New Wave Guide Study

[AJINFLA]

The upper passband of the mid isn't hard to do well. I'm using the 12m in an LR2 passband easily. There are a number of drivers that can do both ends well. Hard cones need not apply, but SS or Seas 5" non-metal drivers can easily handle LR2 at 120/2050. Baffle step will always be an issue, but that's true even of any standard crossover as well. The Duelund is only a slightly altered 2nd order behavior from any standard LR2, what I think really caught SL's interest.

I would say some, if not most, but there are exceptions to almost every rule. Here would be my candidate: Neofone
I need more drivers like I need a hole in the head, but those look intriguing for a not unreasonable price. Someone snag a pair and send them to you, Mark or Zaph .

The tweeter is more limited as I see it. There are certainly many drivers that can do LR2 at 2.05K, but that precludes all 3/4" tweeters, my preference. This forces one to choose a tweeter with limited dispersion. The only exception may be the new 6600 and 7100 from SS.

Preaching to the choir on that one. I went with the XT19 many moons ago for just that reason. Doesn't seem to matter to deaf guys like Zaph :P , but I think I can hear it . Did you catch this tweeter I linked? Tymphany 3/4" neo
Keep your eyes on Alpine, etc.

What's more intriguing to me is that it can be done passively. SL suggests that using an LT on the mid and tweeter could be useful as well, though I'd hesitate to do that on the tweeter even if going active.


They look like good candidates.


No problem. I keep wanting to start with "Deu".

Ya, with very careful selection of drivers, passive is certainly a reality, although you know I favor active so that even more drivers can be selected.
Deu all you want .

cheers,

AJ

[Dennis H]

BTW, the 12m should be able to cross higher than 2k so you could move all the frequencies up a bit. Summarizing SL's recipe:

Pick the center freq F
Woofer, two LR2 at .38F
Mid, one LR2 highpass at at .28F and lowpass at 3.6F
Tweeter, two LR2 at 2.62F

[JonMarsh]

Different strokes...

SL's apparently doing something different with the base transfer function calculation, by his own description using an 8th order polynomial, not I believe the 4th order version described usually by Steen. I'm in communication with a gentleman in Sweden who's worked with Steen Duelund, and he is surprised by SL's transfer function with a Woofer/tweeter crossover at -32 dB, too.





When I run the numbers for a Duelund three way using his published Equations, I get this...






With aleph set to 4.5, and center frequency of 500 Hz as in SL's example, the woofer-tweeter crossover calculates to -26 dB. SL is apparently using an 8th order polynomial derivation (Duelund squared?) which changes the Q in the transition region so that the approach to 24 dB/octave slope on the high and low pass is faster, resulting in a -32 dB crossover.

Because of the range of overlap contributing to significant output, it's my proposal that the driver performance must be lean and controlled to at least -18 dB level, to avoid degradation of the frequency response and audible effects from linear and nonlinear distortion.

Looking at the midrange in this alignment, this would require a bandwidth from 45-50 Hz on the bottom end to 5-6 kHz on the top end. Perhaps a 5" SS Revelator 15M/4531? 15W/4531? (don't like that impedance bobble at 800 Hz at all) Audio Technology SK130-308?

Tweeter, well, the 66000 comes to mind, though possibly one of the ring radiators could even be made to work, maybe even the XT25.

This wouldn't be an easy alignment to do, hard for me not to think that dropping the coefficient and moving the center frequency up a bit might be more realizable. Think about BSC... Will think about that some, but probably not this weekend! ;^)

[capslock]

Here would be my candidate: Neofone
I need more drivers like I need a hole in the head, but those look intriguing for a not unreasonable price. Someone snag a pair and send them to you, Mark or Zaph .

Hmm, judging from the impedance curve, inductivity is higher than I'd like it to be. What kind of units is Sd = 9.503m?? From their description, this is supposed to be a 15 cm driver.

