You see this is not at all my experience and if what you said were true then my claim of inaudibility would be false. I almost never see a speaker that is not dominated by 2nd and 3rd order. Some years back a client of mine did a gauge reliability study of the Kilppel system using a single driver measured several times. The second and third orders that it calculated were stable, but nothing higher. These higher orders have inherently built in limitations in mechanical systems because of inertia and the like. Very high forces are generated by high orders of nonlinearity. In an electrical system theses forces are voltage and there is almost always excess voltage available to generate these higher orders. In a mechanical system these forces are difficult to generate since the system does not have much "force" to spare. This is why mechanical systems tend towards low orders of nonlinearity.
Your example tweeter was unique, I have never seen this level of distortion in any driver that I use. I would be tempted to call it broken, but without first hand knowledge of the conditions, I would refrain from that conclusion. But if you have other examples of drivers which have higher orders comparable to the 2nd or 3rd and cannot be called broken, then I would like to see that data.
Even a transfer function with a pure 2nd order harmonic distortion (which is highly unlikely in a real system) creates a "nice" intermodulation distortion product/products if excited by 2 and more frequencies. Well known CCIF test. So, a tweeter with several % of 2nd order harmonic distortion creates audible difference tone in this test. This results in colorations of the sound. So, we should not end up with linear issues only and rely on masking.
I agree that it is dominated by 2nd and 3rd order. But dominated does not automatically exclude higher harmonics. With higher resolution of our measurements, we find those higher harmonics quite easily.
Attachments
If the kind help of gedlee did not help so far, please point in details to what needs clarification. It must be some of the terms I used, otherwise it should be clear as.. icelandic spring water!
Think scope and SA and what you can see with those (oscilloscpe and spectrum analyzer).
Even IM products are masked - the lower the order the closer the products lie to the masking signal, thus they are lower in audibility in all cases.
And the harmonics mask the higher harmonics, meaning that the orders must go even higher when there are dominate low orders present.
Of course higher harmonics exist, the point remains, and I find that I have t keep saying this, that does not mean that they are audible. All the tests that I have done say that they are not. If you have scientifically valid tests that say otherwise then fine.
And the harmonics mask the higher harmonics, meaning that the orders must go even higher when there are dominate low orders present.
Of course higher harmonics exist, the point remains, and I find that I have t keep saying this, that does not mean that they are audible. All the tests that I have done say that they are not. If you have scientifically valid tests that say otherwise then fine.
so what i perceive as audible distortion in horns in comparison to a direct radiator given that spl and frequency response have been matched/equalized in fact doesn't exist? or simply is considered inaudible? i guess i need to remind myself that my perception is flawed but given that i can hear that something is different regardless of level (not max or at clip levels) if i can't call it distortion what do i call it? (and don't say it's a delusion)
It is "distortion" and I hear it as well, always have. But when I tried to measure and quantify it I found that the nonlinear stuff cannot be claimed to be the culprit. I believe that the culprit is in fact group delay problems in many horns caused by HOMs and internal reflections. These are linear distortions that have a nonlinear perception. You perceive them more at higher SPLs just like you would if they were nonlinear distortion. That's why people have always associated them with nonlinear distortion.
It has taken me almost 40 years to get to this point of understanding. I wrote my first paper on horns some 35 years ago. Then I have written several more since and a book, all on this topic. I used to think, just like you, that it was THD that causes these horrible sounds, but after doing the work, I have to conclude that it is not nonlinear distortion that is the problem. It took me decades to come around to believing this, maybe in another decade or so you will too.
People who hear my waveguides almost invariably comment on how much they sound like a good dome tweeter - only unlike a dome tweeter they have some 20-30 dB more headroom and better directivity.
