Hello Jean-Micheal,
It is true that different people will perceive and describe dynamic quite differently, this is why it would be interesting to get these into one room and watch the confusion.
There is a problem when drivers get close to the extremes of BL and Km, the nonlinear characteristics sort of causes high frequency harmonics to vary depending on where the VC is at the time, this creates unpredictable perception. I have heard some design configurations to sound audiophile dynamic, but the music as a whole does not sound right, and to make things more complicated, stores get really excited saying that this will sell, like a foxy lady with a fancy chassis.
What I have learned listening with different auditors is that people have there own way of pointing out very specific issues when listening to a system. This was very valuable to me. I also look at TV sets the same way, I see different problems with different TVs regardless of price. Some seem amazingly flat but colorful, some seem to have depth, but the color not right, then there are those that leave too much shadow in players move around causing blur. Sometimes I just wonder why they don't take care of those technical issues, but then there is a magic word I remembered, differentiation! Marketing wants to see differentiation, that is what they learn. If everything looked perfect enough, then the only thing to compete with is price. This is the last situation companies want to be in. Regardless whether audio or video, I personally find a way to get performance as close to the feeling the original performance as financially practical, and really hope that consumers are picky enough to do that. This is the only way companies that put forward the effort will excel. But right now, the user base is diverse on preference, and at lease a small group really want to feel the music, this is where I spend lots of effort to accomplish because it is what I wanted ever since I got my first audio equipment.
Listening is a very important part of development, and there is really no shortcut to acquiring the necessary experience. If one gets the experience, one can listen and make a good decision on what tests to make to verify the source of a sound quality problem. Lots of professionals will dismiss certain audibility issues by hanging their hat on somebody else's research, I think it is the wrong attitude.
Is 120 db important? Well, I believe it is not that absolute number that counts, rather it is what happens in that range. I trust that there is less problem in the digital technology, but rather the limitation starts from converting digital to analog throughout the analog section. Quite often people might need to do something to the source recording to compensate or compromise for those limitations. I have spent a significant amount of time trying to get an amplifier to work reliably at different listening levels from very low level up to listening levels challenging the limits of the drivers. As a mater of fact, it was even necessary to tune the frequency response in a way that will provide maximum usable SPL range of the driver providing a balance between the bandwidth. It is also necessary to want the same clean and natural sound at high volume levels as well as low. Power supply matching and layout played an important role here, again, lots of things considered that don't show up in research and textbooks. Being an audiophile my self, I found myself with a maze of cables, each attracted me at different times, but finally I had to go and get into that area as well.
Quite interestingly, most research focus on very niche parts of technology, and my specialty is integration of technology, this is why it seems easier for me to get a feeling how to integrate some ideas, understand the fundamental nature of the issue, then integrate the understanding to develop solutions. Things would be easier if a strong team were in place, but that really only happens by chance. I would say Steve Jobs was pretty lucky. Apple products are moving closer to my liking since I got my first Apple II+, at the time I got it! it cost pretty much the same I would pay for a MacBook now.
For Beyond the Ariel, I really think better integration of the analog section and good listeners with combined efforts are necessary if anything is going to come out of it. I do not see any reports on improvements and missed hits on this project, and from the discussion going on, it seems that the project is in a standing circle where things are changing, and there is no idea what part is right. This is quite different from how I am used to see work done. Generally, we will have a set of design guidelines which are established based on technical reasons, these criteria can change if there is a technical reason for the change, and in the process of development, more design criteria will be added as experience is accumulated. What this process does is that it reduces the cross effects of technical improvements allowing the improvements to be accumulative. For example, always using stiff diaphragm is a criteria, yes. It may cause problems. But it is then necessary to solve the other problems rather than shy away and revert to soft material. I am still interested to see what will result from this thread.
As I have said previously we are probably all in general agreement. I certainly support the views you project here.
I believe we must follow the science, e.g your example stiff diaphragms.
There are a major number of definitive elements that must each and everyone be considered. And that we have to judge which are singulary or plurally most important. It seems to be a general shifting of each parameter up the scale so that the integrated whole is achieving a more truthful sound to be reproduced to give the best aural experience.
