Geddes on Waveguides

"Getting the principles to work over such large bandwidth is just not feasible. An optical lens operates only over less than a octave".

Earl,
Here you very quickly answer the question, this type of lens will not function in the intended way, it is not capable of functioning in an audio application of more than a decade. So how am I wrong that the premise was a mistake trying to use this optic refraction method applied to audio? It was a misapplication of a know function.
 
Utter Nonsense

The AES journal that published them did not agree with you, but your objection is to be expected, noted and filed where it belongs.

I am not alone in the observations I made here, so you must have a very large place to put the multiplicity of dissenting peer reviews. Presumably that is a large and dark location, well hidden from view.

Quote from the paper abstract [1]:

"Analysis of the results shows that the differences in frequency response between drivers are statistically significant, whereas differences in playing level, and therefore nonlinear distortion, were not significant. This unexpected result implies that nonlinear distortion is not audible under these test conditions, and it leads to important conclusions regarding the design objectives of compression drivers."

Whether you like it or not, the human ear will be the final judge of what is acoustically different and superior, and what is not. What the data will tell you only, is why. If you cannot hear the differences between the outputs of different compression drivers, that fact does not mean nor prove that others cannot. Furthermore, the findings of your own study, is a refutation of the claim you made here.

WHG

Reference:

[1] Subjective Testing of Compression Drivers
AES E-Library Subjective Testing of Compression Drivers
 
Last edited:
Comparisons of drivers

"The only thing that I object to is making something out of nothing. Every subjective test that I have done implies that distortion in a compression driver is irrelevant, that the waveguide response dominates the perception making all drivers virtually identical in that regard, and that virtually nothing of what is usually measured will ever show anything that will adequately differentiate subjective perceptions of one driver from another. "

Earl,
Are you just comparing audible distortion of one comp driver vs another on the exact same waveguide OR does this also include frequency response?
I was under the impression that different compression drivers with different exit angles require the waveguide throat to be optimized for the specific exit angle.

Example: comparable 1" from, say, JBL and B&C mated to exactly the same waveguide should measurably vary a bit in their frequency response.

Now, whether such variance is audible or deemed significant to listeners - that's another story.

Bruce
 
"Getting the principles to work over such large bandwidth is just not feasible. An optical lens operates only over less than a octave".

Earl,
Here you very quickly answer the question, this type of lens will not function in the intended way, it is not capable of functioning in an audio application of more than a decade. So how am I wrong that the premise was a mistake trying to use this optic refraction method applied to audio? It was a misapplication of a know function.

You said that the "principles" were wrong, not the "premise". It was a naive application of the principles, but the principles were not wrong.
 
Earl,
Are you just comparing audible distortion of one comp driver vs another on the exact same waveguide OR does this also include frequency response?
I was under the impression that different compression drivers with different exit angles require the waveguide throat to be optimized for the specific exit angle.

Example: comparable 1" from, say, JBL and B&C mated to exactly the same waveguide should measurably vary a bit in their frequency response.

Now, whether such variance is audible or deemed significant to listeners - that's another story.

Bruce

One of the most audible things in a waveguide is the internal reflections. The matching of the driver to the waveguide CAN BE one of the most critical aspects of this problem. It need not be however. But these are things that I would put in the category of "waveguides", not "drivers".

The distortion was not deemed to be audible at all across compression drivers, but the - un-equalized - frequency response differences were. When the frequency response differences of two drivers are equalized to be the same on the same waveguides, the results indicate that these two systems are not audibly different.
 
Earl,
Sorry about the semantic problem there but it does seem that we have come to the same final conclusion that the slant plate and circular stacked plate idea really was an ineffective method to control dispersion over the audio bandwidth of these horns. I would also add that the horn designs used had an extremely small angular dispersion themselves which made the problem even larger. Some of those horns couldn't have had much more than 15 degrees of included angle. Perhaps that was intentional as that would have caused a large diffraction at the sharp end of those small short horns?
 
Lens Notes

whgeiger,
On the point of little research going on with slant plate combinations with horns you are probably correct that there is little. I think most concluded early on that the premise of using optic principals was a flawed concept. That is what I understand and have been told by others who's opinions I hold highly.

>snip<
The frequency band over which a divergent acoustic lens is effective is determined by obstacle spacing for the upper bound and lens extent (size) for the lower bound.

A parallel can be drawn for driver/horn bandwidth as well. Here driver throat diameter (at phase plug exit) and horn mouth extent (size) are the limiting determinants.

Furthermore at lower frequencies, there is no beaming in the radiation pattern to mitigate as the lens becomes acoustically transparent; so the fact that lens affectivity is bandwidth limited is irrelevant.

While extremes are to be avoided, wide dispersion patterns at high frequencies are achievable where needed through deployment of an acoustic lens.

What is needed is a comparable effort to bring this technology into the 21st Century.
 
Last edited:
Wh,
As in anything in life there are few absolutes so you are probably correct that with enough research and development you could make these combinations of horn and lens work. There may be other things that you would have to consider as with everything else such as materials used for the lens itself, perhaps something other than the aluminum used and many other factors would need to be researched. I can think of many factors that would need to be looked at, but that could be a many years research program to understand the many changes that could be effected.
 
