ZDL

[gedlee]

I contend that when the energy largely responsible for the diaphragm's motion is stored energy, the response to voltage input becomes non linear. Anyone who has bothered to use an active crossover or equalizer to boost response by even a 4 or 5 db at or near the driver's resonance band has witnessed a significant increase in distortion with a modest increase in actual acoustic gain at the selected frequency. The system is highly non linear because of the phenomenon of energy storage.

I contend that the energy in a speaker is almost entirely energy delivered to it by the source. The only place where this would not be the case is at a high Q resonance where the energy cannot leave as fast as it is delivered, which is, of course, why the response rises. It rises until the disipated energy equals the input energy - equlibrium - is this "stored energy"? I think that it is largely symantic. In a good driver, where there aren't any significant high Q resonances in the pass band virtually all of the energy in the system is that from the source. In the ideal case where there are no resonances with Q's above critical, then ALL of the system energy is from the source, nothing is retained for anything longer than a 1/2 period, it all gets disipated in less time. To me this means NO energy is "stored".

As to its being linear, your example is simply circumstantial. Sure the distortion will rise for some loudspeakers with a 4-6 dB boost at resonance, but not for all of them and not so in general. But of course it all depends on what the level of 0 dB is. At a high enough level ALL speakers will distort with 6 dB gain and at a low enough level no speakers will distort with a 6 dB gain. Your example is simply ill defined and does not prove your case.

 

[gedlee]

Earl, i do the simulations.

Thanks, I am only interested in frequency response - the rest can be assesed from that.

 

[DualTriode]

Hello All,
Thinking and summarizing in my head some of the flaming discussions here. If there is a systematic approach it is to list the assumptions and design goals and to start there. As the design progresses the flaming discussions are interesting rabbit trails at best otherwise distractions. Over time I have observed that some are unwilling or unable to leave their comfort zone. In terms of speakers one comfort zone is one amplifier, equalization is not allowed, every watt counts and Q:= 0.707 is the only value allowed. O.707 optimizes a level frequency response and bass extension. Another comfort zone is a brave new world of digital crossovers time domain corrections and equalization to compensate for lower Q’s . Then there is the rabbit trail of linier vs. non linier excursion of the cone. The T&S parameters assume small signal and small cone excursions, there is no nonlinearity here. Turn up the power and the cone suspension becomes nonlinear at the limits of excursion. Perhaps equally nonlinear at both ends of the travel perhaps not. There are plenty of rabbit trails here for the unwilling and the unable to travel.

Hey wait a minute, jumping to the other comfort zone briefly. The release information for the new version of UE states that it can program a Berhinger DCX2496 DSP crossover does anyone know how that is done?

BTW
Feelings trump facts, ask my wife. My point of view is that her feelings are just plain wrong or untested assumptions.
DT
All just for fun!

 

[mige0]

I contend that the energy in a speaker is almost entirely energy delivered to it by the source. The only place where this would not be the case is at a high Q resonance where the energy cannot leave as fast as it is delivered, which is, of course, why the response rises. It rises until the disipated energy equals the input energy - equlibrium - is this "stored energy"? I think that it is largely symantic. In a good driver, where there aren't any significant high Q resonances in the pass band virtually all of the energy in the system is that from the source. In the ideal case where there are no resonances with Q's above critical, then ALL of the system energy is from the source, nothing is retained for anything longer than a 1/2 period, it all gets disipated in less time. To me this means NO energy is "stored".

As much as we have a single point where we widely agree on - "stored energy" as an ill term in general that is - here again you get wrong on the time issue - plus your old habit comes through to confuse quality with quantity issues.

*If* you like to pick apart those "from where belongs" the energy in the system (which by itself is a more than crazy attempt IMO) and state any further that for you the "energy storage" exists only at higher than critical dampening Q's, this is - sorry - nonsense.
Simply because your limit of what is labeled "dissipation" (IMO another crazy attempt of characterization having developed in this thread) is by no means any logically - it could be set at whatever threshold just as well making it completely meaningless.

