soongsc said:
The acoustical impedance of a straight tube is constant with frequency when the tube is infinitely long, or damped to supress the reflected wave. If the tube is finite, it has to be well terminated over the entire frequency range to avoid reflections and resonances.
Adding a non-flaring tube to the throat of a horn will lower the cutoff of the system, but it can easily introduce lots of reflections. And it gets worse the longer the tube is. A very short tube can have beneficial effects on the throat impedance, but you can't increase the length very much before running into trouble. You kind of get the quarter-wave resonance of the tube superimposed on the throat impedance of the horn.
You can play with Hornresp to see the effects. Even if Hornresp can't simulate multiple-segment horns incorporating OSWG segments, you can approximate the shape with a short exponential horn in front of a cone. First try to get close to the throat impedance of an OSWG alone, then add the tube in front of the combination.
regards,
Bjørn
Kolbrek said:
Of course a flared mouth will improve things. I would never do a waveguide without one. Is the throat intended to be 1.4"? This will raise the "cutoff" (if one insists on using this term). The 1" will go down to about 700 Hz.
If you add the driver flare, which is fair, you really do need to add the whole driver model itself since it is a highly coupled system. I do see a resonance about where you show it, but never as great as what you show. And the axial hole is there but never as deep as you show it. With a adequitly flared mouth, and a better driver model your BEM predictions would match what I see pretty accurately.
The improved directivity is obvious. I'm surprised that you didn't mention that
Hello Bjorn,
Thank you for those meaningful results.
The reponse of the waveguide as you shown
http://www.diyaudio.com/forums/attachment.php?s=&postid=1636476&stamp=1224494870
should be compared with the response of smoothest response of the AH425:
http://www.diyaudio.com/forums/attachment.php?s=&postid=1634455&stamp=1224236330
Most probably someone will say that those holes and peaks in the OS waveguide response are not audible...
Best regards from Paris, France
Jean-Michel Le Cléac'h
Thank you for those meaningful results.
The reponse of the waveguide as you shown
http://www.diyaudio.com/forums/attachment.php?s=&postid=1636476&stamp=1224494870
should be compared with the response of smoothest response of the AH425:
http://www.diyaudio.com/forums/attachment.php?s=&postid=1634455&stamp=1224236330
Most probably someone will say that those holes and peaks in the OS waveguide response are not audible...
Best regards from Paris, France
Jean-Michel Le Cléac'h
Kolbrek said:
Earl,
OK may we ask Bjorn to do some more homework and put a termination at the mouth of the waveguide with
radius edge equal to 1/4 wavelength at 2000Hz (or less if you want...).
You'll notice that on the response of the few Le Cleac'h horns Bjorn simulated we never can find the hole in the axial reponse that John Sheerin described and that you explained as HOMs in the Le Cléac'h horn...
Best regards from Paris, France
Jean-Michel Le Cléac'h
OK may we ask Bjorn to do some more homework and put a termination at the mouth of the waveguide with
radius edge equal to 1/4 wavelength at 2000Hz (or less if you want...).
You'll notice that on the response of the few Le Cleac'h horns Bjorn simulated we never can find the hole in the axial reponse that John Sheerin described and that you explained as HOMs in the Le Cléac'h horn...
Best regards from Paris, France
Jean-Michel Le Cléac'h
gedlee said:Not a fair comparison since the OS was abruptly terminated in an infinite baffle with no flare and the AH425 was not. Lets at least "try" and be fair OK? This discussion is starting to sound "a little" one sided
Jean-Michel
You misquoted me as usual.
The axial "hole" is not due to HOM within the device, but is in fact due to diffraction at the mouth termination. Only a very large radius can make it go away, but any radius makes it less seveer. AND, lest you forget, I do not recommend sitting on axis to speakers and hence this hole is not within the listening area. Its effect on the power response is minimal due to it very small spatial extent. This hole is not seen in your devices simply because you have a very large mouth flare radius. But if an exponential horn is terminated abruptly into a flat baffle, it too will exhibit the axial hole. My forthcoming Audio-X-Press article will show this.
I would suggest a radius of 4 " at the mouth since that is what I use and there is lots of data available for a comparison.
You also failed to note the marked improvement in the directivity of the waveguide. So the question continues to be: Is the hole more or less audible than a lack of constant directivity? I would contend, and most researchers would agree, than constant directivity is the better tradeoff. You, of course, will disagree. This is a very old discussion.
