I.ve been reading a book on current drive of loudspeakers and one of the interesting things spoken of in that book is the properties of the suspension and changes as proposed of breaking in a speaker and changing the fs of the driver. What was found was that even though there was a shift in the fs, usually lower as the driver was exercised, and I note at not a very high level but enough to make a change, that after a time, and sometimes a fairly significant amount of time the actual original measured response returned. What this tells me as a plastics guy is that the material has what we call a memory and any deformation from excursion only changes the physical parameters temporarily. This does go along with Frank's, fas42's habit of running his speakers hard for a given time before listening to them critically. A temporary shift will occur that will be restored to original after the system has time to relax, I would attribute this to the heating of the material while it is being used. I will qualify that with the understanding that at some point we can permanently change those values if we exceed the capability of the material to return to its original state driving the device to a very high excursion. I do understand the nature of a spider, it is made of different forms of fibers, generally cotton or such, but the resin is a phenolic resin, a hard stiff material in any type of thick cross-section that can be changed by physically cracking or breaking the resin into discrete sections held together with the fiber reinforcement. This would be a permanent change in physical properties. This is an area that I plan to do some serious experimentation with, changing the base chemistry to create a more linear and stable end product. One of the last areas of loudspeakers that really haven;t had any serious changes in a long time.
I'm sure that Sy can expound on this material more than I can as I have not had to deal with phenolic resins on a professional level in plastics molding but do know others who have worked in that area of material science and application development.
I intend to work with some other materials that have the potential to further linearize this very non linear aspect of all current commercially made speakers that I am aware of.
I'm sure that Sy can expound on this material more than I can as I have not had to deal with phenolic resins on a professional level in plastics molding but do know others who have worked in that area of material science and application development.
I intend to work with some other materials that have the potential to further linearize this very non linear aspect of all current commercially made speakers that I am aware of.
Eperando,
Nice work on flattening the impedance rise at resonance with your LCR parallel circuit. My question for you is why not also correct the rising rate of impedance I see as the frequency rises. I do see it is really benign on your graph but wouldn't this also flatten the rising output at higher frequencies that you seem not to want when you create your downward slope? Would flattening the rising impedance rate flatten the output curve of the driver or any driver for that matter if the typical rising rate was not an issue?
Nice work on flattening the impedance rise at resonance with your LCR parallel circuit. My question for you is why not also correct the rising rate of impedance I see as the frequency rises. I do see it is really benign on your graph but wouldn't this also flatten the rising output at higher frequencies that you seem not to want when you create your downward slope? Would flattening the rising impedance rate flatten the output curve of the driver or any driver for that matter if the typical rising rate was not an issue?
I said "It is before to set the zobel to flaten inductance at HF.". Means the HF comp was not done yet when i was working on the motional. But, of course, i do the complete compensation on each speaker i use.
btw: "Enceinte" means in the same time "Pregnant" and "Enclosure", "speaker box", or 'Baffle".
Last edited:
keantoken, next you may have to bench test a few speaker units.
My measurements show that Fs decreases with signal amplitude (still within ‘small signal domain’, i.e. less than ¼ of max excursion).
George
Keantoken,
I would imagine that there is no temperature component in those static calculations, I would imagine that this is one of the missing factors in those simulations. Take a mechanical spring and rapidly cycle that spring and you would measure a temperature rise from mechanical motion. Now is there enough energy vs surface area of a spider and any temp rise may be insignificant but I would have to look at that with some IR measurements to qualify that phenomena.
I would imagine that there is no temperature component in those static calculations, I would imagine that this is one of the missing factors in those simulations. Take a mechanical spring and rapidly cycle that spring and you would measure a temperature rise from mechanical motion. Now is there enough energy vs surface area of a spider and any temp rise may be insignificant but I would have to look at that with some IR measurements to qualify that phenomena.