Besides that, your first sentence statement is ... - I mean - did you already read what you wrote?
Besides that, your second sentence statement .........
So is it for the third sentence statement .....


Oh boy
Michael

 

[DualTriode]

I contend that when the energy largely responsible for the diaphragm's motion is stored energy, the response to voltage input becomes non linear. Anyone who has bothered to use an active crossover or equalizer to boost response by even a 4 or 5 db at or near the driver's resonance band has witnessed a significant increase in distortion with a modest increase in actual acoustic gain at the selected frequency. The system is highly non linear because of the phenomenon of energy storage.

Thanks for finally getting specific about what you actually had a problem with instead of simply hurling labels at me.

Hello,
Where you are losing my consensus is here. I do not believe the effect you are describing is nonlinearity. I would agree to call it a loss of accuracy but more specifically I would call it intermodulation.
The speaker tends to ring and ring louder near the “resonant” frequency. Here is an example sit down at the piano keyboard as i just did and hold down the D key just above middle C and strike the D sharp key, the sound produced is an modulation of both the D and D sharp strings much the same as a jazz player looking for that ¼ step between the keys. That in my mind is not nonlinearity but is still a real effect.
If it is indeed broken and needs to be fixed perhaps a speaker with a lower Q and equalization to correct for the loss in amplitude.
DT
All just for fun!

 

[fntn]

I contend that the energy in a speaker is almost entirely energy delivered to it by the source. The only place where this would not be the case is at a high Q resonance where the energy cannot leave as fast as it is delivered, which is, of course, why the response rises. It rises until the disipated energy equals the input energy - equlibrium - is this "stored energy"? I think that it is largely symantic. In a good driver, where there aren't any significant high Q resonances in the pass band virtually all of the energy in the system is that from the source. In the ideal case where there are no resonances with Q's above critical, then ALL of the system energy is from the source, nothing is retained for anything longer than a 1/2 period, it all gets disipated in less time. To me this means NO energy is "stored".

As to its being linear, your example is simply circumstantial. Sure the distortion will rise for some loudspeakers with a 4-6 dB boost at resonance, but not for all of them and not so in general. But of course it all depends on what the level of 0 dB is. At a high enough level ALL speakers will distort with 6 dB gain and at a low enough level no speakers will distort with a 6 dB gain. Your example is simply ill defined and does not prove your case.

You confuse the issue with a lengthy diatribe about how long it takes to dissipate stored energy. The physical q is defined as the ratio of the energy that is stored (energy that cyclically converts from potential to kinetic) in the moving mass's motion versus the energy dissipated - PER CYLCE. In the end, no other timing aspect is relevant since physical Q is not defined over multiple cycles or some arbitrary time. The energy dissipated after stasis is reached much be equal to the energy supplied by the amplifier - obviously. This is precisely the problem I've been focusing on. Low energy supplied, low acoustic energy output - high energy stored in the moving mass's resonant motion.

You stated earlier that critical damping was achieved at .7 . That was incorrect as most any industry standard publication will attest. Critical damping means Q = .5. Unfortunately, most alignments are tuned for a Q of .707 to optimize bandwidth extension. This makes them susceptible to higher levels of energy storage and higher levels of 2nd and 3rd harmonic distortion. When one attempts to equalize or boost output in this region, the lack of control exerted on the cone (think of it as the ratio of amplifier power versus stored energy in the moving mass) produces higher non linear distortion than an alignment that is critically damped. Because the transmission line configuration focuses on critical damping and second order rolloff below resonance - it represents the most ideal of alignments in terms of reduced distortion and extended bandwidth.

I only offered my anecdotal experience with regard to the use of equalization and increased distortion. I never suggested that in all cases, EQ presented problems. To support my position, I merely cited Linkwitz' position with regard to investigating the presence of non linear distortion when doubts exist about the potential usability of EQ in any given circumstance. Clearly, for issues related to amplitude problems external to the transducer - EQ is likely to produce excellent results with little or no problems as demonstrated by the specific examples of Johnk. In neither case, were high levels of non linear distortion present either before or after the application of EQ - that aspect is very telling as DewardH seemed to suggest. With Joachim's build, I suspect that EQ will prove very valuable in successfully mitigating the effects of diffraction. Essentially, all I'm saying is that enthusiasm for UE has to be tempered with some realities. After a long winded discussion, it does appear that Johnk has begrudgingly acknowledged this as well.