Last edited:
Earl,
Since you keep insisting "all the tests that I have done say that they (higher harmonics) are not audible", I feel compelled to again point out your test's voltage levels covered only a 6 dB range, tested at three levels, 14, 20, and 28 volts, (12.25 to 50 watts into 4" diaphragm drivers mounted to plane wave tubes) resulting in only a 6 dB range difference, and only a 6 dB range of distortion from approximately 10 to 18% at 1 kHz, and 18 to 31% at 6 kHz.
Since those levels of distortion were not compared to a low distortion level (like 2.83 volts) or the reference recording there would be little chance of anyone telling them apart. The source recording (a live recording of Talking Head's "burning down the house") has plenty of distortion to start with. You have still presented no explanation of how subjective distortion perception results could be derived from that protocol.
The protocol used in Klippel's test allows users to determine at what level various distortions can be detected with a variety of music and test tones in a variety of speaker types.
Listening Test
The distortions included in the test are Bl(x) (the variation of the Force Factor with displacement), CMS(x) (the variation of suspension Compliance with displacement, LE(x) (the variation of voice coil Inductance with displacement), para-inductance, thermal power compression, jump out effect of the voice coil, nonlinear compression of the amplitude, and the complex interaction between the different nonlinearities .
The test results clearly indicate that a majority of listeners can detect distortion at levels that were lower than those used in your test, and depending on the source and speaker type, that hundreds of listeners detected distortion as low as .1%.
Have you conducted any comparative tests with multiple individuals that prove linear distortions (such as HOMs) that have a nonlinear perception are perceived as "more horrible" sounding than nonlinear distortion?
Having taken several of the Klippel tests, they seem indicative of real world speakers with distortion at various levels, and that distortion at the level of the drivers in your test is certainly objectionable (if the source is not already distorted) to me.
Do you have some reason to believe the Klippel test results are not indicative of the perceptions of the thousands of people that have taken it ?
Art
Attachments
Last edited:
well i hope that i understand all this better before another decade passes because by then age related hearing loss might make it moot.
i don't think that high levels are necessary to make horn distortion audible or that increasing level makes it more perceptible,in some instances i have noticed that until a certain minimum level was surpassed i couldn't (or wasn't) able to have what i would call "linear operation" or proper response, more succinctly a range of spl over which it behaved "linear"
in trying to further my understanding of horn design doesn't narrowing the coverage angle (increasing the directivity) have the penalty of increased distortion? or is that something else i'm wrong about?
i don't think that high levels are necessary to make horn distortion audible or that increasing level makes it more perceptible,in some instances i have noticed that until a certain minimum level was surpassed i couldn't (or wasn't) able to have what i would call "linear operation" or proper response, more succinctly a range of spl over which it behaved "linear"
in trying to further my understanding of horn design doesn't narrowing the coverage angle (increasing the directivity) have the penalty of increased distortion? or is that something else i'm wrong about?
Sure you can, it is easy. You just double tap the name, and type in the new extension. It will ask if you really want to. This works whether you have the Finder set to display the extention or not.
dave
You can't change the name of a file that is downloaded in an iPhone though - that is where I am having problems. It's supposed to be the same OS core on all Mac or i products I thought.
Where to Arts post go? Oh # 109, it jumped back while I was writing.
The Klippel tests are contrived and not real tests of real speakers in situ. It would never be accepted as a valid data point in the scientific literature because of its lack of controls. Would these same result stand up under scrutiny, I don't know. I know our test was rigorous and short of your complaint that the original signal is too distorted to tell any distortion (which you have no proof of), the test was well controlled. If it does not agree with other tests then I am not sure what to say except to ask if these other tests were as well controlled. If you are trying to get me to agree that my test was all wrong because of the source material, well, that's not going to happen. It is what it is and I'll stand by it.
I am not sure what this means, I don't think that I ever said it, and of course no tests like that have ever been done on a scientific basis - no one cares.
