These questions are hypothetical and theoretical - I'm not actually building this system, just trying to understand it.
Suppose we have a two-way system with a second order crossover at, let's say, 3 kHz. If I understand correctly (and that's a big "if"), then the low-pass driver's response will lag 180 degrees behind the high pass driver's response.
Note that at 3 kHz one wavelength is about 4.5 inches long, so 1/2 wavelength would be 2.25 inches long.
Let's look at three different cases:
1) Tweeter in front: Let's say the drivers are flush-mounted on the vertical face of the enclosure, and that the woofer's voice coil just happens to be 2.25 inches behind the tweeter's. In this case, at the crossover frequency the woofer lags the tweeter by 180 degrees due to the crossover, and by another 180 degrees due to the driver offset (which is really a fixed time delay, but corresponds to 1/2 wavelength at the crossover frequency), which will put the woofer's output a full 360 degrees behind the tweeter's at the crossover frequency. Is this correct?
2) Drivers aligned: Next, let's precisely align the voice coils in the vertical plane. Now the woofer's output lags the tweeters' by 180 degrees, even though their outputs arrive at the same point in time. So in order to not have a deep null at the crossover frequency, we reverse the polarity of one of the drivers so that now they are operating in sync (I hesistate to say "in phase" but maybe that would be accurate) at the crossover frequency. Is this correct?
3) Woofer in front: Finally, suppose we physically move the tweeter back so that its voice coil is now 2.25" behind the woofer's voice coil. In this situation, at the crossover frequency the woofer is 180 degrees behind because of the crossover but also 180 degrees ahead because of the physical offset, so now we connect the drivers with the same polarity and their output is indeed in phase in the crossover region. Is this correct?
It might be said that configuration 1) is wrong in both time and phase; that 2) is right in time but wrong in phase; and 3) is wrong in time but right in phase, at least in the crossover region. Or, maybe I fail to understand the principles involved and how they interact.
Now I would think they'd all do a decent job of passing a 3 kHz sine wave. But which would do the best with a square wave, or a step response, or an impulse response?
Assuming no cabinet diffraction issues, which alignment would be the most likely to sound coherent and image well? (Yeah I know first order crossover would probably be better, but I'm trying to figure out what would be the optimum configuration for second order crossover).
My thanks to anyone who takes the time to comment.
Alidore
Suppose we have a two-way system with a second order crossover at, let's say, 3 kHz. If I understand correctly (and that's a big "if"), then the low-pass driver's response will lag 180 degrees behind the high pass driver's response.
Note that at 3 kHz one wavelength is about 4.5 inches long, so 1/2 wavelength would be 2.25 inches long.
Let's look at three different cases:
1) Tweeter in front: Let's say the drivers are flush-mounted on the vertical face of the enclosure, and that the woofer's voice coil just happens to be 2.25 inches behind the tweeter's. In this case, at the crossover frequency the woofer lags the tweeter by 180 degrees due to the crossover, and by another 180 degrees due to the driver offset (which is really a fixed time delay, but corresponds to 1/2 wavelength at the crossover frequency), which will put the woofer's output a full 360 degrees behind the tweeter's at the crossover frequency. Is this correct?
2) Drivers aligned: Next, let's precisely align the voice coils in the vertical plane. Now the woofer's output lags the tweeters' by 180 degrees, even though their outputs arrive at the same point in time. So in order to not have a deep null at the crossover frequency, we reverse the polarity of one of the drivers so that now they are operating in sync (I hesistate to say "in phase" but maybe that would be accurate) at the crossover frequency. Is this correct?
3) Woofer in front: Finally, suppose we physically move the tweeter back so that its voice coil is now 2.25" behind the woofer's voice coil. In this situation, at the crossover frequency the woofer is 180 degrees behind because of the crossover but also 180 degrees ahead because of the physical offset, so now we connect the drivers with the same polarity and their output is indeed in phase in the crossover region. Is this correct?
It might be said that configuration 1) is wrong in both time and phase; that 2) is right in time but wrong in phase; and 3) is wrong in time but right in phase, at least in the crossover region. Or, maybe I fail to understand the principles involved and how they interact.
Now I would think they'd all do a decent job of passing a 3 kHz sine wave. But which would do the best with a square wave, or a step response, or an impulse response?
Assuming no cabinet diffraction issues, which alignment would be the most likely to sound coherent and image well? (Yeah I know first order crossover would probably be better, but I'm trying to figure out what would be the optimum configuration for second order crossover).
My thanks to anyone who takes the time to comment.
Alidore
I am very interested in this thought experiment. I posted a similar question recently but was unable to get any constructive feedback.
