Panomaniac,a 10% difference might arise by careless throwing of +/- 5% tolerance components in the crossovers.However,a serious manufacturer will never do this especially on higher cost products.Instad,he will match his components on a pair level,achieving tolerances close to zero for the pair.This attitude of some to consider manufacturers as crooks or fools.................
Tarasque,
Since most loudspeaker drivers are operating near or at "stall", and have relatively low BL, how significant would the back EMF actually be?
Bob Carver has used this to great effect in his Sunfire subwoofers, but he uses the backEMF of a specially designed woofer to counteract the current draw from the amplifier. In effect the amp "sees" a 4 ohm load, and produces over 100V across it, or 2500W, but the back EMF reduces the current so that the load is only drawing approx 300W.
In order for the back EMF to be significant, however, the magnet is 10x larger than a standard subwoofer driver, and the Xmax is over 1.25 inches!
Also, you mention that the "full loop gain" has to be considered, but I'm not sure what the feedback mechanism would be to allow signal back into the input stage of the output transistors?
Since most loudspeaker drivers are operating near or at "stall", and have relatively low BL, how significant would the back EMF actually be?
Bob Carver has used this to great effect in his Sunfire subwoofers, but he uses the backEMF of a specially designed woofer to counteract the current draw from the amplifier. In effect the amp "sees" a 4 ohm load, and produces over 100V across it, or 2500W, but the back EMF reduces the current so that the load is only drawing approx 300W.
In order for the back EMF to be significant, however, the magnet is 10x larger than a standard subwoofer driver, and the Xmax is over 1.25 inches!
Also, you mention that the "full loop gain" has to be considered, but I'm not sure what the feedback mechanism would be to allow signal back into the input stage of the output transistors?
SY said:
And manufacturers find the best load and power level they can!
It was just an example. Progress IS being made. I do have some nice AP power measurements that show differences in amps. You know, THD vs. power from 20Hz to 20Khz. Those show some interesting facts.
But I don't know of any FR measurements that are done into a reactive load with anything other than a sine wave. Can you point me to some? They would be very informative.
Any Stereophile review.
edit: Well, not any review. But most of them.
Here's a few examples:
http://stereophile.com/tubepoweramps/404ear/index4.html
http://stereophile.com/tubepoweramps/670/index5.html
http://stereophile.com/solidpoweramps/1103pass/index4.html
http://stereophile.com/solidpoweramps/1093bryston/index3.html
edit: Well, not any review. But most of them.
Here's a few examples:
http://stereophile.com/tubepoweramps/404ear/index4.html
http://stereophile.com/tubepoweramps/670/index5.html
http://stereophile.com/solidpoweramps/1103pass/index4.html
http://stereophile.com/solidpoweramps/1093bryston/index3.html
Here is what I think...
The audio chain has to only deliver the acoustical energy that it is suppost to deliver...
What do I mean by that?
- Have a look at a pulse response of a speaker. after the pulse is gone, the diagram is still moving.
To move the diagram, its mass needs to be accelerated and decellerated. This causes "smearing" when the stored energy is released acoustically. A well desighned amplifier could help dissipate this energy.
- A similar effect is caused by the backemf in power amplifiers. Also the stored energy in semiconductors is contributing as well as all distortion energy
- Again similar effects are seen in sub-optimal desighned DA converters where digital filtering and oversampling is used.
Why would you care about that?
If you look at the laudest part of the audio signal compared to the average signal you will start to understand that music is very VERY dynamic. As soon as your system is not capable of loosing the energy stored in the reproduction chain fast and without acoustical trace, the system will produce acoustical "smear" and will not sound nice.
Did you ever manage to smoke your tweeter? I will bet that that was caused by an amplifier that was unable to deliver enough peak energy to the speaker. The distortion rises and so the amount of average energy fed to the tweeter, hence it starts smoking
How do you measure the stored energy in your system?
I have no clou!!!
I don't either, thus my hurried edit. I haven't read a Stereophile in some years, but the measurements were always the most interesting part, especially when a reviewer raved over a dog and Atkinson would have to do some shufflin'.
p10, that's what I'd expect from a quality manufacturer and that's better consistency than Audax or Dynaudio drivers that I've bought from hobbyist dealers.
