Large midrange for OB??? Scott G ?

[MBK]

Awrite, then. I suppose I have to go ahead at some point... I'm just afraid my ignorance in horn design might make me waste other people's time (the ones with the tools).

I fiddled a bit with the design, made it into an insert for a flat baffle, with a recess in the back to accomodate a disk shaped adapter to bolt the driver on (with this shallow horn, there is no space for a flange and so any mounting bolt would have to go through to the front side of the horn, so I made space for a backplate disk that the driver attaches to, which can then be screwed to the backside of the horn/WG). Model attached...

Data still the same: 1.4" throat for compression driver or 1" dome+surround, WG has OS throat section for 120 degrees nominal coverage, merging into LeCleac'h flare with nominal 380Hz cutoff. Diameter ca. 14", depth ca. 3", expected to keep pattern control at least down to 1kHz (800 Hz hoped-for). One shall see if these predictions beat astrologers and crystal balls...

[454Casull]

Can you really get waveguiding properties by applying the OS equation at the throat only?

[MBK]

The OS profile turns into a conical horn very quickly (read: straight line profile). The deviation from "straight" is then in the fractions of a millimeter. From then on all that's left to do with it is to terminate it somehow, into a flat baffle, or a complete roundover. One could choose a simple large radius, and many do just that. For this design I chose to merge the contour to a LeCleac'h profile of a size such that angle and diameter would exactly line up with the near-conical section of the OS WG, because the LeCleac'h, just as the OS WG profile, claim that the wavefront remains perpendicular to the horn walls at all times. Since it's physics it's all approximations of a complicated reality though (the physicists will wiggle in horror, but then again, I'll say, Lorenz, turbulence, chaos, I rest my case).

[454Casull]

That's all well and good. I just didn't see any straight sections.

[ScottG]

Lynn, I bet we hear VERY much the same. And while I like different designs as an exercise, I wouldn't be at all surprised if we like very similar attributes in a loudspeaker. (..again, I personally use a fullrange low mass driver (lower passband limited) with a pair of dipole midbasses and a supertweeter.)

The BBC school of design is actually quite in "keeping" with most of what I was suggesting, as was the Ariel.

Strange? But consider that both designs are wide dispersion at all freq.s but the top most octave. Figure 4 shows the horizontal off-axis response of a BBC monitor design:

http://stereophile.com/standloudspea...th/index3.html

Its not a constant coverage design.. BUT it is a fairly wide dispersion design, AND there isn't a substantial loss of spl's off axis near the crossover (i.e. the horizontal response is reasonably uniform). Nor is the design a Horn-based one.

OK then.. why the suggestion of constant coverage for a Horn?

The perception of freq. balance and discontinuities in sp-levels (particularly near the crossover), as well as maintaining sp-levels throughout most fundamentals (specifically in the upper octave response) to "stabilize" imaging. (..not insignificantly, it also allows for a larger compression driver which in many circumstances can provide better sound.)

The first problem - freq. balance:

Paul's most recent post starts showing the subjective problems with the first problem (..and I believe thats with a LR 4th order slope and a fairly wide dispersion design, both of which can mitigate the problem).

Disregarding issues of hearing horn mouth problems, what happens is that the apparent sp-level of the horn/compression driver sounds significantly elevated when compared to the partnering direct radiator driver. The more directive the horn is - the worse the problem. At BEST it gives a forward sound with a reduced soundstage and compressed imaging. At WORST it becomes ear-bleedingly "bright" or "sharp", provides virtually no soundstage and no depth to images (..with nothing but the horizontal separation apparent - i.e. classic "pong" stereo).

Now you can, (to an extent) "get around" this problem IF you keep that highly directive tweeter out of the range of fundamentals. Alternativly, you can increase the the in-room freq. balance with either a bipole or dipole operation. I believe Bastani's do both - a high crossover freq. for the compression driver and an open back dipole radiation. Still, its suboptimal - precisely because it asks far to much from the 12 inch fullrange driver (..which I'm betting sounds a LOT like a planar design thats point-source rather than line-source sounding.)

To add to this complexity - there are horns/waveguides that are wide dispersion designs near the bottom of their passband, but are NOT wide dispersion as freq.s increase (..i.e. they become increasingly directive). The waveguide that Paul uses is a minor varient of this. Strangely this can compound the (sp-elevated) problem IF a steep crossover is used AND the partnering driver is not of a similar dispersion pattern near the crossover. Even if well designed, the user virtually always must listen off-axis, otherwise the sound is to "bright". (..it not a matter of "do I want to or not" as it sometimes is for directive direct radiators, but rather "I MUST" listen at least some off-axis.)

Here is an example of a poorly designed loudspeaker with this problem (note figure 5):

http://stereophile.com/floorloudspea...51/index5.html

I'd bet that MOST of the bright character Fremer describes is NOT attributed to the elevated response, but rather IS attributed to the dispersion problem from waveguide to driver at 1.9 kHz and its difference from off-axis responses above and below this wide dispersion.


