Why should they be symmetric? I see no reason to assmue that they would be and lots of reasons they wouldn't be, but the real answer is, as usual, "it depends" - what is the driving signal? If its a simple turn on, turn off then the rise and fall ARE NOT symmetric, but I could design a signal envelope such that they were.
I don´t understand all that paranoia about control...
In any design, the control you get is the control you design for. Calculate Xcursion at a given Freq and Spl and you will know if things are under control enough or not, whatever the box and the number o peaks on the impedance curve...
Or am i missing something ?
It was sarcasm.
I never believed in profiling as the main means for decision, I also understand that misuse of such is really descructive rather than productive. Focus on the issues rather than become a cult follower would provide benefits to the technology.Now he's reduced to quoting himself.
I've learned a lot from John over the years. Anyone who may question his bona fides should go to his website and review the papers he's written over the years. I used his Excel spreadsheet, that he now provides for free, to design my current dipole system, using the UE as the crossover.
Then go to the website of fntn and check his background. Actually, I know of none and in fact only know of a web moniker. Here at diyAudio and over at the PE board, there's zero by way of profile info. Total web anonymity.
It's sad to see some of the nonsense of the sort paraded by fntn. Those of you who are still confused remain so because of that nonsense. Yet it is allowed to stand by most here. Sad, very sad.
You'll have to decide who you think is providing accurate information.
Dave
In the real world, nothing is perfectly linear. Often linear conditions can be assumed in a reasonable range of application so simplify a problem and get reasonably accurate results.
Whenever we did analysis on anything, we always start out with linear model just to make sure that the design is in the ball park of what we are aiming for. This minimizes time to develop an implementation topology. Once this is done, depending on how critical the vibration is to the final design, we decide whether it's more cost effective to just make and test and fix, or go into non-linear modeling.
Last edited:
Sorry about that. Once I get involved, the thread is most likely to be on the first page every day.
Last edited:
True, but if the system isn't linear within such conditions then you can throw away virtually all of your engineering undergrad texts because they are all based on linear systems theory. We assume linearity simply because to do otherwise is akin to starting everything all over from scratch.
If you read about nonlinear systems (and I suggest the book by Martin Schetzen if you dare to attempt this) then you realize how difficult things become very quickly.
Then there is the approximation of "small nonlinearities" - this is were we can at least assume that the nonlinearities are such that the cross terms are negligable. To go beyond this you get into a level of math that is simply untrackable by humans and must be done numerically.
Hi,
There are papers by Jens Blauert (mostly obscure and in German) and of course Dieter Ennemoser's book added to which are many personal conversations. These together with hearing physiology research results informed my precise choices of frequency.
BTW, the speaker originally had a 2KHz 4th order LR (acoustic slope) filter and sounded rather good with it, but in certain rooms there where real problems with sound quality. The later "brickwall filter" crossover cured this problem and sounded even better.
In the interest of disclosure, i normally only use first and zeroth order crossovers...
Ciao T
There are papers by Jens Blauert (mostly obscure and in German) and of course Dieter Ennemoser's book added to which are many personal conversations. These together with hearing physiology research results informed my precise choices of frequency.
BTW, the speaker originally had a 2KHz 4th order LR (acoustic slope) filter and sounded rather good with it, but in certain rooms there where real problems with sound quality. The later "brickwall filter" crossover cured this problem and sounded even better.
In the interest of disclosure, i normally only use first and zeroth order crossovers...
Ciao T
That is exactly what I mean. So rather than just talk about linearity, it makes more sense to discuss how much non-linearity presents effects to what magnitude. Klippel presentations have some BL non-linearity audio effects demonstration, although it was a VERY non-linear case, but we got the idea.
Confusion over the presence of linear conditions or lack thereof seems to be very prevalent - even among folks who fancy themselves as "experts". As Linkwitz seems to do, I maintain that a vigilant attention to detail as to the physical cause of any particular observed distortion is warranted before we assume it can be readily dealt with with simple tools like equalization. When in doubt, the standard tests establishing the properties of additivity and homogeneity should be applied. Linkwitz's warning about conducting tests for the presence of non linear distortion is in my view warranted and correct. The assumption (as appears to be the case with others in this thread) that the bulk of measured distortion is linear and can be corrected with tools like EQ is not grounded in reality. Moreover, most simple forms of distortion testing do not adequately reveal non linearities that are very much a part of the systems we work with on a daily basis. The results we hang our hat on are indeed very much skewed by our particular perspective and focus.
The point of my simple "mass on a perfectly linear spring" example was to explore the question of whether nonlinearity is inherent in the simplest and most ideal of resonant systems, or whether nonlinearity kicks in only when we deviate from such ideality. The consensus, although FNTN appears not to agree with this position, seems clearly to be that there is nothing inherent in resonant systems that makes them nonlinear. That's certainly my understanding.
A real world case that I think supports this position is quartz tuning forks of the type used in watches. They are resonant systems with Q values of several thousand because of their very low damping. If you drive one of these piezoelectric devices on resonance with a small sinusoidal signal (5 mV) and measure the amplitude of the response, and then double the drive (10 mV) and remeasure, and then repeat this doubling and measuring many more times, you'll find that over a large range of drive voltages the response varies linearly with the drive voltage. That's linearity.