Earl,
Sorry about the semantic problem there but it does seem that we have come to the same final conclusion that the slant plate and circular stacked plate idea really was an ineffective method to control dispersion over the audio bandwidth of these horns.

I agree completely. It was simply the use of the terms that I objected to - a "principle" is not the same as a "premise". Its just a "research scientist" thing about being precise.
 
Let Your Ears Do the Listening.

You know there may be a lot of critiques that can be leveled - and have been - but let me ask "Where is your scientific (and reviewed) subjective study that finds a different result?" Don't have one? Then it is pretty cheap to critique the only one that exists because you don't happen to like the results.

When you offer up a study as proof of an assertion, then it becomes more than just a study, and is open to much closer and rigorous scrutiny. Quite honestly, I do not see the need to have 20 or so neophytes tell me about the distortion they are not hearing. What interests me, is the distortion I am hearing, what is causing it, and why. Obviously low levels of it are not audible and masked by the louder signal components that are not so mangled. The other consideration is that most music already contains fair amount of loudspeaker distortion products due to the use of amplified instruments that typical listeners have become conditioned to expect. So, is it any wonder that the study results are such as they are? Conduct the same study using pure tones, solo acoustic instruments, and trained listeners and you can expect an entirely different set of findings.

WHG
 
On the audibility of differences among compression drivers...

I'll wait for the results of your study. Until then I'll stick with those that I have.

In general, I hate to wade into what could easily become sterile polemic, but in all honesty, and with all due respect, this kind of response and stubborn defence of that particular study of yours sounds rather unscientific to me.

Incidentally, I too am a Ph.D. (albeit in a different branch of science: environmental chemistry), and I am fairly familiar with the methods of scientific research. The way I see it, and I think many fellow academics would agree, is that if one particular study produces a given set of results (such as your inaudibility of the differences among compression drivers) under a specific set of test conditions and/or assumptions (i.e. using a short excerpt of a multi-track pop recording, and a relatively unexperienced panel of listeners), then THAT is what it proves, i.e. that a relatively unexperienced panel of listeners is unable to distinguish among different compression drivers reproducing a short excerpt of a multi-track pop recording.

It is however unreasonable (and thoroughly unscientific) to liberally extrapolate those results beyond the strict conditions and scope of the test, and claim that the same study "proves" that compression drivers are indistinguishable tout court.

Respectfully,

Marco
 
In general, I hate to wade into what could easily become sterile polemic, but in all honesty, and with all due respect, this kind of response and stubborn defence of that particular study of yours sounds rather unscientific to me.

I'm not a Ph.D. But in my field, what Dr. Geddes is saying makes perfect sense. Let me put it in the language of my field, the law.

Earl did a study, got x results, using a methodology was sound enough to survive the peer review process, and published his results in the publication of record. Through that process, he's established a rebuttable presumption that his results are correct. That presumption can certainly be rebutted by introducing evidence to the contrary. But to be strong enough to rebut the presumption, the new evidence needs to pass a similarly high threshold (blind listening, equalization to remove confounding FR differences, statistical analysis etc.) as the evidence that gave rise to the presumption in the first place. Thus far, that hasn't happened, so the presumption stands.
 
You guys have it backwards. In psychoacoustic studies using non-impulsive, 'classical music' style signals, the perception threshold is much higher than using 'modern' music. This may come as a surprise, but 'Burning down the house' (or was it something else?) is one of the most revealing excerpts in 'distortion' testing (that's why they use it). Take a (modern, variable bitrate) perceptual coder and put through it a 'solo violin' and some modern music. Guaranteed is that the violin piece will have higher compression ratio, e.g. you can inject more noise/distortion to the signal without it being perceptible.

Also you forget the magnitude AND ranking of the problem. So a panel of 30+- can't reliably detect something, then you can safely file that under 'mostly irrelevant' and concentrate on those things that have been shown to matter. Frequency response; early reflection spectra, level, smoothness, direction, density; power response; group delay in 600-4000Hz range. If you strongly disagree with this, then it's time for some proof to the contrary. Its not like Geddes is alone in this. All industry professionals, who have studied distortion perception, came to the same conclusion and moved on.
 
The only thing that I object to is making something out of nothing. Every subjective test that I have done implies that distortion in a compression driver is irrelevant, that the waveguide response dominates the perception making all drivers virtually identical in that regard, and that virtually nothing of what is usually measured will ever show anything that will adequately differentiate subjective perceptions of one driver from another.
__________________

While there may or my not be scientific reasons to support the above claim I think the "underlying assumption" here is the use of this particular wave - guide design by Earl.

A frugal understanding of the characteristics of this device suggests that a significant amount of the sound from the wave guide at the listening position is reverberant and is linear over a wide response bandwidth.

In the case more conventional and widely used horn types used in the recording industry, ie the Tad style radial horns, the reverberant proportion of the sound at the listening position changes to a larger portion of direct sound as frequency increases. Nor does the sound have to travel via a cross section of foam.

Using the correct test method, test results of beryllium as a material published elsewhere has proven to be a superior for compression driver diaphragms over titanium and aluminium. This is elementary physics if one cares to understand the break up modes of a compression driver diaphragm. Subjectively the matter has been proven beyond doubt to be the case.


One could therefore argue the case for the claim below is valid but only valid in this instance:

"that the waveguide response dominates the perception making all drivers virtually identical in that regard"
 
Last edited: