WMcD

Hmmm, why does anyone assume there is an improvement to be accomplished. The K-Horn works very well. It is a bit annoying that there is some assumumtion that the the design is not optimised. Gil

Dear djk, Thanks for your remarks. All input is welcome. But, gee, you seem to reach some harsh conclusions in a brusk way. I don't take issue with the citations to literature. They say what they say. However. . . . Measurements reveal better than you imply. There is reason for this. A lot of PWK's work has been to work engineering tradeoffs in mouth size. And getting good bass response out of speaker of reasonable size is the goal. The most analytical study of mouth size IMHO is from D.B. Keele in his AES paper on optimal mouth size. He points out that bigger is not necessarially better. And, from his paper, if you look at the numbers for the "optimum" bass near cut off, the Klipsch family of horn loaded woofers are not quite optimum. None the less. In rooms, the mouth load is better than in half space. In a corner, there is a respectable match; even at a wall floor boundry, it ain't half bad. If you look at the reflection coefficents for corner placement, we're in the ball park. An early patent by Thuras on the WE 555 driver has something to say on this also. Essentially, a good choice of driver and reactance anuling back chamber can make things work better than a simple look at mouth size would indicate. You cite to PWK's paper on the early K-Horn. That was his EARLY work and impressions. In my read, it does not support a conclusion that the modern K-Horn or LaScala are dullards in the low end. Measured specs are to the contrary. PWK's AES paper on the prototype of the LaScala showed otherwise. Specs on the K-Horn are otherwise. Regarding the K-Horn, Heyster's review in Audio magazine shows it has good response to the high 30's. The bottom line is that PKW's designs are the result of a thoughtful analysis of mouth loading, the affect of the finite horn, and careful selection of the driver. Put together, it all works very, very well. It is all based on sound engineering and measured results. Mr. Olsen never said that K-Horns or LaScalas don't work as advertised. The bottom line is that K-Horns and LaScala's do work well (1) below Fc and (2) despite having small mouth dimensions. Gil

Dear Klewless: The PWK calculation for the back chamber is for the exponential horn. Also, it assumes an infinite horn. In later patent specifications he points out that the chamber for a finite horn might have to be smaller. This is because the mass load from the finite horn is higher. Thus there is more "reactance" to annul. Also, Don Keele has pointed out that the compliance of the driver has to be taken into account. The driver has its own spring-e-ness. (Sp?) Overall, the calculations are starting points from which to experiment. Text book calculations show that other "families" of horn expansions, like hyperbolic, exhibit other levels of mass reactions. In fact, the exponential is a subset of the hyperbolic. Again, it is difficult to generalize what will actually show up in practice and what is optimum. One important point is that we ARE dealing with the type of bass horn which uses a back chamber. Not all do. The VoT types have a bass reflex chamber. However, it seems that PWK is a fan of the sealed chamber. The people at Jensen reached similar conclusions about its merits. From what we are listening too, there is not much room for argument. This is PWK's design choice. He does it well. Gil

Kerry, I wanted to check the calculations on the back chamber volume. PWK says it is 2.9 x throat area x length to double. So, the scaling factor shows up three times in the equation and the back chamber does scale. Just as you say. Gil

Volvo, The figure of 66% makes some sense to me. 3/4 inch plywood scales down to 1/2 inch. And the 15 inch driver scales down to 10 inch, as you point out. Also, with the 66 % reduced linear dimensions, we are talking about something reasonable in bulk. Five of these are under construction. I owe all you folk some photos and curves. Maybe in a month or so. Making a matching midrange out of plywood was no small problem. Picking a bass driver has been a real problem. I noticed that the drivers used by Klipsch tend to be ones suitable for a larger throat and thus are looking into a smallish throat. Also Klipsch seem to stick to a 97 dB output. By Keele's parameters, the high frequency limit is low, but Keele doesn't account for the effect of the back chamber, which increases the Fs of the system. Naturally, the proof of the pudding is in the eating, and the outcome here has to be measured. I'll keep you advised. Gil

Kerry, My calculations agree with yours, the expansion does "scale." I think that is also true of back chamber volume, but I'm gonna look at it again on paper to be sure. I've been assuming that the LaScala flare doubles every 12 inches, which gives an Fc of 61 Hz or so. Regards, Gil

Nice project. I'm thinking about a 66% scale down along the same lines. It is annoying that the THX, Dolby spec for "small" calls for 80 Hz. That is in the ballpark for your project, and a bit low for mine. The present hand wringing here is to find a suitable bass driver. I think we're on the same wavelength. If you have a suitable sub, and the HT system is set to "small", it should be possible to adopt the LaScala design for the mains and surround. Share your thoughts. Gil

Tony, Thanks for the very nice drawing. To complete the picture, can you tell us the interior dimension of the height of the bass horn. Regards, Gil