I haven't read a Stereophile in some years, but the measurements were always the most interesting part, especially when a reviewer raved over a dog and Atkinson would have to do some shufflin'.p10, that's what I'd expect from a quality manufacturer and that's better consistency than Audax or Dynaudio drivers that I've bought from hobbyist dealers.
I had a look at the measurements. Love those simulated speaker load FR plots!! Ha!! didnt have one for the Bryston.
Might go a long way in expaining why amps sound different. All the amps where pretty flat 20-20K into pure resistive loades. But not into a reactive load. How does the loudspeaker translate this wavey plot into sound energy? Is it close to linear? How do cables fit into that?
The other tests were pretty standard. THD vs power, S/N. IMD spectrum. All very informative.
Just goes to show what happens when a complex load is used instead of a simple one. What happens when a complex signal is used? I'd love to see a test like that. But that's another thread.
Might go a long way in expaining why amps sound different. All the amps where pretty flat 20-20K into pure resistive loades. But not into a reactive load. How does the loudspeaker translate this wavey plot into sound energy? Is it close to linear? How do cables fit into that?
The other tests were pretty standard. THD vs power, S/N. IMD spectrum. All very informative.
Just goes to show what happens when a complex load is used instead of a simple one. What happens when a complex signal is used?
I'd love to see a test like that. But that's another thread.
The reason you get those wavy curves is because of the output impedance of the amp . At the speaker terminals you will get this even if the speaker cables have a significant resistance and the amp has very low output impedance.
The output impedance of the amp ( and speaker cable impedance ) along with the speaker system impedance acts as a voltage divider . If either one or both are not purely resistive , you will get a wavy curve for the signal level at the speaker terminals.
Can make amplifiers sound different assuming other parameters are very similar at normal listening levels.
The output impedance of the amp ( and speaker cable impedance ) along with the speaker system impedance acts as a voltage divider . If either one or both are not purely resistive , you will get a wavy curve for the signal level at the speaker terminals.
Can make amplifiers sound different assuming other parameters are very similar at normal listening levels.
My appologies if this seems OT, however I can't help but see the similarities between this "blind test" and what we've been discussing all along.
Expectation cannot be ignored!
http://zappinternet.com/index.php?video=qujCzaYluQ
Expectation cannot be ignored!
http://zappinternet.com/index.php?video=qujCzaYluQ
Hi Panomaniac!
Your post #1019 - apology for not quoting; trying to keep space down.
You got me on 2 points! Firstly sorry for not including complete analysis, will come back on that. Then mentioning that we are on "opposides of the earth", I had visions of a totally wasted argument on my side. You got me on that one! Appreciate your humour.
Firstly I do take it that whatever disagreement there exists between science and subjectivists (and I am not labelling you as one), we will be respectful enough not to force basics (like the Law of Gravity, Ohm's Law etc.) into untenable moulds. My quoted fractional dB figure is purely an Ohm's Law result. The parameters existing re cables are real measured ones, easy as far as resistance is concerned, and comparatively so regarding L and C (at least determining worst case). It is hardly possible with modern instruments to make a significant mistake there. Again I will suffice to say that measured L and C in loudspeaker cable do not come into the picture until close to or in the MHz region. (The figures are buried somewhere in the past 1018 posts, but I will state them again should you want.)
The consequence is that we can ignore the L-C cable parameters as far as audio is concerned, despite what promotion says. Now about the dB. One can take a real life cable resistance for not too long domestic use as 0,2 ohm - again measureable. The lowest dc resistance in an 8 ohm loudspeaker driver is of the order of 5.6 ohm. If we ignore amplifier output impedance (i.e. =0) and series resistance of cross-overs (again for worst case), then the minimum total resistance through which the signal current will flow must be 5,6 ohm. Adding a cable to that of R=0,2 ohm, makes it 5,8 ohm. The ratio of 5.8/5,6=1,036, or converted to decibels by the equation of 20 log [1.036], gives 0,3 dB. That will then be the maximum difference that such a cable can make, OK?
