Large midrange for OB??? Scott G ?

[mige0]

Hi

At US$230 from E-Speakers, SoundEasy V13.0, combined with the M-Audio 192/24 PCI card, is starting to look good as a MLSSA replacement. I've already got the M-Audio card mounted in my HTPC-format PC, now it's time to order the software and learn to use it.

For a good mic you might also have a closer look at the Earthworks page.

http://www.earthworksaudio.com/3.html
"http://www.earthworksaudio.com/3.html"

I own, use and love the M 30BX (battery powered) and the QTC 1 (around 45 kHz / –3dB) type.
Both are reliable and great performers not necessarily for measurement only.

Greetings
Michael

[Paul W]

You guys have been busy! I’m just catching up, so some random thoughts…

Regarding the baffle shape, I believe there are valid reasons for both symmetric and asymmetric baffles. With a symmetric baffle, diffraction effects are equal for both the left and right sides so the crossover may require more EQ for flat on-axis response. However, whatever is done in the (perhaps more challenging) crossover results in identical left and right off-axis response. With +/- diffraction impact, asymetric baffles can be less challenging for flat on-axis response (easier crossover EQ) but left and right off-axis response will be asymmetric. Does this have a significant impact on imaging? Dunno. John K and SL both seem to favor symmetric baffles, but other respected designers go asymmetric...for slightly different reasons, both seem valid choices.

A null at a tear drop edge would require symmetrical front-rear radiation…which isn’t likely considering basket interference etc. Even if driver radiation were symmetrical, a real-world driver’s acoustic center is not centered within the depth of the baffle (except for ESL etc). However, if the tear drop were reversed (with the edge near the driver) there would be minimal cavity resonance at the driver plus a large roundover at the baffle perimeter for both the front and rear wavefronts. If the distance and roundover are large enough, short wavelength wavefronts will dissipate with distance traveled along the baffle surface. Long wavelengths should cancel when they meet regardless of where they meet on the front-back path.

MJ…interesting observation regarding the feathers on an owl’s wing...they are scary quiet! An infinitely large roundover seems best, but also infinitely impractical…so we need alternatives! Are you suggesting kerf slices in line with the direction of wavefront travel (perpendicular to the sides of the baffle)?

MBK
More food for thought on compression drivers…for a passive implementation, we’re going to scrub off 10-15db of sensitivity to get flat response out of the larger format drivers. A supertweeter seems like an alternative but, in a recent post on Audio Circles, John K reminded me why I cringe every time I see a ST about a foot closer to the listener than the compression driver below it…without the proper delay, arrival time is off by multiple wavelengths! So, that is a rather loopy way of getting back to the BMS coaxial compression drivers. I hate the thought of doing two filters feeding choppy impedance curves, but coaxes would optimize both sensitivity and arrival time. I’ve only seen a few subjective comments on the BMS coaxes, but they were all positive…and I’m happy with the 4552 within it’s limits. Cost is far less than 18S, particularly considering a good ST. I like neo drivers, but I can’t see $100 per driver difference between the 4590 and the neo. Just food for thought…

Did you see the quote in one of the attachments in Lynn’s thread? “Before about 700us, the origin of a sound can be disturbed by a reflection. This error in location is a function of the relative level of the reflection. Sharp-edged rectangular speakers produce strong reflections in exactly this delay region and so are not capable of stereophonic fidelity.”

Out of curiosity, I wheeled one of the BMS 18N850s out to my concrete “launch pad” for LF CSD. Here it is at 100mm which seemed to give a fairly clean driver display. I wanted to do short range ground plane but didn’t have a safe way of holding the baffle at such a steep angle. Too expensive to stick a hole through the cone!

Anyhow, just FYI…

[MBK]

Hi Paul,

agree about the symmetric / asymmetric trade offs. That's something to try out I suppose, I usually prefer symmetric, question is, for a large horn / baffle, the feathering's function, besides mouth reflection dampening, would be to mimic room reflections outside the 700 us window - like a diffuser. BTW haven't seen the quote (yet) but I do remember the number from another source, one of the reasons why I made my baffles so wide (24"), to get at least close to that figure. To really reach it the baffle would have had to be 36" wide...

BMS coaxes - they got several points to go for them - they reach lower than usual 1.4" or 2" drivers (300 Hz!) and avoid the beaming issue, plus supertweeter conundrum. Though I have seen comments (Troels Gravesen?) on how in the 10k+ region time alignment is a) futile (wavelength too short, a few degree off axis listening and the alignment is off anyway) b) not necessary because we can't hear offsets in this region. Active x-o and amplification is not a problem for me, choppy impedance might however indicate other issues.

On the downside for the coaxes, BMS price is cheaper than 18S but not quite a bargain yet, it's more like +- 80 USD or so. Also, where does the money go - more parts (BMS coax) or optimized diaphragm etc (presumably 18S 1.4"). Price is usually not completely arbitrary, since we are outside of audiophile cloud cockoo land. And to actually use that fantastic 300 Hz lower limit would require an imperial sized horn.

