Beyond the Ariel

[Jmmlc]

Hello Lynn,

The expansion of the wavefront area for a Le Cléac'h horn follows the general formula for the hyperbolic horns.

See formula (11) in Bjorn Kolbrek' excellent paper "Horn theory" published by AudioXpress:

http://www.audioxpress.com/magsdirx/...olbrek2884.pdf

For the DIYer the T value can be chosen from 0 to a value over 1.

T = 1 is for an exponential expansion of the area of the curved wavefront. With most high efficiency drivers this results in a slightly upwarding slope of the frequency response curve.

For T =0 you'll obtain a horn with a catenoidal expansion of the area of the wavefront. It is rarely used (but simulations of Le Cléac'h horn with T = 0 shows very few reactance above the acoustical cut-off (but a narrow and tall peak of reactance at acoustical cut-off frequency)

Martin Seddon's Azura horns generally use T from 0.7 to 0.8. This gives a very flat frequency curve on axis. Give a look to the measurements:

http://www.azurahorn.com/Yamaha_on_204.pdf

http://www.azurahorn.com/Yamaha_on_204.pdf

Values of T lesser than 0.7 may be interesting with a driver having (when loaded) a resonance frequency near the acoustical cut-off of the horn.

Low T value also results in longer horns.

Large T values over 1 lead to profile having similarities with conical horns.

Estimation of the acoustical cut-off frequency of a horn from the perimeter length of the mouth is IMHO useless for quasi-infinite horns like the Le Cléac'h horn (or with the Kugelwellen if mouth opened at more than 360°). I call quasi-infinite horns, horns whose mouth shape is able to reduce drastically the reflection of waves (and diffraction at the edges of the mouth).

Measurements shows that if a Le Cléac'h horn profile is calculated with a full mouth curvature (360°) then its cut-off frequency is exactly what was introduced in the calculation (by my spreadsheet or by HornResp ).

The crossover solutions I studied are not devoted to the Le Cléac'h horn specifically. My goal was to obtain low phase distortion and very constant "in coincidence response curve" (this differs of the "power response curve".

Best regards from Paris, France

Jean-Michel Le Cléac'h

Originally posted by Lynn Olson


Jean-Michel, I want to thank you for joining the thread - as the designer of the Le Cleac'h profile, I give particular weight to your remarks. I was curious myself about the air-load reactivity of the larger vs smaller horn, and the possibility of "wasting" some of the potential midrange bandwidth of the 1.4" format 288 compression driver due to the 550 Hz horn being just too small to take full advantage of it. (If that's true, perhaps the best application of the 550 Hz horn is to mate with 1" format compression drivers.)

I've been curious why the 340 and 550 have the cutoff frequencies they do, since the 340 is just about twice as large (in every dimension) as the 550. At first glance, it would seem like either the 340 is really a 275 Hz horn, or the 550 is really a 680 Hz horn.

Martin Seddon tells me that with a 1-3/8" (35mm) throat and mounting plate (to match an Altec 288), the AH-340 is 349mm long, has a 600mm overall diameter, and is 500mm across when measuring the portion of the horn-mouth that is at a 90-degree angle to the central axis. With the same size mounting plate, the AH-550 is 168mm long, and has a 300mm overall diameter (and I'm guessing is 250mm across where the flare reaches 90 degrees).

I'm curious what the "T" function of the equation represents - is that the net overall flare-rate of the entire horn, or a ratio that represents possible shapes with either more or less curvature than a Tractrix? Does this control the sharpness of the acoustic highpass filter?

I plan to use an elliptic highpass function with the notch tuned to the Fs (in the horn) of the 288 driver, in order to control out-of-band excursion as efficiently as possible, but without the time-domain penalties of a high-slope 24 dB/oct crossover. This filter topology should allow freedom from having to "tune" the horn cutoff and rear-chamber volume in order to form an acoustic highpass filter - this was necessary when the Altec 12 dB/oct crossover was all that was available, but today we can use more sophisticated filter functions to control excursion and optimize time-domain response.