Regarding the often denigrated use of "simple" sine waves for testing: A mathematician named Fourier showed that any periodic signal can be respesented by the right series of sine waves of appropriate magnitude and frequency. Limited by 20 KHz this becomes relatively easy. However improbable this appears, we are listening to sine waves and however wonderfully complex music may look to us on an oscilloscope, the audible frequency is limited to 20 KHz and the above theorem is true. (There is hardly room for error here too; it has been used thus since the beginning of audio.) Furthermore, any half-decent amplifier does not exhibit sudden jumps of amplitude and phase angle. A frequency/phase plot pretty well gives us the relevant information.
This is not theory-up-in-the-sky, Panomaniac. It has been verified with real-life situations ad nauseam. The models we use for synthesis are deduced from real-life products and checked against that. Especially in the digital era this is not at all difficult.
Regarding damping factor (also a parameter of exaggerated practical importance): The classic definition of loudspeaker impedance/amplifier output impedance is meaningless. There are other resistances in the circuit. Even if the amplifier output impedance is zero, there will be at least the driver dc resistance still in the circuit as pointed out above. The real damping figure will never be lower than about 1,5! In this context it is also clear that practical cable effect cannot make any meaningful difference. (This has also been illustrated by digital analysis using real circumstances and music.)
As an aside, the only difference that could be made here is by using combined negative and positive feedback (often called motional feedback) to make an amplifier's output impedance negative, thus cancelling some of that 5,6 ohm. But that is a very specialised design case.
I cannot recall asking for "proof" by cable believes, and I have done my share of really puzzling tests. I only said that if under identical conditions cable differences cannot be identified by the same subjects, then they probably do not exist!
I myself have experienced tests where measured and audible results did not "agreee". But that is too wide a field to explain here "in ten easy lessons"; hopefully you will appreciate that. What were the conditions, what was measured and how...... But you very truthfully said that often "they were not measuring everything that's important". That does not mean though that it is not possible to do so! Looking at published measurements and comparing with the listener's comments (e.g. in Stereophile) it is mostly quite evident what the situation was.
I am again occupying more than my fair space. Just touching on those figures with the many zeros after the decimal point. That was most likely total harmonic distortion (THD), which is also of limited meaning especially with semi-conductor devices. It brings in the importance of higher order harmonic products - but that requires a chapter on itself.
Hope this somewhat restored your confidence in us vilified, lonely engineers! Thanks for your post.
Regards
Edit: Oops! Somewhere in the middle: The real damping figure will never be higher than 1,5 ....
Your post #1019 - apology for not quoting; trying to keep space down.
You got me on 2 points! Firstly sorry for not including complete analysis, will come back on that. Then mentioning that we are on "opposides of the earth", I had visions of a totally wasted argument on my side. You got me on that one! Appreciate your humour.
Firstly I do take it that whatever disagreement there exists between science and subjectivists (and I am not labelling you as one), we will be respectful enough not to force basics (like the Law of Gravity, Ohm's Law etc.) into untenable moulds. My quoted fractional dB figure is purely an Ohm's Law result. The parameters existing re cables are real measured ones, easy as far as resistance is concerned, and comparatively so regarding L and C (at least determining worst case). It is hardly possible with modern instruments to make a significant mistake there. Again I will suffice to say that measured L and C in loudspeaker cable do not come into the picture until close to or in the MHz region. (The figures are buried somewhere in the past 1018 posts, but I will state them again should you want.)
The consequence is that we can ignore the L-C cable parameters as far as audio is concerned, despite what promotion says. Now about the dB. One can take a real life cable resistance for not too long domestic use as 0,2 ohm - again measureable. The lowest dc resistance in an 8 ohm loudspeaker driver is of the order of 5.6 ohm. If we ignore amplifier output impedance (i.e. =0) and series resistance of cross-overs (again for worst case), then the minimum total resistance through which the signal current will flow must be 5,6 ohm. Adding a cable to that of R=0,2 ohm, makes it 5,8 ohm. The ratio of 5.8/5,6=1,036, or converted to decibels by the equation of 20 log [1.036], gives 0,3 dB. That will then be the maximum difference that such a cable can make, OK?
