Hi,
Recently I've locked into the lobing issues for a normal multiway speaker. I've written a Matlab program that can simulate the vertical polar pattern and frequency response at any off-axis angle for a TM, MTM or TMM configuration. Crossover point, slope and phase difference between driver can be arbitrary chosen.
All configurations exhibit serious off-axis irregularities, where MTM and TMM are worst. Even with a 24dB crossover slope the off-axis response of a TM have a dip of at almost 60dB (at 30deg). The depth is mathematical and in real life it will be less, but some 10dB at least is to be expected I guess. At 20 and 40 deg off-axis the dip is some 7dB. The case I simulated had a crossover point at 2khz and the width of the dip is about a half octave, with 12dB slope it becomes an octave wide.
MTM configuration is best when following d'appolitos rules by having one wavelength between adjacent drivers and third order filter. This way I get a quasi spherical vertical dispersion pattern at crossover frequency, but the lobing at lower frequency gets worse due to the 2 M's. In my case, with crossover at 2khz, I get a 15dB deep dip centered at about 1khz which is about one octave wide. At large off-axis angles (60deg) irregularities at crossover point is starting to appear aswell.
Now you think, who is listening so much off axis? Well, at least in my room the ceiling bounce will be at approximately 60deg vertical off-axis. This bounce will be almost perfect since my ceiling is of concrete. With a softer ceiling you might get some loss but it can't be much. I also have a floor bounce at about 30deg but there I have a large table...
For a TM config the center frequency of the dip is the same regardless of off-axis degree, only the depth change. For an MTM config the frequency change somewhat but not to the extent where one can expect smoothing of the dips.
No matter how I look at the problem there HAVE to be some serious midrange distortion due to this for all normal multiway systems.
For obvious reasons a fullrange driver will not have this problem. Dispersion at higher frequencies may be worse than a tweeter, but the midrange is intact, all reflections are (more or less) undistorted in the midrange.
A very common comment of fullrange driver is "lovely midrange"... I strongly believe it has mostly has to do with lack of off-axis irregularities and not so much with crossover phase shift, doppler distortions or similar theories.
I’m on the verve of putting together a active filtered MTM speaker using two Seas L15 and a Scan-speak tweeter crossed at 2khz. I’m seriously considering exchanging the tweeter for a Fostex FE103E fullrange and lowering the crossover point to 500hz. This way I get almost no off-axis irregularities…
Please comment on this!
Regards Jesper
Recently I've locked into the lobing issues for a normal multiway speaker. I've written a Matlab program that can simulate the vertical polar pattern and frequency response at any off-axis angle for a TM, MTM or TMM configuration. Crossover point, slope and phase difference between driver can be arbitrary chosen.
All configurations exhibit serious off-axis irregularities, where MTM and TMM are worst. Even with a 24dB crossover slope the off-axis response of a TM have a dip of at almost 60dB (at 30deg). The depth is mathematical and in real life it will be less, but some 10dB at least is to be expected I guess. At 20 and 40 deg off-axis the dip is some 7dB. The case I simulated had a crossover point at 2khz and the width of the dip is about a half octave, with 12dB slope it becomes an octave wide.
MTM configuration is best when following d'appolitos rules by having one wavelength between adjacent drivers and third order filter. This way I get a quasi spherical vertical dispersion pattern at crossover frequency, but the lobing at lower frequency gets worse due to the 2 M's. In my case, with crossover at 2khz, I get a 15dB deep dip centered at about 1khz which is about one octave wide. At large off-axis angles (60deg) irregularities at crossover point is starting to appear aswell.
Now you think, who is listening so much off axis? Well, at least in my room the ceiling bounce will be at approximately 60deg vertical off-axis. This bounce will be almost perfect since my ceiling is of concrete. With a softer ceiling you might get some loss but it can't be much. I also have a floor bounce at about 30deg but there I have a large table...
