Hello,
I want to use a Visaton tweeter (KE 25 SC) in a three way speaker.
To optimize the directivity I would like to mount the tweeter from the inside an use the case material on the front as a waveguide.
This should transfer from the 25mm round diameter of the tweeter to an elliptical outlet of the "horn". The depth of the horn should be arround 19 - 40mm.
1. Where can I find some formulas to calculate the growth function of such a horn?
2. Is there a cheap simulation tool available that can calculate the transfer functions?
Plz don't worry about phase shift or additional delays.
I want to use a Visaton tweeter (KE 25 SC) in a three way speaker.
To optimize the directivity I would like to mount the tweeter from the inside an use the case material on the front as a waveguide.
This should transfer from the 25mm round diameter of the tweeter to an elliptical outlet of the "horn". The depth of the horn should be arround 19 - 40mm.
1. Where can I find some formulas to calculate the growth function of such a horn?
2. Is there a cheap simulation tool available that can calculate the transfer functions?
Plz don't worry about phase shift or additional delays.
I can't help you with the elliptical proposal, but I did do some studies of the results of various concentric, shallow waveguides with a rear mounted tweeter. The tweeter was the Peerless HDS, with the flange removed. Perhaps they will be of some value to you. The results were reported here:
DIY Waveguide experiments
DIY Waveguide experiments
Hello Dan,
thanks for the interesting report on your concentric WGs.
How did you measure the frequency response? Soundcard?
I am interested in a simulation of the "horn" (waveguide). A former collegue wrote his phd in offline simulating a driver and a horn independently - so he can estimate with a transferfunction of a horn and a driver the total transferfunction.
But mainly I am interested in optimizing the directivity a bit. But just guessing the right geometry will lead to an undefined (worse) design I fear.
Any recomendations?
thanks for the interesting report on your concentric WGs.
How did you measure the frequency response? Soundcard?
I am interested in a simulation of the "horn" (waveguide). A former collegue wrote his phd in offline simulating a driver and a horn independently - so he can estimate with a transferfunction of a horn and a driver the total transferfunction.
But mainly I am interested in optimizing the directivity a bit. But just guessing the right geometry will lead to an undefined (worse) design I fear.
Any recomendations?
My measurements are taken with SoundEasy. My approach was the only one I was capable of doing myself, which was to test what I can relatively easily construct with the tools I have. In fact, that was part of the idea. I wanted to test what an average DIY guy could quite reasonably build with typical tools and materials he would have for building a loudspeaker.
As far as theroizing what different shapes, sizes, etc. before hand, I would not be of much help.
As far as theroizing what different shapes, sizes, etc. before hand, I would not be of much help.
Short waveguides like this on dome tweeters will have only marginal effects as has been shown. To truely control the directivty the waveguide needs to be much bigger and it should be feed with a flat source rather than a round one. Then you can get some very highly controlled polar patterns. These shallow waveguides look a lot better than they perform.
I'd say judging the performance of shallow waveguides depends entiely upon your goals. In my case I wanted them to do a couple things. On was to counter the naturally rising frequency of the Peerless HDS. That was accomplished by the waveguides as shallow as 1/2" deep. A second possiblility was to provide a boost a the low end so that when equalized out in the passive crossover, the low end of the tweeters performance was improved, including potentially reduced distortion levels. This was accomplished to small degree with the shallow 1/2" waveguides, but the boost was only a couple decibels. the third thing was some possible improvement in off axis directivity, which is visiby quite obvious when compared to the flush mounted tweeter, especially in the 3/4" deep waveguides. The 3/4" deep WG's provided more boost as well and consequently enhanced low end performance. To my eye, with the 3/4" waveguides, the off axis response and directivity is better with the rounded profile than with the two I did with flat sections, though I'm sure this is not true as the get deeper and wider.
