http://www.avsforum.com/avs-vb/showthread.php?t=1148476&page=11
--Titta vad jag hitta om stereo system compensation(-discussion)---
#313 | Link
rblnr
Advanced Member
Join Date: Mar 2001
Location: New York, NY
Posts: 766
I asked Ingvar about the 'non-schoolbook function of the crossover' btw -- here is his answer:
- What is meant by 2-7khz, non-schoolbook function for crossover type. Can you elaborate?
Well, it is about many things, but to start with, the crossover has many jobs. Accomplishing the cross over function moving the energy from one driver to another when frequency change and protecting the tweeter from overloading is just two of them. These are some of the other:
1. To communicate with the drivers with an impedance optimized to lower distortion and minimizing compression. (The distortion of the QM10 is more then 4 dB lower in the midrange and up, than if the drivers had been driven directly from amplifiers! (i.e. had been a conventional active design. The thermal compression is less than half. This is also true for the active version that is on its way, since it is not a classical active design. It preserves all the above mentioned advantages from the present crossover design.)
2. The crossover in the QM10 does also control the vertical dispersion so that the timbral properties are modulated to make the speakers disappear as sounding object in the soundstage. It also enables the speaker to perform the same task for listeners positioned at different heights (like in a stepped multi row seating). To perform that “trick”, the speaker has to change character with height, to avoid letting the ears detect them as sound sources, and that trick should optimally be covering the range 800 – 7 kHz, but it’s just not possible with the QM10, because the tweeter just can’t perform at such low frequencies. In practice, it does not mean all that much since the ears are easily fooled by the dominating higher frequency clues, which for psychoacoustic reasons wins over the lack of height modulated information between 800 – 2000 Hz.
3. A large part of the stereo system compensation is performed by using the baffle shape, or rather by letting the corners of the speakers add some timbral shaping anti phase reflections. Actually, this part of the performance can only be done in this manner, out in the acoustic decoder world which is entered by the loudspeaker. Attempts to do it in the uncomplicated one dimensional “trafer link-world” before the speaker is impossible, since any timbral change made there would be shaping the sound radiated in ALL directions from the loudspeaker. However, only the sound going direct from the loudspeakers to the listeners must have the stereo system correction (to ensure that a psycho acoustically linear response is at hand, i.e. that the response at the ear drums is the same from the phantom images, as they would have been from real life sound sources being moved though the sound stage) since reflected sound is reaching the pinnas in random directions naturally, and are in no need of compensations from stereo system timbral flaws. BUT – the baffle of the speaker can only do that much. The rest of the task is to create an “invisible sound” and must be done by other means. One of them is performed by the crossover, controlling not only the amplitude from each driver, but also the way the drivers work together. But not even that is enough. In the QM10, the woofer actually has to come back into action at around 6,5 kHz, adding a part of the correction aimed basically only frontally, than can not be done by the much to dispersive tweeter.
Here is a link to a few of the other questions I asked (se nedan)
__________________
see my reviews/articles at
http://www.positive-feedback.com
HT/Music w/Scaena
-------------------Mera av samma goda Särimmer kaka-------------------------
n a followup to his review, Robert Learner interviews Guru QM10 designer Ingvar Ohman to find out just how these black boxes with 4" woofers do what they do...
Robert: The drivers are specially made for the QM10 -- what are some of the distinctive features of them?
1. The membrane is made of a dual layer pulp-composite to make sure that the sound travels slow enough in it, to make the radiated acoustic wave from it bend enough to cooperate with the tweeters generous dispersion.
2. The motor system is of a fairly long stroke design (+/- 5 mm lin.) for its size and for its wide band purpose. This is essential for its ability to play low frequencies without causing severe modulation of the midrange (a boomer design is more difficult to do than a pure squawker).
3. The surround has a half roll that is minimized, also to make sure that the modulation effects are haold down. The bigger half roll, the more the radiating area will change with the stoke (also at small strokes), so I always try to make the surround as small as possible, big enough to cope with the maximum linear stroke of the motor, but not more than about 50% more.