For a TM config the center frequency of the dip is the same regardless of off-axis degree, only the depth change. For an MTM config the frequency change somewhat but not to the extent where one can expect smoothing of the dips.
No matter how I look at the problem there HAVE to be some serious midrange distortion due to this for all normal multiway systems.
For obvious reasons a fullrange driver will not have this problem. Dispersion at higher frequencies may be worse than a tweeter, but the midrange is intact, all reflections are (more or less) undistorted in the midrange.
A very common comment of fullrange driver is "lovely midrange"... I strongly believe it has mostly has to do with lack of off-axis irregularities and not so much with crossover phase shift, doppler distortions or similar theories.
I’m on the verve of putting together a active filtered MTM speaker using two Seas L15 and a Scan-speak tweeter crossed at 2khz. I’m seriously considering exchanging the tweeter for a Fostex FE103E fullrange and lowering the crossover point to 500hz. This way I get almost no off-axis irregularities…
Please comment on this!
Regards Jesper
Your math seems good, but are the conclusions correct? The lobing characteristics of certain configurations (MTM, dipole come to mind) are often touted as benefits. The ultimate FR (_at the listening position_) of ambient (off axis) radiation is also going to be heavily dependent on room treatment and geometry.
I would posit that given a suitably wide sweet spot it is more important to have a smooth transition to the off-axis FR than to meet some artificial ideal.
I would posit that given a suitably wide sweet spot it is more important to have a smooth transition to the off-axis FR than to meet some artificial ideal.
jesper said:
This is interesting! A collegue of mine has done similar matlab measurements and simulations and also came up with narrow dips off axis near the crossover, as I recall it. His system is a two-way with a crossover near 2kHz I think.
I also agree with you that the comparison with dipoles is not really valid, since the dipole has a large frequency range in which it is directed (correct me on this if I am wrong. Is there maybe a range some 0.3-2kHz when it isn't directed?).
As I recall, the recommendation with regards to placement of the drivers in an MTM configuration was less than 1/2 or 1/4 (can't remember) wavelength. The idea is that this causes line source behaviour.
Have you listened to an electrostatic panel before? I think you will find that many such panels have narrow frequency variations of as much as 20dB in the ON-AXIS response (not to mention off-axis
)..
The matter of reflections is dependent on the distance to the ceiling, walls and floor. Depending on the timing of a reflection, it may be perceived as (a) part of the sound, (b) part of the ambience, or (c) an echo.
Have you listened to an electrostatic panel before? I think you will find that many such panels have narrow frequency variations of as much as 20dB in the ON-AXIS response (not to mention off-axis
)..The matter of reflections is dependent on the distance to the ceiling, walls and floor. Depending on the timing of a reflection, it may be perceived as (a) part of the sound, (b) part of the ambience, or (c) an echo.
I've listen to electrostatic panels, and they sure do beam. The dips at low angles, say 10-20 degrees are not my concern, it only gives a narrow sweet spot, reflections are mainly from the back wall and bounce away from the listner. It's the 30-60 degrees which bounce off the ceiling and floor that are bugging me.
Electrostatic panels beam over a wide frequency since they are panels and dipoles, they have a "figure eight" radiation pattern. They should have little ceiling bounce. Line source behave similar, beam over a wide frequency range.
TM and MTM are mostly omnipolar in the midrange.
High frequency floor bounce can probably be perceived as ambience. The wavelength of a 2khz wave is 0.17m. The length difference between direct wave and ceiling bounce in my room is maybe 2-3m, some 12-17 wavelength, it might be enough to be perceived as ambience, I don't know. Does anyone have a good figure? For the 1khz dips in the MTM though it's only 6-9 wavelength....
/Jesper
Electrostatic panels beam over a wide frequency since they are panels and dipoles, they have a "figure eight" radiation pattern. They should have little ceiling bounce. Line source behave similar, beam over a wide frequency range.
TM and MTM are mostly omnipolar in the midrange.
