AVS Forum banner

Hey guys...we need a little rallying here...

1M views 13K replies 544 participants last post by  dane10 
#1 ·

DOPE from HOPE Toe In.pdf 159k .pdf file

 

 

 

Setup of WG Speakers.pdf 206k .pdf file

 

 

 

Introducing...

logo design contributed by avs member omegaslast. thanks omega!


This thread has morphed from a rallying effort to encourage Parts Express to carry a waveguide into a full blown research and development project coordinated by Erich H to create a new one. On top of that, Erich H and contributors developed a couple new compression drivers, flat pack enclosures, and so much more. While I will provide some updates from time to time, the latest can always be seen by subscribing to this thread or visiting Erich's website:
http://www.diysoundgroup.com/

and the Official Project List in the forum.


The primary product that has emerged is the SEOS waveguide (Super Elliptical Oblate Spheroid describes the profile of the waveguide's curvature). The design objective was to provide good horizontal coverage for near constant sound anywhere in the room (controlled directivity for the audio wonks) and vertical coverage that would minimize floor and ceiling bounce, while minimizing the problems associated with traditional 'horns' (internal reflections and such).

 

If you don't know why controlled directivity waveguides are useful, don't worry, attached is a pdf that explains it. 

The design objective has been met. This waveguide provides the high dynamics and SPL capability of a compression driver without any of the "horn sound" that plagues other models.




Off axis performance is spectacular:



Several more pics here:
http://www.diysoundgroup.com/media/c...astic3_1_3.jpg
http://www.diysoundgroup.com/media/c...astic2_1_3.jpg
http://www.diysoundgroup.com/media/c...astic4_1_3.jpg


It is available in several options. The first and the premium option is high gloss fiberglass. The second is a poured concrete type product that costs about half as much. The third is the budget option, an injected molded plastic horn. All three will perform similarly.


The waveguide is available in 6" 8" 10" 12" 15" and 18", though not all models are available in all materials. Throat diameter is 1".


Here is some eye candy of the fiberglass models:








The fiberglass waveguides are available in most any custom color.



12" SEOS measured. DE250 driver, 2m distance, about 5' off the ground.

Horizontal patterns (measured in 7.5 degree steps, 0 to 90 degree range, curves unsmoothed):
http://www.avsforum.com/avs-vb/showthread.php?t=1369780




 

Member "java" is an early adopter and did a great job with his build.

 



Ground up build thread can be found here:

http://www.avsforum.com/t/1417294/seos12-2512-build

Member Brad Horstkotte's interpretation, SEOS 12/360 over a JBL 2226J woofer:



 

Build thread here: http://www.avsforum.com/t/1422024/seos-12-dna-360-jbl-2226-j-build

 

Erich has some flat packs coming together.  Here is an example of how simple they are to assemble:

 



More bwaslo progress...SEOS12 and Deltalite 2512 in a test box. Crossover was designed using measurements from the "lesser clone" driver. Here is an overlay of the 1m unsmoothed frequency response with that design using both the "lesser clone" and the DE250:
http://www.avsforum.com/avs-vb/showt...0#post21851410



The clone (and to a lesser extent, the "better clone") has a funny notch at 8.5kHz, but goes higher in frequency; DE250 doesn't have that notch. (Not visible in this design): The "lesser clone" doesn't go quite as low as the DE250 or the "better clone", but still very close.


Here are 90ish dB distortion sweeps. First the DE250:



Then the "Lesser Clone":


Not much to get excited about in either, both pretty clean. BTW, as a sweep, this is quite loud.




And another pic of the waveguide (plastic prototypes on the left and right, fiberglass in the center). The plastic SEOS 12" is being set up for injection molding, so the per-unit cost will be VERY affordable.






AVS member bwaslo has been running AE TD15M/SEOS 15/B&C DE250 and has amazing results in an active crossover setup. As of this update, he is working on a passive crossover network.




A 'prototype' build of his can be seen here: http://libinst.com/junk/TD15M%20SEOS15/photo2.jpg




AVS member AudioJosh has some pics of a build here:
http://www.avsforum.com/avs-vb/showt...8#post21711808




Here is a corner friendly concept: http://www.audiokinesis.com/images/p..._ak_prisma.jpg




As for one of the compression drivers in development, it is called the BA because it is so large, aka big ass, relative to the DE250.




