Horn Directivity, Small Listening Rooms and 2" Midrange Compression Drivers.

[kevinkr]

A little OT: I'd be very interested in hearing more about the directional characteristics of conicals vs tractrix if you have any insights you'd like to share. Using Hornresp with the 350Hz design I am working on it looks like they start to become extremely directional above 4kHz or so. Dropping from roughly 80 degrees just below 2kHz all the way to about 20 degrees at 8kHz. My design geometry not surprisingly is not that different from the AH! 700 except it has a 2" throat.. This really doesn't look like the right approach to cover this much range.. Tractrix better presumably? I'm currently using a 400Hz exponential over this range that sounds surprisingly good in fact.

Thanks, Kevin

[Hoerninger]

A little OT: I'd be very interested in hearing more about the directional characteristics of conicals vs tractrix

Although it is not close to your interest it is worth to have a look at the "Kugelwellen- Trichter" (spherical horn). It is chorter than a comparable tractrix horn and it has a wider dispersion. (The conical horn is worst in this aspect I suppose.)
All information about the Kugelwellentrichter you will find in this forum.
____________
Peter

[kevinkr]

Although it is not close to your interest it is worth to have a look at the "Kugelwellen- Trichter" (spherical horn). It is chorter than a comparable tractrix horn and it has a wider dispersion. (The conical horn is worst in this aspect I suppose.)
All information about the Kugelwellentrichter you will find in this forum.
____________
Peter

Thank you Peter, I am starting to look into this. Looks very interesting.. I am such a newbie at all of this and have a huge opportunity to get some custom horns built and want to make sure I make the right choice, once cast it is over.. The conical seems to have serious directivity control issues covering the frequency range I need to cover..

[yggdrasil]

The directivity of a horn is not only dependant on the profile (tractrix, conical, exponential...) but also very dependant on the layout(circular, square, bi-radial...).

There are lots of info out there, but unfortunately none of it is really easily accessible, and I find most of the information in bits and peaces.

[Hoerninger]

The directivity of a horn is dependant on the layout(circular, square, bi-radial...).
I find most of the information in bits and peaces.

It is even worse for the DIYer as there is no comprehensive theory. So there are only two choices:
1. make your own R&D or
2. use an approved design .

A "crude rule of thump" is demonstrated by this sketch:


____________
Peter

[kevinkr]

I seriously appreciate the assistance, perhaps we should start a new thread or perhaps a mod can move these posts to a new thread and let us know? I didn't intend to hijack the thread..

OT again: I am going to borrow a couple of different types of horns and give them a listen, one is a tractrix, another a smith.. This should help me to decide what sort of directivity I need. I noted the tonal balance in my system is a little better (tolerable?) slightly off axis than on - which may mean that the conical might actually be a good choice?? I'm sitting slightly below the vertical mid point of the horns and it sounds nice, (tonal balance correct) on vertical axis not so nice.. Sensitivity is higher than the old set up and I am now running pretty near the bottom of the attenuators - this I will need to address at some point. Seems like more uniform dispersion would be the right way to go with the subsequent adjustment to the X-O to address the sensitivity change.

[yggdrasil]

I can split the thread. At which post, and with which new heading? Already the second post is way off topic, but in another you are wandering right now....

[withTarragon]

A little OT: I'd be very interested in hearing more about the directional characteristics of conicals vs tractrix if you have any insights you'd like to share. Using Hornresp with the 350Hz design I am working on it looks like they start to become extremely directional above 4kHz or so. Dropping from roughly 80 degrees just below 2kHz all the way to about 20 degrees at 8kHz. My design geometry not surprisingly is not that different from the AH! 700 except it has a 2" throat.. This really doesn't look like the right approach to cover this much range.. Tractrix better presumably? I'm currently using a 400Hz exponential over this range that sounds surprisingly good in fact.

Thanks, Kevin

If I may interject, sonce this is a topic near and dear to me. The Conical horn is a CD horn, that is, the freq respsonse on-axis and off axis are comparable (as is the the Oblate spheroid, manta ray, bi-radial, some electro voice etc). However CD horns do not have a flat freq respsonse and will require some EQing to boost the high end. The tractrix is not a CD horn and it will beam as you go up in freqeuncy.

The term conical is tricky and typically mis-used. The expansion rate is "conical" and the cross-sectional area will increase by the square of the distance (IOW, twice the distance then 4 times the increase in area). Not all "cones" will necessarily have this rate of expansion (it depends on the angle). Again, the term conical refers to the rate of expansion and not the contour of the sides.

The other problem is that attaching a concial flare (say a "cheerleader's megaphone of the proper angle) will not make a good horn since there would be a sharp discountinuity in the flare (contour) at the throat where it meets the driver. Consequently, conical horns will have some secret recipe that transitions the flare betwen the conical portion and the throat/driver. The designer will (or may have) also have some secret recipe for transitioning the end of the conical section to the mouth. This is usually some rounded over geometry at the exit of the mouth. Without this last feature, the horn will do poorly at low freqeuncies (since horns are usually undersized for practical reasons) and the horn will have mouth reflections (especially at lower freqeuncies). In sum, a "conical horn" may deviate at the throat and mouth by design.

