Beyond the Ariel

Building and tweaking as you go ?

Lynn Olson said:



Frank, there are several ways to go with this concept - none of them wrong, just different. First, pick any widerange driver you like, 8, 10, or 12 inches. Second, choose the bass-assist drivers - one or two, 12 or 15-inch, or a combination of 12 and 15-inch.


Lynn, others,

Appologies to you and others in this long thread if this was already covered (start reading evety post carefully but jumped through last 40 pages or so ... still took me 5 hours but couldn't let go.. ;))

What was again the reason 8'' was the smallest "appropriate" wide-range ?

The reason I'm asking: Building as you go strategy: How about starting around a "well known" field coil like the Fertin 20EX (bit smaller than the recommended 8'', but read very good things about it): First in a minimalist OB, then adding a woofer, then another woofer, then a horn tweeter or a ribbon, etc, etc...

IOW learning and tweaking as one goes and building upon previous stages...

Not so familiar with the larger field coil Fertins, 20EX seems the one most used. I've seen larger field coil Fertin very briefly mentioned in this thread but the 20EX seems the most well known, most model iterations (?), etc...

Others with Fertin field coil experiences, could you pls chime in ?

TIA,

Florian
 
Hi Florian --

I have a pair of the Fertin 20EX in a JELabs OB, and my experience has reflected the comments expressed elsewhere in the thread -- a small diameter widerange driver needs a highpass.

Currently, I run the Fertins wide open + Fostex FT96H with .68uf cap (facing backwards and wired out-of-phase) and they're really great but the setup certainly dictates what I listen to (jazz, small scale classical, chick with a guitar, etc) even though I have a small room and listen at very moderate levels - 75db average c-weighted according to my RS meter.

For a summer project, I had been planning on adding a PLLXO highpass to the Fertin and using a pair of 21" Madison Executioner X21 to cover 30-200hz. However, I recently acquired a quad of Jensen A12 fieldcoils, and I think I might try a 2.5 way with them. I have a pair of Altec 802-8D laying around and am considering getting a pair of 1000hz conical horns from Acoustic Horn, but the high-end is still up in the air (hence I'm following this thread with keen interest to see what Lynn et al comes up with).

Regards,
John
 
diyAudio Editor
Joined 2001
Paid Member
Here is on of my greatest speakermaking discoveries ;)

http://aitwood.com/StoreFront.Asp?WoodType=POPLAR&CATID=10&Section=HALFCYL&wDesc=Half Cylinders (180%20degrees)

Click on : "Millwork Components"

Click on : "MDF Quarter/Half Rounds"

Click on : L Quarter Rounds

or
"MDF Quarter/Half Cylinders"

Or even other titles.

When you get to a list, list, click on a part number and find stuff like this:

A section of a piece 96" long:
http://aitwood.com/StoreFront.Asp?ItemID=8100&wdesc=MDF Prefit (L)%20Quarter%20Rounds

Also 96" long, the photo is just a sample:
http://aitwood.com/StoreFront.Asp?ItemID=4500&wdesc=Half Cylinders (180%20degrees)

http://aitwood.com/StoreFront.Asp?ItemID=7300&wdesc=Radius End Panels
(these are larger panels, the just show a strip as an example)

I known Lynn loves his radius' and this is and easy way to get 'em.

variac
 
I'm really enjoying this thread as there's planty to get the mental juices flowing. Based on my previous enjoyment of the Ariels, I'm going to build whatever comes out as the final design.

When I move back to my home in a year or so, I get back my huge open plan. cathedral ceiling living room. System will be music/HT and as I love horns, will likely be horns front and the BTA O/B's for the rears. However, I also love O/B's at different times and I can see that with some simple switching, and turning my listening chair around, I can probably enjoy both a different times. The horns will be closer too the front wall and the O/B's will be a couple of metres from the rear walls.

Sounds like a plan at the moment.
Sorry for the O/T.
 
Variac said:
Here is on of my greatest speakermaking discoveries ;)

http://aitwood.com/StoreFront.Asp?WoodType=POPLAR&CATID=10&Section=HALFCYL&wDesc=Half Cylinders (180%20degrees)

Click on : "Millwork Components"

Click on : "MDF Quarter/Half Rounds"

Click on : L Quarter Rounds

or
"MDF Quarter/Half Cylinders"

Or even other titles.