Frequency response of the BA driver:
http://www.avsforum.com/avs-vb/showt...5#post21670315






There is also a DE250 clone in the works. Frequency response:

Black is DE250 on 15" SEOS, Red is clone driver on 15" SEOS.






Frequency response:






Also, a very low cost compression driver is in the works. Some early pics:






BWaslo's results of a first SEOS12 design. As built from a PCD design (using OmniMic measurements). The woofer is the (very soon to be announced) Dayton Audio "Designers' Series" 10 inch woofer. The cabinet is an old Advent Legacy II box.

http://www.avsforum.com/avs-vb/showp...postcount=3110






Malcolm, A SEOS12 Center channel (another one by bwaslo):
http://www.avsforum.com/avs-vb/showt...1#post21836431
The schematic for the Malcolm (v1) crossover
Malcolm Crossover Parts List from Parts Express








Full baffle options are coming soon. Prototype shown here:








Erich update:


Besides the 1 piece SEOS-10 baffle, most of the designs right now will revolve around the SEOS-12 and SEOS-15. But they're almost certainly going to be using the DE250 or other similar models. Even if someone comes up with a super high end model, it will likely be with a much more expensive woofer, but still use either the SEOS-12 or 15 and the nicer compression driver that showed a slightly better top end than the DE250.


Keep in mind, that every single option will be good. No one here wants to mess around with anything cheesy, it's just a waste of time right now. So when I say "low, mid, high", I'm talking about pricing, not necessarily sound quality.


It's also going to be hard to break things down into categories because all will be available as DIY kits, but some might eventually turn into completed speakers. But if they do, the boxes I'm thinking about will not be cheap. So you could DIY even the high end stuff, but the really high end will be considered completed boxes, but still the same "design". Hope I explained my plans okay.


Just to get things out in the open:


SEOS-8: There will be at least 2 designs. No 1 piece baffle or complete speaker in the works yet. You have to order the fiberglass or the poured version right now.


SEOS-10: There will be at least 3 designs. Low, mid, high costs. All can use the poured or fiberglass waveguide that you can order right now. Some will also fit the 1 piece baffle. You could use the 1 piece baffle for the cheaper model, it would be up to the person building it. Eventually (if an absolutely great design shows up) I *might* try to get an entire speaker made up for the 10.


SEOS-12: I'm guessing at least 5 designs minimum. With all the different woofer and compression driver combos, it could be even more. Obviously the plastic waveguide, the poured one, or the fiberglass model. I think this and the SEOS-10 will be the main focus. Because of the plastic option, some designs will be very well priced that anyone should be able to afford. Probably a few mid level designs, and maybe 2 higher end designs. If one of those higher end designs is incredible, then maybe a full speaker down the road.


SEOS-15: I know of at least 2 designs. But they're probably won't be a cheaper version here. You can order the poured model or the fiberglass model. I don't think there will be a 1 piece baffle. I doubt a completed speaker like the other sizes would be available.....in the cabinet design I'd like to see. It would just be quite expensive.


SEOS-18: I know at least 2 designs here, possibly 3. Same info as the SEOS-15.





*********************

It can't really be put into words how much work and how much money Erich has invested in this project. He financed the whole thing and invested countless hours coordinating with people around the globe. So when you build a SEOS project, at least post your build and/or drop Erich a thank you note. Also, spread the DIYSoundGroup word around...


As with all projects, this one is not the work of one person.


SEOS is a trademark. The name and logo are copyright (c) 2011-2012. All rights reserved. For commercial use, contact Erich H.

This project is dedicated to the memory of our friend who went by the callsign Zilch.

More data, less wank.


Z's avatar is itself a dedication to the memory of James B. Lansing.






The original post:


As many of you know, Zilch & Co. have created some great crossovers for a horn loaded main that uses the QSC 152i horn. It appears that QSC has cut us off from direct buys of this horn, but Parts Express, as a QSC distributor, may be able to supply our fellow forum members if we can rally enough interest.


Folks, I am asking that you shoot a quick email to Parts Express and let them know how much we appreciate their support of our community and that we would appreciate it if they could continue to provide the QSC 152i horn for DIY builds.