High frequency floor bounce can probably be perceived as ambience. The wavelength of a 2khz wave is 0.17m. The length difference between direct wave and ceiling bounce in my room is maybe 2-3m, some 12-17 wavelength, it might be enough to be perceived as ambience, I don't know. Does anyone have a good figure? For the 1khz dips in the MTM though it's only 6-9 wavelength....
/Jesper
jesper said:
Floor and ceiling bounce occur to early to be perceived as ambience although the ceiling bounce probably lends as sense of height to the presentation. Many believe that reflections occuring within the first 15 to 25 milliseconds are destructive to the first arrival signal and should be absorbed by placing acoustic panels at the appropriate positions.
It all comes down to how we perceive sound and when we "hear" the combfilter effect caused by superimposed waves.
No matter where we are in a room we all sound pretty much the same when talking, there's not much perceived distortion even though there is considerably measurable destructive interference, due to bounce off nearby walls.
I think two sound sources have to be pretty much in minimum phase difference in order to be perceived as one, distorted source. In the bassarea most reflections are minimum phase. Offcasue in the bassarea we also have standing waves to a large extent.
The 2khz ceiling bounce in my room is some 12-17 wavelength delayed from the on-axis one. I think this is perceived as another sound source, at least from lower midrange and up. This source is attenuated, colored in the midrange and has less treble.
/Jesper
No matter where we are in a room we all sound pretty much the same when talking, there's not much perceived distortion even though there is considerably measurable destructive interference, due to bounce off nearby walls.
I think two sound sources have to be pretty much in minimum phase difference in order to be perceived as one, distorted source. In the bassarea most reflections are minimum phase. Offcasue in the bassarea we also have standing waves to a large extent.
The 2khz ceiling bounce in my room is some 12-17 wavelength delayed from the on-axis one. I think this is perceived as another sound source, at least from lower midrange and up. This source is attenuated, colored in the midrange and has less treble.
/Jesper
Just made some measurements on a friends 2-way bookshelf (TM) and my two Seas L15 in a M-M config, and the result correlate very well with my simulated result. There are a 10-12dB dip at approx 3khz in the TM at 30 degrees off axis. At other degrees the dip is noticeable but not so pronounced (some 5dB). Treble start to rolloff at approx 10khz.
The M-M conig have a very pronounced dip at about 1khz of some 10-15dB. The dip doesn't change much from 30-60 degrees off-axis.
So, it's true that a MTM config gives less floor and ceiling reflections, but only so in a very limited frequency range... Anything else is just marketing.
I'd like to have no ceiling reflection, but since it can't be avoided (unless you go for line source or dipole), I prefer undistorted midrange ceiling reflection. This means a fullrange driver instead of a tweeter, crossed over at 500hz.
/Jesper
The M-M conig have a very pronounced dip at about 1khz of some 10-15dB. The dip doesn't change much from 30-60 degrees off-axis.
So, it's true that a MTM config gives less floor and ceiling reflections, but only so in a very limited frequency range... Anything else is just marketing.
I'd like to have no ceiling reflection, but since it can't be avoided (unless you go for line source or dipole), I prefer undistorted midrange ceiling reflection. This means a fullrange driver instead of a tweeter, crossed over at 500hz.
/Jesper
Inhouse Line Array?
I had in mind to combat those problems by building an experimental small line array using 2x5.5'' + Ribbons in an 8+8 small cabinet L&R ceiling suspended stack. 10'' cardioid subs could be providing the lows. It remained an idea due to practical considerations. Maybe I will try it if some brave celibate friend with a big livingroom is going to volunteer at a point...Chinese ribbons are about $ 90 these days retail. A discount is very probable if to buy 16 + 32 midbasses....anybody tickled?
I had in mind to combat those problems by building an experimental small line array using 2x5.5'' + Ribbons in an 8+8 small cabinet L&R ceiling suspended stack. 10'' cardioid subs could be providing the lows. It remained an idea due to practical considerations. Maybe I will try it if some brave celibate friend with a big livingroom is going to volunteer at a point...Chinese ribbons are about $ 90 these days retail. A discount is very probable if to buy 16 + 32 midbasses....anybody tickled?
