It's in the presentation page I posted. He is the author so from his point of view yes SPL and Group Delay dependent.
http://www.gedlee.com/Papers/Sound Quality Improvements in Compression Driver Systems.pdf
Attachments
Hello camplo
WRT the room the only thing you are changing is the differences in power response between the 2 systems. The room is the room.
More about if there were any improvements made by using a "better" material would be easier to hear on which system.
CD vs. Non CD
Rob
the only thing? as if its not critical?
https://www.diyaudio.com/community/...w-distortion-with-a-2-way.334757/post-7604902
I made a good amount of comments here
Unfortunately, I missed too much college level physics and acoustics to understand, much less contribute much to, discussions here on driver/horn design and performance. But I am in Troy Crowe’s queue for deciding among drivers/horns for above my Altec 416-8B sealed midwoofers. The Yamaha JA6681B (plus a tweeter) and Radian 745neobe (two-way) are among those considered. Troy’s measurements on the former, post # 15,275 https://www.diyaudio.com/community/threads/beyond-the-ariel.100392/page-764#post-7483847
In addition to the posted Audio Express link to measurements on the Radian745be, on the page on the above thread above, Pierre describes his listening experiences with these Radians in Athos TH-4001 horns atop TAD sealed woofer; posts 15,266, 15,276.
Along with obviously perfecting the EQ for these Radians, Pierre found that the vanes in the 4001 horns proved beneficial rather than problematic. And while I’ve read of others finding vanes sonically detrimental-including with one using a 6681B/4001 horn combination-comparing notes with Pierre on this might prove enlightening.
I think it is important to point out that this is in the context of many other pages. On the pages before this, he refers to two studies that shows that group delay is harder to hear at low levels. He also points towards his work together with his wife on harmonic distortion, where they confirmed that harmonic distortion is harder to hear at low levels. I believe this was a very well known fact, one that I also mentioned a few posts ago.
Anyway, he then goes on to say this:
He does not say what he compare these SPL levels with. However, as he refers to the audibility of harmonic distortion and group delay on the pages just before as he is putting this into context, it is fair to assume that he is talking about the levels where these types of distortion becomes more audible. At 110-120-130dB, our ability to identify these distortion types are not improving.
I do not recall who said what, but it was stated that this meant that we would not be able to hear HOM at normal listening levels. I noted that I found it very unlikely that he had actually expressed such an opinion. Here is what I believe we can extract from Geddes' document to shed light on the topic.
Here is what he sets out to investigate:
And here is how he concludes after presenting his findings (and this comes after the reference to SPL):
To me, it does not look like he is trying to tell us that HOM is not a problem at normal listening levels.
I also find it interesting that he presents an OSWG as the answer. I do not know all of his detailed findings, but I had the oportunity to look at some analysis done by Mr Kolbrek, which showed that OSWG was about as bad as a conical horn on this. It was not showed or stated that these horns do create it, but the presence of HOM was huge in these two particular horns while it was very low in a comparable exponential horn. Given all the struggles I at that point had experienced with Seos-horns, it was a bit of an a-ha-moment for me when I saw this.
But as I have mentioned so many times now, this will be driver dependent.
I think you've got the same problem as the luminiferous aether: if you can't measure it, (and now it's clear that you're not even trying to measure it), much less trying to determine audibility levels of "HOMs", then engineers should move on and leave the entire subject alone until measurements can repeatably be achieved and audibility correlated to those measurements in some form or fashion. Otherwise, it's the blind leading the blind.
If measurements have been done and correlated to audibility, then I stand corrected. But simulation is not measurement. This is a subject that physicists would chase--like for instance with the EmDrive. (String theory also has the same problems--cold fusion, too.)
I do think those interested in doing engineering should spend their time/energies on measurable hi-fi horn/compression driver performance issues, (including subjective measurements being turned into objective measures via Toole's and Olive's methods)...no matter how much you want to believe someone that's proposing that this kind of distortion "has to exist" and is an audible problem.
Willingness to believe someone or something isn't an engineering topic, but rather a religious topic. I tend to avoid those in my engineering work.
Chris
Last edited:
It's been done.
The first property of HOM, described a few posts back, is group delay. Audibility of group delay has been known for decades. Here is a more modern study of the audibility of group delay with regard to the level of the HOM, the level of the playback and the delay time - http://www.gedlee.com/Papers/AES06Gedlee_ll.pdf
It also affects audibility that a HOM comes to you from a different source angle.
It creates unique room reflections and it creates variations between different room reflections. This can draw attention to them and work against the direct and reverberant fields sounding as one.
