AJINFLA

Hi all (and specifically Jon)

Would love to hear your thoughts about utilizing tweeter waveguides. I believe in low XO high order designs such Jon's, but I see precious little on DIY speaker forums about this subject. I believe the benifits are obvious - large increases in low end output that (when corrected for flat response) reduces excursion/distortion, directivity that closer matches the woofer at XO, etc. But what about the drawbacks? Stored energy issues? It seems that just about every higher end Pro Audio monitor (Genelec,etc.) use a shallow waveguide to increase low end (tweeter) output and control directivty. Most use lower XO's and I would imagine, are thoroughly engineered - beyond the scope of most DIY or even home audio designs.
Your thoughts? (Mark K, I hope your reading - love your speaker pages).

AJ

**For additional info, see this thread http://www.htguide.com/forum/showthread.php4?t=18242

goskers

Unfortunately I don't have any good info for you but I would love to learn more about the useage of waveguides as well.

I hope we have some good info come about.

JonMarsh

This is a good idea and one I'd recommend, especially for higher output systems. Note, by this I mean the shallow waveguides as used by Amphion, SP Technology and others, NOT the wave guides employed by Cerwin Vega, Klipsch, etc.



Review of same:

http://www.enjoythemusic.com/magazi...ptechnology.htm

The drawback is the difficulty of design and manufacture for the DIY enthusiast. I've seen a nice Dipole design with a shallow tweeter waveguide in plexiglass. Frankly, it's hard to imagine CNC or fiberglass layout becoming part of the DIY repetoire of tools in the near future...

I believe I recall a gentleman offering design software for tweeter waveguides, but that could be in a dream or my over active imagination- couldn't find it with a quick Google search.

~Jon

__________________
the AudioWorx
_______________________________

Modula MT
M8ta
Modula Neo D CC
Ardent
Modula MT XE
Modula Xtreme


still coming in 2010....

Ardent Cxi (updated design, new woofers)
Ardent CCX (Compact Center Xtreme)

AJINFLA

Thanks Jon, the dipole design you referred to was mine! A rather crude implementation still under development. The results so far are positive. But I'd like to hear more discussion about the attributes. The complexity added is certainly beyond the realm of most DIY projects, but that sure didn't stop me.
Interesting that when I posted the SPTech link a while back on the Madisound board, the main commenentary was about how the woman in the pic looked superimposed, rather than about the technical details on the site!

Cheers,

AJ

Davey

That damn speaker is blocking part of the view.

Davey.

goskers

So, how do you go about determining the proper depth and width of throat for a certain tweeter?

I am thinking that a 'mold' could be made out of stretch material and fiberglass resin. They do this approach when making some auto kickpanels and boxes.

From the picture it looks as though the mouth is around 7" wide. How would this affect driver spacing for a proper MTM setup?

Thanks

Dennis H

Hey AJ,

Is that waveguide (aren't they really just shallow horns?) you are using a constant directivity profile - i.e. straight taper from the throat and then rounded as it gets out to the mouth? If so, about what angle is the inner straight part?

About construction, all you need to do a CD profile is a big chamfer bit with the proper angle. Cut different-size holes in each layer of the thick baffle and chamfer them before you glue the layers together. Mount the tweeter from the back side.

dwk

My personal audio oddysey has (for now?) brought me to a point that I believe that controlled/uniform dispersion is a key (if not THE key) design element for a speaker. Waveguides are pretty much the only way to get that, so I echo AJ in being slightly surprised that the idea doesn't get more attention.

I'm experimenting with some 60-degree conical midrange horns, along with small neo dome tweeters also in 60-degree conicals. My immediate plan is to move to Geddes profile OS waveguides for the tweeters. I'm using 60 rather than the more typical 90 as I'm in a very small room that requires corner placement if the speakers, and a 90-degree pattern is too wide. My plan for construction is low-tech: stack layers of MDF/ply with rough sized holes to create the basic structure, then 'fill' with bondo and use an accurate profile to sweep out the interior. (this is possibly the ultimate plan for the midhorn too, but this would be a much more difficult construction process)

As far as drawbacks to waveguides, I just don't see many. Stored energy and reflections from the mouth are concerns, although they typically occur at the low end of the range- a large enough waveguide should avoid these problems(as 'should' adopting some Geddes principles) . Phisycal size is larger shichbrings up lobing concerns, but with a lower xover point, it's probably a push.

BTW- I heart the SP Tech stuff at RMAF, and thought they were excellent. Despite a ~900Hz xover to a normal Vifa dome, it worked very well.