The digital paradigm set the whole process back at first decades ago, but that is now changing rapidly with major strides in all the recording and reproduction chain.
The current digital and analogue improvements have finally established a force majeur on the speakers capabilities.
This has helped all types of speaker as the sometime special characteristic say electrostatic, Heil, Horn can be heard closer to its inevitable unique character. Its like our genes being allowed to express themselves more perfectly in new records being set for performance.
So for Ariel the bewildering scope for improvement may start as a circular review until we can see a way move well beyond Ariel.
To take just the horn philosophy do we still believe horns are worthy of more development. I say yes. We finally generally stopped using CRT monitors in favour of LED LCD plasma etc
With the technology revealing the unique properties of the horn speaker it is a make or break time. I would probably never revert to my Quad ELS57 because of its sound and its physical considerations. If I do not make a horn speaker now I probably never will. The horn has to do something no other speaker can do, and I am getting great results with direct drivers.
Last edited:
The part of the horn that attracts me is having a more pure wave front. This is something that direct radiating drivers are physically not so capable of. But also difficult in horn design is trying to get the delayed release of stored energy to decay as fast as possible. Having reviewed some tests it seems to me the task would be at least 3 times the effort of the direct radiating driver I have in mind, but some of the technology is going to overlap.
As I recall Dr. Geddes had mentioned, he has better result with drivers of lower compression ratios. This is reasonable because the compression is a form of storing energy, so higher compression ratios are more difficult to suppress delayed release of such energy. This means we want a driver with only the least amount of compression as necessary. In my mind, I would actually design a special 3" wide range driver and put a wave guide in front of it. I did not take that route because that driver will be optimized for wave guides, and I am not sure what I would do with it alone, and the waveguide itself would be pretty large unless I can design it to blend into a room corner. All these are things that just run through my mind while estimating the effort involved. So putting it later in on is much better for accumulating more driver and diaphragm design experience. But I would also be interesting in listening to other design concepts as well before I get to that.
As I recall Dr. Geddes had mentioned, he has better result with drivers of lower compression ratios. This is reasonable because the compression is a form of storing energy, so higher compression ratios are more difficult to suppress delayed release of such energy. This means we want a driver with only the least amount of compression as necessary. In my mind, I would actually design a special 3" wide range driver and put a wave guide in front of it. I did not take that route because that driver will be optimized for wave guides, and I am not sure what I would do with it alone, and the waveguide itself would be pretty large unless I can design it to blend into a room corner. All these are things that just run through my mind while estimating the effort involved. So putting it later in on is much better for accumulating more driver and diaphragm design experience. But I would also be interesting in listening to other design concepts as well before I get to that.
Hello,
I am sorry to say but this is highly doubful IMHO or the problem is not well enough described.
If we give a look to the electrical equivalent model of a compression driver, there is very few place for any noticeable energy storage...
And apart electrostatic loudspeakers, I would say compression driver acoustical loaded by a resistive load have a far better impulse response that any directly radiating loudspeaker.
The only effect that could eventually be considered as some storage of energy in compresion drivers happens only at very large SPL level and is due to the non linearity of the air itself and at the origin of some H2 distortion added. At domestic level listening this added H2 is generally not audible.
Now may be you are speaking of delayed signals due to the reflectance at the mouth of the horn. If the reflectance at the mouth is large, then multiple replicas of the initial Impulse Response can be observed on the IR response with few miliseconds interval from one to the next and for sur this is audible.
But this can be solved using a horn having very little reflection as the JMLC horn (lecleach horns)...
Best regards from Paris, France
Jean-Michel Le Cléac'h
Digital paradigm has done nothing but move creativity and scientific process forward.
Parallels may readily be drawn between compression drivers and horns to CRT: Compression driver is hot filament, focusing yoke is phase plug, steering plates are horn/waveguide, and phosphor coated screen for which goal is uniformity across extent of radiating surface.
Solid state displays are readily identifiable with direct radiators in concept.
Early bottle necks were complex interaction of interfacing technological and intellectual control, and scale of dissemination and of manufacturing. Control of knowledge is control of power and money. Audio was very low on list. Military uses were at the top. High stakes are in medical field too.
Trickle down landed digital in recording studios as primary entry point into audio, starting renaissance through rest of the audio field.
