Wohooo0000
i cant beleive some one else has seen/heard my tweeters!
also,my tweeters have that little fuzzy bit,its slightly off centre
like they were in a rush to put it on 😛
and the 4 screws are rusted on the front.
and the 4" mids are abit rough too,the surrounds look creased,and one of them moves further than the others😕
so actualy are they not bad!! woahhh
im surprised
im glad also,because i havent taken any DIY measurements,but would like to.
also i want to know im not listening to crap without knowing!
someday,jbl2370 HF + tractix mid, Direct radiator midbass up to the tractix mid and sub (26hz basshorn haha no cant fit,but 2226 or lab12 in my 40hz horn im designing
cheers!
😀 😀 😀 😀
They are much more interesting looking than my phillips tweeters, I'm going to have to get myself a digital camera. I just looked for any pictures of my speakers, and the only one I can find has an AT-AT being attacked by the millenium falcon sort of obscuring it (and the speaker isn't that well exposed)........
Regards,
Tony.
Regards,
Tony.
To SY:
Do you have any new measurements of rise time? What about the Audax/Dynaudio midrange comparison?
Thank you very much
Do you have any new measurements of rise time? What about the Audax/Dynaudio midrange comparison?
Thank you very much
(JPK) I'd like to add something from a different perspective. You have to be careful when evaluating a driver's transient response by any method. Looking at the impulse or waterfall plot for a raw driver can be misleading. If you look at http://www.musicanddesign.com/Stored_energy_1.html about 1/3 of the way down the page you will see waterfall plots of two different 18cm drivers. Obviously they are very different and the raw drivers have very different transient response. Directly below those plots is a waterfall plot of a 2nd order band pass filter. And then below that are waterfall plots of the two drivers connected to crossover filters that yield a net 2nd order band pass acoustic response which is designed to match the bandpass response shown in the previous waterfall plot. As you can see, while the raw driver responses are very different, when connected to the appropriate crossover filter both drivers show a characteristic very close to the targeted band pass response. Both drivers would have the same impact on the system transient response when used with the specified bandpass filter, aside from the small differences at 200 and 1.5k Hz (which could be removed with improved filter design.
The waterfall plots of the raw driver response, or the raw impulse response, only give an idea of how much work might be involved in achieving the desired target acoustic response. Overall, all the little zigzags in the waterfall response are a display of linear distortion or stored energy (one in the same) which must be address in the crossover design. Starting with a better waterfall would hopefully means designing the required crossover filter would be an easier job.
The waterfall plots of the raw driver response, or the raw impulse response, only give an idea of how much work might be involved in achieving the desired target acoustic response. Overall, all the little zigzags in the waterfall response are a display of linear distortion or stored energy (one in the same) which must be address in the crossover design. Starting with a better waterfall would hopefully means designing the required crossover filter would be an easier job.
The two raw drivers are very similar in waterfall plot! At least in the usable range under 3kHz. There is todays typical minimal ringing.
First of all this site shows how bad influence the passive crossover makes.
First of all this site shows how bad influence the passive crossover makes.
This thread has become very interesting! The issues being discussed are really at the heart of technical analysis of loudspeakers.
John, at the risk of taking the thread off topic, I would like to question your statement that speakers are linear.
Your article is a very well writen explanation of the inter-relatedness of time and frequency behaviour of linear systems, but I'm not sure that speakers fall completely in this category.
At very small amplitudes along one specific axis speakers come close to mimicking a linear system. However at the SPL's commonly encountered during music reproduction, the movement of the cone is influenced by the non-linearities of the suspensions, non-linearities in the magnetic system and thermal effects in the voice coil among others. As SPL's increase, these issues become ever more pronounced.
While the impulse response of a linear system captures all of the information available in a linear system, the impulse response of a speaker can never be properly measured. This is equivalent to to saying the speaker is not a linear system. The measured impulse response is highly dependant on not only the mechanical non-linearities already discussed, but more importantly on the orientation of the measurement with respect to the loudspeaker cone. As the measurement is made further off-axis, the impulse response is radically different. Which one is the correct impulse response?
If we accept that a speaker is not really linear, it becomes clearer why it is so difficult to test them in meaningful ways. While deep insights about the behaviour of speakers can be gained through measurements, it must always be remembered that the results of these tests are only meaningful within certain boundrys.
I will venture to suggest that headphones are the only implementation of a speaker that closely approaches a linear system. Diaphragm amplitudes and power levels are very small, and directional effects are not involved due to the close coupling of diapragm and ear. No wonder they can sound so good! They can actually be properly measured! (Although the close coupling introduces it's own set of measurement difficulties)
I can also understand why some aspects of professional speakers sound so good. Compression driver/horn combinations can exhibit very small diaphragm excusions at moderate SPL's and a controlled axial dispersion yielding a more uniform impulse response. Unfortunately they also exhibit strong resonant behaviour and non-linearities due to the high air pressures involved at the throat. Professonal cone drivers can also sound very good because they are typically designed with careful attention to large-signal non-linearities as they are often used at high volumes. Unfortunately they suffer from the same directional issues as other piston radiators.