[Dennis H]

I know SL claims he's using an 8th order polynomial but it doesn't seem to be true. For one thing, that would give you more drivers than a 3-way under the Duelund recipe. When you square the polynomial, you add drivers. For another, as an example, his woofer function is 4th order. I'll let Jon and SL work that one out in an American Idol Genius Throwdown.

b) Woofer lowpass LPD3 = 0.384 / ( s + 0.38 )^4
Four real axis poles at -0.38 which can be realized with two LR2 lowpass filters in cascade (e.g. at 190 Hz).

[dlr]

I might go even higher

BTW, the 12m should be able to cross higher than 2k so you could move all the frequencies up a bit. Summarizing SL's recipe:

Pick the center freq F
Woofer, two LR2 at .38F
Mid, one LR2 highpass at at .28F and lowpass at 3.6F
Tweeter, two LR2 at 2.62F

If I do the tweeter as LR4 @3K, then this gives me 833 for F, thus 233Hz for the woofer. This would still allow me to use the 25w/8565 I now have and use LR2 @300Hz. Moving it down to 233 is a cinch, at least with SoundEasy in DF mode and no impedance interaction. I've already got the 12m doing LR2 @3K, so setting this up shouldn't be all that difficult. Of course I say that without having dug into it yet!

[dlr]

I don't trust the small neo tweeters yet

Did you catch this tweeter I linked? Tymphany 3/4" neo
Keep your eyes on Alpine, etc.

So far any small neo tweeter has had at least one significant issue that would preclude me from using them. Either it's some glitch up high or a QC problem with the stuffing used in the "chamber". I sent an XT25SC to Ken Kantor at his request when I indicated the problem. That was months ago, no response yet.

[dlr]

I think that the key is this line at SL's page

The Duelund 3-way is based on an 8th order polynomial yet the maximum slopes are only 4th order over the first 40 dB of attenuation.

Maybe I'm just missing something here.

Because of the range of overlap contributing to significant output, it's my proposal that the driver performance must be lean and controlled to at least -18 dB level, to avoid degradation of the frequency response and audible effects from linear and nonlinear distortion.

Not quite sure what you're getting at unless it's just target matching. With the many good drivers available, this shouldn't be a big hindrance.

Looking at the midrange in this alignment, this would require a bandwitch from 45-50 Hz on the bottom end to 5-6 kHz on the top end. Perhaps a 5" SS Revelator 15M/4531? 15W/4531?

Precisely the drivers that first came to mind.

Tweeter, well, the 66000 comes to mind, though possibly one of the ring radiators could even be made to work, maybe even the XT25.

Personally, I'd stay away from the XT25. But even the old 9300/9500 line could do rather well, though something such as the Seas 27TBFC or others ought to do as well.

[dlr]

I've gotta take more time when posting

The woofer and tweeter use two LR2 filters each, giving a 4th order rolloff. Unfortunately the DCX can't do that so you'd have to fudge it with an LR4 and some parametric EQ to try to match the transfer function. SE should be able to do it okay.

You're right, the tweeter only needs to do LR4, but that does still eliminate 3/4" tweeters, unless the XO can by pushed up as you noted. In fact, it's kind of what I had in mind. I'd like to use the OW1. I use it LR2 @3K now, so an LR4 ought to be doable to some point below that, not quite sure how down it can be pushed.

[Dennis H]

I thought I'd revive this old thread as there was some discussion of DIY Unity Horns. There's a Unity thread going at AVS and Thomas found the Yorkville horn listed in a parts catalog. I followed up with this and thought it might interest people here.

Wow, good find, Thomas!

With that in mind, I went to the Yorkville site and found the same parts catalog and I also found a service manual for the U15P, the powered version.

http://www.yorkville.com/default.asp?p1=6&p_id=121

So, here are the parts and Yorkville retail prices.

H32 HORN FOR UNITY 15 CABINET $84.00
7402 8R 45W 1" DRIVER BMS 8 R 4550-8R $240.00
7405 5" 8R 30WPGM SPEAKER SEALED BACK $44 (3 required)
7439 15" 8R 800WPGM NEO SPEAKER $300.00

If you were designing your own crossover or going active, all you'd really need to order from them would be the horn and the (proprietary) mid drivers. Assistance Audio seems to have removed their BMS pricing but I did a Wayback Machine look and they were selling the 4550 for $115 in July. That's a very good compression driver but you could probably substitute others. As far as the woofer(s) plenty of home or pro woofers could be made to work well below 300 Hz.

