Is there intrinsic difference sound of horn tweeter vs. conventional tweeter?

[tman]

I'm looking at the Hsu HB-1. One comment people make is that horn tweeters sound bright. I've heard Klipsch and would agree that Klipsch are slightly bright. Is there any reason that a horn tweeter will intrinsically sound bright or is it a matter of implementation?

Anyone out there who really understand physics of speakers care to comment in detail? Personal opinions on Hsu HB-1 are in other posts. Didn't find anything in forum that discusses the difference in tweeter design.

Thanks,
Tman

 

[tbewick]

I'm looking at the Hsu HB-1. One comment people make is that horn tweeters sound bright. I've heard Klipsch and would agree that Klipsch are slightly bright. Is there any reason that a horn tweeter will intrinsically sound bright or is it a matter of implementation?

I think it's a matter of implementation. I've used Tannoy horn's and they have a relatively warm sound.

Anyone out there who really understand physics of speakers care to comment in detail? Personal opinions on Hsu HB-1 are in other posts. Didn't find anything in forum that discusses the difference in tweeter design.

A good book that I'd recommend is 'The Loudspeaker and Headphone Handbook', edited by John Borwick. It covers loads of stuff in there and has an excellent list of contributing authors - Stanley Kelly, Floyd Toole etc.

http://www.amazon.com/Loudspeaker-Headphone-Handbook-John-Borwick/dp/0240513711

I think horn tweeters have the advantage that the conincident tweeter and woofer prevent 'lobing' of the spatial response of the loudspeaker, i.e. they give a smooth off-axis response over a wide area, which improves the stereo imaging of the speakers. The disadvantage is that they are probably less accurate than high quality conventional loudspeakers which have the separate tweeter and woofer. Objectively, this would probably be related to less flatness in the amplitude response of the horn loudspeaker compared to high-quality conventional designs.

Floyd Toole points to high resolution, 1/20th octave amplitude response graphs of loudspeaker as the most important objective measure of loudspeaker performance. His evaluation of performance relates objective measures to evaluated subjective (double-blind) performance of loudspeakers.

 

[Swerd]

Is there any reason that a horn tweeter will intrinsically sound bright or is it a matter of implementation?

I'm inclined to agree that implementation makes the most difference.

There is an interesting article by a DIY speaker builder named John Krutke (aka Zaph) http://www.zaphaudio.com/Waveguidetmm.html, where he takes a widely used good quality dome tweeter (Seas 27TDFC) and adds a shallow horn, or waveguide as he calls it. He goes into greater detail here http://www.zaphaudio.com/hornconversion.html

To summarize what he says in these articles, there are several potential advantages:

• The waveguide greatly reduces distortion at the low end of the dome tweeter's range. This particular waveguide layout works by raising the on-axis efficiency by about 6 dB in a broad range between 1 and 6 kHz. If the crossover reshapes the response curve back down to flat, harmonic distortion is reduced compared to the same tweeter mounted on a flat baffle.
  
  
• The low end of the tweeter's output has a change in directivity that more closely matches the woofer's directivity. The result is very smooth horizontal off axis response, and a power response that does not have the same deep null that a standard speaker can have.
  
  
• The final benefit of a waveguide is that it moves the acoustic center of the tweeter back - closer to the acoustic center of the woofer. In this case, it moves the tweeter back by about 1¼". This puts the listening axis straight forward in a standard 2nd order crossover design, and eliminates the need for a slanted baffle or the increased complexity of a ladder delay network.

It's importantant to distinguish between a dome tweeter mounted behind waveguide and a true compression horn tweeter. A waveguide is generally shallow and has a wide throat with no compression chamber. A true horn loaded tweeter uses a compression tweeter mounted behind a deeper horn. The result is a much louder tweeter, but usually at the expense of smooth response. They are not often seen in home hifi speakers. Note that the HB-1 speaker is quite sensitive, 92 dB, as you might expect with a horn loaded tweeter design.

 

[mtrycrafts]

Floyd Toole points to high resolution, 1/20th octave amplitude response graphs of loudspeaker as the most important objective measure of loudspeaker performance. His evaluation of performance relates objective measures to evaluated subjective (double-blind) performance of loudspeakers.

And, we have audiophiles who reject audio science and make up their own singular reality

I may have to buy that book.

 

[tbewick]

And, we have audiophiles who reject audio science and make up their own singular reality

I may have to buy that book.

Recently I actually managed to find Floyd Toole's articles on the Harman Audio website:

http://www.harmanaudio.com/all_about_audio/default.asp

You're probably familiar with them but they seem to contain a lot of what he wrote in the loudspeaker book. Again I haven't fully read through these articles yet, but skimming through, at least in some areas, more information is contained than is in the book.

I think that one of the most interesting points is that high Q resonances are less audible than low Q resonances. He states this in the book as well but diagrams are provided in the Harman article. This observation is directly in contradiction with what I've read previously in other books:

'The apparent contradiction between perception and measurement, then, begins with the observation that, as conventionally measured, the frequency response is a “steady-state” measurement, showing the resonance outputs at their maximum amplitude. With music, high-Q resonances are rarely driven to their maximum outputs, and so are less audible than the measurement indicates. The problem is not that the measurements are wrong, or irrelevant, it is that they are non-linearly related to the perceptual mechanism in humans, and therefore must be interpreted.

...An interesting fact now emerges: that the conventional method of specifying the excellence of frequency response - ± x dB – is almost useless unless the tolerance is very, very small. For equal audibility, high-Q phenomena could be ± 5 dB, while moderate-Q resonances could be ± 3 dB and low-Q and other broadband deviations could be ± 0.5 dB. Clearly, frequency response curves must be interpreted, there is no simple “catchall” kind of tolerance specification that is truly meaningful. Such is life.'