Second problem - image quality:

Lets say you have a *good* design. The design is wide in dispersion up to about 3-4 kHz, yet well away from the horn's cut-off freq.. The coverage angle/dispersion is a smooth transition from direct radiator to horn. It IS a little "forward" sounding because of its increasingly directive nature at higher freq.s, but it isn't something that would be deemed as universally objectionable.

Image location in such a design is usually quite precise, often described as "pin-point". Soundstaging CAN be wide (..if not particularly deep), IF you have the loudspeakers "toed-out". In fact with this design (and "toe-out"), imaging can be placed (recording dependent) "outside" of the loudspeaker's boundaries. Again though, field of depth is often compromised.

Now such a design probably uses a smaller diameter compression driver. To an extent this allows for less directivity as freq.s increase. There are however "prices" to be paid here.

Smaller compression drivers tend to sound more "compressed" than larger drivers at lower freq.s. This limits the ability of the loudspeaker to reproduce a 3D image (..or what reviewers typically refer to as "palpable presence"). IMO the suspension just isn't compliant enough with these smaller designs (even after "break in"). Sure, it dampens resonances for clean time domain behavior, but at the same time it also introduces signal loss. The same can be said for ring radiators like the BMS drivers.

IMO the "best" of the breed with regard to reproducing a 3D sound is:

1. a larger compression driver (typically 2 inch format)
2. a compliant suspension (relative to a compression driver anyway) after "break in" (..usually with a mylar surround).
3. specifically using high quality aluminum with its very low loss material level (ESPECIALLY for a given profile), but with a moderately high ability to react to change. (Note Titanium is fairly similar to Aluminum, slightly more loss but also with slightly more ability to change - HOWEVER the profiles that are often used with this material are substantially more "lossy" than they otherwise could be - as such, I've never heard a titanium compression driver that sounded good.)

..to me then, Radian is a "no-brainer" ..once the driver has been "broken in" and is used properly.

So getting back to "constant" coverage..

The problem then is that horns with larger diameter drivers have a more difficult time reproducing higher freq.s off-axis (..as indeed almost any driver does). This means that while even a relatively good design with an increasingly directive upper freq. response with a 1 inch driver - starts to become a poor design with a 2 inch driver.

A good constant coverage horn can trade a bit of low freq. wide dispersion to obtain higher freq. wider dispersion. With a limited vertical dispersion pattern (which can be good), it can often become fairly "constant" for most of its bandwidth, and HOPEFULLY most fundamentals. If the response suffers from a loss of "air" or dispersion at the highest freq.s.. well, then thats what a good super tweeter is for. So thats the "deal here"!

So again, while directivity might not seem important - it IS important, especially, when coupling a horn/compression driver to a direct radiator midbass. Moreover for any given horn-to-direct radiator based design there is a moderately precise location for a particular crossover. Its not like saying, I'd like it placed higher in freq. (..or lower). (..a 1st order crossover combined with a 2" compression driver, a low cut-off freq. horn, AND a very narrow vertical exit can be an exception.)

Then there is the subjective bit where we may well have different likes and dislikes..

Runing a 12 inch driver up that high in freq. does give a definite sound. As I mentioned before, the sound (to me) is rather like a point-source planar. Sure, it sounds good.. but thats not the question. What is the question is:

Will a "full range" 12" driver based design (..assuming you could get a properly functioning high freq. filler driver), sound better for you than a REALLY good horn-to-direct radiator sound?

THAT's a tough question - in that I doubt you have a suitable reference for comparison for the REALLY good horn combo (..played back on a system that you consider reference quality).

Quickest *good* answer would be to hop a plane over to Paris and ask Jean-Michel if you could listen to his system. Its a stretch.. but try getting what you can from the subjective comments toward the end of the page here:

http://www.arduman.com/aa/Sayfalar/l...h/lecleach.htm

IMO what would suffer the most would be *ultimate* depth when compared to a "fullrange" 12" design. What you would gain would be clarity and precision (..the latter depending on how close you listen to speakers).

As to size..

Yes, its going to be big AND be even more restrictive on crossover freq.. (heck, you can read the dimensions for the 500 Hz Azurahorn. )

So if this isn't for you then stop now and look for other alternatives.. (and I can think of quite a few there as well).

[MBK]

Hallo Michael,

have Arta too (demo/original sub- 1.0 version) , just didn't think of it .

Usually I use Audiotester, it does almost everything and has nice data saving and curve operations, that make comparisons a lot easier than on Arta (I hear thar Arta is being improved in this respect though).

Audiotester is also maintained a lot but still, occasional problems remain. I think I have a delay problem, maybe also related to my M-Audio Pre with USB1.1 that I use. Signal propagation time is not constant from measurement to measurement and I suspect it just messes up the triggering/timing. So phase and decay measurements are a shot in the dark, but again it may be the M-Audio that's to blame.

[Lynn Olson]

I have to reluctanctly concede the best horn HF I've heard came from 2" aluminum-diaphragm CD's. The two big Altecs (probably A5's) at the San Francisco meet I attended some years ago used theater-size multicell horns and had much more vivid and natural tone colors, along with a lot less A7-type squawk and harshness, than typical 1"-driver horns. Point taken. The 1-inchers sound like they're working hard, and the 2-inchers don't. More importantly, the 2-inchers were a lot more musical.