Last edited:
Fs is the impedance peak, so it will generate much less heat than elsewhere on the impedance curve. Furthermore, any heat that is generated I would not expect to reach the spider quickly if the test doesn't take long. That is just my guess though. If the spider itself generated much heat from friction then I would expect it to wear and break down over time. But, I must be missing something.
I see Paul McG commenting, should I ask him how the noise harvesting works? As I thought loosey-goosey with the concept of loss-less. The frequency domain pictures are essentially meaningless as if there is unused white space below the line. Sorry to report loss-less compression below the Shannon entropy limit is a violation of the second law, as we know it's a bitch.
Which is to say that if I ran say a JPEG image through the process I would get garbage out.
Last edited:
A couple thoughts on suspension non-linearity.
There is no perfect spring or perfect anything in the real world. Really good Springs (music wire) have a fatigue cycle and only last a long time (many operations) if they operate at a very low point on the curve. When you examine a driver suspension, one has a material that also has mechanical damping which is a real issue. The damping means motion is converted by mechanical resistance to heat internally and many materials have temperature dependent properties.
This can appear to be hysteresis, temporary changes in spring constant that change immediately after operation.
Not all problems are manifested equally;
For example, a low Qe, high Qm driver, with a low Fs, when put in a box small enough to raise the Qtb to .6 - .8, will have much less suspension contribution than a driver with a higher Fs and Qtb.
The reason is with the small sealed back volume, the air “spring” is far stronger than the suspension’s “spring”, the air (acoustic suspension) is normally a much more linear spring than a mechanical system.
For bass horn fans, one might notice this is the configuration one has in a modern / proper bass horn.
I wish I had more Time right now but I only seem to get away when I am either stumped or finished with something haha. I also wanted to comment earlier when waterfalls were being discussed as a possible indicator of loudspeaker weakness.
Time is a critically important aspect of things which is not well discussed and measurements like an RTA have no time discrimination and only see total level. I hate to sound like a Dick Heyser fan but I guess I am and would point to his observation that “we” are the ones who assign the frames of reference that help us understand things while nature has no frames of reference.
I see a problem with time, it’s why Group delay is not understood, why some people actually say they don’t like subwoofers because they have group delay.
Our hearing span, the accepted 20-20K represents a 1000:1 span in the size of the wavelength, from a fraction of an inch to many feet. Any filter function or “event” centered at one high frequency will always take proportionally longer, twice as long if an octave lower, 10 times longer if a decade lower in frequency etc. A subwoofer HAS TO HAVE a large group delay, one expected for that frequency and response shape.
So, how can one look at a broad band signal but treat each frequency with the same phase resolution instead of time?
The Wavelet approach is one way, the S Transform is another and in ARTA the cumulative burst decay also presents the time axis of the csd justified as wavelengths or wave periods.
A further investigation in an essentially unexplored area of loudspeaker time tied to radiation can be done with ARTA as well using the modulation transfer function measurements. The MTF is a very common for measuring optical resolution and it’s audio counterpart is used in the language independent predictor of intelligibility called STIpa.
The measurement is graph that shows the depth of modulation at a center frequency (say 1KHz) as it is modulated at increasing rates up to say 30Hz. If you have ARTA, try this one, try measuring up close and compare that to what you see as you move away and include more and more late information.
Ugh, back to work,
Best,
Tom
There is no perfect spring or perfect anything in the real world. Really good Springs (music wire) have a fatigue cycle and only last a long time (many operations) if they operate at a very low point on the curve. When you examine a driver suspension, one has a material that also has mechanical damping which is a real issue. The damping means motion is converted by mechanical resistance to heat internally and many materials have temperature dependent properties.
This can appear to be hysteresis, temporary changes in spring constant that change immediately after operation.
Not all problems are manifested equally;
For example, a low Qe, high Qm driver, with a low Fs, when put in a box small enough to raise the Qtb to .6 - .8, will have much less suspension contribution than a driver with a higher Fs and Qtb.