Each horn sale from them typically results in an additional sale of a woofer, a compression driver, and many crossover components. QSC gets the benefit of proving their good products among a very vociferous group of folks (lots of free online marketing) P.E. has been a long time friend around here so don't go negative however this one turns out.


The email addresses are here:
http://www.parts-express.com/contact.cfm


Thanks.




 
 

Attachments

See less See more
1 31
#53 ·
I was really disappointed to hear the news about the QSC waveguides. I had planned to used them for a first DIY build, but this project sounds very interesting. Why 90x60 instead of round (vertical dispersion control)? And about 17" wide, what horn length/flare rate and what low frequency cut off are you thinking about?
 
#54 ·
There has been discussion about the 17" veritcal distance creating a larger CTC which creates a null in the vertical listening angle.


Of course as you go larger you can go much lower in the XO frequencies and you can use steeper sloped XOs to mitigate that null issue.


I always say pick your compromises
 
#57 ·
Couldn't using a MTM arrangement (like the photos above) move the vertical null further away from the center and make the on axis lobe larger?
 
#58 ·
Fiberglass work with polyester and polyurethane resins is NASTY. With epoxies, much nicer. You just have to pick the right resin system....it is still messy, but will not drive you out of the house/shop/garage with the fumes.


The accurate proto will the the butt-kicker. It will not be easy if home-made (need an ellipse generator router jig at the very least to do a laminated version that has to be smoothed BY HAND, then sealed and waxed in order to generate a female mold that will then be used to generate a male mold that will be used to make the horns), and custom fabricating of the proto will be $$$$$.


JSS
 
#59 ·

Quote:
Originally Posted by maxmercy /forum/post/19547894


Fiberglass work with polyester and polyurethane resins is NASTY. With epoxies, much nicer. You just have to pick the right resin system....it is still messy, but will not drive you out of the house/shop/garage with the fumes.


The accurate proto will the the butt-kicker. It will not be easy if home-made (need an ellipse generator router jig at the very least to do a laminated version that has to be smoothed BY HAND, then sealed and waxed in order to generate a female mold that will then be used to generate a male mold that will be used to make the horns), and custom fabricating of the proto will be $$$$$.


JSS


Definitely, I used HD stuff originally. Now Im spending more $$$ on stuff from online sources like Jamestown distributors. The latest Epoxy I have is ultra clear and orderless, much better!


As for molding using silicone stuff from a source like smooth-on is turning out to be awesome. I just finished using Rebound 25 to create my silicone rubber molding, that stuff is AWESOME. Creating the hardshell around it was not awesome with fibreglass/epoxy because it just does not stick to the silicone


Anyways, sorry about the OT stuff.


FWIW, I do not mind adding funds to a project to get a prototype. I may not have the skills necessary but I can always provide some $$$.



Quote:
Couldn't using a MTM arrangement (like the photos above) move the vertical null further away from the center and make the on axis lobe larger?

Good question, I can not answer that one.
 
#60 ·
I see two potential routes:


1. Homemade version from wood used as a plug for a mold. Frankly I don't see this as feasible with an ellipse, but I'm not a woodworker.


2. Build a 3D model and have a plug machined. This would be more expensive and would require a production run to make financial sense. I'd imagine using Smooth-On Urethane Plastic in silicone molds would be best. I have to believe this would need to be treated like a business albeit a small one.


I wouldn't mess with fiberglass cloth and resin. I'd rather do a urethane pour into a mold. I'm playing with Ruby scripts in Sketchup right now.


Are there any opinions on the ellipse shape? We could do a superellipse if that was deemed better. I haven't given the ellipse shape much thought.


Soho, any interest in modeling an elliptical OS in Akabak or Hornresp?


Edit: And the reason I mentioned the CP385ND is because it was in that list. I'd probably go with the ferrite version because I'm not building for portability and I doubt anyone else here is either.
 
#61 ·
I might know a guy that can create the prototype using a CNC milling machine, I'd have to ask though. If so, I can then create the duplicates. I'll try to call on Monday to see what type of file they need. The company he works for might actually be able to make the duplicates too, I'll find that out as well.