Dr. Bruce Edgar published something in Speaker Builder on how such structures were necessary to prevent the area in the turn from being too great and thus effectively widening the duct. I.e. a departure from the approximation of an exponential expansion. The Speaker Lab near copy of the K-Horn has such pieces at the first turn, i.e. the turn which is actually at the top and bottom just behind the front surface. This is analogous to what you're talking about. The duct there in the Speaker Lab version, like in in your drawing, is about three inches wide. The diagonal piece is solid. The short sides of the cross section of the piece are 1.5 inches long. Therefore the hypotenuse is 1.414 time that. I mention this as suggested dimension to work with. I've seen one explanation from Speaker Lab that this is a matter of "reflecting" treble frequencies. I think that is not quite accurate in view of the wavelengths involved. 400 Hz is about the upper limit and has a wavelengths of over two feet. The other way of explaining it is that the wave has to go a relative long way around the outside of the curve, and a short way around the inside of the curve. You don't want that difference to be anywhere near a half wavelength, because there would be cancellation. My understanding of the design of the K-Horn, Belle, and LaScala is to keep the width of the duct in the turn to a minimum width. The path is split into two symmetrical structures. Otherwise, one path could have a duct width of six inches. The radius of the outside of the turn would then be greater and thus would be closer to half a wavelength. The split path may well reduce the purported problem. Especially since there is only one bend in the LaScala. So, that may be a reason for the piece's omission. However, putting the brace in is probably, IMHO, a good idea and can't hurt. It would be interesting to see how the turns in the Jubilee are designed. It seems the bass horn is being used up to 800 Hz where such considerations have a larger role. Gil

Unless this fellow has run tests, I doubt there is much basis to say one or the other design is "better" in any substantial way just because of the type of magnet. From what I read, the units (I assume, both round and square) go though more QC at Klipsch and EV before they get the the Klipsch label on them. So, Klipsch is hand picking the ones which produce better curves as detected by an objective test. It is a matter of production variations. Gil

John P, Gee that was some very nice investigatory work. Thanks for taking the time to do the work and post the results. If I may offer a thought. The geometry of panel resonances are a bit beyond me right now. But, basically, on the side wall there are three secured sides (back, top, and bottom) and a non secured front side. It may be that you placed the end of the clamp at the midpoint of the front of the side panel. That would create a node at the center of the front edge of the panel. Of course, the whole thing is better damped, which is probably what you saw. However, it would create some symetry between the upper half and the bottom half. Therefore, there are now two identical systems which peak at their own characterist frequency, thus reinforcing each other's output. A "fix" would be to place the clamp a bit off set from the center, perhaps at the three fifths position toward the bottom or top. Now you have a different, and perhaps better, distribution of the resonances. There is most of the damping, but not so much overlapping. It is something to experiment with, and doesn't cost anything. If you try it, let us know. Thanks again for the scientific info. Regards, Gil

Well, we've had reports of a bad connection at the autotransformer. You might check that. Gil

Regarding scaling up a LaScala it could be done. But consider what happens if you scale it up by two to get an Fc down near 35 Hz. You will have a throat which is four times as big and will have to use four drivers. Also, you will be dealing with a cube which is four feet on a side and will not fit through a standard doorway. We can look to a Klipsch design of a bass horn with a theoretical 32 Hz Fc. That is shown in US Patent 4,210,223 to Gary C. Gillium and P.W.Klipsch. It looks a lot like the Klipsch theater woofer. This has a throat area of 172 square inches (using two K-43 drivers). The cross section doubles every 23.3 inches. The mean sound path length is 72.2 inches. The mouth area is 1472.6 square inches. The back chamber is 11,200 cubic inches. Regards, Gil

Gentlemen and Ladies: It is interesting to note that PWK has often expressed exponential growth of the area of his horns in terms of "in what length does the area double." This is a typical function of exponentials. A bacteria or group of rabbits will double by reproduction in X time. Of course, it works the other way too. Radioactive decay has a half life of X time. PWK calculates that the area of the exponential horn will double in 1/18.1 of the cut off wavelength. You must convert the cut off frequency to a wavelength. The result is that the K-Horn cross sectional area doubles (or halves) every 16 inches and has a cut off frequency of about 48 Hz. The LaScala doubles (or halves) every 12 inches and has a cut off frequency of about 70 Hz. Consider that the LaScala bass horn is 24" by 24" by 24" box. Take the throat area as 78 square inches. The length of the two ducts are as follows. Going across the back there is a length of about 12 inches. Total area at the turn should be 156 square inches. Going half way to the front is another 12 inches and area should be 312 square inches. Going the next 12 inches to the mouth gives an area of 624 square inches. Consider that the mouth is approximately 24" by 24" or 576 square inches. So the numbers work out pretty well. Regards, Gil

Gentlemen: The actual throat area of the K-Horn and other horn loaded speakers using the K-33 is approximately 78 square inches. It is true that the motor board has a 3" x 13" slot. However, this commmuicates to two 3" x 13" ducts. The question is thus why PWK uses the slot. It appears to me this is so that a chamber is formed by the concave structure of the woofer. There would not be much of a closed structure without the slot because the throat is approximately half the area of the diaphram. This allows for a resonant circuit to be formed by the compliance of the front chamber and the mass of the diaphram. This technique allows an improvement in response above about 200 Hz. Leo Baranek discusses this as applied to midranges. I'm quite convinced this is the reason for the slot. You will probably find that the exponential equations for the K-Horn and LaScala work well if you assume a 78 square inch throat. Regards, Gil