In the case of the second order crossover a 90 degree phase shift occurs in both the HP and LP sections, and that combines for a total of 180 degrees, creating a null at the crossover frequency. You can cure that by switching polarity of one of the drivers. But then time aligning is also required to maintain perfect phase relationships, and to do that you align the voice coils on the vertical plane.
You can accomplish the same thing as far as phase is concerned by leaving the drivers electrically out of phase and off-setting the voice coils by 1/2 wavelength, but that addresses only phase and not time. The crux of the question is whether you can hear the time lead/lag in that instance. Studies have shown that you can detect differentials as small as .003 seconds or so in the midrange, but the lead/lag with a 2.25 inch path differential would be around .00016 seconds, so it's not likely in this case.
The accepted procedure is to always phase align electrically, and to time align physically if one wants to go to the trouble involved. If you don't physically time align you also will have a phase zit but again in most cases it isn't audible anyway.
You can accomplish the same thing as far as phase is concerned by leaving the drivers electrically out of phase and off-setting the voice coils by 1/2 wavelength, but that addresses only phase and not time. The crux of the question is whether you can hear the time lead/lag in that instance. Studies have shown that you can detect differentials as small as .003 seconds or so in the midrange, but the lead/lag with a 2.25 inch path differential would be around .00016 seconds, so it's not likely in this case.
The accepted procedure is to always phase align electrically, and to time align physically if one wants to go to the trouble involved. If you don't physically time align you also will have a phase zit but again in most cases it isn't audible anyway.
BillFitzmaurice said:
Are you sure about this?
At the XO, there's a 180 phase difference, so connecting the tweeter backwards gives a correct phase @xo frequency. If the accoustical centers are aligned
You speak about a tweeter in reverse polarity, AND non aligned accoustic centers, this seems wrong to me
Alidore,
you are correct in all three examples you bring up.
The thing is that if you have a textbook 2nd order slope for both drivers at 3k and pull the tweet back to get them in phase at 3k, then you´ll have non flat summation above and below the crossing. And no example with true 2nd order slopes will have a good pasing of square waves/impulses though ther will be differences between the three examples.
Bill,
you are basically right but; "The crux of the question is whether you can hear the time lead/lag in that instance."
It´s not the time itself that is interesting here but the issue I bring up above. You will simply have ripples/non flat response.
Bricolo,
the drivers/legs will be both be 90 degrees away iow a 180 degree phase difference between the drivers, one connection summing flat and one creating a deep null.
/Peter
you are correct in all three examples you bring up.
The thing is that if you have a textbook 2nd order slope for both drivers at 3k and pull the tweet back to get them in phase at 3k, then you´ll have non flat summation above and below the crossing. And no example with true 2nd order slopes will have a good pasing of square waves/impulses though ther will be differences between the three examples.
Bill,
you are basically right but; "The crux of the question is whether you can hear the time lead/lag in that instance."
It´s not the time itself that is interesting here but the issue I bring up above. You will simply have ripples/non flat response.
Bricolo,
the drivers/legs will be both be 90 degrees away iow a 180 degree phase difference between the drivers, one connection summing flat and one creating a deep null.
/Peter
BillFitzmaurice said:
i've always wondered about this method thinking "but the voice coils don't make any noise" they only give motion to the thing that does
fair play when it comes to aligning for a tweeter as there's next to nothing in it but say you're using one of the new breed of high Xmax drivers you're going to be way out on any cone you're using due to the extra length of coil involved
would you say the edge of the dust cap would be better or somewhere else entirely?
But the voice coil does make a noise! Sound is vibration, and voice coils vibrate. This sound is conducted to the cone, where it is then coupled to the air. If air was incredibly dense, you wouldn't need a cone at all - the voice coil alone would couple well enough to the air.synergy said:
My understanding here is limited, so I simply theorize here. I own a pair of time aligned speakers, and have also played around with time aligning the speakers by making test cabinets that could have the tweeters moved, and then measured with my RTA and computer. I dont have the ability to truely test what goes on, the resolving power of my equipment is terrible.
Anyway, I have often wondered if simply placing the tweeter's coil at the same plane as the midranges would actually time align the speaker. I had a few things I wondered about with this. First, I think high frequency waves can travel through the air faster than low frequency waves. One reason I think that is that bass waves disrupt the air more, and thus create more turbulance. Normally turbulance causes something to be slowed down, so I would guess bass runs slower. Im not sure of that, but looking at wave theory it just seems to make sense to me. Like I said, my understanding of this is all very limited.