I'll also looking at the filter function you have described on your web-page - whatever gives the best performance for the application (which in this case is mating with a 12" woofer, either Altec 414, Tone Tubby, or 18Sound). If the 340 delivers an extra half or two-thirds of an octave compared to the 550, that's a considerable advantage in terms of matching with the 12" woofer, regardless of crossover topology.

[Anglo]

Lynn

Thanks for the response. I'll show you some measurements and maybe we can work from there?

++++++++++++

>Beaming sounds unnatural to me<

I am aware of beaming, I have a 15" wideband driver that is run fullrange. I know, scary. Now what exactly sounds unnatural to you with regards to beaming?

Thanks

[JoshK]

I abhor the head-in-a-vise affect of beaming. Move your head a fraction of an inch while listening and the FR changes. No thanks. But then again I am a type #2 & type #3 type.

[Anglo]

Thanks Josh

Tell me, how can one measure beaming? Would it be safe for me to say, I should put the meter on axis and just measure one driver, the one that would present beaming?

Thanks

[Magnetar]

Originally posted by Anglo
Thanks Josh

Tell me, how can one measure beaming? Would it be safe for me to say, I should put the meter on axis and just measure one driver, the one that would present beaming?

Thanks

The easiest way is to put on some music and move your head around. If the sound changes a lot then the speaker is beaming. Panel speakers and big non-CD horns are the worst for this in my experiance. It really screws up the 'power response' too, very unnatural and fatiguing to me. Seems some people live with it though. Big headphones unless you move

[Lynn Olson]

Thanks all, much appreciated real-world experience and info. Anglo, the only measurements I can get much use from are FFT's with at least a 5 mSec window, floor reflection removed through acoustical means, and NO SMOOTHING for the computed frequency response. I appreciate seeing the time data with the selected time window (5 to 8 mSec is useful), and the associated FR data that is calculated from the time data. (This is how I present the Ariel data, for example, so the derivation is always clear, along with the choice of window size.)

If the floor reflection is left in the time data, then the CSD is useless, and the FR data is cluttered with comb-filtering that makes fine resolution impossible to read. 1/3 or 1/6 octave smoothing removes the comb-filtering artifacts but also removes evidence of diffraction and narrowband resonances as well. This kind of smoothing dates back the early-Fifties "pen-damping" days at Altec and JBL, and had the very convenient benefit of concealing defects in the horn and compression driver. Now that we no longer use B&K swept-sinewave techniques with a mechanical chart recorder, there's no longer any reason to conceal fine frequency data. (Pixel damping?)

See my measurement article at Nutshell High Fidelity for the kind of data I find useful - pay particular attention to the artifacts generated by comb-filtering from the floor reflections. With spare bits of household items (coats, pillows, blankets, etc.) you can do pretty good job (20 dB) of attenuating the floor bounce, and get much better data as a result.

JMMLC, thanks again for the detailed design criteria for your horns, and the parameters that you've chosen to optimize. I guess I would fall between the Group 1 and 2 listener, although I don't much care for the "giant headphone" effect of some of the biggest horn systems.

The big systems I've heard (with Lowther drivers) had almost no depth reproduction at all, but I could lay that at the door of the Lowthers, with their very uneven wavefront going into the horn thanks to the whizzer driver. A rough, incoherent wavefront at the input end will surely be far worse when it comes out the other end, and that in turn will have severe effects on time-domain and polar-pattern performance.

Conversely, an effective phase-plug, and avoiding diaphragm and surround breakup, should allow the horn to operate closer to what theory indicates should happen. (The validity, or invalidity, of Webster horn theory being a matter for separate discussion.)

I don't think it's fair to blame the horn for defects in driver performance - a horn is not going to magically transform a driver into something it's not. Compression drivers with Mylar diaphragms will sound a lot like Mylar-diaphragm dome tweeters, just many dB louder with more headroom, and the same for all other diaphragm and surround combinations. So front-horn Lowthers still sound like direct-radiator Lowthers, only spatially bigger, louder and with more headroom - and the potential for additional horn coloration.