I realize these don't offer the same level of improvements that your deeper, wider waveguides do, but they also don't have some of the physical limitations and they do provide generally positive improvements over the typical flush mounted dome tweeter. IMO, it's not everyone's cup of tea to have a 10"-15" waveguide in all their speakers. How they look is also important to some folks. In addition, these canb be constructd by virtually any DIYer. To each his own.
I realize these don't offer the same level of improvements that your deeper, wider waveguides do, but they also don't have some of the physical limitations and they do provide generally positive improvements over the typical flush mounted dome tweeter. IMO, it's not everyone's cup of tea to have a 10"-15" waveguide in all their speakers. How they look is also important to some folks. In addition, these canb be constructd by virtually any DIYer. To each his own.
You can have both great sound and great aesthetics, IMO. I'm sure some don't care what their car looks like, since they only drive it, or what their home looks like since they only live in it, etc. Personally, I certainly care how what I create looks like, including speakers. I care what my speakers sound like even more, but I prefere not to settle aesthetically. YMMV.
For those who didn't look at the link, here is a graph comparing the results of the flush mount Peerless HDS tweeter at 0-15-30-45-60º off axis with the same tweeter mounted in a 3/4" deep waveguide with a 3/4" roundover. The latter was laid over the flush mount HDS results, but 15db higher. It is easy to see the boost provided and the directivity gained by this easy to construct waveguide, which takes nothing more than a 1-1/4" forstner or spade bit in a drill and a router with a 3/4" roundover bit, which most of us speaker builders already own, so it might not cost you a dime to implement. The differences are quite obvious. You can easily see how the rising response of the HDS was countered, how the low end was boosted by about 5db at 4khz and how off axis directivity was improved. When equalized to a flat responce in the crossover the benefits are not insignificant.
For those who didn't look at the link, here is a graph comparing the results of the flush mount Peerless HDS tweeter at 0-15-30-45-60º off axis with the same tweeter mounted in a 3/4" deep waveguide with a 3/4" roundover. The latter was laid over the flush mount HDS results, but 15db higher. It is easy to see the boost provided and the directivity gained by this easy to construct waveguide, which takes nothing more than a 1-1/4" forstner or spade bit in a drill and a router with a 3/4" roundover bit, which most of us speaker builders already own, so it might not cost you a dime to implement. The differences are quite obvious. You can easily see how the rising response of the HDS was countered, how the low end was boosted by about 5db at 4khz and how off axis directivity was improved. When equalized to a flat responce in the crossover the benefits are not insignificant.
What is difficult for is judging the effects of different types of horns just from description.
Using a horn adapts/transforms the acoustic impedance of the speaker from motion into sound presure. In your case the idea was to reach a lower corner frequency (while the driver performs lower magnitude), to optimize frequency response on axis and reduce distortions a bit.
The loud-speakers case plays also a big role in directivity. A small speaker cabinet will bent the acousic waves differently than a big cabinet. This also influences directivity.
I know the e.g. Klein + Hummel is jusing waveguides for their studio monitors to avoid reflections from the mixer rack in sound studios.
So my main goal is shaping the directivity - less acoustic energy for the room to reduce hall effects and get a rather neutral behavier while increasing a homogeneous directivity in horizontal space.
Reducing harmonics through less motion of the driver is fine for me as well. Lower corner frequency is nice to have.
What is not critical for me is a smooth frequency response as I can correct that.
What approach can you recommend?
Using a horn adapts/transforms the acoustic impedance of the speaker from motion into sound presure. In your case the idea was to reach a lower corner frequency (while the driver performs lower magnitude), to optimize frequency response on axis and reduce distortions a bit.
The loud-speakers case plays also a big role in directivity. A small speaker cabinet will bent the acousic waves differently than a big cabinet. This also influences directivity.
I know the e.g. Klein + Hummel is jusing waveguides for their studio monitors to avoid reflections from the mixer rack in sound studios.