I am building an elliptical waveguide that will not have this hole. Not that it will have any signifcant audible benifits mind you.
You misquoted me as usual.
The axial "hole" is not due to HOM within the device, but is in fact due to diffraction at the mouth termination. Only a very large radius can make it go away, but any radius makes it less seveer. AND, lest you forget, I do not recommend sitting on axis to speakers and hence this hole is not within the listening area. Its effect on the power response is minimal due to it very small spatial extent. This hole is not seen in your devices simply because you have a very large mouth flare radius. But if an exponential horn is terminated abruptly into a flat baffle, it too will exhibit the axial hole. My forthcoming Audio-X-Press article will show this.
I would suggest a radius of 4 " at the mouth since that is what I use and there is lots of data available for a comparison.
You also failed to note the marked improvement in the directivity of the waveguide. So the question continues to be: Is the hole more or less audible than a lack of constant directivity? I would contend, and most researchers would agree, than constant directivity is the better tradeoff. You, of course, will disagree. This is a very old discussion.
I am building an elliptical waveguide that will not have this hole. Not that it will have any signifcant audible benifits mind you.
mige0 said:Lynn - you will have to do *all* audition with your eyes closed
Michael
Sadly, my own experience is that the most beautiful loudspeakers don't sound very good - the JBL Ranger-Paragon & L100s, the Acapella's, and a bunch of others. Conversely, some of the best-sounding DIY systems left like a collection of leftovers from the movie sets of Frankenstein and The Matrix. So if I have any visual bias, it's probably in the other direction.
Kolbrek said:
I'm glad you published the simulations - thank you very much! They reflect e-mails that we exchanged privately several months ago:
1) What would the impact of the 288 internal throat be on the complete system?
2) What horn profile would be the best overall choice, given that most compression drivers have slightly flared throat sections between the phase plug and the exit from the driver?
There are a few drivers that claim to have no internal throat sections, but I don't want to be limited to just JBL - besides, JBL's are not my first choice on purely subjective grounds.
So the internal throat of the compression driver has to become part of the overall horn design, like it or not. As we know, sharp changes in profile create acoustic misterminations (reflections) that cannot be corrected electronically, since the error is in both time and spatial domains.
As a direct-radiator guy, I'm aware of the bad effects of cabinet-edge reflections; in horns, the problem is much, much worse, affecting not only imaging but tonality, perceptions of harshness, and outright "horn" coloration in the most severe cases. As I see it, acoustic misterminations (and internal reflections) and problems they create in the time domain are the most serious problems of horn loudspeakers.
Internal reflections have to be kept to a minimum for the complete system, from the back plate, to the diaphragm, to the entrance and exit of the phase plug, the internal driver flare, any potential misalignment between the driver and mounting plate of the horn, the internal profile of the horn, and the termination between the horn and the free air of the room. MLS techniques allow a pretty detailed examination of the reflections, so they can be examined, and if possible, removed.
That's why I'm going to be looking at impulse response vs angle pretty carefully, and seeing if some angles have more rapid decays than others. Impulse vs emission angle is almost never specified for professional loudspeakers, so I'm curious what I'll find. For example: the simulation of the AH-425 shows smoothest response 30 degrees off-axis; in the time domain, though, the answer may be different, and that will take direct measurement. In addition, the 18Sound XT1464 sound-reinforcement horns should be a useful cross-check against the AH-425 Azurahorns.
Bjorn, when your computers cool down, I'm curious how the AH-425 would have turned out with a T=1.34, instead of the T=0.707 that I selected. The higher T ratios come closer to a conical profile, with a different set of tradeoffs in loading vs dispersion.
P.S. The TC14WG49-08 has a large peak at the top of the working range - this is what I try to avoid in my loudspeakers, unless the driver has a set of sterling virtues that offset the hassle of the notch filter. The Ariel was intended to a minimalist loudspeaker, with a simple crossover that took advantage of drivers that didn't need equalization. Once you get into equalization and notch-filtering, it's not an Ariel anymore, but a different kind of system.
"The JBL 243X drivers still have a flair through the phase plug (based on my measurements), even though there is no 'throat section'."
Hello John
Where are they hiding it?? If you look at a driver with the screen and thermally conductive gasket removed you can see that the screen rests on the top of the phase plug.