Dennis H

Here's an article on constant directivity horns I found a while back. Click the arrow at the bottom to go to the next page.

http://www.studio-systems.com/audio...Clifford/95.htm

Bullet points I took from it (and some other articles).

You'll need EQ - could be passive.

Throat diameter determines HF response. Smaller extends higher. With a regular dome tweeter, you need to balance that against causing too much compression and/or cavity resonance if the throat is smaller than the hole in the tweeter's faceplate.

Mouth diameter determines LF response. Bigger extends higher. Low cutoff is about where a wavelength equals the mouth circumference.

Angle determines the spread.

JonMarsh

That's what I thought, too. Anyway, I can see a slight shadow from the fill flash to the photographer's left, and the hair against the tree leaves in the background would be difficult to do digitall, so I think she's all there. Just my 0.02.

A lot more natural looking gal than the cheesecake usually used for these kinds of shots- more like the girl next door, IMO.

~Jon

__________________
the AudioWorx
_______________________________

Modula MT
M8ta
Modula Neo D CC
Ardent
Modula MT XE
Modula Xtreme


still coming in 2010....

Ardent Cxi (updated design, new woofers)
Ardent CCX (Compact Center Xtreme)

JonMarsh

Think how it would work with a "good" tweeter...

Of course, I shouldn't say that just because it's derived from a Vifa or made by them- so are my Hales Transcendence. But there's a lot of variation with what might be inside that puppy...

I agree whole heartedly about the controlled directivity thing, which is one of the advantages of dioples, especially in the bass.

Peerless even has a low cost tweeter with an attempt at this functionality, and I had some Audax "waveguides" of 6" construction back in the 80's which were intended to do the same things, but obviously wouldn't work at as low a freqeuncy as some of these newer efforts. It's not new, in a sense, but the size issues about implementing it properly make it unattractive for some kinds of designs and systems.

I think AJ is on an interesting track, because the larger panel desired for a dipole is a natural for that, and tweeter flare is one of the problems to be surmounted in doing a fully optimized dipole as regards controlling/limiting off axis response- making it look like it tracks the on axis, just lower in level.

~Jon

__________________
the AudioWorx
_______________________________

Modula MT
M8ta
Modula Neo D CC
Ardent
Modula MT XE
Modula Xtreme


still coming in 2010....

Ardent Cxi (updated design, new woofers)
Ardent CCX (Compact Center Xtreme)

JonMarsh

Here's a picture of their bigger ones....




There's another one with what looks like a sub/stand, like BESL does, called the "Revelations"

__________________
the AudioWorx
_______________________________

Modula MT
M8ta
Modula Neo D CC
Ardent
Modula MT XE
Modula Xtreme


still coming in 2010....

Ardent Cxi (updated design, new woofers)
Ardent CCX (Compact Center Xtreme)

JoshK

AJ,

Got any pictures of that dipole w/ waveguide? I'd love to see what it looks like.

I also echo gosker's question about how do you make the spacing close enough to avoid lobing?

Dennis H

I think AJ has said on another forum that the waveguide is giving him 8-10 dB more output at 1kHz so it lets you lower the XO frequency to reduce lobing.

An aside, referring back to Feyz's tweeter excursion calcs, I can see this being one of those situations where a Cauer would do better than an LR4. Once the frequency drops below the horn tuning, acoustical output drops pretty fast so you'd probably need one less electrical pole to get LR4 acoustical and excursion could still be a problem down low.

goskers

As for the Geddes principles, has anyone read his book? I believe the website is gedlee.com

It's 70$ so definitely not cheap but if there is some good useful info inside then I may take a look.

As a side note to Dennis's response; maybe a DEQX would come in handy here to keep things under control.

JonMarsh

Hah! That's the site I was trying to remember! With software also treating waveguides.

I've been curious about buying his "Speak" software, becuase it includes a copy of "Audio Transducers". But I don't have any knowledge of what he does beyond what's on the site. The price isn't that much, but you know, $75 here, $300 there, before you know it, we're talking the price of Praxis.

Would be curious to hear any feedback from anyone more familiar with his publications or software.

~Jon

Now, for those "civilians" out there, I should warn you that high end engineering texts tend to be quite expensive, because of the limited audience and the amount of work preparing the books (a little more proofing is required than just running the spell checker twice) (take a look at the book costs in the IEEE library).
His emphasis on directivity and polar response is interesting, and I suspect it would be $300 well spent for a serious designer. Note, this software does NOT implement a transfer function optimizer. Now, I understand that would disappoint a lot of folks, but I see some wisdom in this- there's a tendency to rely too much on axial SPL optimization.