Now before folks beat up on me for narrowly defining the term "conical", check Olson's text book or Berenek's. The disinction does matter, not in terms of directivity, but terms of loading/efficiency

Good luck on your project and I am curious how it comes out and how you solve the various problems that always seem to creep up.

[kevinkr]

If I may interject, sonce this is a topic near and dear to me. The Conical horn is a CD horn, that is, the freq respsonse on-axis and off axis are comparable (as is the the Oblate spheroid, manta ray, bi-radial, some electro voice etc). However CD horns do not have a flat freq respsonse and will require some EQing to boost the high end. The tractrix is not a CD horn and it will beam as you go up in freqeuncy.

The term conical is tricky and typically mis-used. The expansion rate is "conical" and the cross-sectional area will increase by the square of the distance (IOW, twice the distance then 4 times the increase in area). Not all "cones" will necessarily have this rate of expansion (it depends on the angle). Again, the term conical refers to the rate of expansion and not the contour of the sides.

The other problem is that attaching a concial flare (say a "cheerleader's megaphone of the proper angle) will not make a good horn since there would be a sharp discountinuity in the flare (contour) at the throat where it meets the driver. Consequently, conical horns will have some secret recipe that transitions the flare betwen the conical portion and the throat/driver. The designer will (or may have) also have some secret recipe for transitioning the end of the conical section to the mouth. This is usually some rounded over geometry at the exit of the mouth. Without this last feature, the horn will do poorly at low freqeuncies (since horns are usually undersized for practical reasons) and the horn will have mouth reflections (especially at lower freqeuncies). In sum, a "conical horn" may deviate at the throat and mouth by design.

Now before folks beat up on me for narrowly defining the term "conical", check Olson's text book or Berenek's. The disinction does matter, not in terms of directivity, but terms of loading/efficiency

Good luck on your project and I am curious how it comes out and how you solve the various problems that always seem to creep up.

I'm learning something new everyday. I believe I currently lack the sophistication & experience to safely design one of these horns and get it made.. I noticed the treatment at the edge of the horn mouth and wondered at the purpose, I did not know anything at all about the issue of matching the throat contour between the driver and the conical horn. I need to just slow down and do a lot more research before I commit any funds to making anything..

[kevinkr]

I can split the thread. At which post, and with which new heading? Already the second post is way off topic, but in another you are wandering right now....

Such as it is I think the best place to split the thread might be at #10. I didn't really intend to hijack the thread but effectively feel I have..

What I am really trying to get at is what sort of midrange horn, directivity characteristics should I be considering with a large format compression driver in a relatively small room. I think this is subject that hopefully is of interest to others.

"Horn Directivity, Small Listening Rooms and 2" Midrange Compression Drivers.." Open to suggestions here.

Thanks!

[kevinkr]

Thanks, Yggdrasil!

Just a little more background for anyone who might have some insight.. My listening position is about 3m (10 ft) from the speakers. The original set up prior to the arrival of the 2440s used 2420 on 2312 horns with clones of the 2308 diffusers installed. This set up was actually pretty beamy even at the listening position (very narrow sweetspot particularly in the vertical plane) but imaged quite well probably due to the tight directivity.

The room is about 12 x 18 ft (4m x 6m) and ceiling height is 7 ft or about 2.3m. The back of the room to one side extends out another 6 ft or 2 m.. Slightly odd shape. And it is pretty cluttered. (Except in front of the speakers to the listening position) There is a lot of stuff on all sides. It actually is not a bad sounding room..(I have had other rooms that were far worse)

So far it has been suggested that I look at the kuggelwellen-trichler which IMHO is a very interesting design which is a bit large physically although not impossible. Originally I was very interested in the conical, specifically something similar to the AH! 700 or alternately the Edgar 350Hz tractrix.

It is unlikely that I will cross-over below 800Hz so it seems reasonable to design a horn for the 375 - 400 Hz region to reduce size and possibly improve dispersion at the top end. (?) The X-0 are 2nd order butterworth and are -6dB @ 800Hz and should be at least 15 - 18dB down at 400Hz - the filters are heavily resistively loaded by Lpads and hence do not interact strongly with the X-O components I suspect.

I'd really like to avoid the need for EQ if at all possible..

[yggdrasil]

Is this a 2-way or ... The question is how high are you trying to make this driver play?

[Hoerninger]

Journal of the ASA: Acoustical Studies of the Tractix Horn (I and II)

http://forums.klipsch.com/forums/storage/3/1063914/Acoustical%20Studies%20of%20the%20Tractix%20Horn%2 0(I%20and%20II).pdf
____________
Peter

[yggdrasil]

Here's some more reading: http://www.quarter-wave.com/Horns/Horn_Theory.html

[badman]

It is even worse for the DIYer as there is no comprehensive theory. So there are only two choices:
1. make your own R&D or
2. use an approved design .

A "crude rule of thump" is demonstrated by this sketch:


____________
Peter

This is a fine rule of thump for the dispersion of frequencies where the pathlength in the throat is sufficient to create directionality. As you go below this, when the length is acoustically insignificant, you see the pattern become defined by the second section of the horn. This is why tractrix, exponentials, etc. have the tapering beamwidth- they're continually changing in expansion. The longer the pathlength gets the wider the beamwidth is able to be as it's seeing larger and larger angles. This buys you flat on-axis response without so much EQ but does mean a "lump" power response.