Beyond the Ariel

Well, the diameter of the AH425 is 16.5", and it is matched to a 15" widerange woofer. The measured acoustical crossover is around 700~750 Hz, and the drivers are about a wavelength apart. The AH425 presents a resistive air-load for the compression driver down to about 500 Hz.

The measured (not calculated) phase match between the drivers at crossover is 5 degrees or better; the null with the drivers reversed-phased is 25 dB deep, and the slopes of both sides of the null are symmetric. When put in-phase, response is flat through the crossover region, and off-axis response is smooth as well.

The data Gary has sent me looks looks good for a system using large-format woofers and compression drivers, and the subjective reports are matching the objective data. Definitely in a different class than an Altec A5, more like a JBL Everest system, but with a simpler crossover. The woofer only needed one modest notch filter, and the horn did not require shelf equalization or a notch filter, rather impressive for a 1.4" format driver. Response ripples predictably start above 8~10 kHz, although they are smaller in magnitude than I expected, probably thanks to the damping of the Mylar diaphragm surround of the Radian 745P.

It would be interesting if the TrueExtent beryllium diaphragms were available for the Altec 288 or Radian 745P drivers; this would offer response beyond 15 kHz without a supertweeter. Considering the not-insignificant cost of a beryllium diaphragm, a high-efficiency ribbon supertweeter offers more extended response (beyond 50 kHz for the RAAL) and better horizontal dispersion.



Don't go letting Magnetar get to you.. it's a good design. :)
 
The upper-mid horn is a little startling - surely the dispersion couldn't be more than 10 degrees? With a conical profile, it would drop like a rock outside the dispersion window, so there would be an on-off impression as you moved laterally (or vertically). The directivity indexes of the 4 drivers look wildly different; designing the crossover wouldn't have been any fun.

Cast bronze looks classy, I have to give it that, and the wood midrange horn is a work of art. Would have been nice if there were some measurements instead all the pictures taken from weird angles.



It's not hugely dissimilar from something like this (horizontally).

FaitalPRO - Neodymium Professional Drivers



The TAD ET-703 is a very wide dispersion super tweeter.

I suspect that the TAD, while having a very high freq. crossover, is using a first order design - increasing the combined off-axis response above the treble horn's loss in pattern control.

In other words I think it's a high/wide dispersion design with some narrow-band diffraction around 4 kHz (..that likely ripples "upward" in freq.)
 
Last edited:
.....
What about yellow -> green -> blue -> purple -> Red ? What you want is a scheme like "Jet" which only varies one color at a time, then another one and so on. Otherwise its a complex mess.
....
I would leave out red altogether as a pure color - probably. But yes, I do like starting with yellow as the highest level because it looks the brightest. Then down into green, aqua, blue and maybe black at the lowest point. The idea is to have what looks brightest to the eye represent the high levels and the dark colors the low levels. Unfortunately that does not follow the visual spectrum - the eye does not work that way. I'll try some color gradients and present them for our amusement. =)

'rainbow' like color maps are not very good representing data. Adding many different colors to a graph just makes it a perceptual mess. Taking visual perception in mind, a graph combining saturation and luminance variation could make a simple, very detailed representation of data.
summary on the topic:
Why Should Engineers and Scientists Be Worried About Color?
 
Last edited:
Hadn't considered the LV Vox Olympian might be using a diffraction horn, but it seems plausible. Kind of nasty diffraction delay time, though ... the HF horn looks like it's maybe 12" or so long, which would give delayed pulses just a little less than 1 mSec apart. Not pretty, but then, no diffraction horn is, at least in the time domain. Not much attention is paid to diffraction on any of the horns, so I imagine there might be some tradeoffs in image quality.

As for the overall subjective goal for this long-running project, it is coming into view. Sonics in the general vicinity of the Altec Model 19, the JBL Everest series, and the 15" Dual-Concentric Tannoys. The crossover is not as complex as the modern JBL's, with their multiple notch filters, but higher efficiency (the JBL is 94 dB/metre). Low diffraction and clean time signatures remain a priority, as it was for the Ariel back in 1993. Rapid settling times are unusual for large-format drivers, so I'm curious how this aspect turns out.