So, what's the best compromise... really hard to tell and maybe subjective anyway. One would have to both buy and optimize two drivers before saying anything definitive. Somehow to me, so far, a 1.4" by gut feeling seems the best and simplest compromise. But. Thoughts in evolution...

I'm just in the process of gluing together a block for the first prototype horn, then we'll see how much this approach is worth to begin with...

18" CSD: looks smooth... still I think a lot of the 1ms+ data is backwave... hope I can glue some scrap panels together for a larger baffle and then I retry my woofers.

[Paul W]

My understanding is the erratic impedance curve for compression drivers is usually caused by the horn/WG. The fine print on the marketing sheets says the 18Sound impedance curve is free air, while BMS impedance is measured on a horn. It could be that a "perfect" WG would provide smooth impedance curves.

[MBK]

Sounds plausible.

BTW edit, my 24" baffle does indeed place the first edge reflection outside the 700 us window, the minimum baffle width for this is 18" not 36". Phew. I started wondering why I chose my dimensions as I did.

[ScottG]

Nice CSD Paul! Clearly the larger driver/high gauss/linear suspension.. maintains a more linear short-term decay - and is particularly impressive below 200 Hz.

Note that the driver exhibits the same discontinuities that MBK was concerned with.. this time however those discontinuities go further down in freq.. (..MBK - think about that a little further )

About the symmetrical "tear drop" shape:

Remember the dipole "null" is largely dependent on the path length and the directivity of the driver. Here modeling with the Edge is quite useful. A 16" baffle nets us a maximal null right around 100 Hz, and a -3db point just below 200 Hz (..driver dependent though). The compound (and largely) canceling turbulence at these freq.s will break-up diffraction effects from higher freq.s reflecting from the baffle near this edge. The net result is something near a "disappearing" edge that higher freq.s can't react to, while still maintaining a fairly shallow mounting depth for the driver.

Functionally the overall shape of the baffle wouldn't look dramatically different from the Sonus Faber Stradavari (..though of course the side edges would be significantly tapered by comparison).

About BMS Coaxials:

I don't think that the basic BMS design will give you quite that non-compressed ultra-fi sound when comparing the midrange driver portion of the BMS vs. the Radian 2" in the 800-6400 Hz range. Of course I could be wrong - so the usual caveats apply.

Additionally - the ONLY thing that the BMS design will net you in a low "compression" horn that is practical in size, is a better decay signature near its resonance, and an actually lower resonance. That in itself may or may not be significant - depending on the operating passband for the driver (..including the filter).

Now the coaxial super tweeter..

Again - in a low "compression" horn, the super tweeter will likely be unloaded for a good bit of its passband. Because of this, linear and non-linear distortion goes up significantly. While the sound will still retain that electrostatic-like "speed", the decay will however be compromised - and in fact sound more like an actual electrostat (..as opposed to a ribbon's sound). Also, directivity will still be an issue (i.e. it will "beam").

You can *mostly* get around these problems with a well designed more "compressed" format horn with a deeper throat. Unsurprisingly, such a design will also allow the midrange to extend lower in freq. for a given size.

Considering the above - the BMS coaxial units are better utilized with a horn similar to Bert's Orphean (..though perhaps with greater "round over" like the Azurahorns).

I think then the choice of a BMS coaxial vs. the Radian 2" while utilizing appropriate horn designs comes down to a matter of listener "taste". If you want greater directivity and a lower freq. response that will sound more "detailed" - then go with the BMS coaxial. If you want wider dispersion and a more "laid-back" and "natural" sound with still excellent detail then go with the Radian/large format horn/dipole driver. Of the two - I would caution the user/listener about the former BMS version.. some people (myself included), may like the sound initially, but ultimately be unhappy with it because of the increased "compression" and the resulting decreased directivity.

[Pano]

Originally posted by Paul W
Are you suggesting kerf slices in line with the direction of wavefront travel (perpendicular to the sides of the baffle)?

That's how it is done on the Iwata horns and tunnels.

Jack Bouska talks about towels around the edges of horns to ease the impedance transition. Similar idea. Should apply to the edges of a baffle, though perhaps with less effect than on a horn.

[ScottG]

Originally posted by ScottG
Here modeling with the Edge is quite useful. A 16" baffle nets us a maximal null right around 100 Hz, and a -3db point just below 200 Hz (..driver dependent though).

Note.. that was with about a 40 inch high baffle and driver placement relatively near the floor. (..*perhaps* similar to what would occur in such a design). Of course IF its centered on a 16" by 16" baffle - then the path length vs freq. null more closely matches what you would expect of the null considering the freq. (.i.e it moves up in freq.). But its unlikely that the driver would be used that way when considering the overall design (especially if used with a super tweeter).