Regarding the often denigrated use of "simple" sine waves for testing: A mathematician named Fourier showed that any periodic signal can be respesented by the right series of sine waves of appropriate magnitude and frequency. Limited by 20 KHz this becomes relatively easy. However improbable this appears, we are listening to sine waves and however wonderfully complex music may look to us on an oscilloscope, the audible frequency is limited to 20 KHz and the above theorem is true. (There is hardly room for error here too; it has been used thus since the beginning of audio.) Furthermore, any half-decent amplifier does not exhibit sudden jumps of amplitude and phase angle. A frequency/phase plot pretty well gives us the relevant information.
This is not theory-up-in-the-sky, Panomaniac. It has been verified with real-life situations ad nauseam. The models we use for synthesis are deduced from real-life products and checked against that. Especially in the digital era this is not at all difficult.
Regarding damping factor (also a parameter of exaggerated practical importance): The classic definition of loudspeaker impedance/amplifier output impedance is meaningless. There are other resistances in the circuit. Even if the amplifier output impedance is zero, there will be at least the driver dc resistance still in the circuit as pointed out above. The real damping figure will never be lower than about 1,5! In this context it is also clear that practical cable effect cannot make any meaningful difference. (This has also been illustrated by digital analysis using real circumstances and music.)
As an aside, the only difference that could be made here is by using combined negative and positive feedback (often called motional feedback) to make an amplifier's output impedance negative, thus cancelling some of that 5,6 ohm. But that is a very specialised design case.
I cannot recall asking for "proof" by cable believes, and I have done my share of really puzzling tests. I only said that if under identical conditions cable differences cannot be identified by the same subjects, then they probably do not exist!
I myself have experienced tests where measured and audible results did not "agreee". But that is too wide a field to explain here "in ten easy lessons"; hopefully you will appreciate that. What were the conditions, what was measured and how...... But you very truthfully said that often "they were not measuring everything that's important". That does not mean though that it is not possible to do so! Looking at published measurements and comparing with the listener's comments (e.g. in Stereophile) it is mostly quite evident what the situation was.
I am again occupying more than my fair space. Just touching on those figures with the many zeros after the decimal point. That was most likely total harmonic distortion (THD), which is also of limited meaning especially with semi-conductor devices. It brings in the importance of higher order harmonic products - but that requires a chapter on itself.
Hope this somewhat restored your confidence in us vilified, lonely engineers! Thanks for your post.
Regards
Edit: Oops! Somewhere in the middle: The real damping figure will never be higher than 1,5 ....
Guy's...A nice story that noone believed afterward.
During my time as researcher with the dutch PTT, I have arranged dubble blind tests with different speakers Synthese 1, Quad EL, and appogees (these were visible) different cables and different amplifiers.
We had a VandenHul cable, Monstercable and ordenary mainz cord. All of the same gau.
We also had a beard tube amplifier, the Aitos OTL, the Accuphase solid state amp and the Kenwood L01A.
The volumes of the different amps and speakers was carefully matches.
One of the goals was to see of speaker cable and/or amplifier couls be recognized in a doubble blind test.
Each combination was made active with different types of music 4 different times.
The outcome was no one out of 28 people was able to recognize any of the the cables. in total only 2 groups of amlpifiers were recognized, a group with high loopgain and a group with low loopgain.
Is this interesting or what?
(The test was conciddered reccognized if one coud recognize 3 out of 4 or more occurences)
During my time as researcher with the dutch PTT, I have arranged dubble blind tests with different speakers Synthese 1, Quad EL, and appogees (these were visible) different cables and different amplifiers.
We had a VandenHul cable, Monstercable and ordenary mainz cord. All of the same gau.
We also had a beard tube amplifier, the Aitos OTL, the Accuphase solid state amp and the Kenwood L01A.
The volumes of the different amps and speakers was carefully matches.
One of the goals was to see of speaker cable and/or amplifier couls be recognized in a doubble blind test.
Each combination was made active with different types of music 4 different times.
The outcome was no one out of 28 people was able to recognize any of the the cables. in total only 2 groups of amlpifiers were recognized, a group with high loopgain and a group with low loopgain.
Is this interesting or what?
(The test was conciddered reccognized if one coud recognize 3 out of 4 or more occurences)
It can easily be argued that response past 20 kHz is important, as there are difference products in music created above 20 kHz that fall into the under 20 kHz range and are audible. And if they are missing the ear can tell.
