Horn vs. Waveguide

[gedlee]

And all used the "entry level" Selenium drivers, too, no? DE250 with the Selenium throat adapter certainly sounds better to me, though I haven't done more than a cursory listen thus far. Also, the regular resonances Earl shows in his measurements may well be generated by the Selenium drivers rather than the waveguides. That they appear in all of the planes is certainly suggestive.

I have only use the DE250.

Where is these "regular resonances Earl shows in his measurements "? There are two resonances in the passband, one is the driver and the other a standing wave across the mouth, they are unrelated and not correlated with each other. There are no "regular resonaces". (There is another driver resonance below the HP point and I do null out this one, but only because not doing so affects the crossover summing.)

 

[pos]

Also, the regular resonances Earl shows in his measurements may well be generated by the Selenium drivers rather than the waveguides. That they appear in all of the planes is certainly suggestive.

To illustrate what Zilch says, here are some curves run by member 4313B on the audioheritage forum:
( http://audioheritage.org/vbulletin/showthread.php?t=19857 )

The curves are taken on axis, with no eq.

First, the 435Be (the compression driver used in the K2 S9800) and the 476Be (the one used in the Everest 2) on the horn of the K2 S9800 :

We can clearly see a lot of ripples in the response, but the 476Be looks smoother.

Now the same two drivers on the PT-F95HF, the 1.5" version of the econowave:

 

[ZilchLab]

Where is these "regular resonances Earl shows in his measurements "?

"The JBL horn has a substantial amount of internal reflection and diffraction as is evident throughout its frequency response curves."

The nasty +/- 0.5 dB ones here:

 

[soongsc]

This was my first take on horn/waveguides.

An externally hosted image should be here but it was not working when we last tested it.

And the measurements: concaved dome driver on axis, Horizontal, Vertical.
Scale is 5db/div

An externally hosted image should be here but it was not working when we last tested it.

 

[gedlee]

"The JBL horn has a substantial amount of internal reflection and diffraction as is evident throughout its frequency response curves."

The nasty +/- 0.5 dB ones here:

All I can say is that you are not clear in what you are talking about. I thought you were talking about "waveguides" which is my device. At any rate even the JBL "horn" used a DE250, I believe (thats the only driver that I have). Unless the device came with a driver attached. Maybe it did, I don't remember.

Hey, if you want to use that horn be my guest. Just don't expect me to praise it when its not noteworthy. +- .5 dB Really?

 

[Robh3606]

With all the discussion about the vertical nulls I thought this might be useful to some. If you want to see where your nulls are at the crossover point all you have to do is calculate them using the driver spacing and wavelength at crossover. This is taken from an AES paper Improvements in Monitor Louspeakers by DAVID SMITH, D. B. KEELE, JR., AND JOHN EARGLE J.AudioEng.Soc.,Vol.31,No.6, 1983June


Rob

 

[gedlee]

Unfortunately that formula is only accurate for point sources not for sources with directivity. Its approximately correct, but that's all. Directivity at the crossover point changes everything and it can be very difficult to predict because everything is changing for both drivers (magnitude and phase change with both angle and frequency for both drivers). Basically its the kind of problem that only a computer can deal with.

 

[MartinQ]

As mentioned in the first post (and in the referenced PDF) the driver used on the JBL PT-F95HF was from Selenium. Zilch, (or anyone) do you recall if that was a D210Ti or D220Ti?

I wonder how much of the response is influenced by that driver.

 

[ZilchLab]

I wonder how much of the response is influenced by that driver.

D220Ti; Earl did not have any thread-on drivers, so we sent him those, which was fine, because that's what the project uses, and the measurements reflect the actual performance everyone is enjoying.

"Resonances are resonances," according to Earl, whether they are generated by the waveguide, the driver, or the two in combination.

When I see a horn with a regular period of peaks and dips in either the frequency response or the impedance curve, but usually both, then you can be sure that it will sound collored. This has to be looked at both on-axis and off axis as very often the on-axis curve will look the best, but looking at all the axes will tell. Peaks and dips at the same frequency in ALL the curves is a serious problem.

According to Toole, however, to the extent that the peaks and dips appear uniformly on- and off-axis, and in all planes (which these do,) they are amenable to EQ. Before resorting to that, however, ascertaining their source(s) might provide insight into other potential solutions, if they are in fact problematic....

 

[ZilchLab]

All I can say is that you are not clear in what you are talking about. I thought you were talking about "waveguides" which is my device.

I'm sorry you are confused; Newman was talking about the JBL PT waveguides you measured for us.

At any rate even the JBL "horn" used a DE250, I believe (thats the only driver that I have). Unless the device came with a driver attached. Maybe it did, I don't remember.

We have refreshed your recollection on this point now, I presume.

