So Fourier was wrong? A music signal satisfies the criterion for Fourier analysis: no more than a finite number of finite discontinuities; a music signal has no discontinuities. If you put the music on continuous repeat play then it even satisfies the requirement for a Fourier series (rather than a transform): periodicity.45 said:
I never said it was. You said it was, when you used the word "proportional". Maybe you didn't realise what you were saying?
Sorry! He claimed that a system with two different linear forces creates "non-linear" distortion, which is nonsense.jcx said:
It might be instructive to look at global NFB in tube amplifiers from historic perspective. Looking through many 1930s and 40s schematics, I haven't seen a single pentode output amplifier with global NFB of the kind we are accustomed too. Not in car radios, home radios, and even what was considered "high end" of the time.
Philco Radio Schematics. Philco. Free Printable Wiring Diagrams Database
Antique Radio Forums • View topic - Magnavox speaker hint
Examples of "high end" are Magnavox Regency Symphony and Belvedere radio consoles. I had a chance of listening to one of these radios, and the sound was gorgeous in all departments, including tight articulated bass. The amplifiers in these radios were PP 6L6G or PPP 6V6 - with no global NFB.
So why the advent of global NFB in the 50s and 60s?
I think there were two reasons. First, invention of acoustic suspension and tuned port enclosures. These enclosures, which exploited resonance at low frequencies, allowed good bass reproduction from a smaller size cabinet. The paper by Thiele and Small, providing theoretical base for enclosure calculations, ushered the belief that boxed speaker is the only design option, and that earlier designs are passe. Boxes need a lot of electric damping and require amplifiers with low output impedance.
The second was birth of Hi-Fi. Hi-Fi required bandwidth in excess of the audible range and distortion levels in fractions of percentage points. Anything less than that was looked down upon with contempt as Lo-Fi. Since global NFB is very handy in achieving these goals, it became not only the mainstream design practice, but something synonimous with Hi-Fi.
Old time pentode amplifiers without global NFB worked well because they were used with the kind of speakers that is different from what we use today. First, enclosures were open baffles, not resonant boxes. Drivers were large diameter (12" or 15"), and small excursion (Xmax). Speaker cones were lightweight by today's standards, and cone suspension had better damping than today's foam or butyl surrounds. Magnavox 12" full range speakers used in their consoles had small (about half inch) light voice coils. The best of these drivers used field coil magnets. A field coil powered by low impedance supply can be regarded as shorting turn around voice coil providing a degree of electric damping. To sum up, these were low moving mass, low excursion mechanically dampened drivers used in non-resonant enclosures.
Philco Radio Schematics. Philco. Free Printable Wiring Diagrams Database
Antique Radio Forums • View topic - Magnavox speaker hint
Examples of "high end" are Magnavox Regency Symphony and Belvedere radio consoles. I had a chance of listening to one of these radios, and the sound was gorgeous in all departments, including tight articulated bass. The amplifiers in these radios were PP 6L6G or PPP 6V6 - with no global NFB.
So why the advent of global NFB in the 50s and 60s?
I think there were two reasons. First, invention of acoustic suspension and tuned port enclosures. These enclosures, which exploited resonance at low frequencies, allowed good bass reproduction from a smaller size cabinet. The paper by Thiele and Small, providing theoretical base for enclosure calculations, ushered the belief that boxed speaker is the only design option, and that earlier designs are passe. Boxes need a lot of electric damping and require amplifiers with low output impedance.
The second was birth of Hi-Fi. Hi-Fi required bandwidth in excess of the audible range and distortion levels in fractions of percentage points. Anything less than that was looked down upon with contempt as Lo-Fi. Since global NFB is very handy in achieving these goals, it became not only the mainstream design practice, but something synonimous with Hi-Fi.
Old time pentode amplifiers without global NFB worked well because they were used with the kind of speakers that is different from what we use today. First, enclosures were open baffles, not resonant boxes. Drivers were large diameter (12" or 15"), and small excursion (Xmax). Speaker cones were lightweight by today's standards, and cone suspension had better damping than today's foam or butyl surrounds. Magnavox 12" full range speakers used in their consoles had small (about half inch) light voice coils. The best of these drivers used field coil magnets. A field coil powered by low impedance supply can be regarded as shorting turn around voice coil providing a degree of electric damping. To sum up, these were low moving mass, low excursion mechanically dampened drivers used in non-resonant enclosures.
No Fourier was right for what he did. What he didn't do was to explain music!
If the relative amplitudes of the fundamental and the overtones change you get another music! It's not the same thing and it's not just a mathematical thing.
Your problem is that you want to talk about things that you don't know. As usual....
Once the driver has been excited using a low Zout amplifier there are two forces generated: one proportional to displacement (in reality is not truly proportional, it is only for small signal and it is a good approximation for high quality drivers but then things become too complicated surely for you that ignore the basics) and the other proportional to velocity that simultaneously move the cone in the opposite direction whose combination generates non-linear behavior. If the amplifier has infinite Zout the force proportional to velocity is zero.