Allen, I know you're trying to resolve an issue of some importance to the hi-fi horn loudspeaker community here (i.e., amateurs--not wolves in sheep's clothing), but note that the paper you posted does not even mention higher order modes (HOMs).
Here is a direct quote from the paper:
I've been familiar with that linear distortion audibility metric since at least 2010 (i,e., a long time ago now). The paper itself still begs the question: where are the measurements of HOMs?
__________________________________________________________________________________
Is this what you intended to discuss? I don't believe that will be very fruitful.
Chris
Here is a direct quote from the paper:
I've been familiar with that linear distortion audibility metric since at least 2010 (i,e., a long time ago now). The paper itself still begs the question: where are the measurements of HOMs?
__________________________________________________________________________________
If this is the alternative metric, then where are the high resolution GD plots? I assume that we're not talking exclusively about the excess GD (the leftover GD in excess of the Hilbert Transform of the magnitude response). I can show this data, but it shows values well below audibility levels (i.e., 1 ms, an audibility threshold which is 100 times higher than the typical measured GD):
Is this what you intended to discuss? I don't believe that will be very fruitful.
Chris
Hello Chris
I agree I have never been able to definitely "see" them in either a GD plot, as posted, or in a waterfall where I would expect to see it.
Rob
You're not likely to see much looking at a CSD or GD plot as shown.
The linked study is relevant because it looks at sound with both a direct part and a secondary source. HOM fit into this category as they are a linear distortion and they have a delayed component and a physical displacement. It doesn't take much for these factors to add up.
Tell me more about why you feel that's the better way..
The linked study is relevant because it looks at sound with both a direct part and a secondary source. HOM fit into this category as they are a linear distortion and they have a delayed component and a physical displacement. It doesn't take much for these factors to add up.
Hello Allen
Why? You can clearly see the HF break-up modes in the secondary resonance or a compression driver in a CSD? That's stored energy why wouldn't you see HOM's? After all a waterfall is just another way to look at the Impulse.
Same thing with Group Delay? They are defined as group delay so why wouldn't you see it? Again another way to look at the impulse.
Rob
HOMs don't need to be very loud to be interesting, since they have other factors contributing to their audibility. They may also be spread across the spectrum. They could get lost against the background of the direct sound.
Let's not forget too that a CSD plot is showing only one angle, which doesn't help matters.
Let's not forget too that a CSD plot is showing only one angle, which doesn't help matters.
Your absolutely correct that it is only one angle. I would expect some to show? You are measuring on the listening axis.
Are you saying they are scattered through out the radiation pattern? If that is so they would be "concentrated" with a horn/waveguide of narrow directivity as they can't be beyond the radiation pattern.
Rob
Are you saying they are scattered through out the radiation pattern? If that is so they would be "concentrated" with a horn/waveguide of narrow directivity as they can't be beyond the radiation pattern.
Rob
HOM is well documented and understood among those who have actually worked on the topic, and it was so before 1920 when G. W. Stewart documented the formation of HOM in conical horns based on physical experiments. The experiments were sound performance related. This was far from the last time this was investigated, and there are so far no experiment that has disproved this.
The document that @AllenB posted does not investigate HOM, but it does investigate the relationship between listening level and audibility of different types of distortion. That relates to the audibility of distortion that was discussed earlier. We get a very interesting answer here, the investigations were done at 71-80dB. This is not regarded as loud, but the ability to hear distortion will naturally increase towards the higher end of this scale. Then, at least, we know that the SPL part of the equation was not stating that we need extreme SPL levels. This document came before his powerpoint presentation, and he did refer to it just before his conclusion where he puts SPL as a factor for audibility of HOM.
Then we are left with two things. One is the very existence of HOM, the other one is if it is audible. If one denies any of these, it will require some kind of documentation to back things up, or else it is just like "We do not know everything, therefore God."
If the existence of HOM is denied, it is the same as claiming that any horn has a perfect simple wavefront. This goes against a vast amount of sciende, so some kind of proof will be required. One needs to prove that wide dispersion, wide bandwidth horns do also have a perfect wavefront, and it would also be required to demonstrate why all simulations are wrong in order to add any credibility to the case.
If the audibility of HOM is denied, then this too goes against the vast amount of science in the field. Here too, some kind of proof will therefore be required. One has to show a horn with documented high levels of HOM, and prove that noone can hear it. One also need to demonstrate how others got it wrong in order to be able to defend such a claim.
Just pulling "I don't believe in it" out of the hat and requiring others to prove you wrong just does not cut it.