I visited the Radian site and the 850-PB compression driver seems to be the best one there, with smoother response than the 950-PB neodymium alternative. They do mention the 950-PB has lower distortion, though - choices, choices. Any thoughts on this?

I note that Radian recommends a 1.2 kHz crossover at 12 dB/octave. This begs the question - which and what size horn? I see Azura makes two reasonably obvious choices: the AH-340 and AH-550, the 340 having a diameter of 600mm while the 550 has a diameter of 300mm. What exactly does the larger horn with a lower cutoff accomplish - less HOM's and polar-pattern variation since the Radian compression driver is coming in at 1.2 kHz, and the horn cutoff is thus further away?

To maximize vividness of tone color, as you mention in the other forum, the 12" Alnico Tone Tubby is a good choice - the "plateau" region is not that important since the crossover is going to knock it out anyway. The new 15" Tone Tubby is a candidate for floor-mounting and the 1/f compensation driver.

[MBK]

454Casull,

here is a profile.

[ScottG]

..and those multicells are FAR more flawed by comparison (..excellent drivers though).

Ah, into the weird world of driver spec.s! Its not just the impedance that can be misleading.

Here the problem is that the 1.2 kHz 2nd order is in part based on the horn they measure with it. The same is true for the freq. response as well. Unfortunately I can give you no better answer than that it is likely the more expensive unit sounds AND measures better with a better horn (and filter depending of the type of horn). (..note that the same holds true for TAD as well - the 2001 vs the 2002.) On the other hand I seem to remember someone mentioning that there is an .."audiophool" (..ehm... cough..) ENTHUSIAST working at Radian that would likely give you an unbiased response - so you might try contacting them to see if such an individual exists and could give you more information. (..after all, why spend more for less if thats the case?)

The end result is that larger horn buys you a lower crossover point and increased clarity as a result of the extended passband for the compression driver. As a result it also increases the number of drivers likely to work well (..at least technically) with a typical horn, because it gives a bit more freedom for using a 15" driver.

On the other hand though, you have heard the cr@p that people are using for drivers.

Geddes may have an excellent driver for low freq. extension, the correct polar response for the crossover, and tons of power handling - but "delicacy"/"finesse"/"nuance"? No, that 15" BMS driver was made for a "stage sub" for a reason, and it certainly didn't have anything to to with any of the above descriptors.

To go farther you have to think of low mass for a given sd. Both 12" drivers you are looking at have that (..and actually extend into what I term as the "exotic" range of drivers).

IMO, IF you get the right midrange driver and the right crossover in conjunction with an excellent horn & compression driver - then the "need" for a lower freq. passband for the compression driver is moot, UNLESS you are striving for VERY HIGH spl's (averaging over 105 db).

In addition to this however, is also the question of the driver's problems at resonance. Though I'm not sure, I suspect that the Radian's resonance is not nearly as clean as the BMS, and so a crosspoint of 1.2 kHz is more than an octave away and will effectively be inaudible with a good 2nd order driver (..in that there will be "masking" by the midbass driver).

The short answer then is:

Option 1: If you want to push the crossover from 1.2 down to a limit of 900 Hz,(probably more like 960 Hz), with a 2nd order electrical - then you would spec. the larger horn. (Actually, without regard for the resonance of the driver, the Le Cleac'h roundover exit profile should be able to go quit a bit lower than the mid 900's.)

Option 2: If you are OK with a 2nd order from 1.2 to 1.3, AND want the benefit of a more extended off-axis response at higher freq.s, then go with the smaller horn.

Pair it with either of those 12" drivers and I'd pick option 2 every time.

Of course I'd still opt for a CD design if possible.

EDIT: I suspect the 15" tone tubby would work very well for your open baffle compensation.

[mige0]

Hi

MBK, I also like and use AudioTester - its really good, easy to use, packed with features, not pricy and well maintained.

http://www.audiotester.de
"http://www.audiotester.de"

but as for CSD I didn't manage to come around with it either (I have different external soundcards too choose from) and so for CSD measuremants ARTA is the better choice for me.

By the way for those who are interested to perform stored energy measurements in the way like Siegfried Linkwitz does, I asked Ulrich Müller to implement a stimulus alike some time ago. As far as I can see, dipole and open buffle discussions are a lot connected to "philosophical" aspects of stored energy.

These dayes he finished his work and came up with AudioTester version 2.2b / build2. The signal can now be choosen under "Dual Sinus" with "Multiply" checked.
In the HELP file I also saw that stimuli can be stored to files, which is interesting when you would like to create CD's for testing.

I din't have time to check it out in detail as I am currently don't have access to my equippment.

Ivo Mateljan told me that he will come up with the next version of ARTA in June...

Both, Ulrich Müller and Ivo Mateljan are very responsive to feature requests and definitly like to get some good and precise feedback about their software.

Greetings
Michael