Beyond the Ariel

[Lynn Olson]

Originally posted by angeloitacare
hello Linn

How about a driver alignment as on my drawing? Similar to Jack Bouska's speakers ?

Not a fan of big MTM's, sorry. I learned my lesson on the Ariels - getting the subjective response to be flat took several months. I hear the same problems with big commercial MTM's that I experienced with the early versions of the Ariel - except spread over a wider frequency range, extending down into the midbass.

Aside from subjective balance issues, I just don't like the sonic perspective of big MTM's - there's something unnatural-sounding about having bass drivers hanging high up in the air. This is purely a personal reaction, though - other people seem to like big MTM's just fine. I don't.

I found even a little bitty one like the Ariel quite a handful to straighten out subjectively. It's the sound of the array itself - the MTM coloration is still there with the tweeter shut off, and from what I hear at shows, is much worse with bass drivers much further apart.

Big MTM's remind me of line arrays. They just don't sound real to me - and I've heard a lot of them, dipoles, in cabinets, etc. What I hear is a lack of cohesion, a sort of "hifi-ish" disjointed sound with many drivers seeming to be going in different directions at once - the Dahlquist DQ10 being the limit-case example of this. A violin or a singer never sounds like it is coming from several places at once, but that's exactly what I hear when I listen to a large-format MTM or large-area line source.

This sensitivity is probably the result of working for several years on quadraphonic systems with phase-shift matrices, and becoming keenly aware of the vague, swimmy sound of excessive phase angles between pairs of speakers. I found in practice that phase angles between speaker pairs should never exceed 90 degrees, and for sharp, cohesive images, phase-angle difference should be kept within 5 to 10 degrees.

This was with pairs of speakers several feet apart. When I started working on speaker crossovers several years later, I found the phase-difference sensitivity applied to drivers only a few inches apart. That was a result I didn't expect.

The only real exception is at the very highest frequencies, say above 5 kHz, where tight inter-driver phase control seemed to be less audible, and pretty much went to no audibility at all in the 7~8 kHz range.

What's always difficult to explain is that sensitivity to phase (in the absolute sense, referred to the original recording) and sensitivity to inter-driver phase differences, are completely different things, and sound completely different as well.

I am only moderately sensitive to absolute-phase variations, and hear them as a shift in timbre, and sometimes as a sense of "real" vs "unreal" in terms of a non-specific wideband coloration. In contrast, I hear inter-driver phase differences as an immediately obvious sense of incoherence that breaks an instrument into several parts - a grossly unnatural and obvious coloration that never occurs in real life.

This "incoherent" quality is most evident on pink-noise, very audible on symphonic and choral music, but can be rather hard to hear on audiophile-style recordings with sparse instrumentation and a "dry" acoustic - which is why I never use audiophile recordings for serious assessment. My primary tools for assessment are measurement, and large-scale recordings with plenty of ambient content. I'm not listening for this-or-that tonal coloration, which are mostly a matter of balance, but whether entire aspects of the performance are grossly altered or absent entirely.

When measurements and subjective impression diverge, it can happen two ways:

A) Good measurements (flat, good impulse response, low distortion, etc.) but the driver sounds unmusical, with a coarse, grainy, mechanical sound, colorations that do not occur in acoustical music. This is rare, but when it happens, the driver probably has an assembly fault - a decentered, nonconcentric voice coil, or a defect in the spider assembly. Since voice coils are always a little decentered - which puts more energy into one side of the cone compared to the other - this is a matter of degree, and checking whether the other driver has the same fault. It is not unusual for the jig at the factory to be miscalibrated, so all the drivers coming off the line have the same fault. I suspect the defective-jig problem is common with SLR lenses, since I seen multiple copies with the left side sharp and the right side fuzzy, or vice-versa with a different brand of lens. The mechanical construction of the spider can also create some very odd-sounding low-level colorations - this is where I most wary of professional SR loudspeakers, since I have no idea how they will sound at power levels much lower than they are designed for.