Of course if you don't use a super tweeter (or if its coaxial), then you could always use a midrange "filler" above the horn/compression driver. There you could come close to maximizing your cancellation and at least "shoot" for something approaching constant directivity down lower in freq..

Anyway.. the "disappearing" edge concept still applies.. It does however become increasingly moot when you have less and less baffle to work with.

[Paul W]

Hey Scott,

The compound (and largely) canceling turbulence at these freq.s will break-up diffraction effects from higher freq.s reflecting from the baffle near this edge.

Assuming turbulence exists in some regions, the bass would have to be in incredibly fortuitous sync with a singers voice to cancel HF diffraction signals resulting from her song. I agree the baffle could look very, very good, but don't think it will function as envisioned...perhaps in isolated instances, but not across the musical spectrum. But, I've been wrong before and will be a zillion times again

I don't think that the basic BMS design will give you quite that non-compressed ultra-fi sound when comparing the midrange driver portion of the BMS vs. the Radian 2" in the 800-6400 Hz range. Of course I could be wrong - so the usual caveats apply.

Radian is not on radar without reasonable data. It is very unusual that an Engineering department doesn't have basic test data on file. Heck, SEAS publishes much more useful information for $30 tweeters! For now, it's between 18S and BMS.

Additionally - the ONLY thing that the BMS design will net you in a low "compression" horn that is practical in size, is a better decay signature near its resonance,

The BMS also allows placing the ST acoustic center near the correct distance from the listener. ST combined with a 2" exit will either be wavelengths too close to the listener, or, fire toward the back of a larger mid/tweet horn.

Again - in a low "compression" horn, the super tweeter will likely be unloaded for a good bit of its passband.

This agrees with the BMS distortion plots. However, the only 2" exit distortion plots I have also show higher relative distortion at higher frequencies...apparently no difference.

If you want wider dispersion ... then go with the Radian/large format horn/dipole driver.

Got me there...why will a 2" exit have wider dispersion than a 2"/1" coax?

panomaniac,
Do you have a link to the kerfed Iwata horns? A quick Google didn't turn up any kerfs.

[Lynn Olson]

Latest CSD is looking good, you can actually see the driver taking off at 1 kHz. 18Sound as usual comes through. Looking forward to auditioning these drivers, they look like serious players, and a welcome departure from the low-efficiency world of Vifa, Seas, and Scan-Speak.

Regarding Edge, which I've playing with the past week, it doesn't show the floor reflection. This is a big deal when the baffle extends down to the floor, as my design will. So far, I've simulated the floor by simply creating a mirror-image of the baffle, complete with drivers, underneath the "real" baffle. Since the floor is close to 100% reflection, this seems like a valid approximation.

Where it gets a little weirder is when the baffle slopes back a few degrees, something I'm thinking of doing so the bass and widerange drivers have the same arrival times for the listener. Modeling in this in Edge can be done - to the first approximation - by moving the simulated "mike" downwards (to the center of the baffle), but the images in the mirrored floor reflection are altered. The mirror-image drivers move further downward to simulate the increased distance from the listener, but the mirror baffle doesn't. Weird.

Interesting comments about the BMS coax. For a moment I thought you were talking about the BMS coaxial cone/horns, which have pretty scary-looking curves. The BMS coaxial CD, by contrast, looks fairly easy to use, and based on the published FR curves, it doesn't even look necessary to lowpass-filter the midrange CD. A carefully shaped 3rd or 4th-order HP for the supertweeter looks like it would slide right into the intended FR curve.

One thing that's sure to be different between the 18Sound, BMS, and Radian is the choice of diaphragm material. Measurements don't really disclose just how different these sound - and a horn, at best, will simply magnify the characteristic sound of the material. The BMS is a flexible plastic ring radiator, the 18Sound either plastic or Titanium, and the Radian is aluminum with a plastic surround. All different sounding, no matter what the curves say.

(I did have Mike at Radian e-mail me some data from their in-house comparative tests, and in the critical 10-20 kHz region, the Radian 850-PB is quite a bit smoother than the JBL Professional 2" driver, and comparable, just a little different, than the 2" TAD beryllium driver. Distortion measurements have a similar ranking. No, I don't know why they don't publish more comprehensive technical information on their website.)

The one good thing about supertweeters is that deep notches in the FR at 8 kHz or above are not readily audible, unlike around 3 kHz, where they are easily heard a "phasey" artifacts. It's annoying to measure ST crossovers because the mid and ST are typically several wavelengths apart, so moving the mike makes all sorts of ripples in the space of a few inches. But you'll find with pink-noise drive, the nulls - that are so visible on a spectrum analyzer - are surprisingly difficult to hear, even if you're moving around and listening directly for them.

Part of this is because this HF region is the domain of the outer ear, the pinna, and recordings don't have any height information on the disc - except by accident, as a result of spaced-microphone random phase relations.