__________________
the AudioWorx
_______________________________

Modula MT
M8ta
Modula Neo D CC
Ardent
Modula MT XE
Modula Xtreme


still coming in 2010....

Ardent Cxi (updated design, new woofers)
Ardent CCX (Compact Center Xtreme)

dwk

I have the 'Transducers' book, and the polite description would be 'dense'. It's basically a math/engineering text. I have an undergrad degree in physics, and did grad-level work in CS in computer vision which included some DSP, and I'd be lying if I claimed to be able to follow all of it. (although some of that is just being very stale - school was a while ago)

From what I've read, the Home Theater book is more practical.

No experience with the Speak software, but it looks tempting. I suspect it'll have a learning curve that makes Sound Easy look simple, though:-)


Well, yeah, but imagine how CHEAP my setup is :-) Seriously, at about $12 for the tweeter + horn, you can't go wrong. Once I get through the basics and get some decent measurements, then I'll spring for the good stuff. Likely the XT19 that AJ is using, but we'll see -maybe the RS28, but a 1" unit might not be as suitable.



Yeah - if I weren't literally backed into a corner with my speakers, I'd go dipole from where the midhorn tails off (~300/400) on down as it should bea good match. A 2nd order gradient would be even better, but that'll have to wait for a while.

Kingdaddy

Those are easy to make on a lathe, just turn it like a bowl.

Paul H

We've seen your lathe work kingdaddy - I would be reluctant to accept your definition of easy on a lathe - many people might find that level of easy very hard

Paul

goskers

dwk,

What insiightful tips could you give while trying to pursue a design for a tweeter with an xo around 1K?

Any general comments about design that you can recall for us?

Thanks

After doing some quick calculations it looks as though a target xo point of 1.5K would warrant a mouth of 8.5". That's fine but how do you figure throat length and what the heck do you do about the lobing issues that will be created by midwoofer seperation in an MTM arrangement?

Davey

How 'bout something like this for some experimentation? Cheap.

http://mcm.newark.com/NewarkWebComm...?SKU=54-580&N=4

Davey.

AJINFLA

Wow, the thread certainly took off while I was at work today. Here's my current state of development.



I'd be lying if I said I've done much to it lately, as I now live in a condo. The neighbors might frown mightily about speaker testing! Heck, my "demo's" of the HT for friends makes them think that all hell has broken loose next door.
The waveguide used is a CD design identical (I think) to the MCM model. It has a gradual curvature from throat to edge, and is 6.5" dia.
If you think about it, the faceplate of many tweeters form (albeit) a very shallow horn or waveguide themselves. I do recall the Peerless model that attempted to control the directivity of the driver, but not by much. Still better than nearly 180 deg dispersion crossing over to a beaming cone driver.
I tend to look towards developments in the Pro Audio world (real Electrical Engineering) much more than I do the home audio (high end), where Marketing Engineers tend to dominate. Innovations are largely absent, with the exceptions of a few. The SPTech is a prime example of solid engineering at work. Of course SL's design is also driven by such an approach, thus, was the guideline for my design. I have never e-mailed him to ask his thoughts about my approach, but maybe one day I will.
Speaking of HT, it's almost movie time

Cheers,

AJ

Dennis H

That would be the circumference so the diameter would be a bit under 3". Then you'd probably like to make it bigger because you'd like the horn cutoff to be below your crossover frequency. And bigger yet to allow for rounding over the corner to minimize diffraction effects.

Working the other direction, say you have a double-layer baffle, 1.5" thick. And say the hole in the tweeter's faceplate is 1.25". If you use a 45 degree chamfer bit in the router to make your horn, it will be 1.25 + 1.5 +1.5 = 4.25" diameter. Round the edges over and you're somewhere around 5" total. Add a third layer to the baffle and you're at about 6.5" total. Add a fourth, 8"..... etc.

goskers

Hey Dennis,

That method would certainly give you an even dispersion but what about a shallower depth, greater angle on the throat, flared outer rim and other variables that will control the polar response to a greater degree?

I do not know anything about waveguides so I am trying to learn from others about what changes affect what parameters.

Thanks

Dennis H

Greater angle will give you more dispersion. Obviously the limit is 180 degrees which is no horn at all.

dwk

Well, I'm far from an expert, but I'll throw some thoughts out there. Hopefully some discussion will result. Most of this is just my understanding from reading Geddes stuff.