[ThomasW]

I have a call into a local Yorkville dealer. They'll let me know monday or tuesday if the horns can be ordered separately..

[JoshK]

I'm all over it if you can buy the horns and mids seperately!

[ThomasW]

I was looking at old posts on the ProSound web, someone posted that no Yorkville won't sell the U15 horns separately nor will the sell the horn assembly with the drivers...

[Dennis H]

Thomas pointed out in the other thread that you probably don't need their mids either. They use sealed back mids so they can share the enclosure with the woofer but you wouldn't need that if you built an enclosure for the horn.

[ThomasW]

I talked very briefly with Jon today he said nope to using open backed mids unless they mounted in a very small sub-enclosure. Has something to do with needing resistive loading so the midranges don't unload trying to operate below the horn's lowest frequencies.

It's a moot point if the horns themselves aren't available...

[ergo]

Some news from Seas also - a new website and their version of the tweeter with WG is published

http://www.seas.no/index.php?option=com_content&task=view&id=184&Itemid=179

The WG is pretty small, but some directivity control still achieved. And the response is very nice and clean (if the 26kHz peak is not counted).

Ergo

[rc white]

Waveguides

Hello, I am new to this forum and am writing in because I am the one who wrote the waveguide article at Rod Elliot's site.

The first thing I would like to point out is that the devices I have described are quite different from the ones described by Geddes.
The OS type is specifically intended to have a plane wave incident at its throat, and to "bend" this into a spherical one, the plane wave being derived from a compression driver.

The waveguides I described are of the sort found initially in the products of Genelec and KEF etc., these are specifically intended to be driven by dome transducers with no throat compression.

Doing a simple Huygens construction will show that the wavefront from a dome driver is already curved and at least for a hard dome will remain so up to high frequencies. Research done by KEF for its uni q drivers indicates that the optimum dome is one that has an included angle equal to that of the wave guide,(they use 80 degrees). If this is done the emerging wave is at right angles to the wall and if the dome was actually pulsing then it is theoretically capable of launching a one parameter wavefront, (in practice the dome vibrates back and forth and this causes some scattering).
If the wall consists of a cone then the wave front will propagate in a very near mono modal fashion, unfortunately a true mono modal waveguide can only be infinitely long, and we need to get sound out of it so must cut it off at some point, and this is where our troubles begin.

Research done at the University of Adelaide shows that in the conical horn with flared mouth,(very similar in geometry to the ones described by Keele and Geddes), the radiation is dominated by the Bessel modes of the mouth area, in my article the 3kHz. "two section" waveguide I described is of this sort, and the off axis plots show an anomaly in the 8-10kHz. region that corresponds to the first 'mixed" Bessel mode of the mouth region(1,1).
Better results are given by the three section device and this is essentially because the wavefront has "left the wall" before it gets to a section of the waveguide large enough in diameter to excite these modes to any appreciable extent.
The key to understanding these waveguides is that they essentially operate in the acoustic near field of a spherical wave and the plane wave assumption of Websters equation is not relevant.
This post is getting a bit long so I will leave it there.
I hope I have helped to elucidated some points that have been raised.
RCW

[neuro]

I find the concept of the Unity Horn appealing. A high efficiency point source from 300Hz-20kHz. While eagerly waiting to see if the horn section might be available separately, I had a thought. What of the BMS 4590 (or other coaxial CD)? Doesn't this accomplish the same goal in what might be considered a simpler fashion? Will one approach have inherently lower distortion (or some other benefit)? It seems like this is comparing apples to apples, especially if the 4590 is considered in a horn with the same geometry as the Unity Horn (without holes for extra drivers of course). Am I mistaken though? I certainly suffer from a lack of knowledge on the subject of horns and compression drivers, so any thoughts would be welcome. I think that the three mids and a 4550 are in the same ballpark price wise as a 4590.