4. The motor system of the boomer driver is also strong, and actually, the pressure factor is very high (higher than most 18" PA-drivers!) with it's about 176 pa/V, and that means that it can push a sound pressure of 140 dB inside the box, at just about 1.1 volts input signal. This is what it takes to produce the extension you can hear from that little box. (Producing low frequency output is a product of input voltage X pressure factor X box volume X isotermization factor. Besides that, the only enemy is unwanted losses, which I try to keep low.)
5. The strong motor in combination with a relatively light moving system, enables the driver to produce a very high intrinsic sensitivity, which is necessary for a driver in a Guru-system since it incorporates compensations for both the stereo system and the room (thus need to have some head room for the "positive compensations"). The latter means that the speaker is using the room to its benefit at very low frequencies, but also that is has to cope with the non-constructive effects of the room. Most of them can be taken care of by adding some absorption behind the loudspeaker, but there is a range between 200 – 300 Hz where the room will contribute with destructive interferences. That means that the driver must have an intrinsic sensitivity of about 90 dB, to achieve the good system sensitivity (~86 dB) that we want and ended up with.
Robert: Can you give some details on how the helmholtz resonator concept is employed?
I don't know where to start, but the heart of the concept is to treat every aspect of the problem appropriately, which means – treat different acoustic dimensions differently - consider every aspect of the acoustic scene inside the loudspeaker. The inner behaviour of the box itself has to be very dead, to make sure that the mid range can play flawlessly, but at the same time; at the Helmholtz-resonance there must be as little acoustical losses as possible (ideally, 100% of the control should emanate from the boomer driver alone). So it's about engineering of the inner acoustics. Also, there is a considerable advantage in using a slit port in such a small box, and one reason is that in this specific design, the slit port for the H-resonator was easy to design to eliminate the "organ pipe resonance", i.e. the half wave length resonance. Another thing is, that the port length in combination with the port diameter if using a cylindrical port - would exhibit a severe shift of tuning frequency when coming close to overloading*, but the shift from the slit port is very small due to the lack of H/V-symmetry. The height is very small, and thus the rim radius can be small too, but still expanding the surface area significantly at the end.
And I always try to make port designs that will come close to overloading! That might sound strange, but in my experience, having a port overloading at approximately the same level as the woofer (or in this case boomer) does, will benefit the transient behaviour. Having a driver overloading before the port does, tends to colour bass transients because the Q of the H-resonator will rise significantly at driver overloading (since the mechanical impedance is rising when the surround is passing its linear stroke) causing an unarticulated ringing bass. It will not rise to its intrinsic Q (which is 7-9) but way above the much lower Q that is at hand when controlled by the woofer working within its linear range. In short: The port must take over the controller-job from the driver, when the latter is starting to overload.
Robert: What is meant by 2-7khz, non-schoolbook function for crossover type. Can you elaborate?
Well, it is about many things, but to start with, the crossover has many jobs. Accomplishing the cross over function moving the energy from one driver to another when frequency change and protecting the tweeter from overloading is just two of them. These are some of the other:
1. To communicate with the drivers with an impedance optimized to lower distortion and minimizing compression. (The distortion of the QM10 is more then 4 dB lower in the midrange and up, than if the drivers had been driven directly from amplifiers! (i.e. had been a conventional active design. The thermal compression is less than half. This is also true for the active version that is on its way, since it is not a classical active design. It preserves all the above mentioned advantages from the present crossover design.)
2. The crossover in the QM10 does also control the vertical dispersion so that the timbral properties are modulated to make the speakers disappear as sounding object in the soundstage. It also enables the speaker to perform the same task for listeners positioned at different heights (like in a stepped multi row seating). To perform that "trick", the speaker has to change character with height, to avoid letting the ears detect them as sound sources, and that trick should optimally be covering the range 800 – 7 kHz, but it's just not possible with the QM10, because the tweeter just can't perform at such low frequencies. In practice, it does not mean all that much since the ears are easily fooled by the dominating higher frequency clues, which for psychoacoustic reasons wins over the lack of height modulated information between 800 – 2000 Hz.