I no longer believe decades of Altec and JBL propaganda that horns somehow "damp" diaphragm and surround resonance - I see no measured evidence for this claim, and I don't hear it either. The "personality" of the diaphragm, surround, and magnetic-system linearity is always going to be there, impressed on the time, spectral, and distortion domains. I am even more skeptical of claims of "perfect" diaphragm materials with "inaudible" coloration. We've heard that before with titanium, now it's beryllium that's the new wonder material. It measures better, true, but does it sound better?

Marc Henry (of Music Concrete fame) is reporting good results with the 340 Hz horn and the big Radian 950 (2" throat, 4" aluminum diaphragm with Mylar surround) compression driver. JMMLC, are there any other drivers that are personal favorites - old and new?

[Jmmlc]

Lynn,

May be that's the legend of the picture:

http://www.homecinema-fr.com/forum/a...php?pic_id=798

that lead you to speak about Lowther.

In fact, the legend of the picture is erroneous and the system on the picture don't use a single Lowther driver. This system was demoed by Marco Henry at the last Hifi show in Paris few weeks ago and the drivers are 2 x 46centimeters PHL on open baffle (in the corner of the room) to reproduce the lowest frequencies. 2 x Altec 515 in small front horns for bass, one Radian 950 (from memory) on the large circular 200Hz horn and then a small BMS compression driver on 500Hz horn for the high mid + treble.

It was the best sound of the show.

About 2" drivers, a friend of mine use a JBL2441 but with a Radian diaphragm and I could listen his system and appreciated it a lot (far better than with the original JBL diaphragm).

Best regards from Paris, France

Jean-Michel Le Cléac'h

Originally posted by Lynn Olson

JMMLC, thanks again for the detailed design criteria for your horns, and the parameters that you've chosen to optimize. I guess I would fall between the Group 1 and 2 listener, although I don't much care for the "giant headphone" effect of some of the biggest horn systems.

The big systems I've heard (with Lowther drivers) had almost no depth reproduction at all, but I could lay that at the door of the Lowthers, with their very uneven wavefront going into the horn thanks to the whizzer driver. A rough, incoherent wavefront at the input end will surely be far worse when it comes out the other end, and that in turn will have severe effects on time-domain and polar-pattern performance.


Marc Henry (of Music Concrete fame) is reporting good results with the 340 Hz horn and the big Radian 950 (2" throat, 4" aluminum diaphragm with Mylar surround) compression driver. JMMLC, are there any other drivers that are personal favorites - old and new?

[Anglo]

Hi

Thanks Josh and Magnetar for your response. I do know what beaming is, it is the unnatural part that got me intrigued. Of course when sitting in the "sweetspot" the sound is fine and yes once you move out of that spot, well--- it is a different story! I understand what you mean by unnatural when it is explained this way. Again, thanks and yes I agree.

The thing is the slam and weight that the big (see here 15") wideband driver gives makes any other " less beaming" driver (see here smaller) seem obviously smaller. Instruments lose weight and they sound empty, somewhat filled with helium air at times and tone also lacks density.

Those aspects bother me, but again if I could tame the beaming I would probably be "happier". Any suggestions for this would welcomed.

Is there anywhere in this thread where the beaming of a big driver used as a wideband covering the most critical part of the sound i.e 200hz to 4Khz has been talked about?

++++++

Lynn, o.k. thanks. I will do such measurements and address the floor bounce. Looking forward to addressing some issues.

Thanks,

[John L]

Gosh, this has got to be the most successful thread I have EVER seen! It is impossible to go through it all in a practical sense.

So, let me ask: Are all the items that go into the Ariel, still current for purchase?

[Robh3606]

"Is there anywhere in this thread where the beaming of a big driver used as a wideband covering the most critical part of the sound i.e 200hz to 4Khz has been talked about?"

I will post a LEAP plot for you of a 15 on an IB board. There is quite a bit of beaming going once you get above 1K. You also use this old nomgraph. I like the LEAP plots better.

Rob