Still strongly considering an Altec/GPA version with Alnico 416's (notch filter retuned for this driver) and Alnico 288's. There is a certain quality to Alnico-magnet drivers with underhung voice coils that I like. I've asked Alexander of RAAL what might be responsible, and he mentions the gaps of the classic Altecs operate quite close to saturation, and this changes the sonics.

My version may well have the resistive-vent cabinet looking like a scaled-up Fonken, or the full-scale Jensen. Gary's real-world results will guide what I do with my version. One thing I like about resistive-vent enclosures is they are relatively insensitive to dynamic Q variations, so the bass has a somewhat different quality than classical T/S closed or vented boxes.
 
Last edited:
OK, then it should be easy since I am only asking for one solution!

Let's say we start with yellow at the top, which is 255 red and 255 green in RGB. Should I assume that it should end in red?, no probably blue by what you say above. My point to Pano would simply be that it is no simple matter to come up with a color algorithm that would produce a continuos array of colors such that the luminance would be a continuos slope. In fact, it may not even be poossible.

What about yellow -> green -> blue -> purple -> Red ? What you want is a scheme like "Jet" which only varies one color at a time, then another one and so on. Otherwise its a complex mess.
The perceptual research, some of which is referenced in the old IBM paper linked in Daniel's post, is that changing hue does not work well for expressing amplitude variations. A hue change is a good indicator for data that has a clear threshold (for example, indicating sea level in cartographic elevation model data) but otherwise it is better to vary the luminance and saturation. Key is to ensure that the luminance variation for the hue(s) chosen are monotonic as that has been found to give the best results. That is not trivial, but is certainly possible and is used. Blue makes the smallest contribution to luminance, its variation has the lowest perceptibility, so hues made from red and green are better. There has been a lot of research into this for visualisation of medical and seismic data, googling will turn up various papers on it.
 
Thanks Dan!

After settling on the values for the Zobel, I added the low pass filter using calculated values for a 2nd order Bessel at 700 Hz. Of course, the lift from 1.5 kHz to 4 kHz was still there with all its ripples, but the whole thing was rotated clockwise. The next step was to knock down the first big bump with a narrow notch filter. I experimented with various notch tunings, widths and depths. Best results were achieved with a filter centered at 1,720 Hz. IR improved as well.

There was still a problem, though: I wasn't reaching -6 dB until about 900 Hz. Experimentation with added capacitance and inductance eventually resulted in a curve that hit the target rolloff and had the desired shape. The filter was now electrically tuned for -6 dB at 420 Hz! Thank goodness for Arta.

Best of all, the IR and CSD plots looked better than ever. The 2.5 ms wrinkle, while still present, is now a tiny fraction of its former self. I am very pleased with where it ended up. I think it's as good as I'm going to get with a 2nd order filter on a 15" woofer.

Here's the FR curve:

LowerXover1fr.jpg


Here's the impulse response:

LowerXover1_IR.jpg


And here's the CSD:

LowerXover1_csd.jpg


Finally, here is the crossover schematic:

Xover1.jpg


Gary Dahl

Thanks Gary!!!

Was there any on and off axis response measurements done yet?
 
Administrator
Joined 2004
Paid Member
Superb paper - thanks: Just what I was looking for.

Ditto! Most excellent. I hope it's something you can use, Earl.

I did not take long last night fooling around with a dozen or more color gradients to discover pretty much the same things the IBM crew did. Luminance and saturation are key. And most color gradients don't show subtle variation well. The 2 below are the best I could come up with quickly, and you can see they aren't great. Better then JET , tho IMO.
The IBM stuff is superb.
 

Attachments

  • example-1.png
    example-1.png
    5.3 KB · Views: 791
Hello Panomaniac,

Your upper colormap is quite similar to the one used for the spectrogram in Adobe Audition (it also inspired from the Hot colormap...).

I doubt the usefulness of your second as our eyes (and our computer screens) are weak inside the green range.

See here attached, a screen shot of a spectrogram from Adobe Audition.