When you get to a list, list, click on a part number and find stuff like this:

A section of a piece 96" long:
http://aitwood.com/StoreFront.Asp?ItemID=8100&wdesc=MDF Prefit (L)%20Quarter%20Rounds

Also 96" long, the photo is just a sample:
http://aitwood.com/StoreFront.Asp?ItemID=4500&wdesc=Half Cylinders (180%20degrees)

http://aitwood.com/StoreFront.Asp?ItemID=7300&wdesc=Radius End Panels
(these are larger panels, the just show a strip as an example)

I known Lynn loves his radius' and this is and easy way to get 'em.

variac

Totally cool and worth repeating. Although not drawn on my chintzy drawings, I'd like a 270-degree radius on the sides and back surface of the baffle - in other words, if the baffle is 3/4" thick, then the radius is also 3/4", and sweeps around all the way to back edge, close to the driver. This reinforces the side of the baffle (a little, anyway) and makes for a smooth front edge and a curvy surface on the rear.

augerpro said:
I may have missed it but how do you prevent combing with 12" mids and large horn tweeters?

I live with it and don't worry about it. The Ariel has 5.5" Vifa drivers in a pretty big MTM with a 3.8 kHz crossover, and I think the image and spatial quality is close to superb - certainly a lot better than the commercial alternatives.

The vertical polar pattern is going to be quite a bit better in the new system, without the annoyance and tweeter-location limitations of MTM. Let me tell you, MTM's have issues which go ignored in the usual happy-talk design literature.

This may sound crass - I apologize in advance if I hurt anyone's feelings - but I think a lot of the people who go on and on about how terrible it is to have drivers too big, or too far apart, or exceeding a made-up rule about driver size vs crossovers - have never phase-optimized a crossover, and wouldn't know where to start.

I can tell you right now if the phase angle between the drivers isn't carefully optimized over the whole crossover region, you will never have good driver integration, even if the drivers are very small and touching each other. This is something I learned from Laurie Fincham of KEF in 1975, and he was right. Inter-driver phase optimization is absolutely essential for good image quality - and also getting the vocal range right. And yet I see very little written about it, probably because it's a rather tedious process, and doesn't lend itself to a "rule of thumb".

Why am I going on about this? Well, phase angle is important between the stereo pair, which of course are several feet apart. If it is less than 10~20 degrees, you will hear a tight central image (if your speakers are phase-matched). If the phase speads out to 90 degrees, it will be spread across the whole space, with transients blurred laterally, starting in one speaker and ending in the other. If the phase angle open up to 120 degrees, then the central image starts to fade out and weaken, and you get "detenting" near the speakers. At 180 degrees, you're sitting on the centerline of a null, and it sounds really weird and strange.

Well, guess what, the same exact thing happens when the drivers are an inch apart, instead of several feet. You just have to get a bit closer, but you'll hear it just the same, if you know what to listen for. This is why I avoid crossovers that produce a 90 degree phase angle between drivers - the real-world acoustical crossover always has ripples in it, and in the rippled areas, phase angles increase considerably, producing nulls over a fraction of an octave, a very unnatural effect indeed.

Do lobes get narrower as crossover frequency increases relative to driver distance? Of course. But what about the steering of these nulls? That is a direct function of the phase relation between the drivers, which is the vector sum of the electrical crossover and the driver's acoustical response. An innocent-looking ripple at 3 kHz can kick the vertical polar pattern up, then down by several feet - in the narrow region between 2.8 and 3.2 kHz. Not so innocent, is it? (I invite comparison with the Ariel, measured at the same distance, and with no smoothing applied.)

In the example shown, the artifact is a result of a rigorous adherence to a minimalist HF crossover used with a midrange driver run right through its first HF resonance.

Personally, I regard this as a defect, plain and simple, too bad about the great reviews, high-profile market presence, and US$20,000 price tag. From a production standpoint alone, replicating this exact defect at exactly the same frequency is going to be very difficult. In practice, the ripple and the null are going to be a little different in every speaker they make. It's hard to make nulls all alike - since they are cancellation artifacts, they are extremely sensitive to tuning and production variations.

As mentioned in previous posts, I put tone quality first, well ahead of worries about polar patterns. I cross drivers in the regions where I can get the smoothest and most uniform phase transfer between drivers - this means using notch filters to remove peaks in the crossover region (where they are very undesirable), or making the (small) peak part of the transfer function of the filter. Either approach is valid, but the second approach doesn't take as many parts, although tuning can be a little touchy.
 
Minimum Phase

I need to amplify the point of the previous post - in a less contentious manner, so here goes.

Let's talk about minimum phase. My definitionn is a pretty poor one, and I'm sure the other contributors will have a far more exact definition compared to this ex-technical-writers definition. But I will jump anyway, and give you a broad-brush picture to illustrate what I was talking about in the last post.

In a minimum-phase system - electronics, speakers, etc. - there is an identity between the frequency response and the phase response. One dictates the other. In a non-minimum-phase system, that identity is broken, and the phase response does not correlate directly (or perhaps in any way) with the frequency response.

In the world of electronics, non-minimum phase is rare, unless you use a specialized gizmo called an "all-pass filter". These have flat frequency response, but spin the phase through 180, 360, or more degrees. These funny little circuits were widely used back in the quadraphonic days to create 90-degree phase shifts for encoders and decoders. (You'll see them used all through my Shadow Vector patent. They measure perfectly flat, but do really weird things to square waves.)

