In the 80's I did a system with horn mids & tweets. they were hung with fishing line from the ceiling with the voice coils aligned vertically. From the side the woofer, mid horn & hi horn fronts were all miss alingned but the voice coils were streight up and down. If I pulled the string on the back of the Mid horn moving it farther away from the listner the Image across the sound stage collapsed and it almost came to the point of sounding mono. I did this test with at least 50 different people listening to the system and everyone noticed the difference no matter if they were an Audiofile or thought there clock radio was the best sounding thing they owned. I find it very hard to believe that moving speakers in and out of alignment is not audible.
This is the mid-high PA horn system that I use to play with. It requires *four* first-order phase shifters in the tweeter for proper alignment (horns are really evil). Crossover is 1600Hz 4th-order Linkwitz-riley. Result: You can no longer tell where is the tweeter and where is the mid, it's perceived as a point source on-axis thanks to optimum summing, even when you listen with your face just 30cm away from the horns.
By the way, these horns are something like 8 or 12dB down when it comes to radiate towards the walls, floor and ceiling, so even an untreated room becomes mostly anechoic.
By the way, these horns are something like 8 or 12dB down when it comes to radiate towards the walls, floor and ceiling, so even an untreated room becomes mostly anechoic.
TwisterZ said:
What do you think would happen if you moved the entire speaker away from the listener? Then the stereo-image would collapse too.
This is not necessarily because the frequencies may be not exactly in phase: because the speaker is further away from the listener, the level is simply down a bit.
Yes, I used both averaged pink noise FFT (SpectraLAB) and listening to tune the phase shifters. The criteria is just to obtain the deepest and widest notch in the crossover region on-axis when tweeter polarity is reversed. That guarantees a nice "single-driver-like" sum when the right polarity is restored.
I do agree that time-alignment might be less important than flat response and low non-linear distortion.
But the ones that say it doesn't matter at all are usually those manufacturers whose speakers are not transient-improved (I don't like the term transient-perfect since there is no such thing in reality).
PHASE matching is really important in terms of proper on-axis summation and controlled lobing - nothing to disagree here. This should definitely taken care of.
But TRANSIENT-improved designs offer more clarity and better imaging than non-improved ones. There are also sources saying that these have improved dynamic response. I can imagine that the latter is specially true in the sub - 500 Hz range where most of the power of large percusssion instruments lies.
Regards
Charles
But the ones that say it doesn't matter at all are usually those manufacturers whose speakers are not transient-improved (I don't like the term transient-perfect since there is no such thing in reality).
PHASE matching is really important in terms of proper on-axis summation and controlled lobing - nothing to disagree here. This should definitely taken care of.
But TRANSIENT-improved designs offer more clarity and better imaging than non-improved ones. There are also sources saying that these have improved dynamic response. I can imagine that the latter is specially true in the sub - 500 Hz range where most of the power of large percusssion instruments lies.
Regards
Charles
Something you all might find of interest:
I recently got some 4" woofers and the little dayton Nd tweeter for a budget MLTL MTM bookshelf with separate sub, and what I usually like to do when I get new speakers is hook them up and listen to them full range. Except tweeters which get a cap. For this initial listen I just set the speakers on the table and have a quick listen. I played with different tweeter positions and I noticed significant (night and day) differences with about a 1/4" of misalignment-maybe a little more, but not much.
I know that this is not exactly scientific, but I attribute my results to one, the frequency of the overlap was in the critical midrange increasing audibility and two there was alot of overlap resulting in multiple point sources radiating the same signal with the induced delay. In other words the audibility of the delay was enhanced by the drivers reproducing identical signals in the pass band as opposed to the narrow 24dB/oct stop band that they used.
I would say that the importance of time alignment is related to crossover frequency and slope, therefore in some cases its trivial but in others it could easily make or break the design.
I recently got some 4" woofers and the little dayton Nd tweeter for a budget MLTL MTM bookshelf with separate sub, and what I usually like to do when I get new speakers is hook them up and listen to them full range. Except tweeters which get a cap. For this initial listen I just set the speakers on the table and have a quick listen. I played with different tweeter positions and I noticed significant (night and day) differences with about a 1/4" of misalignment-maybe a little more, but not much.
I know that this is not exactly scientific, but I attribute my results to one, the frequency of the overlap was in the critical midrange increasing audibility and two there was alot of overlap resulting in multiple point sources radiating the same signal with the induced delay. In other words the audibility of the delay was enhanced by the drivers reproducing identical signals in the pass band as opposed to the narrow 24dB/oct stop band that they used.