3. A large part of the stereo system compensation is performed by using the baffle shape, or rather by letting the corners of the speakers add some timbral shaping anti phase reflections. Actually, this part of the performance can only be done in this manner, out in the acoustic decoder world which is entered by the loudspeaker. Attempts to do it in the uncomplicated one dimensional "trafer link-world" before the speaker is impossible, since any timbral change made there would be shaping the sound radiated in ALL directions from the loudspeaker. However, only the sound going direct from the loudspeakers to the listeners must have the stereo system correction (to ensure that a psycho acoustically linear response is at hand, i.e. that the response at the ear drums is the same from the phantom images, as they would have been from real life sound sources being moved though the sound stage) since reflected sound is reaching the pinnas in random directions naturally, and are in no need of compensations from stereo system timbral flaws. BUT – the baffle of the speaker can only do that much. The rest of the task is to create an "invisible sound" and must be done by other means. One of them is performed by the crossover, controlling not only the amplitude from each driver, but also the way the drivers work together. But not even that is enough. In the QM10, the woofer actually has to come back into action at around 6,5 kHz, adding a part of the correction aimed basically only frontally, than can not be done by the much to dispersive tweeter.
Robert: How do the above (and anything else) contribute to the huge soundstage and powerful base these squat little boxes throw?
I've touched some of it already, but I can go into some more detail I guess. One of the things I've tried to accomplish is to give the speakers the power to communicate not only to a single listener in a "sweat spot" position, because even if I do like the word "sweat", I'm not very fond of the word "spot". How many ears can be fitted in a spot? Not even two! And in real life music situations - I've never seen people fight to get the sweat spot chair!
Obviously, something is very wrong when such a need is accepted in the world of reproduced music. So what I try to accomplish is speakers that can be listened to basically from any point within the width of the speakers themselves. That does basically mean, that the speakers, besides what is already mentioned regarding vertical dispersion properties, also must have a phase stabile behaviour within a fairly large horizontal window, and they also need some horizontal directivity. If all these qualities are at hand, they will perform to the task of opening a holographic window rather than a 3D-window, to the recorded world. At least if toed in so that the time-intensity-trading can be balanced to make the HAAS effect loose its typical dominance. Normally, that means that a listener sitting in the middle, will actually see the outer sides of both speakers.
Regarding performance in the low frequency range, it is all about throwing nothing away. I do not know how many times I've heard people saying that the QM10 seem to have a performance that defies the laws of nature. That is not true of course, but I try to get really close to the limit of what is possible (closer than most are used to hear, and that can obviously create the impression that something defying the laws of nature is going on).
There is a mathematical/physical limit to what can be accomplished that basically says that a bigger box can go either deeper in the bass, or be given a higher efficiency, a speaker that has a higher efficiency must sacrifice its size or extension, and a speaker that has a better extension must have sacrificed its size or its efficiency. But the limit differs depending on type of box. An optimal H-resonator is better than a normal symmetrically damped bass reflex box, which is better than a transmission line box, which is better than a closed box… Most designs have more than a few dB to the limit, some have 10 dB to go. I try to get much closer than that. Also, designing the speakers to work close to a wall, and at a specified height above the floor, makes it possible to take the support from these surfaces for granted, and the speaker can use the equation in a… twisted way. For normally designed loudspeakers (made with no room taken into consideration) each octave of extension costs 9 dB, but not when room support is a part of the equation (and speaker wall reflections above the constructively interfering range are damped), you get at least 9 dB for free (which in itself can be used for getting an extra octave of range). But also that extra octave typically gets you down into the range where the room is starting to be a reactive acoustical load (when more and more of the surface reflections are starting to get in phase which each other) and you can actually get even more than the extra octave, without scarifying the sensitivity. At least in medium sized rooms.
Robert: Thanks for the time!
Comments
it's always interesting to hear the designers perspectives in their work... than again, put ten in a room and you'll get ten different answers! I do have to say the Guru's sound unlike anything I have heard at their size...