If anyone here can create a file giving exact measurements of what most DIY guys might want, that would help of course. I'm just beginning to learn about the waveguides, so as of now, I really don't know which ones have the most demand. My work season has finally slowed down, so I can devote a decent amount of time to getting this done if you guys want to.



There is one thing that could cause a problem. I'm sure Zilch would test the one that was made, but I think we would really have to make sure it was going to test pretty good. The prototype work wouldn't be super cheap and each change would require a new prototype. And I think some of the WG's that look like they'd do okay don't perform as well as people thought? At least I think that was the case.
 
#62 ·
Doing a wood plug prototype would take tons of time, as you would be cutting many ellipses in wood, gluing them all together and filling and sanding to shape.


Unfortunately, using a CNC mill, unless it is a super one with LOTS of tool time will leave tooling marks that will also have to be hand sanded to final shape. With the wood option, you can at least see all the wood 'corners' as you are sanding to use a guide that you have sanded far enough....


Either way, you are looking at having someone that has great skill to do the final sanding....


Maybe a combination of using a CNC router table and using it to cut 1/8" wood into many ellipses would be a good compromise? Fill the 'corners' with something like Bondo (silicon microballoons and a low-shrink resin system, take your pick) and sand to final shape......then the part-making can begin....use a good, viscous resin for the 'finish', then use pigmented resin and choppped strand mat and resin would be pretty cheap and effective. you can then layer reinforcing cloth and resin on that to strengthen the build, and then peel-n-seal....the interface for the compression driver would be then added...


JSS
 
#65 ·
I'd go with 1/8" or 1/4" for the areas that change less (near the throat), 1/8" or even 1/16" for the areas that change the most (near the mouth). A bolt-on driver would be easier than a screw-on....Just make the horn onto the flange you will use to bolt the driver onto.....


For cost savings, use hardboard instead of wood for the initial plug. Do it in an insulated and climate controlled shop to avoid thermal and/or hygroscopic expansion problems.


Make the female mold from the original plug w/ fiberglass, then use the fiberglass 'master' to cast some silicone (or fiberglass) 'slave' plugs you can send out to individual users. Do not keep using the plug, just use it to make masters as needed. Too much work is in the plug to be using it often.


I'm sure there are other ways to do this, but this is how I would go about it with limited tooling and equipment...


JSS
 
#66 ·
I was thinking it could be machined from a more durable material than wood (metal or plastic). It would likely cost more and it would require recouping that cost. Bottomline is that this won't be a $15 horn but if it is the only elliptical OS horn available to DIYers it could be worth something.


I didn't buy the QSC's simply because they are cheap. I would have probably paid $100/ea for them simply because there aren't many good options out there. The 18Sound horns aren't what I'm looking for. Geddes stuff is way to expensive and axisymmetric. DDS is hit and miss plus it is expensive. QSC was doing us a favor really and it probably doesn't make business sense for them to sell to DIYers. You definitely can't count on something like that. Now Zilch has made a bunch of nice designs for a very limited number of people.


I know some people did buy the QSC simply due to its price. What percent would have bought one if it were $100? How about $100 and significantly improved?
 
#67 ·

Quote:
Originally Posted by coctostan /forum/post/19548223


Are there any opinions on the ellipse shape? We could do a superellipse if that was deemed better. I haven't given the ellipse shape much thought.

I have no clue what a "superellipse" might be.


Look back to Earl's early work in OS, and you'll find the first experiments and publications used an elliptical design. The measurements evidence the on-axis "notch of perfection" in those, as well. That's the primary (stated, at least) reason he is considering alternatives to his current axisymmetric waveguides.


He claims that the (trade secret) knowledge he has accrued during the intervening years MAY allow him to come up with an improved elliptical design. He has not reported any progress with his prototype toward achieving these objectives of late, however....

Edit: http://en.wikipedia.org/wiki/Superellipse


Does not address the symmetry/center-to-center distance issue, apparently.
 
#68 ·
I agree about the goodness of the DE250.

Quote:
Originally Posted by catapult /forum/post/19542391


About matching the CD exit angle to the horn, I don't think it's a big deal. You can build the horn for a 0 degree exit angle and a few strokes with a piece of sandpaper wrapped around your finger will smooth the transition well enough for guvment work. Even Earl uses modeling clay at the transition to smooth things out.