The design goal of Geddes waveguides is to have smoothly tapering response off-axis, falling to -6dB at the 'design angle' of the waveguide. It should continue to taper off outside this angle, rather than abruptly stopping, as this will undoubtedly cause diffraction/ripples etc in the polar map.

The geometry of the waveguide actually seems to do little to taper the response *inside* the design angle. What the waveguide is doing is creating the smoothest possible transition from the throat geometry to the mouth geometry. In the Oblate Spheroidal formulation, the throat geometry is a circular plane wave, and the mouth geometry is radial (ie a cone). In theory, if you feed a *true* plane wave into an OS waveguide, you'll get a uniform response across the mouth.

Controlling the response inside the design angle thus means controlling the velocity profile at the throat, and then allowing the waveguide to transform this to the resultant profile at the mouth. Geddes uses an absorptive foam plug in the waveguide to effect this velocity tapering. My quick thought on this is that since planar tweets like the Neo3 are clamped at the edge, their velocity profile should fall towards the edge. Wuld they make a good candidate for waveguide loading?? (I think the design of the XT19 might have similar properties, but I'm not sure how the 'hole in the middle' will affect things)

The response outside the design angle *seems* to be determined by the mouth geometry/treatment- it acts as a relatvely normal aperature. In other words, make it as large as you can, and smoothly transition this to free space/baffle via a rounded radius.

Geddes doesn't talk much about acoustic gain or loading, since he assumes the use of compression drivers. In a home environment, these are so efficient by themselves that the gain of the waveguide is irrelevent. For normal tweeters, this probably isn't really true, as the gain and loading are what allow pushing the xover freq down. I *think* treating waveguides as conical flare horns is pretty accurate for these purposes. However, since full T/S specs are rarely available for tweeters, using Hornresp is tough. It is worth noting that there are two related but distinct effects of the loading. First is the increase in efficiency. The other is the increased air mass load on the driver - this increases the effective mass, and actually lowers Fs, and I believe this is just as important as the increased output at allowing a lower xover (in reality, the increased efficiency comes from the increased mass load, so it's the same thing from two different perspectives). Unfortunately, I haven't either run numbers or done measurements to see how much Fs changes. I'd guess maybe 30%.

What does this mean in practice? Well, probably not much. I think that if you start from a hole the size of your tweeter running straight out from the faceplate and smoothly transition that to a cone of your desired angle, and put an edge-treatment radius on the edge if the cone, you'll be getting 80+% of the theoretical benefit. In terms of driver selection, I really don't know - I suspect experimentation is going to be needed, unfortunately. I'd guess that the typical 1" metal dome will have too high a mass corner to be practical, but that's a true guess. In terms of Fs, you want as high as possible without being too high - I'd start with a guess that the resultant effective Fs is an octave below your target xover freq.

Gotta run to work. Hopefully this can generate some discussion

JonMarsh

Interesting inputs, your suppositions are in line with what I've been digging up so far...

I suspect a few experiments, including the one Davey dug up, are in order to generate some useful data.

Now, Geddes admits that depending on the driver design, this loading will require some EQ to compensate for the way it affects the driver response- for a direct radiator tweeter, I expect that means it will boost the output in certain areas along with controlling directivity- I suspect the SP Technology folks have done a lot of investigate into that, and appear to be using some type of additional wavguide at the throat. Curiouis and interesting.

Well, since, unlike Kingdaddy, I dont' have a lathe at home, (it would have to compete with my table saw and router table for the same limited space), maybe I'll order a few of those flares Davey spotted, and see what happens experimentally on a large baffle otherwise. Methinks they might be a little smaller than I'd desire for the range I like use, but might generate some useful info.

Unfortunately, AJ's link doesn't take us to his photos- I've seen a few elsewhere, but I'd like to see what he has to share.

~Jon

__________________
the AudioWorx
_______________________________

Modula MT
M8ta
Modula Neo D CC
Ardent
Modula MT XE
Modula Xtreme


still coming in 2010....

Ardent Cxi (updated design, new woofers)
Ardent CCX (Compact Center Xtreme)

goskers

dwk,

If I am understanding your post right it does not matter too much about throat angle or throat length. Is there a certain point at which one should start the smooth transition to the mouth?

I am assuming that the less severe and wider mouth the better?

Can someone describe how the absorbtive foam would be used and where for controlling velocity?

