Loudspeaker Wave-front Propagation

[Deang]

[jt1stcav]

Just goes to show you can't beat a horn-loaded tweeter...

Right?

[Strabo]

It depends.

Are these supposed to have been made by a single wave pulse?

[sfogg]

What was the input signal?

If it was a single pulse the plain horns response doesn't look so good. But on the flip side I'm not sure that says much since the SPL out of each of the tweeters will vary since they all received the same voltage input. Might be more interesting to see how they compare playing at the same relative SPL.

Shawn

[Deang]

Looking at the graphs -- I would guess single pulse.

Shawn has an excellent point. The text says a maximum pulse of 10v was used. Why sure, if the same voltage was applied to each -- then that horn was shattering glass while the others were at a mere whisper. Of course, that would be an absolutely retarded way of doing the test.

[Manuel Delaflor]

Interesting graphs, I wonder why nobody (AFAIK) do that kind of study right now. I would like to read the entire paper, to really understand what is going on.

[sunnysal]

dean, where did you dig that up? I would love to read the whole thing. to see how the room, etc. might afffect the results. it seems kinda nutty to work with 10V input instead of equivalent SPL...no? anyway let us know where we can find this, thanks, tony

[justin_tx_16]

It looks like it was from the Dope From Hope compilation I put into a PDF file.

[Deang]

Here http://www.snippets.org/index.php3 in the Audio Library.

There is no 'whole thing' -- that was it.

[sunnysal]

oh well...I will post on AA to see if anyone there is familiar with the article...I will let y'all know what I find out. tony

[artto]

Figures 5 & 6 are non-existent. Measurement techniques are not specified. The source signal is not specified. Performance specifications for the drivers not specified. The pulse time wavefronts apparently have something to do with missing Fig. 6.

Totally meaningless.

[LynnOlson]

This article originally appeared in the 1976 January/February issue of the Journal of the Audio Engineering Society (JAES). Since I was an AES member at the time, I dug up my old 1976 copy and scanned in the entire article, so the figures make a little more sense. (Note these are 150dpi GIF files, so it'll take a while to appear on slow connections.)

http://www.nutshellhifi.com/library/page1.gif

http://www.nutshellhifi.com/library/page2.gif

http://www.nutshellhifi.com/library/page3.gif

http://www.nutshellhifi.com/library/page3a.gif

http://www.nutshellhifi.com/library/page4.gif

With modern MLS systems, the heroic methods of the original technique would be greatly simplified. You'd still need an X/Y motor for the microphone, or maybe the speaker itself could be rotated on a small motor mount.

[LynnOlson]

Oops, forgot a file. The full set from the JAES article follows. (Warning, big 240 to 350K 150dpi GIF files.)

Page 1

Page 2

Page 3

Illustration

Page 4

Page 5

[g3dahl]

Wow, that's quite an article! Thanks for posting the entire piece. I had to print it out for a good look, but it was worth it.

I wonder if the authors followed up on this article. Perhaps they used it as a justification to dismiss horn speakers. I guess JVC isn't known for horns, come to think of it! But we do know that horns do some things especially well, such as reduced intermodulation distortion and fabulous dynamics.

It will be great if research and experimentation of this kind can be used to further improve the sonic results attainable with horns.

Gary Dahl

[sunnysal]

thanks Lynn, that is terrific. I am gonna print it out now and read through..comments to come. lynn any thoughts on the article? seems to damn horns though my questions remain about why 10v versus SPL matching, the room, etc. regards, tony

thanks artto for that "insightful" comment, I think it was not meaningless just lacking in details since it was partial, it did allow us to go out and look for more...which we know have, thanks to lynn, we would not have even known about the article if it had not been posted by dean, partially.

[LynnOlson]

It's an interesting article, and as far as I know, unique in the AES literature. This is unfortunate, since this article is the most detailed analysis and visualization of horn-mouth diffraction I've come across in twenty-five years. As anyone who's measured horns with MLS or TDS knows, horns can have problems with internal reflections, which appears in the time domain as a slowly decaying pulse-train.

Look at the waveform on the direct on-axis part of the illustration, and you'll see the time response of the speaker. Doing an FFT and converting the data into the frequency domain discards phase information; for example, a triangle and square wave have identical frequency spectra, but are quite different as far as an amplifier is concerned. It does take some interpretation to mentally convert time data into the equivalent frequency domain, and to recognize the presence of reflections.

Of course, having come up with the powerful technique they did, the authors picked speakers with dramatic differences in time and dispersion characteristics. It's fascinating to see the usual vague polar plot replaced by time dispersion vs polar angle. It helps to know what an "ideal" driver would do - which is radiate a single pulse over 360 degrees. Such drivers exist - ionic speakers radiating in free air - but are also very inefficient and generate awkward amounts of nitrous oxide and ozone, neither healthy for the listeners. That's why ionic drivers are mostly lab curiosities instead of commercial speakers.