Cheers, Ralph
John, at the risk of taking the thread off topic, I would like to question your statement that speakers are linear.
Your article is a very well writen explanation of the inter-relatedness of time and frequency behaviour of linear systems, but I'm not sure that speakers fall completely in this category.
At very small amplitudes along one specific axis speakers come close to mimicking a linear system. However at the SPL's commonly encountered during music reproduction, the movement of the cone is influenced by the non-linearities of the suspensions, non-linearities in the magnetic system and thermal effects in the voice coil among others. As SPL's increase, these issues become ever more pronounced.
While the impulse response of a linear system captures all of the information available in a linear system, the impulse response of a speaker can never be properly measured. This is equivalent to to saying the speaker is not a linear system. The measured impulse response is highly dependant on not only the mechanical non-linearities already discussed, but more importantly on the orientation of the measurement with respect to the loudspeaker cone. As the measurement is made further off-axis, the impulse response is radically different. Which one is the correct impulse response?
If we accept that a speaker is not really linear, it becomes clearer why it is so difficult to test them in meaningful ways. While deep insights about the behaviour of speakers can be gained through measurements, it must always be remembered that the results of these tests are only meaningful within certain boundrys.
I will venture to suggest that headphones are the only implementation of a speaker that closely approaches a linear system. Diaphragm amplitudes and power levels are very small, and directional effects are not involved due to the close coupling of diapragm and ear. No wonder they can sound so good! They can actually be properly measured! (Although the close coupling introduces it's own set of measurement difficulties)
I can also understand why some aspects of professional speakers sound so good. Compression driver/horn combinations can exhibit very small diaphragm excusions at moderate SPL's and a controlled axial dispersion yielding a more uniform impulse response. Unfortunately they also exhibit strong resonant behaviour and non-linearities due to the high air pressures involved at the throat. Professonal cone drivers can also sound very good because they are typically designed with careful attention to large-signal non-linearities as they are often used at high volumes. Unfortunately they suffer from the same directional issues as other piston radiators.
Cheers, Ralph
jirka said:
(JPK) The CSD of the bandpass filter is for an active bandpass filter. The driver plus filter responses are for the drivers connected to active filters designed to yield the same acoustic bandpass response. The CSD is a function of the characteristic of the BP response. You may wish to look at the second page of the discussion http://www.musicanddesign.com/Stored_energy_2.html
to see the effect different crossovers introduce on an otherwise flat, DC to light, system.
The point I am trying to get at is that it is necessary to look at the system response rather than just the transient response of one component. Consider a hypothetical 3-way speaker system with 1st order acoustic crossover. If correctly designed the transient response of the system at the design point will exhibit no crossover induced transient distortion. However, the transient response of the 1st order bandpass nature of the midrange driver will be far from perfect.
What the individual unfiltered driver does is on little consequence if its filtered response can be shaped to the required acoustic target.
I would also suggest that the water fall plots for the two raw drivers are far form "very similar".
For the original post. Transient response is determined mainly by frequency response if the higest portion of FR is not achieved by cone resonance. If the higest frequency portion has cone resonance, it will show up in the waterfall plot as a mountain ridge shape.
Waterfall plots give an overall view on how fast a driver can stop. The more distance between each FR line, the faster the decay. Generally I like to see at least a 12db difference in 0.02ms for frequencies above 1KHz which generally means the sound will be quite clean.
How the plots need to be interpreted also depends on the type of driver the data represents. If you are looking at a cone type driver, the edge may act like a second sound source in a similar fasion as edge diffraction impulses will effect measurements.
Impulses are are interesting because they seem to be able to show whether energy is being reflected back to the coil.
Waterfall plots give an overall view on how fast a driver can stop. The more distance between each FR line, the faster the decay. Generally I like to see at least a 12db difference in 0.02ms for frequencies above 1KHz which generally means the sound will be quite clean.
How the plots need to be interpreted also depends on the type of driver the data represents. If you are looking at a cone type driver, the edge may act like a second sound source in a similar fasion as edge diffraction impulses will effect measurements.
Impulses are are interesting because they seem to be able to show whether energy is being reflected back to the coil.
I have made loudspeaker systems that sounded better as a sum than the quality expected by the quality of their parts.
Many aspects can be controlled. Nevertheless, other things than cone material signatures and various motor distortions can't be hidden especially when the SPL is getting serious. So when budget permits I prefer the better raw components (usually with better measurements profile too). The less of complexity for control is applied, the better the subjective result.