B) Bad measurements (peaks in the response, energy storage in the time domain, distortion at certain frequencies) but the driver sounds (very) good. I've had this happen several times, particularly with audiophile "boutique" drivers. I give the driver credit for good sound, but it remains under suspicion. At the minimum, the crossover design will be tedious, and removing the known colorations a long, and probably unsuccessful, task. A moderate coloration that is acceptable for a few weeks will not usually be acceptable over several years, which is why I remain suspicious of drivers that sound good and measure bad.

I kind of envy the idealism of the approach of designing a speaker from the Platonic Ideals of XYZ dispersion pattern, textbook-perfect crossovers, near-perfect square-waves on the oscilloscope, the lowest IM distortion in the industry, or any other pursuit of the Ideal.

Instead of working down from an ideal, I'm trying to find a few drivers that are efficient, have plenty of headroom, and have colorations that are acceptably low and are musically consonant. Colorations as low as possible would be nice, but the subjective quality of the residue that remains is not a minor concern, since most loudspeaker colorations are neither pleasant nor consonant with musical values.

That's the reason for the modular design. I have no idea which set of drivers will measure and sound the best - and it seems quite likely there will be no "winner", but a set of subjective preferences.

P.S. If people could publish the key dimensional measurements of 12" and 15" drivers I'd be appreciative - I'm looking for outer diameter, diameter of baffle cutout, diameter of bolt circle (and number of bolts), and depth of mounting flange. In particular, I'm looking for the Altec/GPA 414-16A and the Lambda TD-12M and TD-15M.

[nickmckinney]

Originally posted by Magnetar



Do remember the SRO 12 with the huge alnico magnet? The guitarists swear by them -

I had a pair reconed with a doped flexair edge and a vented dustcap with a wad of cotton stuck behind the cap.

If you can build a driver that sounds like these you should be avle to sell a buncjh of them to hifi people - use the paper edge for guitarists and you'll sell lots more - maybe have a hemp cone option

Yeah I rebuilt a few of those, don't remember the details but I remember that tank on the back.

We have talked about making guitar speakers but that is such a subjective area I don't know if we will do it. There are so many different ways to make the cone distort and its all subjective what each person likes. We are looking at paper surround versions of the 755 8" driver for those that want that.

[Lynn Olson]

Continuing on the theme posted earlier about the subjective perception of coherence, that probably begs the question in the reader's mind of whether the new system will have the same problem. Good question!

I break the problem down into two parts: the 700~850 Hz crossover and the 6~7 kHz crossover. Despite the large spacing, all three of the drivers at the 700~850 Hz crossover are within one wavelength of each other, and will have an in-phase crossover where inter-driver phase differences are less than 10 degrees - this is something I do as a matter of routine in crossover design. When inter-driver phase angles are all under 10 degrees, the drivers subjectively merge into a common source.

The HF crossover is another matter. The drivers are separated by 6 wavelengths - a lot. Not only that, the dispersion pattern shifts from conical with a soft edge (Tractrix & Le Cleac'h pattern) to very wide horizontal and narrow vertical. This would be a disaster at a lower frequency, say to 2 to 3 kHz, which falls right in the region of peak sensitivity of the ear.

However, in my previous experience at Audionics, supertweeter crossovers have much lower audibility than expected. The one thing to avoid is a "shelf" transition, where the supertweeter is 1 or 2 dB higher in level than the midrange driver. The rising edge of the shelf subjectively sounds like at peak centered around that frequency, and draws attention to itself. On the other hand, having the "shelf" slightly depressed in level compared to the mid-frequency driver leads to it not being audible at all. So the design tolerance is all in one direction - never raise the supertweeter higher in level than the mid driver!

This is also a very good reason to use a supertweeter with flat and level response, and avoid any type of peaking or response irregularity near the crossover region. This is a common problem with horn supertweeters, which can rather rough in the 5~7 kHz region - this makes setting the level on the supertweeter crossover much more difficult.