Hey, if you want to use that horn be my guest. Just don't expect me to praise it when its not noteworthy.

By no means would anyone expect otherwise, even if it were.

+- .5 dB Really?

OOPS, my error, +/- 1 dB, actually, in each of the curves between 0° and 45°. Do you see some basis in your measurements for suggesting otherwise?

 

[xpert]

Exhibit B: The 18Sound XT1086
These guys got soooooo close. This stupid waveguide is almost perfect, it's frustrating that they dropped the ball. It has constant coverage, the mouth termination is impeccable, it's sturdy, and it's affordable.

So why did they screw up the throat???

-----------

It might not be as uncompromising as a 911, but it's still a great sports car.

Maybe we can all work together and find the audio equivalent of a Mazda Miata; a great design that everyone can afford.

I owned several Porsches. Not that great, but if You look at it as a Volkswagen, o/k.

On topic: You know the JBL2352? It has flat walls and a big flare at the mouth. Nevertheless it sports some diffraction slot inside. The XT1086 measures great. Where are those "HOM"? Making a horn is making compromizes. Look at the result. Which "theories" are featured doesn't make the difference sound wise. What is important is what comes out acoustically. And that could be measured if it was there. So where is the "HOM"?

so long

 

[FrankWW]

Cited from that:

"A THD distortion curve will not reveal this effect,
nor will a frequency response graph. 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."

What I understand from that is: Mr. Geddes obscures once again the fact that the impulse response and frequency response graph are only different pictures of the same thing. They can be calculated (!) one into the other. Your home equipment does that on a daily basis.

IR and FR graphs don't show the same things. That's why folk convert one to the other. The paper was written for professionals who know exactly that. Nothing is obscured.

The methodology that has been deployed here was not specific to HOM. There was no differential analysis whether or not the measurement addresses what it should. Is it nessecary to have group delay and amplitude irregularities both together, and if must it be minimum phase or reflection wise?

You are grasping at straws, I think, but it's difficult to know since the last sentence doesn't make sense entirely. But I'll try to help you: Any horn or waveguide, (even minimum surface), due to reflections from the walls, creates signals which are delayed with respect to the on-axis or direct signal . They present to listener delayed and with amplitude irregularities fused together. These are HOMs. Geddes is testing for the audibility of this phenomenon - not for the delay and amplitude anomalies separated.

In the end, even IF one believes (!) in what Gedlee (incorporation?) presented here as a (non) scientific paper,

Indeed, you are grasping at straws, and in a dishonorable, ad hominem fashion by claiming the paper is "non-scientific". The paper was written in response to work by Olive and Toole, researchers who work for Harmon, and work by Glassberg and Moore, who are very prominent psychacoustic scientists, all of whom present work at AES meetings under exactly the same conditions as Geddes and Lee, and cited by them in the text and notes.

I believe you wrote this in an effort to provoke Geddes to either write something rash or withdraw from this discussion. Shame on you!

You continue:

the conclusion is that ripples of +/- 2dB are no issue:

Your assertion was HOMs are not audible. Now you assert, even if they are, they are not significant:

"Figure 3 – Ratings versus playback level for +2 dB linear distortion with delay as a parameter. These data
are not significant."

At 80 dB! I believe US protocols for human testing limit exposure to 80 dB but musical performances both live and recorded often exceed these levels and it's quite possible +2 dB linear distortion is audible at 90 or 100 dB.

At least with modern designs as EV HR90, HP940 and JBL 2352 ripples of 2dB magnitude are avoided. Most recent designs as JBL PTW and 18s XT1086 are much better.

I searched the net for such figures and could not find any. Perhaps you can provide them since you make the claim. So far you have not produced evidence.

Look at the picture as a whole! Was does GedLee LLC tell You if not to buy their systems, as all others are faulted?

No. They tell you they conclude that subjective testing of speakers is not reliable if its carried out under a low SPL regime.

I should think the message they are trying to put forward, like many other companies which make public their research and technical notes, is that they are serious people who do thoughtful research and apply it to their products. Nothing wrong with that.

This, you write here, is nekulturny and spiteful:

I respect his highly appreciated solution of horn equation. Good work, really! That does not mean that I give up my open mind. remember Shockley, the inventor of the transistor. As far as I know he didn't see a penny for it. Sh* happens!

so long

 

[noah katz]

xpert,

You might as well not say anything else.

Even if you what you said had any validity, which it mostly doesn't, no one with any sense would pay heed to it because it's obvious that your primary motive is to discredt Dr. Geddes.

 

[xpert]

Originally posted by FrankWW
You are grasping at straws, I think, but it's difficult to know since the last sentence doesn't make sense entirely. But I'll try to help you: Any horn or waveguide, (even minimum surface), due to reflections from the walls, creates signals ...