[JoshK]

neuro, not to point you elsewhere but Mr. Danley commented on some of the differences between the unity concept and the BMS coax CD on audioasylum's high efficiency board. There is also a lot of chat about that driver on that board from people who have actually owned and used it.

[Dennis H]

Josh, can you give us the Cliff Notes version of the chat over at AA?

[JoshK]

Geez, not even sure I remember what all was said. I will probably bastardize what Danley said, but I think he said that basically the BMS coax won't do a true unity because of the physical relation of hte mid to tweeter CD's. I.e. I think the time alignment is a bit trickier for one. But then I don't truely understand the whole concept of the unity and its intricacies.

I think those who heard the BMS coax liked it and those who didn't had problems with crossover and once resolved liked it. But I didn't get the sense that anyone thought it was the holy grail either. It sounded a bit like there is a price to be paid for the benefit of coincident drivers. Let's not forget that the real limitations on range of CD is the horn. Too deep and the highs suffer and beam, too shallow and you don't get any low end gain. Its near impossible to get good even dispersion from 300hz to 20Khz.

[rc white]

coaxial waveguides

In regard to coaxial designs, also at ESP in the loudspeaker photos section is a "coaxial" waveguide speaker I built.
The horn is basically a conical type with a flared mouth driven by a pp coned five inch car speaker with a coaxially mounted Ti. dome driver, this has its own waveguide.
The crossover frequencies are at 300-3kHz. and the high frequencies do not have to illuminate the mouth of the larger horn.
The design achieves 90degree horizontal directivity from around 600Hz. to 15kHz. and vertically 45degrees from around 1.2kHz.
There are plans afoot to make a SEAS coaxial and KEF uni q version of this with a elliptical cross section waveguide.

[AJINFLA]

at ESP in the loudspeaker photos section is a "coaxial" waveguide speaker I built.

Link?

cheers,

AJ

[rc white]

http://www.sound.westhost.com/phpBB2/viewtopic.php?t=848

Link to coaxial speaker system description.

[dwk]

I talked very briefly with Jon today he said nope to using open backed mids unless they mounted in a very small sub-enclosure. Has something to do with needing resistive loading so the midranges don't unload trying to operate below the horn's lowest frequencies.

It's a moot point if the horns themselves aren't available...

Jon is right, as always. The Unity relies on the overall tuning of the midrange for correct operation, and the small sealed chamber is used to set Fs and Qts. This could have been done differently of course, but attempting to substitute other mids in place of the existing sealed ones wouldn't work.

[dwk]

Josh, can you give us the Cliff Notes version of the chat over at AA?

From my memory, the major sticking point was time alignment. The Unity uses an xover topology that induces an extra 90 degrees of shift in the tweeter, which delays the wavefront to be exactly aligned with the midrange as it enters the horn. This is even more advanced in the Synergy designs, as Tom achieves constant phase across the passband of the horn. (i.e the Synergy designs pass a square wave)

The BMS doesn't allow time-aligning the wavefront in this manner, since the two drivers are coincident. With a DSP/FIR xover you could of course fix this, and such a system would probably be very good. I'm not aware of a suitable off-the-shelf conical flare that will take either of the BMS coax drivers, though (well, aside from the Orphean)

[rc white]

horn tapping

A conical horn has a radially Dependant flare rate.
The flare rate of an exponential horn can be put as 2k where k is the wavenumber (2pif/c), and if the distance from the apex of a conical horn is r, this point corresponds to the flare rate of an exponential horn if kr=1.
At this point the real and imaginary parts of the impedance are equal and the resistive part of the impedance is .707 of the total, i.e. -3db.

We find that if we drive the conical horn at this point the real impedance increases toward the mouth and decreases toward the throat, the real part of the impedance being the cosine of the angle between the pressure and velocity, this can be put as angle=arctan(1/kr).

It is interesting to note that at 300Hz. the r value is around 180mm. the quarter wavelength at 286mm.