- 'Audio: Science in the Service of Art', by Floyd Toole. Page 12.

 

[GreenJelly]

Conclusion: Results mater, implimentation doesnt.

 

[mtrycrafts]

Recently I actually managed to find Floyd Toole's articles on the Harman Audio website:

http://www.harmanaudio.com/all_about_audio/default.asp

You're probably familiar with them but they seem to contain a lot of what he wrote in the loudspeaker book. Again I haven't fully read through these articles yet, but skimming through, at least in some areas, more information is contained than is in the book.

I think that one of the most interesting points is that high Q resonances are less audible than low Q resonances. He states this in the book as well but diagrams are provided in the Harman article. This observation is directly in contradiction with what I've read previously in other books:

'The apparent contradiction between perception and measurement, then, begins with the observation that, as conventionally measured, the frequency response is a “steady-state” measurement, showing the resonance outputs at their maximum amplitude. With music, high-Q resonances are rarely driven to their maximum outputs, and so are less audible than the measurement indicates. The problem is not that the measurements are wrong, or irrelevant, it is that they are non-linearly related to the perceptual mechanism in humans, and therefore must be interpreted.

...An interesting fact now emerges: that the conventional method of specifying the excellence of frequency response - ± x dB – is almost useless unless the tolerance is very, very small. For equal audibility, high-Q phenomena could be ± 5 dB, while moderate-Q resonances could be ± 3 dB and low-Q and other broadband deviations could be ± 0.5 dB. Clearly, frequency response curves must be interpreted, there is no simple “catchall” kind of tolerance specification that is truly meaningful. Such is life.'

- 'Audio: Science in the Service of Art', by Floyd Toole. Page 12.

Great, thanks. Book marked the link, cheaper than the book.
Now we need to know the Q factors of speakers? He also like to measure at 1/20 octave, great.

 

[Joe Schmoe]

While I can't say which is better or worse (this being personal preference), I will say that they image differently. Roughly speaking, conventional tweeters sound like they are in front of me while horns make me feel like I am "inside" the sound, if that makes any sense.
(FWIW I don't care for the Klipsch speakers I have heard because of their brightness.)

 

[Swerd]

Even though it wanders away from the original posters question about horn loaded tweeters, thanks for the Floyd Toole link.

He has led the way in defining useful scientific measurements in audio playback gear. He has tried to bridge the gap between electrical and acoustic measurments, and human audio perception. In particular, he has shown the importance of blinded listening tests to determine what features or sounds listeners do and do not notice or prefer.

I'll read the full article more carefully later. About the high & low Q matter, is Toole saying that simple frequency response curves do not adequately address this question? Time is also a factor with high Q, and that is better shown in an impulse response curve or a 3D frequency-response-time graph (waterfall plot). Or is he saying that our ability to perceive differences between high and low Q varies with frequency range where it occurs?

 

[GreenJelly]

CAPTAIN you are entering into dangerous territory that will slowly drive us all MAD! Please reconsider your quest into this most insane topic, for we will all certainly go crazy. I recommend we turn around and go Warp 6 out of this dangerous sector.

Damn it... Im not even a fan of Star Trek, though the Next Generation was an amazing show... See, the madness is already seeping in…

In all seriousness, I attempted to build a pair of good speakers. I soon found myself lost in a sea of complexity that astonished and bewildered me. It was too much for me personally to handle. I am a systems analyst and database designer by trade, and am not in fear of mathematics. However, even this scared the… I stopped around a topic very simular to this one.

I wish you luck on this topic, but understand that Ive heard it beign said that it takes a company an average of 20 years to develop and evolve a product into a good speaker. On rare occasions companies have faired much better (some original Thiels were very good), and on other occasions companies have raised the bar. A few times a company stubles down into a pit of dispair and failure (BOSE). It is an insanely complex a chaotic world.

With that said, Im jumping boat, and grabbing the escape capsule.

Good Luck
Mike

 

[tbewick]

I'll read the full article more carefully later. About the high & low Q matter, is Toole saying that simple frequency response curves do not adequately address this question? Time is also a factor with high Q, and that is better shown in an impulse response curve or a 3D frequency-response-time graph (waterfall plot). Or is he saying that our ability to perceive differences between high and low Q varies with frequency range where it occurs?

I haven't read the Harman article fully either, but I believe what he's saying is that on usual music playback, the high-Q's aren't usually fully 'excited' or engaged, whereas the broad low-Q's are excited more often, making them more audible. He pretty much says that the usual interpretation of impulse decay plots where long, high-Q tail-offs are seen to be a very bad thing, is wrong.

I haven't read the papers he references, but I am a bit cautious about this. In the chapter he wrote in the book I read, he said that he likes to avoid using solo instrument recordings. This is because they show much more variation in production than, for instance, orchestral works. I wonder whether this would affect the results, because on average, solo instruments may be more likely to excite a narrow frequency range?

On frequency response curves -

'Another common measurement is one in which the audible frequency range is divided into equal fixedpercentage bandwidths, such as 1/3 octaves, or in which a high-resolution measurement is heavily smoothed, or spectrally averaged, on a continuous basis. These spectral-averaging devices have extremely limited utility in the design and evaluation process. Spatial averaging adds information, whereas spectral averaging removes spectral
details, making curves look smoother and prettier. The sound, though, remains unchanged. Is it any wonder that some people mistrust measurements?'

- again, this contradicts some things I've read before.

Ive heard it beign said that it takes a company an average of 20 years to develop and evolve a product into a good speaker.

This is probably very relevant for horn speakers, since the physics involved in designing high-quality horns is meant to be very complicated.