In speaker drivers, they are minimum phase over most of their working range. Note the use of the phrase "most". When a driver gets close to the breakup region, it starts to depart from minimum-phase behaviour, for the simple reason that there are multiple centers of radiation, some closer to the listener, some further away, and the output of these regions rapidly fluctuates with frequency.

So let's take a hypothetical almost-perfect driver with completely flat response - with one exception. It has a big peak at 1 kHz, and the driver is still in the minimum-phase region.

What does that peak do to the phase response? Quite a lot. At the very top of the peak, the phase response is unaltered from the flat region, but midway up one side, it is retarded, and midway up the other side, it is advanced. Phase shifts of 45 degrees are not unusual with peaks from loudspeaker drivers.

Well, big deal, ripples in frequency response cause ripples in the phase response. No news there. It is a big deal when that peak falls in the crossover region, though. Now the nice perfect 90 or zero degree phase angle (between the drivers) is all spoiled. Since both drivers, and each part of the crossover filter, is minimum-phase, we now have a great big phase deviation right where we don't want it - in the middle of a crossover region.

What makes this phase deviation much worse is when it falls at a fairly high frequency - the drivers are one or more wavelengths apart, so the vertical lobes are starting to narrow, and the response variation falls in a region where the ear has peak sensitivity to response variations - in the 1 to 5 kHz, peaking at 3 kHz. What might be completely acceptable at 500 Hz is not acceptable at 3 kHz.

Small bumps in the frequency response create rapid shifts in driver phase, which is fairly benign by itself, but a big problem when it is in the middle of a crossover region. Why? Because the phase steers the vertical polar pattern up and down - not in the controlled way of the idealized models shown in the speaker-design books, but in an uncontrolled way that occurs over a very narrow frequency range. This does not correspond to anything in nature, and as a result, sounds quite artificial and unnatural.

So it is very worthwhile doing everything we can to smooth the response of the drivers in the crossover region. Small ripples that would be of no consequence elsewhere in the response become very important, since they can steer the polar pattern up and down, or even aim a null at the listener, the worst-case situation.

As for minimum phase of the entire speaker system, well, to get that, you have to go to the trouble of designing a linear-phase system. The easy way is to use 6 dB/octave crossovers, but paradoxically, that can make the whole polar-pattern-steering problem much worse, since 6 dB/octave crossovers have such broad crossover regions. In addition, driver IM distortion is several times worse with 6 dB/octave crossovers, not a minor consideration.

If you use higher-slope crossovers, like the acoustic 24 dB/octave crossovers of the Ariel and many other speakers, then the speaker system as a whole becomes non-minimum-phase. Even the theoretically ideal crossovers behave like allpass networks, spinning the phase through 180 or 360 degrees at the crossover frequency. More crossovers (like a 3 or 4-way), more phase spin, so the square waves get pretty weird looking.

This can be corrected - before the fact, as it were - with digital gizmos like the DEQX, which can analyze the departure from minimum phase and come with an inverse function (in the digital domain). Some people swear by these things, others swear at them. Myself, I've heard the before-and-after of the switch being flicked, and it was barely audible. For other people, it is day and night. There's a fair amount of controversy about the audbility of phase distortion - and if so, at what frequencies.
 
John,

nullspace said:

I have a pair of the Fertin 20EX in a JELabs OB, and my experience has reflected the comments expressed elsewhere in the thread -- a small diameter widerange driver needs a highpass.

Currently, I run the Fertins wide open + Fostex FT96H with .68uf cap (facing backwards and wired out-of-phase) and they're really great but the setup certainly dictates what I listen to (jazz, small scale classical, chick with a guitar, etc) even though I have a small room and listen at very moderate levels - 75db average c-weighted according to my RS meter.

Your setup is pretty much the standard Fertin arrangement I was referring to.

Unfortunately I share (with a vengeance..) Lynn's penchant for large symphonies / orchestrals. As such reproducing properly (i.e. not compressing) those explosive dynamic transients are only second to tonal accuracy in my wish list. In that respect a setup that puts that "chick with guitar" on your lap might have a hard time putting a whole symphonic orchestra there too .. :)

My inquiry was two pronged:

On one hand, if the 20EX can handle _properly_ those large orchestrals (referred to by Lynn in the first post on this thread) when employed in a multiway OB setup.

On another hand gather experiences/impression about the larger Fertin field coils (could find only very,ery few references).

Thanks again,

Florian
 
Crossover design is coupled to lobing and power response, which John K has provided some execllent educational material.