Falling short of the ideal, a simple low-pass function would smooth out and delay the pulse, and you see that with the off-axis time response of the horn/ribbon speaker (almost certainly the famous Decca/Kelly ribbon of the 1950's).

You can play a guessing game which driver is which, but here's my two cents worth:

A) EV T-35 or a God-awful Japanese clone - very likely the latter.

The one and only Decca/Kelly ribbon/horn, a genuine classic and very rare even at the time.

C) Either a good cone tweeter like the Peerless HFC 225 (one of the best tweeters of the day, I used it myself in the Audionics M-32 and TL-30 speakers) or some crummy Japanese copy.

D) The only and only Heil Air-Motion Transformer (AMT).

E) Boy, that could be anything, but I'd guess the dreadful Philips 1" clear-plastic dome tweeter, a real dog, but very popular in the mid-Seventies. Or maybe the KEF T27, still kind of a dog, but more popular in high-end speakers.

F) Maybe the Peerless soft-dome, dunno about that one. Domes in the Seventies were pretty bad, and prone to side-to-side rocking motions that created a lot of distortion in the lower part of the frequency range. The advent of Ferrofluid in the late Seventies helped center the VC in the gap and really improved most dome tweeters.

[sunnysal]

thanks lynn, very imformative. it does make one wonder about the whole "horn thing" though. there was precious little detail in the article about the physical conditions of the test rig/space. I still wonder about the 10v versus matched SPL issue too. I wish we would see more investigation of this nature, I too have seen virtually nothing like this in recent years out of any of the journals. Some stuff published in doctoral thesis' but nothing in professional journals. glad to see you around here to help out, even if you don't use klipsch speakers! warm regards from sunny El Salvador, tony

[LynnOlson]

"Glad to see you around here to help out, even if you don't use klipsch speakers!"

Actually, I just picked up a sharp-looking pair of oak-finish Chorus I's, after hearing Charlie Kittleson's Chorus' at his place a month ago. First pair of speakers I've bought in thirty years (the rest were the ones I designed for Audionics and the Ariels, which grew out of a Positive Feedback article).

Although I've designing medium to low-efficiency speakers since 1975 or so, I'm a newbie to true high efficiency. Rather than try and climb the glass cliff of an all-new high-efficiency system, it's easier to learn from the masters. I've never cared much for JBL - ever - and Altec has never done much for me either. But I've heard enough good-sounding Klipsch speakers over the years that after hearing CK's setup at home, I took the jump and bought the Chorus I's.

I expect this will be an interesting experience, since I essentially learned speaker design at the knee of Laurie Fincham of KEF and the BBC design team, who I visited back in 1975. After the Audionics team returned from the UK, my boss, Charlie Woods, drafted me into finishing the Audionics TLM-200, a four-way transmission-line monstrosity abandoned by the previous designed when he split for parts unknown, leaving no forwarding address. Six months and a 56-element crossover later, the TLM-200 was finished and sprung on the world at the Summer CES in Chicago, and I think we sold a grand total of 12 pairs. Not really my finest effort, but you have to start somewhere.

The more modest Audionics TL-30's, M-32's, M-33's, and T-52's turned out better, though a heck of long way from high efficiency. We were lucky to make it to 86dB/metre with KEF or Audax Bextrene cones and the so-so tweeters of the day. I was startled at the improvement in clarity with the Ariels, and they were merely 92dB/metre. The 101dB/metre Chorus is a whole different world. PWK was right: efficiency matters.

[sunnysal]

Well, I am even more pleased that you have taken the jump into Klipsch. Those are fine speakers. Perhaps this experience will lead you to try Klipschorns eventually, to really hear what PWK was talking about you need to try the Klipschorns, I think he would agree that they were the best, and strangely enough the first, ambassadors for his high efficiency, low distortion concepts. Too bad they are so very picky about room placement! Tony

[LynnOlson]

Well, maybe the Belle Klipsch don't have the corners (and the room) for the big guys.

I'm also carefully reading the series of AudioXpress articles about Bill Fitzmaurice's DR speakers, which have basshorns that make it out to 5kHz! Maybe not the last word in flat response, but still, a basshorn that even makes it 1kHz is kind of amazing. If the DR-horn response can be smoothed out, that would really relax the size and sophistication requirements for the mid horn, letting it cross at Cornwall/Chorus frequencies, around 700Hz or so.

By the way, my current project is smoothing out the top end of the Chorus I mid horn - it has a pair of bumps at about 5.5 and 6.8kHz that are 5dB high. I'll be measuring Gary Dahl's Chorus II's (which arrive on his doorstep today ) and see if the Tractrix does the same thing. If not, then I decide between woodwork on the Chorus I or a notch filter added to the crossover. Back in the bad old days of Audionics, I made notch and inductance-correction filters for each driver even before I did the high and low-pass filters, so that would be nothing new for me. But I also remember there's no free lunch in speaker design - even the most skillful notch filter subtly degrades transparency and the sense of "directness", so you have to avoid over-equalizing the speaker.