I thought the clay was just for filling in any gap, not for determining geometry.


Quote:
Originally Posted by LTD02 /forum/post/19545663


what is the advantage of the CP385ND over the CP380M?

Presuming ND means a neodymium magnet, less weight.


Doesn't the size need to be decided on?


My vote is for a WG matching to a 12" driver with XO close to 1 kHz.
 
#69 ·
"Presuming ND means a neodymium magnet, less weight."


yeah, the nd version is about 1/2 the weight ~7lbs vs ~3.5lbs. for most around here, that is not an advantage. i was more focused on the distortion numbers where the 380m seems to hold a big advantage with respect to almost every drive that i have seen in its category. b&c doesn't provide distortion plots, so draw your own conclusions there...
 
#70 ·
Zilch, do you have a take on why the Geddes OS horns have a null on-axis? Is it a function of the ellipse (in Geddes case a "circular ellipse") or the large CTC spacing or both? The QSC horn is essentially a very large N superellipse or "squircle". It doesn't exhibit this null to the best of my knowledge.


At the very least, an elliptical OS WG should improve on the CTC spacing from Geddes round designs. It could give us a nice low horizontal pattern control without intruding on the vertical with a narrow null.


These are the specs I'd shoot for with the EOS WG:
  • 90x50 pattern
  • Good directivity down to 800-1000hz (~17" wide before roundover)
  • CTC spacing of ~10.5" for a 12" driver (~8" tall mouth with a notched roundover)
  • Match a 15* driver exit angle to work well with the DE250 and CP380
  • Work well with a 1000hz crossover point


I guess I could live with the "Geddes null"/"notch of perfection" if those specs are met at a reasonable price.


Could it be that JBL's non-symmetric horns are addressing Geddes's null issue? It would be difficult to build an odd mouth shape unless we could get a mathematical representation of the specific mouth shape like we can with a circle, ellipse or rectangle.
 
#71 ·

Quote:
Originally Posted by coctostan /forum/post/19550186


Zilch, do you have a take on why the Geddes OS horns have a null on-axis?

Yes, with any symmetrical shape, the diffraction occurring at the mouth exit, whether it's from a sharp edge or a roundover, sums and interferes on axis as a function of wavelength. Perfectly round shapes are the worst in this, and the on-axis notch in Earl's designs is greater than 6 dB deep, and owing to the roundover, presumably, fairly wide. By 7.5° off-axis, it's all but gone, but its presence mandates that his designs do not perform well on-axis, and he cautions not to listen or EQ them there. I suspect that's why JBL and others generally do not use axisymmetrics in their designs -- an axial response anomaly of this magnitude is a deal-breaker in SR applications.


I would advise against locking into any waveguide design so wide as 17" plus. Listen to Earl on this one, and optimize for a total width more on the scale of Abbey. QSC got it right, in my view, at 14" wide with roundover; 15" total width would be a practical maximum. If Brandon's preliminary measurements are correct, that'll get down to 1 kHz for mating with 15" woofers at 90°, and still not be overly wide for use with 12s. Note that Earl's designs are NOT 90° (-6 dB), rather, somewhat narrower. I'm with Noah on this -- 12" woofers are more easily workable, and probably ideal.


Yes, there are those who want to do much bigger, and they are vocal about that, but statistically, they are not "mainstream," and not going to be truly satisfied unless they can get down to 800 Hz, or an octave lower, even, with large-format drivers, added supertweeters, and yada, yada....
 
#72 ·

Quote:
Originally Posted by dwomd /forum/post/19547885


Couldn't using a MTM arrangement (like the photos above) move the vertical null further away from the center and make the on axis lobe larger?

Not in my experience. I tried it with the original Ewave waveguide and some woofers. The problem that comes up is that the total highpass cutoff slope on the waveguide (acoustical + electrical) is pretty steep, so below the point where the horn stops providing significant output you are left with two separate woofers spaced fairly wide apart and both covering around 1kHz, which gives them notches at smaller angles than you'd get if you just used a single woofer and horn which can get closer together. MTM can make your horn's axis the same as the system's axis, but seems to narrow the vertical null-free zone.
 