Rob
Hello John
Where are they hiding it?? If you look at a driver with the screen and thermally conductive gasket removed you can see that the screen rests on the top of the phase plug.
Rob
Attachments
gedlee said:
I used a 1.4" throat because I coupled the waveguide to the 288B internal flare. And I selected the 288B internal flare to make it easier to compare the combination to the AH-425+288B combination, which I have simulated earlier.
But I thought increasing the throat diameter lowered the "cutoff" (or lowest usable frequency), following your equation (23) in your 1989 AES paper. This equation gives fc as 44*sin(half angle) / (throat radius). Is this equation wrong?
To see the system response, yes. What I show now is the transfer function from throat to a point 3m from the mouth, the horn trans-impedance - velocity in, pressure out. This of course exaggerates some things, and you don't get the mass roll-off. It is not the frequency response of the system, I think I have quite clearly stated the simulation conditions.
Not surprising. The 288B have a 64mm long internal flare, not including the phase plug. Modern drivers have shorter internal flare, pushing the first resonance upwards. This clearly reduces the problem.
I can re-run the simulations with a 4-in radius flared mouth, if you wish.
As I have stated several times, there is no driver model used in the BEM simulations. The "frequency response" plots shows the pressure at the field points for a constant velocity at the throat. Given the throat impedance, the response for any driver coupled to the horn can be found.
The improved directivity is obvious. I'm surprised that you didn't mention that
But as you say, it is obvious
Besides, this was not the question. Soongsc asked what happened with a tube in front of the waveguide. I presented two simulations, with and without a slowly expanding throat section to show the difference. This is also why I didn't add a mouth radius.I would very much like to run a simulation of the complete system, including driver model, internal flare, waveguide and mouth radius. It would be great if you could please send me the details, privately and confidential if you wish, you have my e-mail address. Since you have so much measured data, it would be an excellent way of verifying the model.
Best regards,
Bjørn
Hello,
Do I misinterpret you when regarding the classical graph showing the effect of the reflections and diffraction inside the conical horn:
http://www.diyaudio.com/forums/attachment.php?s=&postid=1590534&stamp=1219285883
You wrote:
"Yes, a classic case of HOMs at their worst."
http://www.diyaudio.com/forums/showthread.php?postid=1590563#post1590563
Best regards from Paris, France
Jean-Michel Le Cléac'h
Do I misinterpret you when regarding the classical graph showing the effect of the reflections and diffraction inside the conical horn:
http://www.diyaudio.com/forums/attachment.php?s=&postid=1590534&stamp=1219285883
You wrote:
"Yes, a classic case of HOMs at their worst."
http://www.diyaudio.com/forums/showthread.php?postid=1590563#post1590563
Best regards from Paris, France
Jean-Michel Le Cléac'h
gedlee said:
Jean-Michel
Well you certainly misunderstood the comment.
The far field, as shown in Bjorn's work, and the nearfield or internal sound field of the horn, are not a one to one relationship. The conical horn plot as shown does show serious HOM and a serious diffraction at the mouth, but I don't see how this relates to your comment. Both problems exist in that plot, I commented on one of them. When there is a large reflection at the mouth, as well as a large diffraction at the edges, then these wave will propagate back down the device and become HOMs. This is precisely why the foam is such an advantage as it significantly attenuates these reflections and diffractions. But the axial hole is a far field issue and depends only on what the velocity distribution is in the mouth, not on what the HOM are doing within the device.
I hope that this is clearer now.
Well you certainly misunderstood the comment.
The far field, as shown in Bjorn's work, and the nearfield or internal sound field of the horn, are not a one to one relationship. The conical horn plot as shown does show serious HOM and a serious diffraction at the mouth, but I don't see how this relates to your comment. Both problems exist in that plot, I commented on one of them. When there is a large reflection at the mouth, as well as a large diffraction at the edges, then these wave will propagate back down the device and become HOMs. This is precisely why the foam is such an advantage as it significantly attenuates these reflections and diffractions. But the axial hole is a far field issue and depends only on what the velocity distribution is in the mouth, not on what the HOM are doing within the device.
I hope that this is clearer now.
Robh3606 said:"The JBL 243X drivers..."