The rule-of-thumb is to choose a supertweeter crossover high enough that switching the supertweeter on and off is only slightly audible, select a supertweeter with very flat response close to the crossover frequency, and be careful to never set the level higher than the mid driver. The last rule-of-thumb is commonly violated in many audiophile speakers in the desire to give the speaker more aggressive and obvious HF - more "zing", more "detail", more sparkle, particularly to draw attention away from a peaky midrange. Reviewers all seem to love speakers with exaggerated HF, and the industry responds with sizzly-sounding speakers. Boom-and-tizz have been big sellers since the Fifties, so I don't expect it to change any time soon.

Will the driver array sound coherent? I can't predict that in advance, since so much depends on time-alignment and the exact inter-driver phase relations (for the 700~850 Hz crossover). I'm pretty sure an off-the-shelf 6, 12, 18, or 24 dB/octave active crossover would give poor results, since the chances of the inter-driver phase angles falling within a 10-degree window would be small. In practice, different parts of the slopes (close to the crossover and further away) have to be adjusted so the drivers retain their phase relationship within a one-octave window, not just at the exact crossover frequency.

[Variac]

Magnatar: R-H also have :

CCH1500-6 HORN,CMPLX CONIC,1500Hz,60x40,1"
CCH1500-9 HORN,CMPLX CONIC,1500Hz,90x40,1"

1500 Hz horns . I guess not as expansive vertically..

[FlorianO]

Magnetar, all,

Originally posted by Magnetar
It depends. They do load lower and are usable to around 1200 HZ vs 2000 with the XT120.

They also have better vertical radiation. My panels center point is 36 inches and I find they are more livable for me, sitting down, standing up or laying on the couch..

As far as coloration the XT120 is pretty darn low IMO, these are better. They do not sound like a dome, ribbon, AMT cone or a horn. The little RTR's electrostatic tweeters are better (I'm using a GREAT driver the Emilar EA175) on the top treble where the shimmer and silk is. The Emilar is only good out to 15K so it might be the very top I'm missing.

So for the money the XT120 is the best deal if you don't need to go as low or don't care too much about the vertical pattern. Both are excellent horizontally. IMO compared to other horns or'waveguides' the RH is still an outstanding bargain for 200 a pair more - it does SOUND better.

I am thinking Renkus Heinz does have a better horn in it's stable. I'm not sure which one - maybe the 90 by 40 large format model that loads to 500 cycles..

Any positive experiences/can recommend a low-coloration WG that is usable down to 1k ? Any experiences with the larger XT1464 and its more limited directivity (too much so for home use ?) ?

[Magnetar]

Originally posted by Variac
Magnatar: R-H also have :

CCH1500-6 HORN,CMPLX CONIC,1500Hz,60x40,1"
CCH1500-9 HORN,CMPLX CONIC,1500Hz,90x40,1"

1500 Hz horns . I guess not as expansive vertically..

Those should work well with an eight or ten crossed over higher. Much like the XT120

Originally posted by FlorianO
Magnetar, all,



Any positive experiences/can recommend a low-coloration WG that is usable down to 1k ? Any experiences with the larger XT1464 and its more limited directivity (too much so for home use ?) ?

No, I use horns that low. The tractrix is low coloration if you don't mind head in the vise listening position or use progressively smaller horns as you go up in frequency. The XT1463 seems to be too narrow of radiation for my listening habits/preference.

[fred76]

Originally posted by Lynn Olson

This is also a very good reason to use a supertweeter with flat and level response, and avoid any type of peaking or response irregularity near the crossover region. This is a common problem with horn supertweeters, which can rather rough in the 5~7 kHz region - this makes setting the level on the supertweeter crossover much more difficult.

Hi All,

Speaking of horn supertweeters. Are there any 16 ohm horn tweeters in current manufacture usable for hifi (to be used ~7kHz or more)? I mean exluding the exotic$ from Ale and Goto or plasma etc. Thank you.

