JoshK said:
Hi Josh
They were the same in the theater, except that the ear level is targeted at more like the middle of the two sources.
gedlee said:
Your critique of the $9.90 JBL PT waveguides you measured for us, topic of this thread, relies in large part upon your assertion of a design and performance superiority inherent in your axisymmetric approach.
Examined in detail, that assertion does not stand up quite so well to scrutiny as you might prefer, particularly in the light of recently posted vertical polar measurements, so it is understandable that you now desire to minimize its import.
Your designs are not a "target," rather, the standard for comparison you cite, and it is only reasonable that they be subject to analysis and discussion here....
" Horizontal orientation does nothing to affect the position of vertical lobes and nulls."
Obviously the null moves with the speaker, so are you saying the vertical lobe has no falloff to the sides?
Obviously the null moves with the speaker, so are you saying the vertical lobe has no falloff to the sides?
Asymmetrical verses axisymmetrical
I'm not even talking about Geddes speakers. I'm talking about round horns in general. Personally, I wouldn't use them and I've made my case why. But I have lots of friends that use round horns.
Earl uses round horns, so I would sort of consider him to be a "round horn guy", but that's not all there is to him. Like I said, I have a lot of other friends that use round horns too. I wouldn't make that choice, but I sure don't like how heated some of these discussions get. One of my best "audio buddies" is Duke LeJeune, the guy standing in the photos with Earl. We've been friends for a long time. He uses round horns, but said he started getting interested in CD horns after one of my Crossover Electronics 101 seminars. Seems like he told me he is using my crossover topology in his speakers too.
So all of us that do the uniform directivity thing are seeking essentially the same goals. We're kindred spirits, I think. We just have slightly different trade-off priorities. I think the round OS guys want reduced HOM more than they want good verticals, and the rectangular / elliptical guys want good verticals more than they want reduced HOM. I wouldn't say that either side totally dismisses the other, and both would prefer to have the lowest HOM and the best verticals too. It's sort of like how sometimes you can increase power and gas mileage, but usually you're slightly favoring one over the other.
This discussion isn't all about Earl. It's about horns verses waveguides, and to me it's also about symmetrical verses asymmetrical. My thinking is that I like the catenary flare, but that I also like a horn that isn't very tall. That means an elliptical would be a good way to go. Then again, if the elliptical has weird directivity in the vertical - if it loses it rapidly within the passband, then a shape that loses directivity in the vertical more gracefully, like maybe tractrix might be a good compromise. Catenary side walls, tractrix on the top and bottom. You can make this in an elliptical or rectangular shape.
There are a lot of bad choices, sure, but there are a lot of good choices too. Perhaps we should limit our discussion (and our tone) to those things we feel are good, the reasons why and the possible trade-offs. I don't think any of us is so arrogant as to think we have the monopoly on "absolute best".
I'm not even talking about Geddes speakers. I'm talking about round horns in general. Personally, I wouldn't use them and I've made my case why. But I have lots of friends that use round horns.
Earl uses round horns, so I would sort of consider him to be a "round horn guy", but that's not all there is to him. Like I said, I have a lot of other friends that use round horns too. I wouldn't make that choice, but I sure don't like how heated some of these discussions get. One of my best "audio buddies" is Duke LeJeune, the guy standing in the photos with Earl. We've been friends for a long time. He uses round horns, but said he started getting interested in CD horns after one of my Crossover Electronics 101 seminars. Seems like he told me he is using my crossover topology in his speakers too.
So all of us that do the uniform directivity thing are seeking essentially the same goals. We're kindred spirits, I think. We just have slightly different trade-off priorities. I think the round OS guys want reduced HOM more than they want good verticals, and the rectangular / elliptical guys want good verticals more than they want reduced HOM. I wouldn't say that either side totally dismisses the other, and both would prefer to have the lowest HOM and the best verticals too. It's sort of like how sometimes you can increase power and gas mileage, but usually you're slightly favoring one over the other.
This discussion isn't all about Earl. It's about horns verses waveguides, and to me it's also about symmetrical verses asymmetrical. My thinking is that I like the catenary flare, but that I also like a horn that isn't very tall. That means an elliptical would be a good way to go. Then again, if the elliptical has weird directivity in the vertical - if it loses it rapidly within the passband, then a shape that loses directivity in the vertical more gracefully, like maybe tractrix might be a good compromise. Catenary side walls, tractrix on the top and bottom. You can make this in an elliptical or rectangular shape.
There are a lot of bad choices, sure, but there are a lot of good choices too. Perhaps we should limit our discussion (and our tone) to those things we feel are good, the reasons why and the possible trade-offs. I don't think any of us is so arrogant as to think we have the monopoly on "absolute best".
Wayne,
Is that a response to my question?
You seem to be taking great pains not to answer it; I can only conclude that to do so would weaken your case against round horns.