Even a very good digital controller like BSS or XTA ($$$) has a demonstrably distinct sound signature especially in a High - End concept.
Typical multistage op-amp analog circuits are lo-fi in my experience.
Many aspects can be controlled. Nevertheless, other things than cone material signatures and various motor distortions can't be hidden especially when the SPL is getting serious. So when budget permits I prefer the better raw components (usually with better measurements profile too). The less of complexity for control is applied, the better the subjective result.
Even a very good digital controller like BSS or XTA ($$$) has a demonstrably distinct sound signature especially in a High - End concept.
Typical multistage op-amp analog circuits are lo-fi in my experience.
Hi Ralph,
Of course you are correct. Ultimately the nonlinearities alter the design picture. But the hypothesis that a loudspeaker can be built using active or passive crossover filters which have fundamentally time invariant transfer functions is based on the assumption (requirement) that drivers behave over much of there operating range as linear devices. In that regard we can use time invariant active and passive circuits to shape the response to what we desire with some degree of confidence. On the other hand, if drivers were highly nonlinear the task would be impossible. For example, if we attempted to measure the frequency response of a highly nonlinear system it is possible that we would obtain significantly different results if the response were stimulated with stepped sine, swept sine or MLS methods. It’s not unlike good amplifier design. Hopefully a good amplifier starts with a circuit that has low linear and nonlinear distortion over most of it operating range before feedback is applied. Feedback is then applied minimally to clean up the response.
Your point about off axis measurement is a good one. Describing the variation of response with position would generally require nonlinear mathematics. However, at a given spatial position the response is still fundamentally linear in its behavior.
Of course you are correct. Ultimately the nonlinearities alter the design picture. But the hypothesis that a loudspeaker can be built using active or passive crossover filters which have fundamentally time invariant transfer functions is based on the assumption (requirement) that drivers behave over much of there operating range as linear devices. In that regard we can use time invariant active and passive circuits to shape the response to what we desire with some degree of confidence. On the other hand, if drivers were highly nonlinear the task would be impossible. For example, if we attempted to measure the frequency response of a highly nonlinear system it is possible that we would obtain significantly different results if the response were stimulated with stepped sine, swept sine or MLS methods. It’s not unlike good amplifier design. Hopefully a good amplifier starts with a circuit that has low linear and nonlinear distortion over most of it operating range before feedback is applied. Feedback is then applied minimally to clean up the response.
Your point about off axis measurement is a good one. Describing the variation of response with position would generally require nonlinear mathematics. However, at a given spatial position the response is still fundamentally linear in its behavior.
Thanks for your reply John!
I agree with your comments but I still feel that taking the impulse repsonse at a static position in the near-field of a single driver doesn't represent the listener's experience of a speaker system in a meaningful way. In a typical domestic situation the listener is usually located, at least for the lower frequencies, in the far-field, beyond the critical distance. The power response of a speaker being of major importance for real speakers in domestic listening rooms and for 'main' studio monitors. The added complications of a multi-driver system does make the task of optimisation something close to impossible. Additionally, room modes dominate the speaker impulse response at the lowest frequencies. The on-axis impluse response simply doesn't tell you how a speaker will sound.
I think that a lot of the confusion surrounding crossover types is a direct result of the impossibility of achieving a good impulse response both on-axis and within the directional envelope of the speaker. Some crossover types suit the power response better, others the axial response. Others are compromise between the two.
I think what is needed is some way of integrating the impulse response over a range of off-axis measurements. This is certainly beyond my mathematical scope. Maybe someone has some suggestions?
It also seems necessary to consider the acoustic environment when designing a speaker system, particularly with regard to the crossover. Without knowing whether the listener is near or far field, crossover design is largely guess work. I have often read about how speaker xyz was analytically designed using the standard tools and then tweaked by ear for the power response!
In the case of near-field speakers the designer at least knows how the system will be used by the listener. A predictable close-in listening environment allows the designer to concentrate on the on-axis impulse response and optimise this parameter. No wonder such systems are so popular amongst professionals!
Cheers, Ralph
I agree with your comments but I still feel that taking the impulse repsonse at a static position in the near-field of a single driver doesn't represent the listener's experience of a speaker system in a meaningful way. In a typical domestic situation the listener is usually located, at least for the lower frequencies, in the far-field, beyond the critical distance. The power response of a speaker being of major importance for real speakers in domestic listening rooms and for 'main' studio monitors. The added complications of a multi-driver system does make the task of optimisation something close to impossible. Additionally, room modes dominate the speaker impulse response at the lowest frequencies. The on-axis impluse response simply doesn't tell you how a speaker will sound.