I understand HOM from first principles as I understand the complete calculus perfectly. You could belive that or not.

My question is very simple: If I set up an investigation from what do I know it is related to HOM?

When human beings are involved to test my hypothesis, from what do I know that what they perceive is HOM and not any other. Example given: the difference between both cases "HOM" vs. "non HOM" occures only because the "HOM" case comes with ripples in the amplitude response. Is it the ripples for their own or have these ripples to be escorted by group delay, and if, with minimum phase group delay or with excessive gropu delay? That decides whether it is simply equalizable or not.

You have to admit that I have a bit of an understanding here. And that it isn't against GedLee personally. Just some curiosity.

Is it true that the annoyance of horns - if any - is grounded on HOM that GedLee avoids exclusivley (if You pay)?!

Originally posted by FrankWW
Indeed, you are grasping at straws, and in a dishonorable, ad hominem fashion by claiming the paper is "non-scientific".

Papers for AES conventions do not undergo the standart second reading as those conventionally called scientific. I set "non" in paranthesis, by the way.

Originally posted by FrankWW
Shame on you!

You continue:

Your assertion was HOMs are not audible. Now you assert, even if they are, they are not significant:

Obviously You didn't get that GedLee themselves wrote that the data wasn't significant BECAUSE the effect was to small with ripples as small as +/-2dB. One may conclude, putting all that criticism above besides, that if HOM (??!!) is of a magnitude not to come over 2dB ripple is would be neglegible anyhow.


I really don't understand what the problem with my assertions is. Isn't it Geddes who anytime comes with "scientific truth" and "scientific authority"? So, here it is. It is my personal expectation, that GedLee tries to commercialize the OS waveguide idea. That's completely o/k with me. Why not? But IF "science" is used to advertise it, it should stand scientific criticism to some extent. For instabce I never wrote that the Summa wasn't a worthwhile invest. I never wrote it sounded bad, You know. Keep things straight, please.

by

 

[soongsc]

...

I really don't understand what the problem with my assertions is. Isn't it Geddes who anytime comes with "scientific truth" and "scientific authority"? So, here it is. It is my personal expectation, that GedLee tries to commercialize the OS waveguide idea. That's completely o/k with me. Why not? But IF "science" is used to advertise it, it should stand scientific criticism to some extent. For instabce I never wrote that the Summa wasn't a worthwhile invest. I never wrote it sounded bad, You know. Keep things straight, please.

by

Seems like a head on confrontation.But to some degree, I can understand it is a difficult issue. If I were he, I would not give the complete answer, but would reveal some aspects to consider. Basically Earl has addressed how he looks at the impulse; although that is not the whole story, it is a part of the information. For those whom are well versed in the physics in this aspect, it should not be too hard to figure things out.

 

[gedlee]

According to Toole, however, to the extent that the peaks and dips appear uniformly on- and off-axis, and in all planes (which these do,) they are amenable to EQ. Before resorting to that, however, ascertaining their source(s) might provide insight into other potential solutions, if they are in fact problematic....

This is true, and I do believe that a great many of the peaks and dips could be controlled with EQ. But thats not the whole story. When there are many peaks and dips then there have to be significant internal reflections. Whenever there are internal reflections there also has to be HOMs created as a result. The HOM cannot be EQ'd because they are both non-minimum phase and non-lumped parameter (or one-paramter if you like). The HOM wave effects are different across the wavefront or put another way the effects will be different in different directions. Hence HOM cannot be correct with any EQ. This IS the crux of the problem. ONLY the acoustical design can minimize the HOM.

So while the actual "peaks and dips" may not look bad to you, their existance is strong evidence of the existance of HOM which will always have a negative effect on the sound and cannot be corrected.

If I were still doing research I would certainly look to ways to better evaluate HOM and diffraction effects (they are basically the same thing) as the techniques for defining them is seriuosly lacking. But I haven't had a chance to do any research in a long time. One often meets a crossroads where you have to decide to be the geologist who finds the gold, or be the miner who digs it out. Here I think that I have done both - but I put on my miners cap daily.

I think that John's point here is well taken. I can present all the theory in the world about the sound quality of horns and waveguides, but in the end they need to be heard to really confirm the theory. All I can say is that no one who has heard my waveguides has ever claimed to have heard better.

The JBL device is a real bargin, of that I have no doubt, but I would bet my reputation on the fact that no one would find its sound quality to be better than a well done waveguide with foam. But then the later isn't $10. Thats the tradeoff and the cost of being state-of-the-art.

 

[gedlee]

You are grasping at straws,

At 80 dB! I believe US protocols for human testing limit exposure to 80 dB but musical performances both live and recorded often exceed these levels and it's quite possible +2 dB linear distortion is audible at 90 or 100 dB.

Frank

Thanks for taking this on. I refuse to deal with such foolishness.