Is that a response to my question?
You seem to be taking great pains not to answer it; I can only conclude that to do so would weaken your case against round horns.
breez said:
I'm not overly concerned about the group delay inherent in the filter. Let me describe what I think the issue is.
Suppose you have the system flat (amplitude and phase) on-axis. Then if you are off-axis vertically, the path length difference to the two sources will cause a relative phase shift between them. At the crossover point the phase of the sum will have to transition from the LF source phase to the HF source phase. The steeper the crossover the faster the transition, hence the bigger the spike in group delay.
Admittedly I don't know at what point this becomes a problem but I have to imagine that at some point it is. Suppose you use some real brick-wall filters. Then at more than say 10 degrees off-axis vertically you have a huge spike in group delay. At that point a huge portion of the radiated sound has a huge spike in group delay.
Ok, there's definately some speculation above, but that's the reason why I suggested a more moderate approach (say LR8 or LR16). With those you can make the holes in the amplitude response smaller without making huge spikes in the group delay off-axis. Thoughts?
Where are the nulls located in this ~9.5" c/c spaced 12" woofer implementation of the subject JBL waveguide? How wide is the vertical lobe?
http://www.jblpro.com/catalog/support/getfile.aspx?doctype=3&docid=665
http://www.jblpro.com/catalog/support/getfile.aspx?doctype=3&docid=665
noah katz said:
When you move real far to the sides, you'll see nulls appear out there too, at certain frequencies. This has more to do with the directivity of the direct radiating midwoofer, and is a function of its size.
The vertical nulls are all about path length differences between woofer and tweeter.
At the center of the forward lobe (which may or may NOT be along the baffle normal), the woofer and tweeter are exactly in phase with one another in the crossover band. Pick a frequency in the overlap band, and there will be a position (hopefully out in front of the speaker) where there is zero phase difference between woofer and tweeter. This marks the center of the forward lobe. In actuality, this position shifts a little through the overlap band because of the phase shift of the crossover. However, given a reasonably narrow overlap, the center of the forward lobe can be thought of as a fixed position.
Now move upwards from this position and you are getting further away from the woofer, closer to the tweeter. This can be expressed in percentages of a cycle. At high frequencies, it may even represent a multi-cycle shift, but for speakers like we are talking about, the crossover is carefully designed so the shift is less than a cycle through a fairly large angle. When the movement puts the listener closer to the tweeter by exactly 1/2 wavelength, that's the point where the upper null forms. Moving downward, when you get to the place where the woofer is closer to the listener than the tweeter by exatly 1/2 wavelength, the lower null forms.
What you'll see when you do measurements is the sound is very good through a range of +/- 1/4 wavelength. When the difference gets closer to about 1/3 wavelength on either side, the nulls begin to form. Between about 1/3 and 1/2 wavelength, the null grows from a shallow dip to a strong notch. Then as you travel further going from 1/2 to 2/3 wavelength, the notch subsides, becoming more of a dip and then relaxing. By the time you've moved all the way out to a full cycle shift, there is no notch, in fact, the wavefront is (generally) constructive again. However, this is a full cycle shift, and you would probably consider that to be "dirty". Sine waves will combine constructively even if passing through a full cycle shift, but aperiodic waves won't.
Wayne,
I understand all that; not sure why you feel the need to keep repeating the basics and not address the question.
Now that I think about, as you move off axis horizontally, all of that applies as long as the woofer and tweeter have the same horizontal directivity, so toe-in won't help.
I understand all that; not sure why you feel the need to keep repeating the basics and not address the question.
Now that I think about, as you move off axis horizontally, all of that applies as long as the woofer and tweeter have the same horizontal directivity, so toe-in won't help.
"Suppose you use some real brick-wall filters. Then at more than say 10 degrees off-axis vertically you have a huge spike in group delay. At that point a huge portion of the radiated sound has a huge spike in group delay."
I don't follow.
If the filters are truly brick-wall, the overlap range is infinitesimally small so anything bad that happens is of no consequence.
Given the way our hearing works, if the overlap range is restricted to 1/12th octave or less it should be inaudible.
I don't follow.
If the filters are truly brick-wall, the overlap range is infinitesimally small so anything bad that happens is of no consequence.
Given the way our hearing works, if the overlap range is restricted to 1/12th octave or less it should be inaudible.
Rybaudio said:
That the situation is not simple, I whole heartedly agree. While the steeper filters do make the transition narrower, the phase rotates faster. The nulls are more sensitive to phase than anything else and this rapid phase rotation could be a problem.