[rc white]

driving tapped horn

Further to my post of yesterday..
If you take an Eminence Alpha 6, put a rear volume of .5 litres a front volume of .2 litres, and tune this to 1kHz. via two 20mm. holes, modeled in win isd you get a single ported bandpass box with a +12db. peak at 1.2kHz. and a 3.5db. peak at a roll of point that starts around 300Hz.

Based upon the principle that below cut of a horn acts like a single ported bandpass box, driving a 60degree conical horn with above described device at a point 280mm. from its apex now gives us a bandpass box with a 5.3 litre front volume and a 280mm. diameter port, (this is the volume and area of the mass reactance area of the horn from driving point to apex), this gives us a single +13db. peak at 386Hz.

Combining these two gives an overall result that looks vey much like the two octave +12db. result described in the unity horn white paper with the cut off moved up to 300Hz. , giving the 300-1.25kHz. result specified for the Yorkville unity horn.

This is of course a crude model and I haven't built a prototype to try it yet, but it does accord with theory and predicts pretty much the characteristics what is already made and measured.

[Paul W]

I remember seeing polar plots for the BMS 4590 somewhere on the Internet (don't remember which horn)...but I've since lost the link. Does anyone have a link?

[kevmurray]

Waveguide Math --> Flare

Hello, I am new to this forum and am writing in because I am the one who wrote the waveguide article at Rod Elliot's site...
RCW

Hi Mr. White and welcome to the forum. I immediately recognized your chosen handle because I was studying your article on Rod Elliot's pages last week. I have been trying to gather the math into a spread sheet to help me make quick adjustments and see the results. In this I'm finding it difficult to get things straight. I never seem to get reasonable results regarding the flare.

I'm aware your two examples are different in that regard because one approximates a two angle flare and the other a three, but my math always results in poor transitions from conical region to flare. The exit angle of the conical region (always 45* from axis for a 90* waveguide) is never the same as the entrance angle of the flare. It seems an ellipse-arc rather than a circular-arc is the only way to achieve a 45* entrance angle and 90* finish. Is the flare intended to be "fudged" during construction to smooth things out? I hope I'm being clear. ops:

Kevin

[rc white]

waveguides

The mouth section of the two section waveguide consists of a circular arc that meets both the inner section and the baffle at right angles, the slopes of both inner section and baffle are the same as the slope of the circular arc where they meet.
Some people have complained that the published formula for the two section device doesn't work properly, (it seemed to work when I checked the article), and have come up with alternative ones, the one in the article I got from Matlab whilst attempting to make comprehensible to others my back of the envelope scribblings.
The major thing is that it is a combination of a straight line and a circular arc that is an asymptote of a continuous curve.

[kevmurray]

The mouth section of the two section waveguide consists of a circular arc that meets both the inner section and the baffle at right angles, the slopes of both inner section and baffle are the same as the slope of the circular arc where they meet..

When you say right angles don't you mean parallel? I'm not trying to nit pick, I just want to make sure I'm following you.

Some people have complained that the published formula for the two section device doesn't work properly, (it seemed to work when I checked the article), and have come up with alternative ones, the one in the article I got from Matlab whilst attempting to make comprehensible to others my back of the envelope scribblings.

I had to play with the formulas a bit to achieve the correct order of operations. In MS Excel this required a lot of brackets to make things work properly.

The major thing is that it is a combination of a straight line and a circular arc that is an asymptote of a continuous curve.

When you say continuous curve do you mean the mouth? I would have said the baffle face to make things simple but then again in construction it's not really asymtotic anyway. The flare must transition to the baffle at some point, or did you mean something else entirely?

Thanks for helping me out with this!

[EdL]

For anyone wishing to obtain an excel spreadsheet which depicts a waveguide shape, drop me a line. JoshK did most of the work.

[rc white]

waveguides

Basically if a straight line meets a circle at a right angle then the line is tangent to it, (it has the same slope), in the waveguides described as two section devices the radius that is the transition between the conical inner part and the baffle is at right angles i.e. tangent to both the inner section and the baffle.

This is not a continuous curve in that it has one function from throat to mouth i.e. it is an inequality, but does have the property that it has a slope with no discontinuities at the points where the functions meet, the property of a spline.