 

[fntn]

Hello,
Where you are losing my consensus is here. I do not believe the effect you are describing is nonlinearity. I would agree to call it a loss of accuracy but more specifically I would call it intermodulation.
The speaker tends to ring and ring louder near the “resonant” frequency. Here is an example sit down at the piano keyboard and hold down the D key just above middle C and strike the D sharp key, the sound produced is an modulation of both the D and D sharp strings much the same as a jazz player looking for that ¼ step between the keys. That in my mind is not nonlinearity but is still a real effect.
If it is indeed broken and needs to be fixed perhaps a speaker with a lower Q and equalization to correct for the loss in amplitude.
DT
All just for fun!

I personally have witnessed 2nd and third harmonic distortion rise more than 7 db after bumping up gain 4 db on a speaker at approximately 150hz to help compensate for minor baffle step dropoff. This is a very common scenario. When harmonics are generated - it's referred to a non linear distortion. Your experience may be different. What I just told you is but one example and I'm sure there are many, many others. I'm not making any of this up. I have better things to do with my time.

 

[Joachim Gerhard]

Earl, i retrospect i am happy that you alerted me to do the simulation. The Peerless SLS P830668 i am using is now produced by Tymphany and the parameters i find on their webpage are cosiderably different. They call that woofer now Rev.1 so it has being changed. I end up with more box volume then i thought and the isobaric simulation makes no sense at all to me. I will now run in a woofer and measure the parameters myself. That needs some time.

 

[Joachim Gerhard]

Running in the woofer.

 

[soongsc]

Joachim, I thought measurement and generation of T/S parameters would be a standard procedure in any development effort, especially in the driver selection phase, and sample check for every batch of drivers that may come in. You surprise me.

 

[Joachim Gerhard]

Soon, usually with woofers from Peerless i had excellent agreement with the published parameters and the German importer measures them again. I stubled over this Rev.1 anouncement and then i found changed parameters. Actually they are not that diffent but the Qms went up and the Qts too. So this driver has less mechanical loss what is a good thing. The price is higher volume i need. In this bigger cabinet the -3dB frequency is lower. It seems that the driver was optimised more for closed box alignement. One situation i have at the moment is that i do not find enough time for DIY. I would have put this woofer into a test cabinet anyway and at that point in time i whould have found out. Sure, this is no excuse for this faux pas.

 

[DualTriode]

Hello,
Where you are losing my consensus is here. I do not believe the effect you are describing is nonlinearity. I would agree to call it a loss of accuracy but more specifically I would call it intermodulation.
The speaker tends to ring and ring louder near the “resonant” frequency. Here is an example sit down at the piano keyboard as i just did and hold down the D key just above middle C and strike the D sharp key, the sound produced is an modulation of both the D and D sharp strings much the same as a jazz player looking for that ¼ step between the keys. That in my mind is not nonlinearity but is still a real effect.
If it is indeed broken and needs to be fixed perhaps a speaker with a lower Q and equalization to correct for the loss in amplitude.
DT
All just for fun!

Hello All,
Outside the comfort zone and off the mark.
Intemodulation distortion is nonlinear distortion caused by two or more injected tones. When a single test tone excites and interacts with a natural resonance that is nonlinear distortion plain as simple even with perfect cone suspension. My hat is off!
I would like to see a demonstration / results.
If it is indeed broken to the point it needs to be fixed perhaps a speaker with a lower Q and equalization to correct for the loss in amplitude.

DT
All just for fun!

 

[Joachim Gerhard]

I did some simulation with the parameters i found on the Tymphany website. I will double check tomorrow when the woofer is run in. What works best is a single driver in a closed cabinet of 70 liters that is 100% stuffed. For simplicity sake i will try that first. Still i plan to mount the driver magnet up. See the simulation at 1W and the maximum output at 25W. The linear peak to peak throw is given at 16mm.