The document that @AllenB posted does not investigate HOM, but it does investigate the relationship between listening level and audibility of different types of distortion. That relates to the audibility of distortion that was discussed earlier. We get a very interesting answer here, the investigations were done at 71-80dB. This is not regarded as loud, but the ability to hear distortion will naturally increase towards the higher end of this scale. Then, at least, we know that the SPL part of the equation was not stating that we need extreme SPL levels. This document came before his powerpoint presentation, and he did refer to it just before his conclusion where he puts SPL as a factor for audibility of HOM.
Then we are left with two things. One is the very existence of HOM, the other one is if it is audible. If one denies any of these, it will require some kind of documentation to back things up, or else it is just like "We do not know everything, therefore God."
If the existence of HOM is denied, it is the same as claiming that any horn has a perfect simple wavefront. This goes against a vast amount of sciende, so some kind of proof will be required. One needs to prove that wide dispersion, wide bandwidth horns do also have a perfect wavefront, and it would also be required to demonstrate why all simulations are wrong in order to add any credibility to the case.
If the audibility of HOM is denied, then this too goes against the vast amount of science in the field. Here too, some kind of proof will therefore be required. One has to show a horn with documented high levels of HOM, and prove that noone can hear it. One also need to demonstrate how others got it wrong in order to be able to defend such a claim.
Just pulling "I don't believe in it" out of the hat and requiring others to prove you wrong just does not cut it.
Your absolutely correct that it is only one angle. I would expect some to show? You are measuring on the listening axis.
Are you saying they are scattered through out the radiation pattern? If that is so they would be "concentrated" with a horn/waveguide of narrow directivity as they can't be beyond the radiation pattern.
Rob
He is not talking about the angle out of the horn. He is talking about the angle towards the listener. Some energy will arrive in line with the driver, while other energy will arrive from a bit off this axis. Our ears are brilliant at locating separate sound sources, while this is a very complex field for microphones.
We could turn this on its head too. How would you go about to document the following?:
1: All sound from a horn arrives from a single point in space.
2: The wavefront emiting from the horn is a perfect simple wavefront.
It is not an easy task.
That too. I was just saying that a multi-angled plot opens up new opportunities to visualise the data. You can trace a mode varying in angle as it changes frequency, whereas on a single plot that might only be a small bump in the response.
All of these replies have confirmed what I have suspected: no one is actually measuring anything remotely related to HOMs. I really don't care who you quote: until someone starts to measure what they're talking about (i.e., not simulate), the second part of the story--audibility--cannot be ascertained, and guys, at the end of the day, that's all that really matters--interesting as the subject may be to some. The rest is BS until that gap starts to close.
If I've missed the golden paper that shows direct measurements of HOMs in the type of horns we are talking about (mainly K-402s in this particular part of this now-very old and outdated [2022] thread), then please direct me to that paper (directly, not indirectly--a link will do).
Believe it or not, I was willing to let the issue drop off if the conversation about HOMs dropped off. That's not what occurred, above.
Chris
If I've missed the golden paper that shows direct measurements of HOMs in the type of horns we are talking about (mainly K-402s in this particular part of this now-very old and outdated [2022] thread), then please direct me to that paper (directly, not indirectly--a link will do).
Believe it or not, I was willing to let the issue drop off if the conversation about HOMs dropped off. That's not what occurred, above.
Chris
Last edited:
I don't see anyone questioning their existence. The question is audibility and the measurement. We have a summary line in the paper that say's
"A careful look at the impulse response might yield the best insight, however this has not been quantified. Methods for measuring the nonlinear effects of our subjective perception are currently under investigation."
Well the Group Delay curve and the CSD are derived from the Impulse specifically to see and quantify Group Delay specifically and CSD where you can see resonances and delayed energy.
These are standard measurements yet we can't see HOM's
If we can't reliably measure it we can't quantify how significant it is WRF audibility.
You can't prove anything if it can't be measured or quantified.
So yes they exist.
So round and around we go.
Rob
I'll simplify all your lives by saying HOM's are mainly just diffraction related back wave that emits when waveguide profile changes too suddenly and sound diffracts, be it inside the device or at the mouth. The back wave emits backwards where the sound came from, thus back inside to the device, and reflects from inside the device back out.
All this shows up as direction and frequency dependent undulations in frequency response plot. Or group delay plot, as it's sound arriving after direct sound. Now, this shows easily in simulation because the source can be anything ideal so the waveguide contribution to diffraction is clear. If there is more in frequency response measurement it must be from the compression driver, or from junction of the two, and would show up with appropriate delay. So, if you have a waveguide that has proper roll over on the mouth as part of nice continuous profile all the way from phase plug one doesn't see any wiggle in frequency response to any direction up until top octave, where wavelength is smaller than the throat and dominated by what's coming from inside the driver.