So my main goal is shaping the directivity - less acoustic energy for the room to reduce hall effects and get a rather neutral behavier while increasing a homogeneous directivity in horizontal space.
Reducing harmonics through less motion of the driver is fine for me as well. Lower corner frequency is nice to have.
What is not critical for me is a smooth frequency response as I can correct that.
What approach can you recommend?
In my experience bigger is better. You will get an effect from even a small waveguide, as noted, but the larger the device the more well controlled the directivity will be and the lower in frequency and the greater gain you will achieve - all desirable things.
The cabinet is not a strong factor on directivty except for a surface mounted dome , in which case the edge diffraction will affect the directivity. Even a small waveguide helps this scenario.
To be clear, I would ALWAYS use a waveguide as a surface mounted tweeter is a bad idea. How large and deep it is depends on what you are trying to achieve (looks or performance, slight improvement or tight control, ease of fabrication, etc.) But for the record, waveguides look "good" to me, not ugly as is being implied above, but appearance truely is a subjective thing.
The cabinet is not a strong factor on directivty except for a surface mounted dome , in which case the edge diffraction will affect the directivity. Even a small waveguide helps this scenario.
To be clear, I would ALWAYS use a waveguide as a surface mounted tweeter is a bad idea. How large and deep it is depends on what you are trying to achieve (looks or performance, slight improvement or tight control, ease of fabrication, etc.) But for the record, waveguides look "good" to me, not ugly as is being implied above, but appearance truely is a subjective thing.
Dr. Geddes will speak more appropriately about this I'm sure. But what I see with the graphs of the shallow waveguide, and is consistent with my own tests, is that it doesn't improve directivity low enough. It drops the point from roughly 8K to 4k, but you need to have control down to below the crossover point. While the shallow guide helps, clearly, and could even lower distortion a little bit, its going to offer no appreciable gain (thus no appreciable lowering of distortion) as well as offer no control of the directivity, where the tweeter will act essentially omni-directionally. It makes the solution a sort of "Better than nothing, maybe?" type solution, rather than an actual viable alternative to large waveguide designs. My experience has been that the lower the crossover point (while avoiding compression distortion and all listening levels) the better. I find the speakers have a better polar response, have a much better sound stage, and sound more coherent. With my focal designs, just dropping the crossover point from 3.4khz to 1.8khz had a drastic effect in this regard, but the focal can not play loud enough at 1.8khz without audible power compression. A waveguide would need to be roughly 8-10" in diameter and pretty deep to offer enough gain to say 1.5khz in order to make that work. Again, I would also want pattern control to the crossover point, and it would have to be at least that large (Elliptical would probably be more like 8" wide by 6" tall).
I will try and post results when I can, but I have made a qaveguide similar to dlneubec's idea, using rings of MDF. I made mine with a 2" roundover bit and a 3/4" roundover bit. The entry is rounded over with the 3/4", then the 2" makes up the rest. I also tried to flatten the mid-section flare a little via putty and sanding. The diameter is 5" at the flare exit point, and the depth is I believe 4". I was hoping for a more compact horn that would fit in a speaker using 6" drivers, but alas, it did not offer the gains or pattern control I needed (only good to 3khz).
I will try and post results when I can, but I have made a qaveguide similar to dlneubec's idea, using rings of MDF. I made mine with a 2" roundover bit and a 3/4" roundover bit. The entry is rounded over with the 3/4", then the 2" makes up the rest. I also tried to flatten the mid-section flare a little via putty and sanding. The diameter is 5" at the flare exit point, and the depth is I believe 4". I was hoping for a more compact horn that would fit in a speaker using 6" drivers, but alas, it did not offer the gains or pattern control I needed (only good to 3khz).
Some years ago I had seen a horn design made of MDF rings for a spherical wave horn. So in principle I can also increase the depth for the elliptical type of horn without using a much thicker fron material.