I think that a lot of the confusion surrounding crossover types is a direct result of the impossibility of achieving a good impulse response both on-axis and within the directional envelope of the speaker. Some crossover types suit the power response better, others the axial response. Others are compromise between the two.
I think what is needed is some way of integrating the impulse response over a range of off-axis measurements. This is certainly beyond my mathematical scope. Maybe someone has some suggestions?
It also seems necessary to consider the acoustic environment when designing a speaker system, particularly with regard to the crossover. Without knowing whether the listener is near or far field, crossover design is largely guess work. I have often read about how speaker xyz was analytically designed using the standard tools and then tweaked by ear for the power response!
In the case of near-field speakers the designer at least knows how the system will be used by the listener. A predictable close-in listening environment allows the designer to concentrate on the on-axis impulse response and optimise this parameter. No wonder such systems are so popular amongst professionals!
Cheers, Ralph
Hi Ralph,
Room interactions are a completely separate issue than whether the system is fundamentally a linear system. I would expect that we all agree that the sound of the room is probably a dominant factor in the listening experience. And power response is of course another importance issue. But again it has more to do with the choice of drivers, crossover point, speaker configuration, etc. than whether or not the system is linear. That the system behaves as a linear system over its intended operating range is still a prerequisite. However, the on axis response of a speaker (impulse or otherwise) does tell you a lot about the speaker since it tells you what the spectral content of the direct sound is. And that will impact the sound. Whether or not the listener is beyond the critical distance, and how important the direct sound is, is really very dependent on the setup, room size, RT60 etc. In my room I am seated at just about the critical distance. Certainly most monitoring is done at less than the critical distance.
If you want an average impulse response of the speaker then you could start with the power response and perform an inverse FFt on it to get a corresponding impulse response. But this actually masks the contribution of the direct sound and more heavily weights the off axis response. Consider a situation where the speaker sounds too bright. You need to know why. Is it because the speaker has an inherently flawed response with a peak in the 5K area, as would be revealed by the anechoic, on axis response, or is the problem because the room is to live? If the problem lies with the on axis anechoic response then that is where it should be addressed and doing so will correct any imbalance in both the direct and reverberant field. However, if the problem is with the room and the reverberant field then it is the room which should be corrected. I think this is a common problem at many live venues, let alone home environments. If the latter case were corrected via changing the speakers response, assuming flat on axis response to start with, then the sound may be corrected so that overall it doesn't sound hot or bright, but there will be something lost since the direct sound will no longer contain the correct spectral balance. Such a correction could/would result in a loss of detail such understanding vocals.
But all this is getting way off the original topic of waterfall plots and transient response.
Room interactions are a completely separate issue than whether the system is fundamentally a linear system. I would expect that we all agree that the sound of the room is probably a dominant factor in the listening experience. And power response is of course another importance issue. But again it has more to do with the choice of drivers, crossover point, speaker configuration, etc. than whether or not the system is linear. That the system behaves as a linear system over its intended operating range is still a prerequisite. However, the on axis response of a speaker (impulse or otherwise) does tell you a lot about the speaker since it tells you what the spectral content of the direct sound is. And that will impact the sound. Whether or not the listener is beyond the critical distance, and how important the direct sound is, is really very dependent on the setup, room size, RT60 etc. In my room I am seated at just about the critical distance. Certainly most monitoring is done at less than the critical distance.
If you want an average impulse response of the speaker then you could start with the power response and perform an inverse FFt on it to get a corresponding impulse response. But this actually masks the contribution of the direct sound and more heavily weights the off axis response. Consider a situation where the speaker sounds too bright. You need to know why. Is it because the speaker has an inherently flawed response with a peak in the 5K area, as would be revealed by the anechoic, on axis response, or is the problem because the room is to live? If the problem lies with the on axis anechoic response then that is where it should be addressed and doing so will correct any imbalance in both the direct and reverberant field. However, if the problem is with the room and the reverberant field then it is the room which should be corrected. I think this is a common problem at many live venues, let alone home environments. If the latter case were corrected via changing the speakers response, assuming flat on axis response to start with, then the sound may be corrected so that overall it doesn't sound hot or bright, but there will be something lost since the direct sound will no longer contain the correct spectral balance. Such a correction could/would result in a loss of detail such understanding vocals.
But all this is getting way off the original topic of waterfall plots and transient response.
SY said:
The 30-w54 isn't known as one of the best 12" woofers ever made, fur nuthin'. I've been using a pair for quite some time. Since long before they were deleted.
SY said:
And I should get off my lazy butt and finish the speakers that use the drivers that prompted the original post 😉 They've been sitting there with no backs on them since Sep last year 🙄
I can't beleive it is nearly three years ago I started asking questions about them 😱
Tony.
- Status
- Not open for further replies.