The Klippel tests are contrived and not real tests of real speakers in situ. It would never be accepted as a valid data point in the scientific literature because of its lack of controls. Would these same result stand up under scrutiny, I don't know. I know our test was rigorous and short of your complaint that the original signal is too distorted to tell any distortion (which you have no proof of), the test was well controlled. If it does not agree with other tests then I am not sure what to say except to ask if these other tests were as well controlled. If you are trying to get me to agree that my test was all wrong because of the source material, well, that's not going to happen. It is what it is and I'll stand by it.
I am not sure what this means, I don't think that I ever said it, and of course no tests like that have ever been done on a scientific basis - no one cares.
Last edited:
If you are hearing "distortions" in a horn at all level, then, by definition, it is linear distortion.
Narrowing the coverage angle would make a small incremental increase in 2nd order nonlinearity because of the decreased flare rate, but this would even be hard to measure given the nonlinearity of most compression drivers. It would also increase dramatically with level.
Really!? Is this how childish this tread has become? Does this have even the slightest relevance to the Topic?
turk, I have numbered places in your post that need discussing and are open to empirical observation. An SPL meter or SPL meter Ap for a cell phone is handy.
[1]if it's a badly designed or badly made horn this may be so.
[2]But if that is the case, it will sound even worse the more you turn up the SPL and will eventually drive you out of the room.
Why do I say that? Because a well designed and made speaker won't sound "too loud" until you have it playing way over 100 dB at listening position. The sensation of "too loud" is caused by hearing distortion - no sensation of "too loud", no audible distortion.
[3]This statement is difficult to understand perhaps you could express it in another way.
[4] Historically, horns were were used to amplify sound by making the connection between the driver diaphragm and the air more efficient because electrical amplification was difficult and expensive. Electrical amplification is now cheap for homes and other smallish venues and horns are used in them now also for wave guiding or directivity control. Standards have risen and folk don't like the distorted sound of old fashioned horns.
The distorted sound characteristic of horns is caused by refraction of sounds from the sides, from the edge at the mouth/baffle boundary, and reflection from the impedance of the mouth back to the driver diaphragm.
This refraction/reflection are delayed parts of the signal which comes directly out of the horn to the listener. And, although there may be reinforcement and cancellation of some frequencies, no new harmonics are produced, this is an acoustic phenomenon only. This is a linear process.
Acoustical problem requires an acoustical solution.
Earl Geddes has provided some solutions. He showed from first principles that there is a horn wall profile that produces the least amount of refraction/reflection, and that rounding over the mouth/baffle boundary produced lower level refraction, that ensuring the driver horn connection was as smooth as possible, and that it was possible to diminish the level of of remaining refraction/reflection produced inside the horn by filtering it with open celled foam.
He wanted to call horns developed through his mathematical formula waveguides to distinguish them from the traditional horn developed out of a different equation but that really didn't happen.
So to answer your question, it depends on the profile of the waveguide for the coverage angle how much distortion is produced. It's certainly true that if you want a narrower coverage, the waveguide will be longer.
I'm running out of time. There is a very lengthy thread entitled 'Geddes on Waveguides', in the stickies I think, at at the Diyaudio Multiway Forum. You should look at it.
I think this bit's worth repeating: " I believe that the culprit is in fact group delay problems in many horns caused by HOMs and internal reflections. These are linear distortions that have a nonlinear perception. You perceive them more at higher SPLs just like you would if they were nonlinear distortion" - thanks Earl.
Q: do the 'HOMs and internal reflections' have an analogue in cone speakers? - breakup modes?
Q: do the 'HOMs and internal reflections' have an analogue in cone speakers? - breakup modes?
The sides and the middle of the cone can drive energy off axis when they are not in phase.
The outer and inner cone can form a dipole, which depends on phase for its directivity.
This was with 2.77 Vrms only, at speaker terminals. Amplifier distortion is always checked during these tests to make sure it is a speaker distortion what we measure. I do not think the coil is very close to the end gap, the driver is operated far below its limits.
- Status
- Not open for further replies.