Thanks

AJINFLA

Damn that Yahoo photos, I think my settings keep reverting to private to save bandwidth. I certainly didn't change my pics from public viewing. Ok here we go, lets try this
http://pg.photos.yahoo.com/ph/ajinf...m/ph//my_photos


Hope thats better.

AJ

JonMarsh

Thanks, AJ. MUCH better.

Interesting pics, and nice construction job.

~Jon

__________________
the AudioWorx
_______________________________

Modula MT
M8ta
Modula Neo D CC
Ardent
Modula MT XE
Modula Xtreme


still coming in 2010....

Ardent Cxi (updated design, new woofers)
Ardent CCX (Compact Center Xtreme)

technimac

Karoliina, in Finland, has done a lot of DIY work on DIY waveguide construction. Her site is worth perusing for an "across the pond" look at waveguide development in Europe (mainly Finland) as well as her research into the best tweeter driver for waveguide use.

http://www.karoliinasalminen.com/fo...ewforum.php?f=4

Cheers, Bruce

__________________
"While we're at it" - the four most dangerous words in Home Improvement

JonMarsh

Thanks, Bruce. I came a cross her info a few months ago and downloaded it to the HD- thanks for bringing it up here. I came across her stuff through the speaker design workshop prototype page, which hasn't been updated since August 2004. It's hard to find the time for really involved projects like this...

~Jon

__________________
the AudioWorx
_______________________________

Modula MT
M8ta
Modula Neo D CC
Ardent
Modula MT XE
Modula Xtreme


still coming in 2010....

Ardent Cxi (updated design, new woofers)
Ardent CCX (Compact Center Xtreme)

AJINFLA

Actually Jon, those 6.5" dia. MCM's are probably quite close to what you may want for your designs. A 10-15db increase in output between 1k-1.4k would allow you to either cross even lower with your Cauer filter, or lower the stress on the tweeter, or both! The increase in output is quite linear as frequencies decrease and is easily corrected using DSP. I long ago stopped using passive filters, which you seem to favor. Yes DSP is still (relatively) in its infancy, but I can't see it not being the future. I'm hopping aboard now (the Behringer unit is unbeatable for $250 - and falling) with the hope that more powerful units will become available at reasonable cost in the near future. My waveguide caused a small emphasis in the dip in the XT's response at 12k when measured directly on axis, but disappeared quickly off axis. In fact, the off axis response of the entire system is very smooth, even at extreme angles. Does this matter? I think so. I'd be interested to see the effects when implemented in one of your designs. Using a 1" driver like the new Dayton RS or the Millennium may allow for an even smaller guide for a small increase in (low end) output and better direcivity - more along the lines of what you see on Genelecs, Mackies,etc. Unfortunately cannot help with the limited dispersion of a 1" dome at the very top end. Part of the reason I chose the XT19 over the Millennium, XT25,etc. Does this also matter? To my ears it does. I know few people can hear much above 15k or so, but I could definately hear a difference between the XT19 and big brother XT25, which I am convinced is (mainly) because of the wider dispersion at 15k upwards.
Gotta run.
Here's why - heading over to the Cape tommorow (again). My buddy works for NASA, so I'm cruising the base tomorrow again to see this bad boy (he sent me this pics today). It's getting ready to go again
All this speaker stuff really pales in comparison to the engineering that goes into this rig
http://us.f533.mail.yahoo.com/ym/Sh...0&view=a&head=b
http://us.f533.mail.yahoo.com/ym/Sh...0&view=a&head=b


AJ

dwk

Well, the angle is up to you. It should be chosen so that the angle matches the radiation pattern of whatever you are crossing over to. In most cases, 90 degrees is probably safe. Geddes for example crosses his 15" 90-degree waveguide over to a 15" woofer at about 900Hz, as this is where the patterns match.

As for the transition, it should be done quickly. If you don't, you get a tube in front of the tweeter, and this 'front chamber' rolls off the high end - not good. I'll have to dig up the Oblate Spheroid equation, but I think it's:

r = sqrt( a^2 + (x * tan(theta))^2)

where r is the radius at point x along the axis, a is the radius at the throat (ie x=0), and theta is the half-angle of the waveguide



This has never been well described. Either Dr Geddes has a patent pending so he's staying quiet, or it's simply a competitive differentiator. If you search the AA High-Eff board for 'Geddes and foam' you'll learn as much as I'm aware of. The short version is: low density polyurethane foam (30ppi), which isn't too common. Either thicker or more dense at the sides than in the center, then tweak until you get what you want.