Loudspeakers for domestic use is limited market, easily flooded with commercially manufactured speakers. Large scale commercial interests continue as thrust for further driver development. Automotive manufacture is large market with better predicted demand then domestic as target for new audio systems.
Focus in domestic systems by major manufactures is home theater, with a proliferation of highly integrated systems utilizing advanced DSP techniques and advancing auto EQ systems with increasingly friendly user interfaces.
Time and space for listening at realistic levels to music verges on high luxury making big speakers hard to market. Automobile enables multitasking; with ability to spend time with head in relatively fixed position.
Innovation and art always start with the individual. Delineation of horn/waveguide creations of enthusiasts from commercially arrived at solutions using scientific methods to communicate possible improvement is only productive route. Subjective impressions unquantified and defined by scientific principles indeed leads to fruitless talking circle.
Yes, it is highly important to bring like minded individuals together, and socialization is crucial to human fulfillment, but in scanning through this extensive thread I see little to no sharing via scientific metric.
"dynamics" is simple recent example. Dynamics in audio: principle application of term to primary behavior of audio systems as linear is straight forward. Beyond compression drivers that derive operation based on non linear principles, rest of audio reproduction strives for linearity.
This leaves usage to properties of dynamic range of non linear output systems. Sure is easy to have zero input, and 150dB output with non linear system. Loads of dynamic range. And sure, it is difficult to make highly linear system capable of 150dB output. Lots of driver surface area, easily achieved with arrays. Quantifying performance differences of the two systems in scientific fashion is very straight forward.
Dynamics of non linearity as how do non linearity mechanisms change with signal level; as in "static non linearity" v "dynamic non linearity". This is not my connotation in phrases such as "Speaker A is more dynamic than speaker B." Both forms of non linearity may also be assessed in scientific manner, and metrics for audible differences defined that allow for defined study, isolation of controlling variables and a path forward in terms of speaker/driver implementation.
Linearity in king of fidelity.
A large hemispherical driver array with high linearity may be fixed to mouth of suitable horn, and linearity studied at both mouth and throat. Such a simple idea, it seems to me something that has likely been done. Even so it would be an experimental approach worthy of an enthusiast's repetition as potential means of further enlightenment.
Repetition and duplication are kings of science.
When I study driver performance, linear impulse response is recovered along with harmonic distortions, as these always are revealed in presence of non linear mechanisms. I measure these for differing static drive levels. Audibility of distortions present when measuring typically don't become readily discernible until drive levels are tragically high. Auralization by listening to measurement response recordings normalized in level with headphones often reveal that changes in perceived sound are largely do to non linearity of hearing mechanism. Dr. Geddes has written on his studies based around this auralization technique. Pure sine wave testing is highly revealing of audible harmonic distortion, and indeed in parts of human hearing spectrum very low level harmonic content is easily heard. Likewise the presence of two tones, with one at very low relative level are readily discerned over extensive parts of hearing spectrum.
With broadband signals as with music, distortion discrimination often requires drive levels beyond intended levels of speaker/driver application. As such this becomes a metric for determining general limits to the application of specific speakers and drivers.
Synthesized broadband signals are my preferred approach for measurement and study of non linear distortion. All non linear distortion mechanisms produce intermodulation distortion, much of which is readily audible with speakers/drivers at drive levels that are often well within those of test unit's intended application. Audibility is confirmed by auralization with listening to normalized playback of response recordings that rule out non linearity of human hearing as cause.
Linkwitz is big fan of sine waves with raised cosine envelopes for generating impulse burst signals for exploring non linearity. These bursts are highly similar to bursts that may be isolated from continuous waveform of tripple tone signals with closely spaced frequencies.
I have worked with both types of bursts and find them highly limited due to sparse spectrum of highly non linear amplitude content. Signals are very similar to signals from instruments capable of producing chromatic scales.
Highly percussive instruments produce very broad spectra with numerous components of similar amplitude. Percussive instruments in combination with chromatic instruments lead to audible speaker coloration in highly level dependent manner. Duration and timbrel qualities make quantifiable description difficult if not impossible between disassociated people. It is often possible and relatively easy for one person to play a specific piece of music, or short segment where coloration is personally apparent that changes with volume, or with different speaker, and be able to directly point it out to another person who is there. Garnering together, and quantifying such observations into useful metric may be possible, but apparently close to impossible with musical source signals.
