a look at that, do some comparative measurements, see if it doesn’t look like conditions / loudspeakers which result in better / higher MTF’s sort of correlate to better stereo imaging (when you have a pair of the speakers).
By stereo image, I don’t mean a pleasing wall of sound but more like a large window to another space, one where the location of a voice etc between the speakers is much stronger than the awareness of the R and L speakers being the source of sound producing the phantom image.
Best
Tom Danley
Danley Sound Labs
John K,
I just so happen to agree with all that you have just written. I can not dispute your mathematics or your logic, it all makes sense. And in pro-audio yes the time constant is much longer I was talking about. But remember, even with a correct mathematical analysis of a physical function, sometimes there is a factor left out of an equation. Do I know of one to explain what we are trying to talk about, not at this time. I can only equate it to loudspeaker design and what I know. There are many instances where the lumped sum analysis may be correct in the general sense but it is missing multiple factors in reality. Every single material change and every single mechanical junction is not included and these do make a difference in the final resultant. Just changing an adhesive will change the measured response if you know what you are looking for and look hard enough for that change. But using the standard formulas for analysis will not show those changes. They are factors left out of the formula or a lumped sum is used and this can cause real rounding errors or missed interactions. I am not here to fight and only wish that I had your mathematical background, I am impressed.
I just so happen to agree with all that you have just written. I can not dispute your mathematics or your logic, it all makes sense. And in pro-audio yes the time constant is much longer I was talking about. But remember, even with a correct mathematical analysis of a physical function, sometimes there is a factor left out of an equation. Do I know of one to explain what we are trying to talk about, not at this time. I can only equate it to loudspeaker design and what I know. There are many instances where the lumped sum analysis may be correct in the general sense but it is missing multiple factors in reality. Every single material change and every single mechanical junction is not included and these do make a difference in the final resultant. Just changing an adhesive will change the measured response if you know what you are looking for and look hard enough for that change. But using the standard formulas for analysis will not show those changes. They are factors left out of the formula or a lumped sum is used and this can cause real rounding errors or missed interactions. I am not here to fight and only wish that I had your mathematical background, I am impressed.
JOhn
I do misunderstand the physics, I bounced my model off of my PhD thesis advisor who is a nuclear physicist. He agreed that the output level has to modulate and that all speakers would exhibite the effect some vastly sooner than others.
We are talking different time scales. I do not see a transient in my discussion being less than a ms, if that's your time frame then I have no disagreement. But I do not believe that you can say that a .1 dB change (the JND for level is on the order of tenths of a dB) over a time scale of 100-200 ms is NOT audible, but is certainly possible. I don't think that any of us can say one way or the other without doing the tests (which I hope to do).
Tom
Nice discussion. One thing here though is how sensitive a good crossover design is to the tweeter response both acoustically and elecrically. The changes that I saw in my tests were dominately arround the crossover and then again at the upper edge of the bandwidth. Almost nothing in between.
That is very very interesting. I can certainly see what you are saying and it certainly correlates with my experinces as well. I simply do not think that we can write off this area so easily.
In the MTF test system nonlinearities would also tend to lower the MTF as well as thermal so it might be hard to sort out the two things. I was trying to devise a test that would do that, but found that it would be very difficult.
I understand that, but I disagree. Most people rate the dynamics of a speaker in subjective terms - they just listen to it. There is a consistency across several decades and cultures that high efficiency speakers sound more dynamic than low efficiency. That's a common subjective assessment, so we might ask "why?"
Why do so many people tend to find HE (and horns) more dynamic sounding than other designs? Are they simply wrong? If so, what leads to the error?
John K's ideas sure make sense to me, it does not seem that voice coil heating could be fast enough to effect dynamics. Unless there is more to it than that.
It's an interesting area of study, for sure.
Clearly all these thermal effects etc are completely secondary of importance compared to listening room influence. How otherwise IMP would have gained what it did ?