I was looking at B&C DE35 but they're only available at 8 ohms. The reason I ask is that I have 16 ohm drivers for a 2way and would want to supplement the highs of the main compression driver (which rolls off too early on axis).

Lynn, I just saw an exampe of the new GPA Alnico series you mentioned, now online:

LF woofer 515C

[badman]

Originally posted by Lynn Olson



However, in my previous experience at Audionics, supertweeter crossovers have much lower audibility than expected. The one thing to avoid is a "shelf" transition, where the supertweeter is 1 or 2 dB higher in level than the midrange driver. The rising edge of the shelf subjectively sounds like at peak centered around that frequency, and draws attention to itself. On the other hand, having the "shelf" slightly depressed in level compared to the mid-frequency driver leads to it not being audible at all. So the design tolerance is all in one direction - never raise the supertweeter higher in level than the mid driver!


The rule-of-thumb is to choose a supertweeter crossover high enough that switching the supertweeter on and off is only slightly audible, select a supertweeter with very flat response close to the crossover frequency, and be careful to never set the level higher than the mid driver. The last rule-of-thumb is commonly violated in many audiophile speakers in the desire to give the speaker more aggressive and obvious HF - more "zing", more "detail", more sparkle, particularly to draw attention away from a peaky midrange. Reviewers all seem to love speakers with exaggerated HF, and the industry responds with sizzly-sounding speakers. Boom-and-tizz have been big sellers since the Fifties, so I don't expect it to change any time soon.

Hi Lynn:

I agree with your comments on MTMs and line arrays. I prefer simply upscaled driver arrays when you need more output. The idea of pattern control of MTM or WMTMW on a dipole is rather compelling for a few reasons, but there's no reason to believe that they'll not have the colorations that bother you. The array would in all likelihood disperse the dipole peak and dip, and potentially even help extend bass response, but at what a cost.

I also agree with you on the tuning of supertweeters. Your comments around supertweeters also apply to subwoofers (both of which I almost always use). A lot of audiophiles grouse about subs, but there are very few rigs in which subs are set up properly, as SUBwoofers, meaning below a REAL woofer. When you run it in at 80, 100, 120Hz to cover the area that your midwoofs are already strained in, no surprise it sounds bad. And most are cranked up too loud besides. Get your crossover down below 40-50, and things get much much better, that extra 12dB of suppression of higher frequencies goes a heck of a long way towards eliminating the 'voices from the box' issues. My current sub has fooled me that it's off a few times, until I'm sitting there looking at it and see the sudden cone excursions that I'd hate the mains to be doing.

Anyway, glad to see that you're progressing towards a 'real' build. I'm eager to see your results, the lambda drivers in particular are so sexy, I am excited to see your testing/impressions (as well as Zaph's).

Best of luck with this interesting continued project, and I hope your leg continues to improve.

[Lynn Olson]

Just got a call from GPA today, they will be shipping the pair of 288-16H's and pair of 414-16A's next week. The 288-16H's that I ordered were part of the first 288 Alnico production run at Great Plains Audio.

Interesting to think it's been a third of century since the combination of the 288 aluminum tangential-surround diaphragm, Tangerine/Hendricksen phase plug, and Alnico motor system has been assembled as a 288-16H compression driver. The 414-16A's probably go back even further.

Considering the pent-up demand for Alnico drivers, I hope the good folks at Lambda keep Alnico magnets on their wish-list - I'm not the only Alnico nut out there, from what I can tell. Yes, they sound different - and in a way I like (DHT triode vs pentode kind of difference).

[nickmckinney]

Originally posted by Lynn Olson
Considering the pent-up demand for Alnico drivers, I hope the good folks at Lambda keep Alnico magnets on their wish-list - I'm not the only Alnico nut out there, from what I can tell. Yes, they sound different - and in a way I like (DHT triode vs pentode kind of difference).

Not only Alnico but also some field coil versions (you supply the power supply)