The 80 dB limit is quite correct, and stated in the paper. above this point requires FDA approval and a whole array of added and expensive requirements. Thus we were limited to testing at what is really a very low level. Since the data ALL showed a significant trend towards higher audibility at higher SPLs, there is no doubt that at some point any level of HOM will be audible.

 

[doug20]

Normally we agree on a lot of stuff, but I don't agree with this.

I was a horn lover, until I heard the Summas.

That's why it was such a "Eureka" moment when I realized that Zilch and Xpert haven't heard a modern waveguide. I was in the same boat that they are in, up until four years ago. If it wasn't for hearing the Summas, I'd probably be building an Econo Wave right now!

I believe you've done a single demo of the Summa in public. Did you know that three of the people who were there that day wound up buying or building constant directivity horns?


Doug listened to your Summas and bought constant-directivity Yorkville Unities, designed by Tom Danley.

Lynn Olson listened to your Summas,
and started a project with waveguides.

I listened to the Summas, bought JBL MP215s because they're similar, then stepped up to 'the real deal.'

And those are just the people who I know by name; who knows how many others had their eyes opened that day?

This might be a stretch, but it reminds me of a famous show by The Sex Pistols. It was at a tiny club in Manchester, but it was so mind-boggling that half-a-dozen people in the audience went out and started bands, or got involved in the music business! Some people claim that this show kick-started The Smiths, Joy Division, The Fall, and the Buzzcocks.

http://www.bbc.co.uk/manchester/content/articles/2006/05/11/110506_sex_pistols_gig_feature.shtml

The point that I'm trying to make is that those people at the Sex Pistols show probably had a mild interest in punk rock, but they walked out raving fans, and started bands.

I see the same thing going on with your customers; they're quite enthusiastic but reading about it is academic. You gotta experience it.

Good points.....

My reply to Geddes "Clearly a Horn lover is not going to like what I have to say. None of them have. "

We wouldnt have very long threads of discussions if people didn't like what you have to say

I know your website and all the links from it and on the DIY forums is the ONLY reason Im extremely interested in horns again (Klipsch fan for over 25 years).

The message overall is pretty clear, I don't dwell on how the message is present to long because if its valid I want to learn from it.

 

[doug20]

This is true, and I do believe that a great many of the peaks and dips could be controlled with EQ. But thats not the whole story. When there are many peaks and dips then there have to be significant internal reflections. Whenever there are internal reflections there also has to be HOMs created as a result. The HOM cannot be EQ'd because they are both non-minimum phase and non-lumped parameter (or one-paramter if you like). The HOM wave effects are different across the wavefront or put another way the effects will be different in different directions. Hence HOM cannot be correct with any EQ. This IS the crux of the problem. ONLY the acoustical design can minimize the HOM.

So while the actual "peaks and dips" may not look bad to you, their existance is strong evidence of the existance of HOM which will always have a negative effect on the sound and cannot be corrected.

If I were still doing research I would certainly look to ways to better evaluate HOM and diffraction effects (they are basically the same thing) as the techniques for defining them is seriuosly lacking. But I haven't had a chance to do any research in a long time. One often meets a crossroads where you have to decide to be the geologist who finds the gold, or be the miner who digs it out. Here I think that I have done both - but I put on my miners cap daily.

I think that John's point here is well taken. I can present all the theory in the world about the sound quality of horns and waveguides, but in the end they need to be heard to really confirm the theory. All I can say is that no one who has heard my waveguides has ever claimed to have heard better.

The JBL device is a real bargin, of that I have no doubt, but I would bet my reputation on the fact that no one would find its sound quality to be better than a well done waveguide with foam. But then the later isn't $10. Thats the tradeoff and the cost of being state-of-the-art.

You may not understand why (or care) but I have a better understanding from just that post!

I didnt understand that even what I consider to be small ripples (dips/peaks) in the response could be audible.

 

[Patrick Bateman]

Good points.....

My reply to Geddes "Clearly a Horn lover is not going to like what I have to say. None of them have. "

We wouldnt have very long threads of discussions if people didn't like what you have to say

I know your website and all the links from it and on the DIY forums is the ONLY reason Im extremely interested in horns again (Klipsch fan for over 25 years).

The message overall is pretty clear, I don't dwell on how the message is present to long because if its valid I want to learn from it.

I think the punk rock comparison is fair!

The Sex Pistols came along and threw a grenade into the music scene, and even though their album only reached #106 in the US charts, it's influence is still heard in music that's being released 30 years later.

I'd say it took fifteen years until bands would widely acknowledge their influence (though there were plenty of "underground" acts that embraced the sound quickly.)

I think we'll see the same thing with waveguides; it's been 20 years since they written up in the AES, and by 2019 it will probably be commonplace. It seems to be getting "critical mass" lately.