Posted by: Danny Kaey | March 25, 2008 8:14 AM
agreed, and it's particularly true of speakers I think -- it's where you find the most striking design variation. I find this particularly interesting: ' In the QM10, the woofer actually has to come back into action at around 6,5 kHz, adding a part of the correction aimed basically only frontally' it means that the crossover throws a narrow frequency band back at the woofer that is otherwise in a range entirely covered by the tweeter. Not sure how many speakers do something like this.
Posted by: Robert | March 25, 2008 8:38 AM
I have heard the swedish original of this (INO PI60) and while they sound good one has to wonder about the absurd cost. A couple of nondescript cones in a plastic-covered MDF (?!) box for $2.5 k?! No better than a Snell Type K but 5 times the cost. Ridiculously overpriced for what you get...
Posted by: Davip | April 27, 2008 10:28 AM
Agreed that they aren't cheap, and that the box is nothing special on the outside. That the dollar has tanked doesn't help the price. Nondescript cones -- does it matter what gets you there? What I get most from Ingvar and his answers is that he takes a very holistic approach to speaker design and that he's got a lot of tricks up his sleeve -- tricks that come from deep understanding of how loudspeakers reproduce sound and how we hear it. This is a classic case of 'more than the sum of its parts' to my ear.
Posted by: Robert Learner | April 28, 2008 8:03 AM
Davip (or anyone else) can certainly be forgiven for questioning the value of the QM10. At first blush, one might assume that a speaker with an MDF cabinet, non-exotic drivers, no wood veneer, etc., retails at over $2000 in order render a huge profit margin for the seller. However, a close look at the speaker and what goes into it will obviate such an assumption. For several reasons - prominently the precisely curved vent opening, absolutely necessary for optimum performance - the cabinet can only be constructed with computer-aided machinery that resides only in first-world countries. The GURU boys have expended a lot of time and energy attempting to locate appropriate cabinet makers in lower-cost locales, to no avail. The expensive paint is also used in Hasselblad cameras - and the space shuttle! Resistant to temperature extremes, this paint is highly scratch-resistant: the GURUs will look fresh well into the future. Also, and more important, the paint precisely seals all seams in the cabinet, preventing resonance. The midrange/woofer unit is Ingvar's proprietary design: it's NOT a stock driver - and it's made in Denmark (the tweeter hails from Germany). Non-stock drivers from Europe are, needless to say, much more expensive than stock drivers made in Asia. Too, the GURUs are limited-production items from Swedish cottage industry. While economies of scale may at some point in the future obtain, for now the QM10s are crafted in small batches, in a first-world country, by people who have universal health care, a squeaky-clean environment, and virtually no poverty in their midst. (If you've never visited Sweden, please know that it's nigh-on a paradise - except for the weather!) Too, as Bob Learner points out, the precipitous fall in the dollar - 30% vs. the krona in just eighteen months - has driven the price of all imported goods inexorably higher. One can only hope that a new administration and Congress next January will take the necessary steps to strengthen our currency. It bears mention that, being an extremely easy load to drive, the QM10 will perform admirably even with cheap amplifiers. At shows, we take great pleasure in shocking attendees with just how good the speakers sound with a $400 integrated amp and CD player, $30 speaker cables, and $95 interconnects. This is a system that the average guy can afford - and that rivals the performance of gear costing multiples of the price. And most important: while $2000+ may seem like a lot of money for what one gets in terms of drivers, cabinet, etc., I'll gladly put the GURU QM10 up against anything at or near the price, performance-wise. Those whose primary criterion is performance will consider the QM10 a bargain at the price. If one seeks a good-looking piece of furniture, something science-fiction in appearance, or something BIG, one is obviously best advised to look elsewhere. Finally, I assure you that both the factory and ourselves as importers are working on a significantly smaller profit margin than is typical for loudspeakers. We really are doing our best to keep the cost affordable!
Posted by: Lars Erickson | May 10, 2008 1:08 PM
-Kanske något av ovanstående avsågs, i HiFi+ eller bara rumors (eng)-