I was thinking of building 3 way speakers with a pair of 4 inch tang band drivers covering 250-8000hz, a scan speak tweeter for 8-20khz, and a pair of 10 inch drivers covering 30-250hz. I want to use such a high 8000hz xo frequency as 8000hz is about 10dB lower then 2k on the equal loudnesss contours.
I am really concerned with lobing effects. What would be the best driver arrangement? I was thinking MTMWW or TMMWW.
Thanks
Alex
I am really concerned with lobing effects. What would be the best driver arrangement? I was thinking MTMWW or TMMWW.
Thanks
Alex
jesper said:
1, You can damp out the ceiling reflection, and BTW floor and side wall reflections are just as damaging to the image. A 1/3 octave measurement at the listening position will reveal to some extent how much the colored off-axis sound is a problem
2. You can use a coaxial speaker and you will get no lobing problems at all.
3. If you are really concerned about directivity, look into high quality constant directivity horns. In a reasonably large room, they can be very hi-fi.
rick57 said:
There may not be any public domain formulae, but there are known formulae. All the math for his waveguide is in Earl's AES paper. I used Excel to model it against Charles Hughes' Quadratic Throat Waveguide (there isn't much difference - and there is a typo in the latter that made it tricky!) which also has published math. I believe all the math is there in Keele's and Ureda's papers too. None of it is easy reading, but if you have the desire you can figure out how to plot the curves.
I've been woried about lobing errors also. All this is very interesting.
I found an acoustic simulator that is sum what helpful in getting a feel for these sort of problems. It's a 2D wave tank JAVA applet http://www.falstad.com/mathphysics.html. there is alot of cool stuff there.
One thing I began to realize is that at these freqs. and size of drivers (5.5 ,6.5, 7 ect) and the listening distances, that the mid drivers are not point sources. With a sourse hight of, say 6.5 inches the the lobing error problems of two mids about 6 inches apart tends to diffuse the lobing errors some what.
So sticking with one T and two M's whats the best solution?
I found an acoustic simulator that is sum what helpful in getting a feel for these sort of problems. It's a 2D wave tank JAVA applet http://www.falstad.com/mathphysics.html. there is alot of cool stuff there.
One thing I began to realize is that at these freqs. and size of drivers (5.5 ,6.5, 7 ect) and the listening distances, that the mid drivers are not point sources. With a sourse hight of, say 6.5 inches the the lobing error problems of two mids about 6 inches apart tends to diffuse the lobing errors some what.
So sticking with one T and two M's whats the best solution?
rick57 said:
Well, at the risk of one more slightly thread-jacking post I'll chime in. I think that discussion of controlled directivity is actually rather on-topic, as it is one of the best ways to solve the problem the original poster is asking about.
Briefly, *any* CD horn reduces to conical as you move away from the throat, and in fact the easiest and most common way to do a CD horn is simply to make it straight-sided. Lots of folks do this and it works OK, particularly in cone-driven mids.
Earls work is aimed at achieving the best possible pattern control, and a large part of that is avoiding discontinuities in the flare, particularly right at the throat. Thus, his OS waveguide is a forumlation that matches a conical flare at the mouth to the exit geometry of the driver at the throat.
There is really no such thing as a CD bass horn. Pattern control starts to break down where the dimensions of the device become comparable to the wavelengths involved, which for real-world devices that fit in the home is in the 250-500Hz range. Below this things go omni and there isn't much you can do other than ensure that there is a smooth transition from controlled to omni radiation.
BTW - I believe that Earl said in another forum that he never patented the OS waveguide forumlation. As a private user/DIYer it doesn't matter though, as patents only apply to commercial use. You are free to copy/clone anything you want as long as it's for your own personal use.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.