It's true that most loudspeaker drivers are less linear than quartz tuning forks, but the fundamental fact remains: resonant systems need not be nonlinear.
And for the record, while I assume the point was made as a light-hearted jab, I'll add that I'm a contributor to Maine's educational system, not a product of it. Any stupidity I exhibit is the fault of genetics, laziness, and the educational systems of New York, Illinois, and California.
A real world case that I think supports this position is quartz tuning forks of the type used in watches. They are resonant systems with Q values of several thousand because of their very low damping. If you drive one of these piezoelectric devices on resonance with a small sinusoidal signal (5 mV) and measure the amplitude of the response, and then double the drive (10 mV) and remeasure, and then repeat this doubling and measuring many more times, you'll find that over a large range of drive voltages the response varies linearly with the drive voltage. That's linearity.
It's true that most loudspeaker drivers are less linear than quartz tuning forks, but the fundamental fact remains: resonant systems need not be nonlinear.
And for the record, while I assume the point was made as a light-hearted jab, I'll add that I'm a contributor to Maine's educational system, not a product of it. Any stupidity I exhibit is the fault of genetics, laziness, and the educational systems of New York, Illinois, and California.
Last edited:
The presence or lack of a high q for a particular resonance does not guarantee high levels of energy storage or low levels of damping at resonance that often result in non linearity. To achieve high levels of energy storage - the levels that result in such non linear conditions as suspension overload and hysteresis, a force vector must be operated through a distance (creating work and energy). The energies and damping involved in quartz oscillation do not involve such conditions - particularly over the range of excitation forces you specified. Moreover, if one were to change the frame of reference - examining the linearity of behavior of the quartz oscillator over a wide range of excitation frequencies, one would see an enormous "non linearity" in the transition from the non oscillation region to the oscillation region. With the loudspeaker, fundamental resonance is considered a region of instability that exists outside the expected passband where such levels of instability are not prevalent. In the quartz oscillator, the "passband" is considered the region of oscillation. The problem that precipitated much of this discussion was the uncontrolled energy present with the driver operating in the region of transition to resonance. The capacitor alone was not capable of controlling the loudspeaker without the LCR trap because of stored energy in this "transition region". You've offered an apples and oranges comparison with the express purpose to prove a flawed point. The devil is in the details.
In many cases linearity is assumed because, even thou it only covers ~5% of problems, it can be handled by math we understand (admittedly great strides have been made in the last 50 years on math to cover non-linear problems)
dave
There you go again. You simply will not stop changing context and making new, false and specifically dishonest claims as to what has been said and I'm tired of it. The one thing, the primary thing that was said is that non-flat FR is a linear phenomenon and that this non-flat FR can be completely and nearly perfectly EQ'ed. You claimed, rather dramatically, that it could not be done. John even went so far as to provide proof with a severely non-flat FR driver.
An externally hosted image should be here but it was not working when we last tested it.
This included the breakup, rather severe breakup. The non-flat range was more than 20db, yet the EQ'ed response was nearly perfect.
An externally hosted image should be here but it was not working when we last tested it.
Now you claimed quite explicitly that filling in a high Q dip would add significant distortion. John pointed out the fact that a crossover cannot add distortion. That was yet another of your claims that is false. He then provided proof with before/after distortion measurements. None of that has changed. The only change was your later equivocation.
Looking back I note that you completely ignored John's second set of measurements, no doubt because they totally contradict your claims. I would ask others here to review the measurements and his page with descriptions and see the evidence for yourself.
Both John and I have repeatedly pointed out that non-linear distortion is not controllable via EQ. You have, however, repeatedly implied that we said otherwise. This is simply being dishonest. I for one would appreciate an honest debate. Up to now, you have not seen fit to do so and I'm tired of that.
You've also mis-quoted Linkwitz as when you said "Linkwitz stated clearly that the application of EQ to amplitude response anomalies is effective only when non linear distortion is not present." Those are your words, not his and are not accurate. Provide a link if you think you have accurately quoted him because I have no doubt that he would not say this. I searched his site and did not find it or anything like it. Were that true, no amount of EQ of any sort could be applied to any driver. But we know that's not the case. Find the reference or retract it.
Since you have a thing for quoting Linkwitz, I'll do so again. Right at the top of the page you referenced he says this:
Either a driver is linear or Linkwitz is wrong with regard to an impulse response.
I'll quote again the first sentence of your reference page:
But you continue to claim that it's not linear. Hmmm.
To add to that, again from Linkwitz:
Stop with the false and dishonest comments.
Dave
Last edited:
You mean you can't address the detail points, avoidance is apparent in your replies.
I've got to say, you have all the characteristics of past posters such as "Auracle" and his other incarnations. Amazing similarities. Ah, it was Villastrangiato. Has he returned?
DAve
Last edited:
This is true. So when we do equalization like used in UE, we have to measure various locations to help identify how to equalize to provide optimum performance. This is much faster than trying to do a full scale non-linear analysis. Equalizing to a simple standard location does not seem to obtain optimum results. Additionally, the noise level during measurement is also going to play a significant role in effecting equalization.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.