If we put another point on the circular arc, four points in total, then the nearest continuous function that fits is the parabola, the three section device can also be looked at this way but in this case one in which the throat coincides closely with the latus rectum.

These represent what we might call asymptotes to the range of curves found by Johansen in his AES paper, (the actual continuous functions that fit being parabolas), to be those that give the closest approach to constant directivity.

Overall what the article points out is that you can make a range of waveguides that have constant directivity, and minimum higher order mode effects using only straight lines and circular arcs, provided these meet with no discontinuity.

[TomK]

Basically if a straight line meets a circle at a right angle then the line is tangent to it,...

A picture's worth 2KB. Perhaps a little more in this case...

[JonP]

Probably a year back I started some work on a waveguide project, sadly "life" has been interfering, as well as laziness.

Anyway, one thing I stumbled across was a paper on a kind of waveguide, and it provoked me to try making waveguides with roundover and chamfer bits. I'm wondering if the paper I had read might be similar to what Mr. White is suggesting.

The idea was called a Quadratic Throat Waveguide. Took a bit, but I found the link: Peavy Waveguide

It's a full white paper writeup, it covers a few horn/waveguide types, then develops to their idea. I got to the point of duplicating the curved throat and matching layer of 45 deg chamfer as a proof of concept, but never built a taller stack, or got to measuring the results.

Hope this provides a few "pictures" for the discussion...

[TomK]

Thanks for the reference, Jon.

I think the drawing on page 8 is very helpful.

[JoshK]

That quadratic throat WG looks awfully similar to Geddes' oblate spheroidal WG. Maybe when I get a few moments of free time I can play with the algebra to demonstrate the differences or lack-there-of (I like the practice).

A while back on this forum, someone mentioned a place that would make custom router bits. I would think if you could get a router bit to do the quadratic throat part that making these kind of WG's would be accessible. The straight sided conical part is sure easy enough if you have patience to build up stacked layers.

[dwk]

That quadratic throat WG looks awfully similar to Geddes' oblate spheroidal WG. Maybe when I get a few moments of free time I can play with the algebra to demonstrate the differences or lack-there-of (I like the practice).

They are very similar, and I'm sure somebody posted a graph comparing the two - it was either here or at diyaudio.

One of the things that Earl said was a factor in his not patenting the OS waveguide is that the math is stable - the OS curve is optimal, but slight deviations away from it don't induce huge changes in the behavior. The Peavy approach is this type of a deviation, and is still 'close enough' to be a reasonable approach. Of course, Peavy DID patent it, which I guess says something.

[EdL]

Josh posted: "....custom router bits. I would think if you could get a router bit to do the quadratic throat part that making these kind of WG's would be accessible"

Exactly. Geddes mentioned the angle of entry for the throat of the Summa was 6 degrees. Earl says the entry angle has a lot to do with the success of a waveguide. He chose the B&C driver for its favorable characteristics. The waveguide in the Summa was built for it.

[ThomasW]

For those with fat wallets Geddes will sell his waveguides including the foam phase plugs for $250ea...

[JoshK]

Dr. Geddes' WG's aren't cheap but are reasonable IMHO. The only reason I haven't jumped on it myself was I was looking to maybe try bigger ones and he was offering to make a small quantity of *one* profile only.

[JoshK]

...Geddes mentioned the angle of entry for the throat of the Summa was 6 degrees. Earl says the entry angle has a lot to do with the success of a waveguide. ....

Which reminds me, I need to add the entry angle to the spreadsheet. I started working on it earlier and then got buried in other stuff.

[EdL]

Josh, You may not know. A copy was forwarded to Dr. Geddes. His only comment was about the entry angle. So in his own special way, he confirmed it is correct.

My thought (for in the shop) is that it is simple to truncate the throat at a chosen intercept angle. To do so in the spread sheet makes my head hurt....so, in the spirit of your avatar's undying stamina...wear it out big boy!

[ThomasW]

I ordered some blocks of open cell reticulated foam. I'll carve them into the appropriate shape and see what effect they have when placed in the DDS ENG-1 waveguide.