If this is true though, then you would need to account for this with the tweeter placement. However, as was mentioned before, many of these differences are so small given the distances we are talking about that hearing it would be next to impossible. My guess is that, if what I said was true, it still isnt going to be audible.
another thing I have guessed, and clearly see in the design of both my JmLabs speakers and other aligned versions. Theil are infamous for the care in propper time aligning. I actually believe that I read mine are not truely phase aligned but only time aligned, or something along those lines. Anyway, if the speaker is tall, and there is a driver above the tweeter, and thus above ear level, then to properly time align, I would think you would need to angle them down. I may be confusing theories here, as it just crossed my mind that this might only have to with the directivity of the driver. Anyway, it seems that there is more to time and phase alignment than some simple wire swapping and distance changes. It would be my guess that is why these modern speaker DSP devices are able to so accuratly align every aspect of the speakers, so they measure so well.
Anyway, I have often wondered if simply placing the tweeter's coil at the same plane as the midranges would actually time align the speaker. I had a few things I wondered about with this. First, I think high frequency waves can travel through the air faster than low frequency waves. One reason I think that is that bass waves disrupt the air more, and thus create more turbulance. Normally turbulance causes something to be slowed down, so I would guess bass runs slower. Im not sure of that, but looking at wave theory it just seems to make sense to me. Like I said, my understanding of this is all very limited.
If this is true though, then you would need to account for this with the tweeter placement. However, as was mentioned before, many of these differences are so small given the distances we are talking about that hearing it would be next to impossible. My guess is that, if what I said was true, it still isnt going to be audible.
another thing I have guessed, and clearly see in the design of both my JmLabs speakers and other aligned versions. Theil are infamous for the care in propper time aligning. I actually believe that I read mine are not truely phase aligned but only time aligned, or something along those lines. Anyway, if the speaker is tall, and there is a driver above the tweeter, and thus above ear level, then to properly time align, I would think you would need to angle them down. I may be confusing theories here, as it just crossed my mind that this might only have to with the directivity of the driver. Anyway, it seems that there is more to time and phase alignment than some simple wire swapping and distance changes. It would be my guess that is why these modern speaker DSP devices are able to so accuratly align every aspect of the speakers, so they measure so well.
Since sound usually travels faster through most of the construction materials than through the air, the alignment of the voice coils isn't an exact one regarding time alignment.
I assume that the alignment of voice coils is just a coarse solution to avoid measuring of the actual delays involved. Since the plane of radiation is frequency-dependant one can only measure to get exact results. As an approximation the coil alignment might be better than nothing. One can assume that for a cone driver the acoustical center is in the middle of the cone at its upper working range, that's where the method might come from.
BTW: 2nd order filters with overlap and a recessed tweeter can give quite good performance in the frequency and time domain but for one axis only unfortunately.
Regards
Charles
I assume that the alignment of voice coils is just a coarse solution to avoid measuring of the actual delays involved. Since the plane of radiation is frequency-dependant one can only measure to get exact results. As an approximation the coil alignment might be better than nothing. One can assume that for a cone driver the acoustical center is in the middle of the cone at its upper working range, that's where the method might come from.
BTW: 2nd order filters with overlap and a recessed tweeter can give quite good performance in the frequency and time domain but for one axis only unfortunately.
Regards
Charles
Yup, Charles is right. This is just an aproximation. No "real science" behind the idea.
About sound speed in air. I do not think there is any significant difference in the speed of low vs. high frequenices (if any at all). Also, turbulence is more likely to reduce the amplitude than affecting the speed I´d say.
/Peter
About sound speed in air. I do not think there is any significant difference in the speed of low vs. high frequenices (if any at all). Also, turbulence is more likely to reduce the amplitude than affecting the speed I´d say.
/Peter
The speed of sound is the same regardless of frequency, at about 344m/s in air. The speed of sound is directly proportional to the rigidity of the material it is travelling thorough, which is why it travels faster in the cone than in the air.pjpoes said:
Maybe I was not clear enough, but that wasn't what I tried to say.
Not all frequencies are radiated by the same part(s) of the cone. As a very coarse rule of thumb one can say that the higher the frequency, the smaller the area radiating it. If one assumes that the higher frequencies are radiated by the area around the dustcap then the drivers are definitely better time-aligned if their coils are aligned than if they are all flush with the baffle.
Most important: The combined effects of driver alignment and crossover will have to be taken into account.
Regards
Charles
The thing with phase and time alignment is that it is all an estimation when attempting to align by physically aligning the drivers. The radiation of different wavelengths actually originates from different portions of the cone/diaphragm but like Bill indicated, this difference is not likely audible to the human ear. The utilization of software allows us to align by the arrival of the pulse and to calculate the phase relationship to the best of our ability. Note that the measured pulse, however, is only at one frequency and that although sound travels at the same speed regardless of frequency, it is eminating from a slightly different point on the diaphragm, hence a slightly different arrival will be obtained based upon the chosen frequency to be measured. These differences are not likely audible (as Bill indicated or as I said before). In general, the target of aligning acoustic centers would be most accurate if done at the crossover frequency as this would promote a more uniform transition from one driver to the next however it does not account for the overlap frequencies which are not only coming from different points on the diaphragm but also from different point sources and combining (i.e. there will always be a combing effect to some degree).