Where are they hiding it?? If you look at a driver with the screen and thermally conductive gasket removed you can see that the screen rests on the top of the phase plug.
Rob
CLS said:
Correct. The exit of the driver is larger than the combined area of the phase plug slots near the diaphragm. The phase plug slots are really where the horn starts anyway - it's just built into the driver. What I've heard anecdotaly is that most CD's have this, but I've actually measured it on the 243X drivers.
Kolbrek said:
Bjorn
Yes you are correct, I did have that backwards, the 1" "cutoff" will be higher, but then it would appear that the simulations are off quite a bit at the lower end as I don't see this kind of "cutoff" in reality. But this could be a result of the driver coupling.
Your stated simulation assumption were clear, but I do disagree with "Given the throat impedance, the response for any driver coupled to the horn can be found. " as this would only be true if the system were uncoupled, i.e. there was insignificant energy exchange between the two systems. This is, of course true for direct radiators, but not for compression drivers. At the higher frequencies where the driver is mass loaded there probably is little coupling, but not at the lower frequencies. The driver and the waveguide are one coupled system at the lower edge and the impedance alone will not work with high accuracy.
I would love to see a simulation of my device. The first problem is obviously the sharp edge at the mouth - this is always a very bad idea and you have shown exactly why that is. The next would be to use a 1" throat, but this is probably less of a a factor than the driver coupling. I am not sure how Steven Kirkups software would handle a coupled matrix at the throat or if this could be done. In my simulations I use matrices throughout so the coupling comes naturally, but a two port matrix coupled to BEM may be difficult. I'd also use a more modern throat length.
Finally I would love to see some sims of the foam, particularly of what theory says should happen. The foam has a very complex function, but is a major audible effect as anyone who has tried it will tell you. The details of its function are obscured by the difficulty of analytical approaches and of measurements within it.
I can supply you with anything that you need. Just let me know. I have fairly accurate lumped parameter models of some compression drivers. These might be handy.
Over here I posted driver electrical impedance curves of a wave guide for free air direct radiating driver, 1/3 of wave guide with abrupt termination, and the full wave guide. I do not know if change in acoustic load would show up in the eletrical impedance or not, but the change in f0 of these 3 configurations seem interesting.
http://www.diyaudio.com/forums/showthread.php?postid=1624926#post1624926
It would be interesting to see if Lynn does any measurement on his horns.
http://www.diyaudio.com/forums/showthread.php?postid=1624926#post1624926
It would be interesting to see if Lynn does any measurement on his horns.
gedlee said:
The driver flare I used probably exaggerates the effect, compared to the internal flare in the drivers you use.
I view it like this: the horn throat impedance forms the load for the compression driver. For steady-state simulations, the average complex throat velocity and pressure can then be found if we know the throat impedance, the parameters of the driver, and the input voltage. Changes in the throat impedance will be reflected in the throat velocity, diaphragm displacement, electrical impedance and so on, so there is a coupling. But do you mean that the reflected wave, upon reaching the throat, should see an impedance equal to the source impedance of the driver, instead of the infinite impedance of the constant velocity surface? This could be arranged quite easily by changing the boundary conditions, I think.
These two points are trivial to change.
I don't think this can be done. The BEM SW needs the velocity and velocity potential to be specified at each boundary element. But if we can find the throat velocity, which I believe we can, given the throat impedance and driver parameters, I think this will get us where we want, unless we need to know the exact velocity distribution at the throat. In that case, it may be necessary to simulate the phase plug as a phase plug, drawing up the slits and all. That would of course be possible, but it requires quite a bit work with the BEM interface on my part. (Currently it generates a horn profile from a set of parameters). Hopefully we can get close enough, for a start at least, by using a plane (or curved) wave front with uniform velocity distribution at the throat.
I don't think this is possible with BEM. Perhaps FEM could do it, but with BEM it seems pretty hopeless.
Great
What I need to know is the following: -The parameters for the waveguide (throat diameter and angle, length from the throat to where the mouth radius starts, and the waveguide angle [theta0])
-The end areas and length of the internal driver flare, including the phase plug, and if the flare is conical or another shape
-The driver parameters, at least Sd, Bl, Cms, Rms, Mmd, Le and Re
-The volume and size of the rear chamber
-The volume and size of the front chamber
I assume the mouth flare radius is 4 inches, as per your earlier posts.
Best regards,
Bjørn