- Elias
exactly
Pano,
There is no question that in a compression driver or a cone driver on a horn that the actual movement of the diaphragms is much less for the same spl output. I think this has a great deal to do with the difference between a horn and a direct radiator in the sense of dynamics. The instantaneous acceleration and the linear travel make a difference here. The rise time that the wave needs to produce the sound is what I think we are talking about. Leave out throat distortion and such and just look at how much less you have to move the diaphragm, things can just happen so much faster. I started with waveguides, was convinced that consumers wouldn't touch horn loaded systems anymore and moved over to direct radiators. But I must say with all the enthusiasm for horns here on this site I am looking at bringing back some of my earlier work. Knowing what I do now that I didn't then I know I no longer have to put up with the nasty horn honk of old, electronics and active crossovers and time correction are much easier to accomplish compared to the days of passive networks and trying to do time correction in a passive method. Long live waveguides and all they bring to the party.
Steven
There is no question that in a compression driver or a cone driver on a horn that the actual movement of the diaphragms is much less for the same spl output. I think this has a great deal to do with the difference between a horn and a direct radiator in the sense of dynamics. The instantaneous acceleration and the linear travel make a difference here. The rise time that the wave needs to produce the sound is what I think we are talking about. Leave out throat distortion and such and just look at how much less you have to move the diaphragm, things can just happen so much faster. I started with waveguides, was convinced that consumers wouldn't touch horn loaded systems anymore and moved over to direct radiators. But I must say with all the enthusiasm for horns here on this site I am looking at bringing back some of my earlier work. Knowing what I do now that I didn't then I know I no longer have to put up with the nasty horn honk of old, electronics and active crossovers and time correction are much easier to accomplish compared to the days of passive networks and trying to do time correction in a passive method. Long live waveguides and all they bring to the party.
Steven
While I don't have experience across "several cultures" I do have experience across (more than) several decades . . . and what I have observed is that every time questions of "dynamics" arise the mode of demonstration is "turn it up". Then whichever speaker wilts first is declared the "less dynamic". It really is just as simple as that. "High efficiency" speakers do not sound more "dynamic" than "low efficiency" speakers (of otherwise comparable quality) if they are played at the same (within the working range of both) level.
It is the realities of our signal capture and delivery systems that determine what "dynamic range" is available . . . and that is superimposed on the noise floor of the demonstration environment. Louder, then, appears more dynamic because it rises further above the background (reference) sound level.
dewarth,
Loudness is always a tricky thing in comparisons of speakers. The louder even if it is only a couple db will often win. But it is more complicated then just matching the spl level. All it takes is for one of the systems to have a peeked or raised area in the frequency response curve for one speaker to win over the other. I hope this is not what we are speaking about with dynamics and I don't think this is what Pano and the rest of us are trying to describe. I would put dynamics more in the camp of instantaneous transient response, very fast rise time in a signal.
Loudness is always a tricky thing in comparisons of speakers. The louder even if it is only a couple db will often win. But it is more complicated then just matching the spl level. All it takes is for one of the systems to have a peeked or raised area in the frequency response curve for one speaker to win over the other. I hope this is not what we are speaking about with dynamics and I don't think this is what Pano and the rest of us are trying to describe. I would put dynamics more in the camp of instantaneous transient response, very fast rise time in a signal.
I don't know. Most of the traditional dogma of what makes good sound quality has been shown to be false, things like THD, etc. that I don't believe that "It really is just as simple as that."
But 105 dB from an Orion without some serious degradation is really optimistic. SL said in a talk that his target was more like 90 dB. I questioned him on this level saying that I thought it was too low. He just said that he didn't listen all that loud. Well that doesn't work for me.
That's fine . . . preternaturally loud (amplified) sound is certainly "in" these days (I hear it oozing from passing cars with distressing regularity) and everyone is of course welcome to their own preference and taste. And there's no question that horns (of the sort you design and sell) do "loud" far in excess of anything ORION (or the newer LX521) can muster.
I'm only questioning whether it's correct to call that "more dynamic" . . .