But as you described by lowering the crossover point you had compressions - what order of your passive filter are you using? I guess that these compressions rather came from unadequate dampening of the low frequency signal. If the horn is too short, the adaption of acoustic impedance is too bad for these low frequencies and the driver will make excessive move without sound pressure.
I am intending to use 10th order digital filter. So the remaining energy below the cuttoff point will be neglectable in my case and so I can extend the frequency range downwards.
But as you described by lowering the crossover point you had compressions - what order of your passive filter are you using? I guess that these compressions rather came from unadequate dampening of the low frequency signal. If the horn is too short, the adaption of acoustic impedance is too bad for these low frequencies and the driver will make excessive move without sound pressure.
I am intending to use 10th order digital filter. So the remaining energy below the cuttoff point will be neglectable in my case and so I can extend the frequency range downwards.
I would agree with Matt that getting the crossover point down is a good idea. I would never use one above 2 kHz and I prefer it to be below 1 kHz. Only the larger waveguides will allow this for all the reasons that he points out.
Regarding crossover slopes you have to remember that excursion is a linear thing in its effects, not logarithmic. Thus once you are more than 10 dB down there is very little improvement that would come from a sharper filter as regards excursion. I would not expect a 10th order filter to be even a marginal improvement over a second order one.
Regarding crossover slopes you have to remember that excursion is a linear thing in its effects, not logarithmic. Thus once you are more than 10 dB down there is very little improvement that would come from a sharper filter as regards excursion. I would not expect a 10th order filter to be even a marginal improvement over a second order one.
An article I wrote regarding shallow dome driven waveguides is here
http://sound.westhost.com/articles/waveguides1.htm
The formula given for the two section axis symmetric type are reliable predictors of the -3db. directivity frequency and can be used for such as 90 x 40 wave guides.
In general the depth is determined by the mouth area and the dome diameter.
Transforming a circle at the throat to an straight walled ellipse or a figure with a rectilinear center section and semicircular ends are both methods I have looked at, (although haven't had time to construct and test yet). The frequency response and directivity should not be much affected by the different geometry, although the later is easier to make since it is constructed of circles and straight lines.
The radiused mouth section can be something of a problem if you want to generate it with a router since the radius is not constant, a constant radius results in a ridge or a non flat mouth.
In general for crossovers bellow the 2kHz. region you need a dome with at least .5mm. linear excursion for the best results.
rcw.
http://sound.westhost.com/articles/waveguides1.htm
The formula given for the two section axis symmetric type are reliable predictors of the -3db. directivity frequency and can be used for such as 90 x 40 wave guides.
In general the depth is determined by the mouth area and the dome diameter.
Transforming a circle at the throat to an straight walled ellipse or a figure with a rectilinear center section and semicircular ends are both methods I have looked at, (although haven't had time to construct and test yet). The frequency response and directivity should not be much affected by the different geometry, although the later is easier to make since it is constructed of circles and straight lines.
The radiused mouth section can be something of a problem if you want to generate it with a router since the radius is not constant, a constant radius results in a ridge or a non flat mouth.
In general for crossovers bellow the 2kHz. region you need a dome with at least .5mm. linear excursion for the best results.
rcw.
Thanks Robert for the good article!
What I expect from my so far experiences and knowledge is:
The influence of the acoustic impedance for the driver will make no big difference, if the horn mouth ends in a round or elliptic shape - important is that the surface area has the same growing function. Correct?
The same acoustic impedance will result in the same excitation and thus in the same frequency response (on axis). Only the directivity is changed as the sound will be more or less bundled in to the room. Also Correct?
So in principle I can use the same functions as you described and modify the growing function in x and y but keeping the surface area in z equal to the sperical horn?
I know I have to care about the edges more that on a spherical horn as they end differently in x and y dimension.
Using a higher order cutoff has two main advantages:
1. Reduce the frequency bandwith in which more than one speaker is active (in cross over band). Two sound sources will create acoustic interferences as their amplitude and phase will be totally different.