 

[gedlee]

Intemodulation distortion is nonlinear distortion caused by two or more injected tones. When a single test tone excites and interacts with a natural resonance that is nonlinear distortion plain as simple even with perfect cone suspension.

IM is simply the result of using two tones as input to the system instead of one - one can use ten tones as is often done. It's the same nonlinearity as harmonic distortion (a single tone) just a different way of looking at it.

"When a single tone excites and interacts with a natural resonance" it produces nonlinear distortion IF and only if the system is nonlinear. If the system is linear then there is no nonlinear output. Basically the "excites and interacts with a natural resonance" is a seperate issue from "nonlinearity" - they are not coupled as I have said many times before. One can happen or the other can happen or both can happen or none can happen - they are not related.

 

[john k...]

Where's the beef?

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.

 

[john k...]

I personally have witnessed 2nd and third harmonic distortion rise more than 7 db after bumping up gain 4 db on a speaker at approximately 150hz to help compensate for minor baffle step dropoff. This is a very common scenario. When harmonics are generated - it's referred to a non linear distortion. Your experience may be different. What I just told you is but one example and I'm sure there are many, many others. I'm not making any of this up. I have better things to do with my time.

Let me guess. You held the output level of the signal generator constant when you measured distortion. Thus the speaker was playing 4dB louder in the region where the EQ boost was applied and consequently the distortion went up. The cause was not the EQ but that speaker was reproducing the fundamental at a 4dB louder level and the driver excursion would close increase by 1.6 and an increase in NL distortion would be expected. Now go back, as I did in my tests, and match the level of acoustic output at the fundamental and look at the distortion. I will not change.

It is like what I said pages ago, unless the acoustic output at the fundamental is matched then this is just saying the driver produces more distortion when you turn up the volume.

This was the point I ws trying to make a long time ago. There are two references, distortion for a constant input level and distortion for a constant output level. If you put any type of EQ between in and out, then distortion for a constant input will potentially change. But if you map the distortion at some frequency, X, vs amplitude of the acoustic output at 1 M you will find that the distortion vs amplitude does not change regardless of what EQ is applied. The nonlinearity of the driver has not changed because of EQ. What changes is the map of the operation conditions. And while 7dB may sound like a lot is is only a factor of 2.23. What is pertinent is what level distortion that multiplies. If 10% goes to 20% not so good. If .1 % goes to .2% who cares?

These statement can not be made without proper context. "I saw a guy jump out of a plane an live." It looses a lot of WOW when I tell you the plane was sitting on the runway.

Anyway, I have exceed my allowed postes to this thear for the day.

 

[fntn]

Here's the beef.

 

[dlr]

Here's the beef.

The window you've used is totally inappropriate. Expand the time scale as you've set HOLM so that it is including both pre-impulse data (improper placement of the start time marker) and much room reflection (improper placement of the stop time marker). Expand the time scale, show the time scale and show the quasi-anechoic data with properly set windows (in-room it can't be much more than 3-5ms) and then we'll see what's really happening. What you show is meaningless as presented. It's indicative of someone unfamiliar with proper use of measurement software.

This also explains why the distortion above 300Hz appears to be improved by 5db in the "boosted" case.

Dave

 

[fntn]

What are you talking about? Distortion is not affected by window size. The specific window was selected to highlight the difference in amplitude at the frequency in which boost was applied.

 

[dlr]

What are you talking about? Distortion is not affected by window size. The specific window was selected to highlight the difference in amplitude at the frequency in which boost was applied.

You've selected "Amplitude" and "Distortion" for the curve display. The lower set would be the 2nd HD from what you show.

There is no high-Q change in FR, it' very broad as shown, but in any case, the FR that you show also was not made properly. Showing before/after is fine, but with the windows you've selected, the FR is contaminated with room response, so it does not isolate and represent the change in driver response.

Both the FR and distortion curves don't present meaningful data as presented. In fact, were that the case, one could try to make the claim that increasing boost at 150Hz in this driver improves the distortion of the driver above 300Hz by 5db or more. That alone should tell you something.

Dave