An interesting take by an engineer on what Fourier has to say
http://www.t-linespeakers.org/oddsends/mrFourier.html
And as a mathemetician who has at least grazed the surface of complexity theory, once you start taking something simple (ie a single sine wave) and add complexity (ie multiple sin waves), all sorts of seemingly strange & wonderous things start to happen (ie the Madlebrot set).
Here is a fairly simple test stimulus that a good friend has been championing for sometim that helps examine the complex....
http://www.t-linespeakers.org/fivecycles/
An externally hosted image should be here but it was not working when we last tested it.
dave
Planet10,
I will not argue with higher mathematics! (I did pass a course in Maths 4, but seeing as that was all of .....er, well, a lo-o-o-ng time ago...)
We are going a little off-subject, but I thought it would be recognised that I was making nominal statements. I believe they are still true if one takes say 30 KHz or even 40 KHz. (By the way, at that point, which normal loudspeakers will still act with accurate amplitude/phase coherence to do justice to your statement?)
The modulated sine analysis must be accepted (though I am a little worried by the contribution of the instant start/end from sine to straight line). But again, what is that attempting to show? By (again simplified) stating that sine wave tests are in order, I was referring to steady state sine waves up to the highest frequency (make that 40 KHz if you wish or whatever, but certainly not much higher). A distortionless analysis would show no higher order harmonic products, while a device contributing anything of its own would show extra harmonic components. So if these are absent, or low enough.....? And also including intermodulation tests?
It would appear that excellent sounding amplifiers show definable behaviour by the above tests. Are you saying that one is still missing something then? Such as? Since one is able to test with any number of frequencies, Mr Fourier's friend's later comments about phase shifts would have been intercepted exactly as they would be by using a single sine wave.
I accept that an e.g. violin could generate intermodulation products from supersonic vibrations e.g. by some non-linear action in the flexing of the wood, etc. But I would then expect these to arrive as the finished audible intermodulation products at the microphone, not as separate (super-audio) frequencies that have to be accurately amplified by the amplifier to intermodulate to audio products later. If that is not the case, how can we hear them in the live situation in the first place? I still see the picture as "limited" by the human hearing response, whatever leaves the instrument, if I make myself clear. If e.g. an amplifier can "process" that to acceptable accuracy, I have difficulty to perceive what we are missing here, Dave.
Regards.
I will not argue with higher mathematics! (I did pass a course in Maths 4, but seeing as that was all of .....er, well, a lo-o-o-ng time ago...)
We are going a little off-subject, but I thought it would be recognised that I was making nominal statements. I believe they are still true if one takes say 30 KHz or even 40 KHz. (By the way, at that point, which normal loudspeakers will still act with accurate amplitude/phase coherence to do justice to your statement?)
The modulated sine analysis must be accepted (though I am a little worried by the contribution of the instant start/end from sine to straight line). But again, what is that attempting to show? By (again simplified) stating that sine wave tests are in order, I was referring to steady state sine waves up to the highest frequency (make that 40 KHz if you wish or whatever, but certainly not much higher). A distortionless analysis would show no higher order harmonic products, while a device contributing anything of its own would show extra harmonic components. So if these are absent, or low enough.....? And also including intermodulation tests?
It would appear that excellent sounding amplifiers show definable behaviour by the above tests. Are you saying that one is still missing something then? Such as? Since one is able to test with any number of frequencies, Mr Fourier's friend's later comments about phase shifts would have been intercepted exactly as they would be by using a single sine wave.
I accept that an e.g. violin could generate intermodulation products from supersonic vibrations e.g. by some non-linear action in the flexing of the wood, etc. But I would then expect these to arrive as the finished audible intermodulation products at the microphone, not as separate (super-audio) frequencies that have to be accurately amplified by the amplifier to intermodulate to audio products later. If that is not the case, how can we hear them in the live situation in the first place? I still see the picture as "limited" by the human hearing response, whatever leaves the instrument, if I make myself clear. If e.g. an amplifier can "process" that to acceptable accuracy, I have difficulty to perceive what we are missing here, Dave.
Regards.
- Status
- Not open for further replies.