This team might also want to select a common baffle modeling CAD tool so design ideas can be quickly shared and tested.

http://www.musicanddesign.com/Power.html


Consideration of Power Response in Speaker Design: Crossover
Choice and Setup.
© 2006 John Kreskovsky

http://www.musicanddesign.com/
 
Hi

if anybody here is using Comsol 3.3 ? I know that Onur Ilkuror has access to this program at his university and he tells me that it is a very powerful tool.

There is also FEMM

http://femm.foster-miller.net/wiki/HomePage
"http://femm.foster-miller.net/wiki/HomePage"

Weaken the field with a coil would increase the Q of the speaker to some degree what might be good for OB but has also trade off's with distortions when leaving the saturation area.

http://www.audioasylum.com/forums/set/messages/37696.html
"http://www.audioasylum.com/forums/set/messages/37696.html"

For other applications it might be good to "overdrive" the field of the magnet, going even deeper into saturation.
I would expect that the sonic characteristic should become more that of the fieldcoil speakers this way.


Greetings
Michael
 
Lynn,

I guess it's time to start building and experimenting...

My full ranger will be the 12" TT Alnico (since I already have a couple).
My preference is to use one amplifier for the full panel and its limiting my driver selections for the mid/bass and bass drivers (Qts > .5)
The new 15" Super Boy Tone Tubby Qts is 238javascript:smilie(':mad:')
mad

Would be Ok to use 2 X 16 ohms 12" TT ceramics for the Mid/Bass and Bass?
Are any other alternatives? Been googling but can't find anything suitable.

regards
Frank
 
fiacono said:
Lynn,

I guess it's time to start building and experimenting...

My full ranger will be the 12" TT Alnico (since I already have a couple).
My preference is to use one amplifier for the full panel and its limiting my driver selections for the mid/bass and bass drivers (Qts > .5)
The new 15" Super Boy Tone Tubby Qts is 0.238
mad

Would be Ok to use 2 X 16 ohms 12" TT ceramics for the Mid/Bass and Bass?
Are any other alternatives? Been googling but can't find anything suitable.

regards
Frank

Hi Frank -

Yes, I'd certainly use a pair of 12" TT ceramics. They'll be tonally compatible with the widreange driver, an important consideration. By intentionally using conventional air-core inductors for the cascaded lowpass filter, you'll be adding an ohm or so of series resistance to the midbass and bass drivers, for once, a good thing.

I've been thinking along similar lines for other drivers, too. If the superb Radian 12" coaxial is used, the obvious choices for the midbass and bass drivers would be the Radian 15" and 18" drivers. If the 18Sound 12NDA520 is used, the obvious choice for the midbass and bass drivers would also be tonally similar 15 and 18-inch models of the 18Sound drivers.

On a closely related topic, I've made a cross-post over in the OB/ScottG thread. The following post by Magnetar certainly has some delicious pictures.

Since there seems to be universal agreement amongst the serious, been-there-done-that compression driver crowd that large format is the way to go - those wonderfully relaxed but super-dynamic mids and lower treble - there's also a following for supertweeters as well.

It hadn't really occurred to me that the Radian 5312 coax could be used with a supertweeter, but since it uses a 2-inch exit/3-inch-diaphragm compression driver, it's not a bad idea. Here's an interesting candidate: the RAAL 140-15D ribbon tweeter.

Extending the discussion over in the OB/ScottG thread, horns are terrific at extremely low distortion and wide dynamic range, but asking for more than a decade of frequency response is going to compromise performance at one end or the other. At the low end, there's the triple catastrophe of excess excursion, horn cutoff, and the standing wave region just above cutoff. At the top, you're pushing what the diaphragm can do, and reach the first-breakup region. The less you ask of the bandwidth, the more you get back in terms of relaxed sound and low distortion.
 
Lynn,

I have been eyeing these 8" Radian Coax's on ebay......

No experiance with them but I'd wager they are sweet if you use a good 15 below them.

a937_12.JPG
 
Talk to Radian...

I cannot recall the details but when I purchased a set of these some years back they (Radian) told me that the plastic cone versions were the ones to get. The ones I have have the stamped frames and the half round surround with the plastic cones.
If the price is $125.00 the pair you could afford to have Radian recone them for you. Remove the comp driver to save some shipping cost. Worth a phone call to Radian to discuss the options.
 
Re: Talk to Radian...

moray james said:
I cannot recall the details but when I purchased a set of these some years back they (Radian) told me that the plastic cone versions were the ones to get. The ones I have have the stamped frames and the half round surround with the plastic cones.
If the price is $125.00 the pair you could afford to have Radian recone them for you. Remove the comp driver to save some shipping cost. Worth a phone call to Radian to discuss the options.


What's wrong with the compression driver?

I prefer paper cones over plastic and cast frames over stamped!

I just bought a pair from him for $ 200.00 - the compression drivers are worth more then that :) At least to anyone looking for a good new 1" aluminum compression driver