IF, I were interested in this problem, I would look into it some more, but quite frankly I think that its academic. Unless someone is actually in a null, and I think that we've agreed that this can be easily avoided with pointing, or stands, or distance - whatever - the nulls aren't that important. The floor and ceiling bounce are never a good thing - only horizontal reflections add ambiance and we basically only localize horizontally. So the simple solution is to just kill the floor and ceiling bounce and the vertical reflections and the vertical coverage angle aren't a big deal - null or no null. But horizontal - now thats extremely important because it basically controls imaging, coloration and spaciousness - those things that I design a good loudspeaker to do. I just cannot see equating the importance of the vertical control with the horizontal control. I'm basically looking at an elliptical waveguide to help with the axial hole and to shorten the enclosure. I really don;t care about the shallower vertical coverage.
noah katz said:
That's certainly how I've always viewed it: equidistance between the acoustic centers of the two drivers defines a plane, which may be tilted with respect to baffle normal, but which extends laterally as well as medially, above and below which the nulls form.
We await Earl's empirical demonstration that this does not occur off-axis in his designs.... 😎
Rybaudio said:
Actually, I think you're on the right track. You may not go all the way to a 48dB/octave crossover or something like that, but being able to modify slopes is part of the design process. You may find filters that improve the performance - that's one of the main areas where designers must optimize.
Here's the problem I've faced (indirectly) when people ask me for recommendations on active crossovers for my speakers. When I design a crossover, I have the physical box and the physical crossover (or extremely accurate model, in the initial development stage) available for measurements. I can optimize the crossover to set everything just right. It's a semi-delicate set of competing priorities that the designer juggles, and it's real easy to get one wrong and screw the pooch.
So as much as I agree that active crossovers are beneficial, I know that if the guy using them doesn't have the measurement equipment and the ability to use it, he's probably not going to hit the mark. Even savvy designers miss it sometimes, so a hobbyist is even more likely to. I'm not trying to be controlling, but I also don't want the guy to get dozens of hours and hundreds or thousands of dollars into a project trying to setup the crossover, only to get mediocre results.
The bottom line for me is, if you're competent with measuring equipment and have a lot of time to do measurements, on and off-axis, then the active crossover is probably worthwhile. It has other benefits too, so go for it. But if you aren't able to do a pretty large battery of acoustic measurements with good accuracy, you can't properly optimize the crossover. You just can't see what you're doing without lots and lots of measurements.
Rough measurement can sometimes be sufficient to get a crossover pegged. An RTA and moving the mic around, a couple phase inversions, and you can deal with the gross trends. Refinement after this is where it gets complicated- the notch filters, etc.
Naturally, you need to spend time to get it REALLY dialed in, but a proof of concept can be done fairly rapidly.
Naturally, you need to spend time to get it REALLY dialed in, but a proof of concept can be done fairly rapidly.
noah katz said:
I thought you were just kind of trolling me. Some of the ways you put things seemed kind of rude, and I was biting my tongue sometimes, not replying may have been my best response.
noah katz said:
"I thought you were just kind of trolling me. Some of the ways you put things seemed kind of rude, and I was biting my tongue sometimes, not replying may have been my best response."
I don't believe my original question was rude, but the answer seems so obvious now that maybe you thought I was baiting you.
"We await Earl's empirical demonstration that this does not occur off-axis in his designs.... "
*That's* a troll.
I don't believe my original question was rude, but the answer seems so obvious now that maybe you thought I was baiting you.
"We await Earl's empirical demonstration that this does not occur off-axis in his designs.... "
*That's* a troll.
ZilchLab said:
Here you go. I only included the lobing at the XO frequency as that's the only one that's really accurate the way I have to kludge the horn directivity with a big tweeter. The 3rd order Butterworth crossover (acoustical response) that JBL uses tilts the lobe quite a bit. I made the same (conservative) assumption as with the Summa that the acoustic center of the tweeter is an inch behind that of the woofer.
Attachments
As an aside, if Earl is using 3rd order acoustical slopes, that could explain the lobing behavior Wayne observed in the Summa. I think the sims show that the XO makes a bigger difference than round vs. rectangular horns. The sims I did of the Summa used 4th order slopes and they tend to direct the lobe forward with less tilt but that may not be what he's using at all.
Very interest, I was wondering about that myself. Those are simultations that you are doing? What kind of polar response model do you have? Waveguides are very difficult to model polar responses.
I completely agree that the crossover dominates the problem, more so that anything else being argued here. And I have found that to get the horizontal right you do have to compromise the vertical somehwat - and I have made clear what my priorities are.
I think that because the phase is so critical, and the phase of both sources changes off axis, probably in complex ways, it might be difficult to say what actually happens. But thats the first good point that I have heard in this whole discussion.
I completely agree that the crossover dominates the problem, more so that anything else being argued here. And I have found that to get the horizontal right you do have to compromise the vertical somehwat - and I have made clear what my priorities are.
I think that because the phase is so critical, and the phase of both sources changes off axis, probably in complex ways, it might be difficult to say what actually happens. But thats the first good point that I have heard in this whole discussion.
- Status
- Not open for further replies.