Even with the advantages of software, there is plenty of room for play as different people favor different approaches - hence the fine tuning after the computer simulation.
Just my two cents,
I am by no means an expert.
Jay
Even with the advantages of software, there is plenty of room for play as different people favor different approaches - hence the fine tuning after the computer simulation.
Just my two cents,
I am by no means an expert.
Jay
Fun with your speakers.
1. Line arrays can measure terribly when a point source microphone is used close to them because of the different travel distances from the driver to the microphone. At a normal listening distance this effect is virtually gone. (Still line arrays sound different than conventional cones, notice I didn't say better or worse)
2. Crossover phase shifts change as you move away form the crossover point making a perfect solution all but impossible.
3. Besides phase shifts crossovers have time delays (group delay) which change at least a little for diffent frequencies making it possible to time align speakers whose offset distance is "incorrect".
4. Driver phase and time delays also change with frequency
5. I believe that traditionally drivers are time aligned with reference to the joint between the voice coil and the cone/dust cap.
So all speakers no matter how carefully designed on the mathmatical end are "voiced", that is to say we build carefully as the design predicts then we adjust anything we want till it sounds right. The trick seems to be in knowing what to adjust.😉
1. Line arrays can measure terribly when a point source microphone is used close to them because of the different travel distances from the driver to the microphone. At a normal listening distance this effect is virtually gone. (Still line arrays sound different than conventional cones, notice I didn't say better or worse)
2. Crossover phase shifts change as you move away form the crossover point making a perfect solution all but impossible.
3. Besides phase shifts crossovers have time delays (group delay) which change at least a little for diffent frequencies making it possible to time align speakers whose offset distance is "incorrect".
4. Driver phase and time delays also change with frequency
5. I believe that traditionally drivers are time aligned with reference to the joint between the voice coil and the cone/dust cap.
So all speakers no matter how carefully designed on the mathmatical end are "voiced", that is to say we build carefully as the design predicts then we adjust anything we want till it sounds right. The trick seems to be in knowing what to adjust.😉
simon5 said:
In the end it comes down to your implementation, your room, and your tastes... it is all about compromises. The set of compromises that work for me don't work for everybody. One of the reasons there are "zillions" of good speakers out there...
dave
Are line arrays better or worse?
There is no simple answer to the question of what sounds better.
There are several kinds of line arrays. Ribbon drivers are a line arrays, planar drivers act much like a line array (electrostatic such as the Martin Logans and/or electromagnertic such as the Magneapans), some people have mimiced a line array (Pipedreams and others) with rows of conventional drivers.
There are one way, two way and three way line array designs.
I've seen prices from planar computer monitors at around $75 all the way to over $100K so there is no answer as to how line arrays sound.
When done well, they sound great but so do well done cone dynamic speakers.
A well built line array driver element typically costs more than its cone driver equivalent.
Typically line arrays have lower distortion but often at the price of lower dynamic range. Dispersion is different making it a problem when mixing line arrays with conventional cones.
Worse yet, both dipole and monopole line arrays exist (signals out back and front or just out the front) further complicating the question as to which sounds better.
All this and more is why there are so many varieties of speakers and speaker designs. If you like lots of bass and loud, line arrays are probably not the best first choice.
Much of this is opinion, there are probably exceptions to every statement. It's a great hobby but it does take some work to exract the most from it.
There is no simple answer to the question of what sounds better.
There are several kinds of line arrays. Ribbon drivers are a line arrays, planar drivers act much like a line array (electrostatic such as the Martin Logans and/or electromagnertic such as the Magneapans), some people have mimiced a line array (Pipedreams and others) with rows of conventional drivers.
There are one way, two way and three way line array designs.
I've seen prices from planar computer monitors at around $75 all the way to over $100K so there is no answer as to how line arrays sound.
When done well, they sound great but so do well done cone dynamic speakers.
A well built line array driver element typically costs more than its cone driver equivalent.
Typically line arrays have lower distortion but often at the price of lower dynamic range. Dispersion is different making it a problem when mixing line arrays with conventional cones.
Worse yet, both dipole and monopole line arrays exist (signals out back and front or just out the front) further complicating the question as to which sounds better.
All this and more is why there are so many varieties of speakers and speaker designs. If you like lots of bass and loud, line arrays are probably not the best first choice.
Much of this is opinion, there are probably exceptions to every statement. It's a great hobby but it does take some work to exract the most from it.
- Status
- Not open for further replies.