Last edited:
And it's not distortion by the accepted definition. This looks like the whole "linear" distortion conversation over again. Any speaker or room will modify the amplitude and phase relationships of an instrument's fundamental to its harmonics, by convention this is not considered distortion. Distortion by a non-linear process can create new harmonics not present in the original.
It is generally called transient distortion. For sure when speaking of loudspeakers.
Yes any speaker will modify relative phases and amplitudes except that some are better (and good enough) and some are worse (not good). Good or not good also depends a lot on how you drive that speaker. This discussion into the discussion started because of the usual myth about low Zout amps and the non-existent problem of damping factor. It is a false problem. A high damping factor is not necessary, rather it surely the one that can achive the worst result!
Perfection would be nice but it does not belong to our world. Fortunately there is the usual threshold that saves us.The room is the same and in fact in early studies people could not detect anything because of the bad acoustics of the room messing everything up. The room problem is possibly more complicated because one needs to minimize that distortion without transforming a normal and familiar environment into an unreal one (anechoic or even anything too close).
Last edited:
Thanks Scott for pinch hitting for SY.
While no guru and not an EE, I have an anecdote which helps me remember the concept.
I was a grad student, upgrading my experimental apparatus handed down from 2 PhD students. I was trying to improve the SNR, and looked at the input and output of some filters (this was long enough ago that you could only practically do high order real time filtering in silicon). At the research group meeting, I said "I'm going to remove these filters, look at what all that nasty phase shift is doing to the signal."
The profs and students all turn to look at me and simultaneously and say loudly "But it's linear!"
Maybe only a nerd would think it's funny...
While no guru and not an EE, I have an anecdote which helps me remember the concept.
I was a grad student, upgrading my experimental apparatus handed down from 2 PhD students. I was trying to improve the SNR, and looked at the input and output of some filters (this was long enough ago that you could only practically do high order real time filtering in silicon). At the research group meeting, I said "I'm going to remove these filters, look at what all that nasty phase shift is doing to the signal."
The profs and students all turn to look at me and simultaneously and say loudly "But it's linear!"
Maybe only a nerd would think it's funny...
What is? Transient distortion is non-linear distortion that is triggered by a transient signal; as Scott noted, it is adding harmonics that were not present in the original signal, at the moment a transient signal occurs.
I don't think that is what you mean when you say 'it's called transient distortion', so it appears your use of the term is wrong. But maybe you can clarify.
Jan
Quite a few people (those who don't get what Fourier and Laplace is all about) mistake the visual representation of linear distortion as "non-linear transient distortion". Say, when you start a sine wave feeding a RC-Highpass, the output waveform initially looks distorted (and you may actually measure some apparant nonlinear distortion when you haven't given the circuit the time to settle, a typical error for SPICE beginners).
I think I stated it clearly. Transient distortion is simply the inability to follow the signal. It's usually represented in the time domain.
This is due to the fact that the driver does not behave linearly and the amplifier is unbable to correct it or makes it even worse. Do you want to call it differently? Call it transient response. In amplifiers it should be the slew rate???
Everything that alters the signal by a certain degree is a form of distortion. It is quite obvious for me.
Maybe you read a different book than the rest of us. I think this discussion leads nowhere.
Good luck to you.
Jan
Good luck to you.
Jan
A music signal is a mathematical thing. If it were not, we would not be able to produce it using physical artifacts (usually known as instruments) and we would not be able to reproduce it using different artifacts (usually known as an audio system).45 said:
You are still clinging to your view that two linear forces can combine to produce non-linearity. This is simply wrong. If you had said from the beginning that the various forces in a speaker are individually somewhat non-linear then I would agree.45 said:
There is still room for a few more in my ignore list so don't push me.45 said:
A high damping factor is needed for conventional speakers, which is all I was saying. Of course there are other ways of achieving the same result or a different result.
Probably you are right the discussion leads nowhere but not because I am reading a different book. I am listening to a different system where a few things are not accepted as impossible limitations to overcome.
If you can demonstrate what a listener feels by Fourier analysis you certainly get a Nobel price. You confuse physical sound with musical sound.
Again. Doesn't occur to you that in presence of back EMF the other term simply becomes non-linear and without back EMF it could be approximated as linear within certain limits. That's the problem that has to be solved in driver design (or driver selection if a simple user) to make the current amplfier work better! All the rest has secondary importance.
Please note who insulted another member first. Using the term distortion in this way becomes so confusing that discussion is pointless. Something that can be undone with a simple linear process is not distortion as the term is commonly used. Music put through an RIAA pre-emphasis returns to what is was going through an RIAA pre-amp. For your example, say you run a 10G data stream through 300' of twisted pair and the eye pattern is totally bad. In many cases an equalizer will fix this almost perfectly.
- Status
- Not open for further replies.