2. Tweeters are normally used above their resonance frequency. A low order filter will let pass residual low frequency components to the tweeter and thus create excitation arround the resonance frequency to the tweeter. Although these components are not audiable from the tweeter it will generate excessive movement and thus additional distortion to the HF signal.
-> Additionally the tweeter will generate Doppler modulation to the HF signals...
What I expect from my so far experiences and knowledge is:
The influence of the acoustic impedance for the driver will make no big difference, if the horn mouth ends in a round or elliptic shape - important is that the surface area has the same growing function. Correct?
The same acoustic impedance will result in the same excitation and thus in the same frequency response (on axis). Only the directivity is changed as the sound will be more or less bundled in to the room. Also Correct?
So in principle I can use the same functions as you described and modify the growing function in x and y but keeping the surface area in z equal to the sperical horn?
I know I have to care about the edges more that on a spherical horn as they end differently in x and y dimension.
Using a higher order cutoff has two main advantages:
1. Reduce the frequency bandwith in which more than one speaker is active (in cross over band). Two sound sources will create acoustic interferences as their amplitude and phase will be totally different.
2. Tweeters are normally used above their resonance frequency. A low order filter will let pass residual low frequency components to the tweeter and thus create excitation arround the resonance frequency to the tweeter. Although these components are not audiable from the tweeter it will generate excessive movement and thus additional distortion to the HF signal.
-> Additionally the tweeter will generate Doppler modulation to the HF signals...
rcw said:
Could you elaborate on what you mean here?
Aoxomox said:
Again, the difference bewteen a lower order filter and a higher order one in this regard are minimal. I see no reason to avoid the tweeters resonance. They are usually well damped and don't see large excursion increases at resonance. Its all about total excursion - in linear values not dB. As long as this does not exceed the capabilities of the unit your fine. With most dome tweeters however this is quite limited, which is why I use compression drivers.
I use a B&C DE250. I've tried the BMS coaxials and don't like them. The two diaphragms don't mix well at the crossover which creates some bad polars at that frequency, which unfortunately is right in our hearings critical range, about 6 kHz. The DE250 is very well controlled up to about 13 kHz, at which point it has a sharp diaphragm resonance. But at this frquency I don't really care.
I will add to Dr. Geddes comment, that regardless of slope, with my particular driver, 1.8khz was just too low. I have modeled the excursion of the tweeter with all sorts of filters and find that, if anything, it makes it worse as the excursion is at its worse around the crossover point. Instead of having a more gradual attentuation of output and thus excursion starting above the crossover point on down, its sharp and narrow, offering little aid in controlling excursion at that point. If the tweeter could play to 1.8khz without issue (focal tweeters are known to have very little excursion) then this probably wouldn't be an issue, but then, I don't see how a 10th order filter would add to anything either. I actually find it quite hard to get good driver integration with such sharp slopes (I've done a lot of work with elliptical slopes), but to each their own.
I'm not sure how others feel about this, but if the crossover is low enough, I don't find a little overlap between drivers objectionable. The integration of drivers tends to be smooth and consistent, the polar response is good, and I hear no signs of poor driver to driver consistency. However, as we go up in frequency, I do find this to be an issue, and steeper slopes do seem to be beneficial. I just try to avoid these areas if I can. Of course, it all depends on your application and design, as I have one right now that isn't avoid this area, and so I use 4th order slopes, but I don't find such designs to sound good with music either.
I'm not sure how others feel about this, but if the crossover is low enough, I don't find a little overlap between drivers objectionable. The integration of drivers tends to be smooth and consistent, the polar response is good, and I hear no signs of poor driver to driver consistency. However, as we go up in frequency, I do find this to be an issue, and steeper slopes do seem to be beneficial. I just try to avoid these areas if I can. Of course, it all depends on your application and design, as I have one right now that isn't avoid this area, and so I use 4th order slopes, but I don't find such designs to sound good with music either.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.