I don't believe this to be true when synthesized broad band signals with spectra having well defined properties of bandwidth and magnitude are used. Such signals are readily produced using FIR filters, and are my preference when studying drivers for potential to color sound perception.
Last edited:
Hello Jean-Michel,
I generally look as CSD to see how fast it decays within something like the first 0.3 ms, then I look at the 3ms range. It seems to me when the horn has very little reflection, the CSD in the upper half of the spectrum is improved. The closer to the low end cut off frequency, the less it helps. While I am sure there are some good designs out there, it is not easy to accomplish.
I seldom use equivalent models. From some of the BEM simulation I have shown in the Geddes thread, what you say about mouth reflection seems true, but I do not remember seeing them in my measurements. Perhaps you have seem them in real measurement?
I have no idea what harmonic sounds like during listening simply because I have not tried to relate a certain harmonic curve with listening impression, to I cannot say whether H2 is audible or not, but when diaphragm displacement volume becomes higher, then you start to have compression. What I think is it is best to keep this to as low as reasonable. Since the low end of the frequency can get close to the mechanical resonance of the driver, some additional attention must be considered. The air volume in the horn acts like mass, and together you have a spring mass characteristic.
Impulse response of direct radiator and compression driver with similar intended pass band are readily equalized to make them identical.
Linearity is fundamental even when non linearities may be added deliberately or unintentionally usually the latter. No debate about the dynamic range issue It is immaterial whether this is 120 dB or 150db other than if you want to hear a spider climbing arouind inside a guitar while it is being played.
Linearity can be analague or digital but we haveto be able to convert back and forth and each time we lose something. Like the spider it may not matter to have digital artifacts or analogue ones below our audibility limits.
These days the economics has become master and sentiment give a boost to some forgotten interest for may say horn speakers, but start offering them as an art form then the market place jets this into the want to have bracket then the money comes in and really develops the goods beyond our wildest dreams. The Ongaku amp may be an example.
But humble simplicity is always there somewhere. Look how simple the current drivers are but we have better materials from aerospace, and military that have paved the way.
Music is analogue either all the way through to our hearing, or it goes through at least an AD DA process. This can only approximate the original analogue sound. The art is to speed up the binary or otherwise coded process beyond the limited bit rate of our neuron synaptical process and hearing components and its filters that digital will sound identical. This has been going on for some years and is now bearing fruit, from the aerospace military developments since perhaps the last war.
Perhaps the next stage is no driver in a loudspeaker but rather a plasma device or other electron movement to sound emssion device comes along. And this may be nearer than we think.
But for now we have enough technology and materisl to improve every existing type of loudspeaker or any other device digital or analogue if the economics of profit is envisaged like happened with Bill Gates.
We cannot yet see what will blow away our current ELS, Horn or direct drivers but NEO magnets are giving a renaissance to direct drivers and for horns and it may be that the horn or may be short wave guide is going to be the winner. Genelec, Behringer (B2031P and A versions) and others.
IBM computers were blown away by PC s and modern economics
That is precisely my take on the situation as well. There is no theoretical support for this claim either.
I don't recall saying that. Not that it is not true. There are no compression drivers with different compression ratios so there wouldn't be any data. In theory a lower compression ratio will reduce the air nonlinearity, but I wouldn't give this factor much weight. The higher compression ratios increase efficiency which lowers a lot of other problem like thermal modulation. I would tend to think, without and data mind you, that the higher efficiency would be my preference over lower distortion (which was not found to be audible.)
I have done those studies and the lower the order of the nonlinearity the less audible it is. that's not a guess there is substantial data to support it. Massive amounts of second order nonlinearity, up to 20-25% were found to not be audible. Tenth order is audible at a very tenths of a percent.
It may be comparatively audible when compared IOW with the source minus the 2nd order HD. It has always been said that 2nd order helps to combat higher order HD.Negative feed back knocks out the 2nd order but the discordant higher order stuff then becomes less tolerable.