The reason is with the small sealed back volume, the air “spring” is far stronger than the suspension’s “spring”, the air (acoustic suspension) is normally a much more linear spring than a mechanical system.
For bass horn fans, one might notice this is the configuration one has in a modern / proper bass horn.
I wish I had more Time right now but I only seem to get away when I am either stumped or finished with something haha. I also wanted to comment earlier when waterfalls were being discussed as a possible indicator of loudspeaker weakness.
Time is a critically important aspect of things which is not well discussed and measurements like an RTA have no time discrimination and only see total level. I hate to sound like a Dick Heyser fan but I guess I am and would point to his observation that “we” are the ones who assign the frames of reference that help us understand things while nature has no frames of reference.
I see a problem with time, it’s why Group delay is not understood, why some people actually say they don’t like subwoofers because they have group delay.
Our hearing span, the accepted 20-20K represents a 1000:1 span in the size of the wavelength, from a fraction of an inch to many feet. Any filter function or “event” centered at one high frequency will always take proportionally longer, twice as long if an octave lower, 10 times longer if a decade lower in frequency etc. A subwoofer HAS TO HAVE a large group delay, one expected for that frequency and response shape.
So, how can one look at a broad band signal but treat each frequency with the same phase resolution instead of time?
The Wavelet approach is one way, the S Transform is another and in ARTA the cumulative burst decay also presents the time axis of the csd justified as wavelengths or wave periods.
A further investigation in an essentially unexplored area of loudspeaker time tied to radiation can be done with ARTA as well using the modulation transfer function measurements. The MTF is a very common for measuring optical resolution and it’s audio counterpart is used in the language independent predictor of intelligibility called STIpa.
The measurement is graph that shows the depth of modulation at a center frequency (say 1KHz) as it is modulated at increasing rates up to say 30Hz. If you have ARTA, try this one, try measuring up close and compare that to what you see as you move away and include more and more late information.
Ugh, back to work,
Best,
Tom
Are you serious?How come you have expensive amps but cheap speakers?
Strange as it may seem, the correlation between the price of a driver and it's quality is shaky at best. The components used in this kit are high end-ish for all intent and purposes. With speakers, 90 percent of SQ depends on system integration and enclosure size and shape.
P.s. this 90 percent is obviously pulled from a dark place, the point being that it is quite possible to get excellent sound from mediocre components, whereas the opposite is true as well.
I wish to say that many here, don't know much more than I do about speaker design, and most, probably less. I only mention this when people pose as speaker 'experts' and have strong opinions about other people's loudspeaker systems. For example, Focal tweeters. Yes they can be bright, but not necessarily. Amp quality and the modifications made by Wilson Audio seems to tame them significantly. The 'peak' was measured at 20KHz rather than 16kHz as mentioned, and I think that this is important. Not perfect, but better than most. At least it EXTENDS to 20KHz. I would prefer more.
Do-you remember this cheap Audax tweeter TW8B with a metallic cone, used by Elipson ?
It sounded very bad ! (omho)
I was not very satisfied with this one too: http://diyparadise.com/w/part-recommendation-fostex-ft17h-horn-tweeter/
Oh, by the way, the curve of my horn is not smoothed more than necessary to see the accidents in a comfortable and efficient way. Measurements for work and design are not commercial ones.
Last edited:
Of course, most tweeters don't even extend to 20KHz, but I think, for serious evaluation of electronics, it is necessary. Actually, I use my STAX electrostatic phones for the most serious listening. They will annoy me after a time, but they are clear! Normally, just like most here, I use a speaker that rolls off, the Sequerra Met 7, and I prefer the rolled off tweeter for extended listening of TV and radio.
I prefer it for all listening, I shudder to think what Frank would do to it. IIRC the much liked in the old days large Advent had like a brickwall filter at 17kHz.
Last edited:
Then why not be at least as suspicious of aging audio reviewers whose ears can't reach even close to 20kHz?
se
- Status
- Not open for further replies.