I would say that the importance of time alignment is related to crossover frequency and slope, therefore in some cases its trivial but in others it could easily make or break the design.
Eva,Eva said:
Thanks! Indeed, very interesting method.
OK, another degree of freedom. If we use a low order xover (say, an inductance for midwoofer and a capacitor for tweeter at simplest case), we obtain rather wide range with simultaneous speakers radiation (smoothing off-axis dispersion behaviour). Are there additional consederartions concerning time alignment at such case?
First, the lower the order of the crossover, the wider the overlap region.
The wider the overlap region becomes, the harder it gets to match on-axis phase response of both drivers within that region (think of horns with peaks and dips near LF and HF cutoff), and the worse the acoustic disaster that happens when proper matching is not achieved.
On the other hand, a wider overlap region tends to produce a wider off-axis cancellation region and thus a narrower sweet spot, something that I consider very useful when dealing with direct radiating speakers and rooms (I consider room reflected sound a very disgusting thing). Indeed, I think that the main advantage of dipoles (and horns) is reduced room interaction due to massive off-axis cancellation.
Also, note that such a simple passive crossover with just a L and a C will become hard to align, because both the impedance rise due to voice coil inductance of the woofer and the impedance peak due to tweeter resonance will introduce additional poles and zeros in the filter resulting in peaks and dips and additional group delays to compensate. Active filtering is strongly recommended (Indeed, I gave up on passive filtering for my horns after some attempts due to very complex impedance curves).
The wider the overlap region becomes, the harder it gets to match on-axis phase response of both drivers within that region (think of horns with peaks and dips near LF and HF cutoff), and the worse the acoustic disaster that happens when proper matching is not achieved.
On the other hand, a wider overlap region tends to produce a wider off-axis cancellation region and thus a narrower sweet spot, something that I consider very useful when dealing with direct radiating speakers and rooms (I consider room reflected sound a very disgusting thing). Indeed, I think that the main advantage of dipoles (and horns) is reduced room interaction due to massive off-axis cancellation.
Also, note that such a simple passive crossover with just a L and a C will become hard to align, because both the impedance rise due to voice coil inductance of the woofer and the impedance peak due to tweeter resonance will introduce additional poles and zeros in the filter resulting in peaks and dips and additional group delays to compensate. Active filtering is strongly recommended (Indeed, I gave up on passive filtering for my horns after some attempts due to very complex impedance curves).
No, it's the opposite. The wider the overlap region, the wider the frequency range that gets screwed up when proper summing is not achieved. For example, high order crossovers have been routinely employed in PA horn applications for a long time now as a poor fix to bad summing and undesirable on-axis lobing.
Also, 8th order filters have been employed to reduce subwoofer and satellite overlapping to the minimum extent in order to avoid location-dependent cancellation in disco and PA applications (imagine that you move through a dance floor with a bass/mid/high cluster hanging in the center and the subwoofer in one corner).
Also, 8th order filters have been employed to reduce subwoofer and satellite overlapping to the minimum extent in order to avoid location-dependent cancellation in disco and PA applications (imagine that you move through a dance floor with a bass/mid/high cluster hanging in the center and the subwoofer in one corner).
Eva,Eva said:
It is PA's history rather arguments
Moreover, I can suggest high order xovers in PA is a way to avoid tweeters overloading. So, I still don't understand why does a case with first order xovers is worse. If we recall own speakers inductance, "crosspoint crossing" becomes even more smooth, that is to say we have less opportunity to identify this crossing. I don't exist as I haven't much experience. Just want to understand 
cs said:
cs said:
It may not make a difference under certain situations.
Without question it makes an audible difference in other situations.
The drivers in my main horns are about 16 inches to the rear of the drivers on my sealed subs, crossing at about 95hz
You can hear the difference. but only when you are actually spinning the delay. You don't notice the error until it's gone. In fact, I had my wife spin dial for delay and as I listened the subs seemed to slowly move up until it sounded as if the subs were located in the middle of the horns about 34" higher.
Eva (and All),
Another question. Has your method (pink noise bounded at cut point frequency range) some advantages in comparison with Step Response plot analysis? I have tried to measure the same 2-way loudspeaker at tweeter axis and ~20 degree toward floor, and results seem to be clear.
Another question. Has your method (pink noise bounded at cut point frequency range) some advantages in comparison with Step Response plot analysis? I have tried to measure the same 2-way loudspeaker at tweeter axis and ~20 degree toward floor, and results seem to be clear.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.