#73 ·
Quote:
Originally Posted by ZilchLab
Yes, with any symmetrical shape, the diffraction occurring at the mouth exit, whether it's from a sharp edge or a roundover, sums and interferes on axis as a function of wavelength....
I've noticed that same notch on some Tannoy coax's and never knew why. Thanks. I assume the less "round" the mouth is, the less this notch exists. For instance a square would have much less than a circle (or rectangle vs ellipse). In that case, a superellipse with a higher N would make the ellipse less "round" and possibly lessen the null.

Quote:
Originally Posted by ZilchLab
I would advise against locking into any waveguide design so wide as 17" plus. Listen to Earl on this one, and optimize for a total width more on the scale of Abbey. QSC got it right, in my view, at 14" wide with roundover; 15" total width would be a practical maximum. If Brandon's preliminary measurements are correct, that'll get down to 1 kHz for mating with 15" woofers at 90°, and still not be overly wide for use with 12s. Note that Earl's designs are NOT 90° (-6 dB), rather, somewhat narrower. I'm with Noah on this -- 12" woofers are more easily workable, and probably ideal.


Yes, there are those who want to do much bigger, and they are vocal about that, but statistically, they are not "mainstream," and not going to be truly satisfied unless they can get down to 800 Hz, or an octave lower, even, with large-format drivers, added supertweeters, and yada, yada....
I found a quote from Geddes on DIYA saying that his ideal was an 18" wide WG mated to a 12" woofer. ( http://www.diyaudio.com/forums/multi...on-thread.html ) He doesn't specify if that includes the roundover or if it is the OS portion only.


What is wrong with a wider WG? Is it extra horn depth making passive crossovers more difficult? Cabinet size?


I'd personally like to take advantage of the DE250/CP380's ability to play down to 1000hz. I'd also like the WG to hold pattern below my crossover point to some extent as well since I'm not going to use a brickwall filter. I guess I'd like something closer to the large format experience without the drawbacks in the higher frequencies. I'm not sure any of this stuff is "mainstream" but you are right it would make for pretty large enclosures. There is probably a compromise to be had.
 
#74 ·
Quote:
Originally Posted by coctostan
I've noticed that same notch on some Tannoy coax's and never knew why. Thanks. I assume the less "round" the mouth is, the less this notch exists. For instance a square would have much less than a circle (or rectangle vs ellipse). In that case, a superellipse with a higher N would make the ellipse less "round" and possibly lessen the null.



What is wrong with a wider WG? Is it extra horn depth making passive crossovers more difficult? Cabinet size?
No doubt there are some that would like the 18" WG, but I bet most people think it's just too big. The end design could be over 20" wide.


As for the less 'round' mouth, do you mean something like this:
 
#75 ·
Quote:
Originally Posted by noah katz
I thought the clay was just for filling in any gap, not for determining geometry.
Well, filling the gap and smoothing out manufacturing defects in the early models.


I think the point about the throat is that having a transition that feels smooth to your finger is more important than having it exactly follow some curve.


Random thoughts, grain of salt, blah, blah:


My QSC horns have an initial angle much steeper than any CD. And the hole isn't even round. I figure I'll need to add a wooden spacer and sand it and the horn to get a nice, smooth transition.


If the initial angle of the horn is too shallow, it's easy to fix with sanding. If it's too steep, you have to add material and then sand. See above.


Augerpro is hearing something he doesn't like with the Beyma and the QSC. He likes a budget B&C better. Throat mismatch?


Tom Danley always stresses that you can't just pick a driver and a horn. You have to pick a driver and a horn that work well together. He says the non-Neo BMS works as well as anything he's seen with conical horns -- which implies horns with a less-than-perfect throat transition.
 
#76 ·
the 4555 is one of z's favorites as well. on the jbl horn (PT-F95HF, Part# 338650-001, $98 ea), that driver's frequency response was really good (+/- 1.5db from about 1500-18000hz or so). in addition, each driver tested had among the most consistent driver-to-driver response that i've ever seen. i have no idea what the throat angle is, fwiw.


the plot used to be here:
http://audioheritage.org/vbulletin/s...313#post185313


looks like z. deleted it. z., where did it go?
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top