My problem with compression drivers is that they can generate in a horn high levels of 2nd order HD although having some does sweeten the sound. Like with SET rather than a push pull amplifier cancels out the 2nd order HD This is academic with the best push pull amps where there is low NFB but a good low distortion design.
Yes, it would apply to both.
The data says that nonlinearities in compression drivers is inaudible.
I don't understand.
The nonlinear orders are all independent of one another, but the harmonics are not.
It can be shown that feedback will decrease the lower orders substantially but can increase the higher orders slightly. This may not be the best audible approach.
Earl,
I would agree with data that says nonlinearities in compression drivers is inaudible at low drive levels used in typical domestic environments, but driven at anything close to manufacturer's rated power levels, the distortion is clearly audible.
You can hear it in the recordings:
http://www.diyaudio.com/forums/multi-way/212240-high-frequency-compression-driver-evaluation.html
At low levels, the compression drivers sound like the recording, at high levels the differences between the drivers become apparent.
Give the recordings a listen, I'd be amazed if you found the high drive level distortion inaudible.
Art
Inaudible to me IMO would mean without a direct comparator that it is not audible.
But if you have the same source minus the 2nd order HD it will sound less sweet by direct AB comparison Depends on level of HD esp with 25%.
Now any 2nd order HD with a violin or piano etc would likely add its own harmonics to the 2nd order HD from the horn and or amplifier enhancing the harmonics of the played note and making it seem slightly brighter or sweeter.
This effect applies with SET that generates much 2nd order HD resulting from low NFB but can make the perception of the sound more enjoyable.
If you get too much from the SET and horn speaker then you can get a significant difference in the sound, maybe this is part of the 'horn' sound we muse over with long horns apart from back reflections
Art
I think that maybe you should read our AES paper first.
We tested compression drivers only, no horns or waveguides. We tested at the thermal limit of the devices, about 140 dB in the plane wave tube (posed some real problems doing that). The correlation between nonlinear distortion and perception was not significant. Frequency response difference were. Its all in that paper. Many objected to the results, but no one has refuted them.
Your observations are exactly what led me to study HOMs. Because, it is obvious that compression drivers on horns sounded worse as they were turned up. But, given the results of our B&C study I had to conclude that it was not the compression driver. It has to be the horn and or the horn compression driver interface. At the time I was not at all clear on what it was.
Then Lidia and I studied the potential for the audibility of HOMs. Low and behold, they DID become more audible at higher SPLs, even though they are a linear phenomena. This means that your results could easily be HOMs. Since you only tested horn driver combinations you cannot conclude which it is. But if you consider that our results indicate that it isn't the driver, then the obvious conclusion is the one that I came to some ten years ago.
Further, attempts to minimize the HOMs and all forms of diffraction have resulted in waveguide loudspeakers that do not sound worse as you turn them up, they just get louder, right up to the point where they are actually painful.
To me the results are conclusive. You can draw your own conclusions I suppose.
Well, then maybe somebody in your wave guide thread mentioned it, or you might have discovered something new.
I am sure it is true, but I seldom reference that kind of information because I cannot relate it with specific music listening sessions, nor do I know the necessary performance data of the equipment used for the tests. It is often true that the learning capability factor of humans is ignored during these kinds of tests. The experience factor has a major influence on any capability to detect flaws, it's like some people will pick out smaller flaws in paintwork than others. One paint shop could not figure out why their enclosure paintwork kept being rejected until I showed him side my side why. Explaining to the enclosure contractor did not work, they just got the paint shop and I together. When it comes to listening, the auditor just tell me what is wrong with how it sounds right there, then the technical issues are dealt with.
The audibility guidelines are good for initial design minimum performance requirements, but cannot be used as a reason for no need to consider in the improvement process.
Last edited:
Questions for Gedlees Company:
http://www.diyaudio.com/forums/multi-way/100392-beyond-ariel-869.html#post3478359
Douglas Self mentioned that development of good audio equipment is a process of reducing flaws. I think when something is audible or not depends the kind of flaws existent in the equipment. Some are going to be more dominant than others, so the less dominant ones are going to be less audible until the more dominant ones are reduced to be less dominant. This is the improvement cycle I generally take.