Best regards from Paris, France

Jean-Michel Le Cléac'h

(I am probably more accustomed to Matlab Jet and HSV colormaps and prefer them. I must add that I am not at all bothered by the fact that their luminosity variation isn't linear. I guess that each different application needs its own colormap).


Ditto! Most excellent. I hope it's something you can use, Earl.

I did not take long last night ... Luminance and saturation are key. And most color gradients don't show subtle variation well. The 2 below are the best I could come up with quickly, and you can see they aren't great.
 

Attachments

  • AdAud_spectro.jpg
    AdAud_spectro.jpg
    13.9 KB · Views: 701
Last edited:
Hello Panomaniac,

Yes, there is some noise on the spectrogram (purple on black) and the pulse is designed "by hand", in order to show all the colors.

Best regards from Paris,

Jean-Michel Le Cléac'h


That's a good one JM. Do I see a pulse with more slightly more energy at the low end? And background noise (magenta) before and after the pulse? If so, it works for me.
 
I'm on page 100 of Bob Cordell's Power Amplifier book. My comments are here. As a former technical writer for Tektronix, I feel it is one of the best technical books I've ever read, even if your interest is limited to loudspeakers or tube amps. Get it, read it thoroughly, and learn a few things about that mysterious black box you've been listening to for all these years.
 
Last edited:
It's a short list:

Bob Cordell Designing Audio Power Amplifiers

Morgan Jones Valve Amplifiers, Third Edition

Morgan Jones Building Valve Amplifiers

Ben Duncan High Performance Audio Power Amplifiers

Ralph Morrison Grounding and Shielding Techniques

Newell and Holland's Loudspeakers - For Music Recording and Reproduction

Earl Geddes, "Audio Transducers" (order direct from Gedlee)

These are the books on my reference bookshelf, along with old-timers like the RCA Radiotron Designer's Handbook (4th edition), Valley and Wallman "Vacuum Tube Amplifiers", HF Olson's "Acoustical Engineering", every issue of Sound Practices, Vacuum Tube Valley, and Speaker Builder magazines, along with Audio Amateur going back to 1971. Also the AES Journal from 1971 through the early 80's.

Unless you are researching the history of audio, the books in the top list are the ones to get. Yes, they are pricey. If that's an issue, start with Newell and Holland, and then get either Bob Cordell's book (if you're into transistors), or Morgan Jones' book (if you're into vacuum tubes). If you are a beginner, get the Loudspeaker Design Cookbook by Vance Dickason.
 
Last edited:
How could you leave out Toole's book? Granted Earl's book has some information on psychoacoustics, but Toole's book has tons of valuable information in it. If you don't know how your brain interprets what the speakers produce, then you have little basis for effectively applying information in any of those other books so in my world it's at the top of the list.
 

ra7

Member
Joined 2009
Paid Member
Unless you are researching the history of audio, the books in the top list are the ones to get. Yes, they are pricey. If that's an issue, start with Newell and Holland, and then get either Bob Cordell's book (if you're into transistors), or Morgan Jones' book (if you're into vacuum tubes). If you are a beginner, get the Loudspeaker Design Cookbook by Vance Dickason.

Thanks Lynn... I'll start with one or two of the ones you suggested.
 
RAAL has also recently announced a dipole ribbon tweeter. I found this early documentation. Personally, I would minimize the baffle area by separating the ribbon's transformer off of the front mounting plate.

www.solen.ca/pdf/raal/DIPOLE140-15D.pdf

A DIY dipole ribbon tweeter is simple to construct from 0.5" x 0.5" NdFeB N50 magnets and some 0.5" square 1018 steel. I build 4" long x 0.7" wide dipole ribbons which can easily be crossed at 1,500Hz, and mount it on top of no-baffle Lambda TD15D or Tang Band W8-1808 mids. Material cost ~ $50. A no-baffle dipole ribbon tweeter has modest area required for good dispersion.

I wonder if a Fish Speaker would sell well in the Pacific North West.
 

Attachments

  • RAAL dipole.jpg
    RAAL dipole.jpg
    36.2 KB · Views: 803
  • Pacific NW Fish Speaker.jpg
    Pacific NW Fish Speaker.jpg
    62 KB · Views: 818