If there are some other HOMs in the system, like the waveguide resonating or something, they would not be visible with all the diffraction going on, which is kind of a huge phenomenon in frequency response plots, could be many db. If one makes a diffraction free (relatively) device, and still see wiggle, it should be easy to test and reason what's the issue. If there is no wiggle in frequency response I'd quess there is no HOMs, definitely there is no diffraction then. No matter which profile horn, OS, exponential, no horn at all, if there is no properly sized and shaped mouth rollover there would be a lot of diffraction showing in any plots. Basically, diffraction makes secondary sounds emit from the device, and if HOMs are secondary sounds, I don't know whats the difference in practice, at least they would contribute to similar perception of sound I bet.
As disclaimer, I don't know what HOMs would actualy be, they seem mysterious voodoo like entity but seem awful lot like diffraction. Unlike HOMs seem to be, diffraction is very easy to detect both in measurements and in very simple simulations, it's relatively easy to reason with and happens with any physical objects, also with waveguides.
edit. simplest possible example a "conical" horn
https://www.falstad.com/ripple/Ripple.html?rol=$+3+512+64+0+0+778+0.048828125 s+2+255+21+2+0.46680807839388144+0+10.001413602849373+99.99941507823934+1+0 w+0+207+91+253+18 w+0+306+91+259+17 w+0+253+20+259+20
Same with extension, non smooth profile with sudden change and it emits the backwave as well.
https://www.falstad.com/ripple/Ripple.html?rol=$+3+512+64+0+0+778+0.048828125 s+2+255+21+2+0.46680807839388144+0+10.001413602849373+99.99941507823934+1+0 w+0+207+91+253+18 w+0+306+91+259+17 w+0+253+20+259+20 w+0+365+132+305+90 w+0+208+90+155+131
Click the links to see it happen.
edit2. mabat posted some ripple tank examples on ATH thread: https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-6714375
All this shows up as direction and frequency dependent undulations in frequency response plot. Or group delay plot, as it's sound arriving after direct sound. Now, this shows easily in simulation because the source can be anything ideal so the waveguide contribution to diffraction is clear. If there is more in frequency response measurement it must be from the compression driver, or from junction of the two, and would show up with appropriate delay. So, if you have a waveguide that has proper roll over on the mouth as part of nice continuous profile all the way from phase plug one doesn't see any wiggle in frequency response to any direction up until top octave, where wavelength is smaller than the throat and dominated by what's coming from inside the driver.
If there are some other HOMs in the system, like the waveguide resonating or something, they would not be visible with all the diffraction going on, which is kind of a huge phenomenon in frequency response plots, could be many db. If one makes a diffraction free (relatively) device, and still see wiggle, it should be easy to test and reason what's the issue. If there is no wiggle in frequency response I'd quess there is no HOMs, definitely there is no diffraction then. No matter which profile horn, OS, exponential, no horn at all, if there is no properly sized and shaped mouth rollover there would be a lot of diffraction showing in any plots. Basically, diffraction makes secondary sounds emit from the device, and if HOMs are secondary sounds, I don't know whats the difference in practice, at least they would contribute to similar perception of sound I bet.
As disclaimer, I don't know what HOMs would actualy be, they seem mysterious voodoo like entity but seem awful lot like diffraction. Unlike HOMs seem to be, diffraction is very easy to detect both in measurements and in very simple simulations, it's relatively easy to reason with and happens with any physical objects, also with waveguides.
edit. simplest possible example a "conical" horn
https://www.falstad.com/ripple/Ripple.html?rol=$+3+512+64+0+0+778+0.048828125 s+2+255+21+2+0.46680807839388144+0+10.001413602849373+99.99941507823934+1+0 w+0+207+91+253+18 w+0+306+91+259+17 w+0+253+20+259+20
Same with extension, non smooth profile with sudden change and it emits the backwave as well.
https://www.falstad.com/ripple/Ripple.html?rol=$+3+512+64+0+0+778+0.048828125 s+2+255+21+2+0.46680807839388144+0+10.001413602849373+99.99941507823934+1+0 w+0+207+91+253+18 w+0+306+91+259+17 w+0+253+20+259+20 w+0+365+132+305+90 w+0+208+90+155+131
Click the links to see it happen.
edit2. mabat posted some ripple tank examples on ATH thread: https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-6714375
Last edited: