Home |
Search |
Today's Posts |
#1
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
(It was suggested that I bring my question over here....)
While so many people run around with their hands in the air talking about this amp and that amp, their distortion measurements, tube-vs-transistor, yadda-yadda-yadda, I'm concerned with something much more fundamental. If the phrase is true "all amplifiers cause distortion," which I believe is the case, my question is simple... How is it created? For instance, let's take the scenario of an all-analog, all pure class-A staged amplifier.. My *assumption* is that in an ideal model, this scenario would generate no distortion, but in using real-world components, distortion is still generated. I understand that there exist what are called "nonlinearities" in the amplifier, where at some input levels, a change of the input voltage causes a particular change in the output voltage, but at some other input level voltage, the same change in voltage (just offset from the original) would cause a different amount of change in the output. So is distortion's root this nonlinearity? And if so, why does this nonlinearity always manifest itself as n-order harmonics? And how does clipping come into the picture? I've read that class-A tube distortion is "more pleasing" because most of its generated harmonic content are low-order fundamentals with a steep rolloff (n 5), but push-pull (transistor-based but even apparent in push-pull tube) topologies have a less-steep rolloff, with harmonics still of decent amplitude even with the higher-order harmonics (n10). Even if that is assumed to be true, what causes the tube to have a steeper harmonic rolloff? One article I read seemed to imply it had to do with a tube being a "high impedance" amplifier. Not sure what that means, if you compare a 30W tube amp to a 30W transistor amp, what's different? Could you adapt a transistor-based circuit topology to act as a higher impedance amplifier? Some quick background-- I've got an EE degree in electrical and computer engineering with emhpasis on the digital realm of circuit design. But I've been trying to go back and "fill in the details" in the analog world due to my heavy interest in audio. So while I easily understand some EE topics, others I may not have as fundamental a grasp on. Any and all input would be greatly appreciated! ...dane |
#3
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Todd H. wrote:
So is distortion's root this nonlinearity? I'd say that's a good way to look at it. And if so, why does this nonlinearity always manifest itself as n-order harmonics? Great question. To appreciate it, math is involved, and that math involved functions that have squared terms in them, among other things. In the time domain, if you have a component whose transfer function introduces And how does clipping come into the picture? Take an input of sine(t). The ideal output would G*sine(t) where G is a linear multiplier representing the gain of the amplifier stage. Clipping results when the the amplifier runs into the supply rail. In the extremest case of clipping your sine wave looks like a square wave. If you do a Fourier transform on a square wave you get a very long equation that shows the square wave as a summation of sine waves all harmonically related t othe original. If the original input is at say 100Hz, the frequency components of the square wave will be weighted sum of 100Hz 200Hz 300Hz 400Hz, ad infinitum. I forget the specifics of the math, but mentally envision an equation that takes the original sine wave, and adds sine waves and successive harmonics. That's where you begin to appreciatiate how clipping introduces new frequenies in the signal that are multiples of the original. And hence, the term harmonic distortion. Thanks Todd for that reply. So let me rephrase so ensure I'm along the right path: - unwanted distortion is due to nonlinearities - some nonlinearities exist even within the "linear" operation range of the amplifier - some nonlinearities exist when approaching the real-world limits (power rails) of the amplifier - It is not that the amplifier does anything special to add distortion in terms of fundamental multiples, but rather when mathematically transformed into the frequency domain (FFT) the distortion is manifested that way. Another question that comes to mind, then, is that if a wave is made up of fundamental pure sines with different phases and frequencies (that makes sense, I knew that one already), I guess I'm wondering why amplifying, say a 1 kHz sine, doesn't introduce some 1.05 kHz sine as some distortion coefficient. It would seem to me that if distortion is caused by nonlinearities, then there must be an infininte collection of possible nonlinearities that could incur the creation of a harmonic of some decimal-multiple instead of whole-multiple of the fundamental. Your thoughts? ...dane |
#4
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
wrote in message oups.com... (It was suggested that I bring my question over here....) While so many people run around with their hands in the air talking about this amp and that amp, their distortion measurements, tube-vs-transistor, yadda-yadda-yadda, I'm concerned with something much more fundamental. If the phrase is true "all amplifiers cause distortion," which I believe is the case, my question is simple... How is it created? This might help: http://en.wikipedia.org/wiki/Amplitude_distortion This article has lots of links to other resources as well. |
#5
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
(Todd H.) writes:
And if so, why does this nonlinearity always manifest itself as n-order harmonics? Great question. To appreciate it, math is involved, and that math involved functions that have squared terms in them, among other things. In the time domain, if you have a component whose transfer function introduces .... non linear terms. x^2 terms and friends. Heh, evidently got lost in thought. -- /"\ ASCII Ribbon Campaign | Todd H \ / | http://www.toddh.net/ X Promoting good netiquette | http://triplethreatband.com/ / \ http://www.toddh.net/netiquette/ | http://myspace.com/mytriplethreatband |
#6
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
wrote in message ups.com... Todd H. wrote: So is distortion's root this nonlinearity? I'd say that's a good way to look at it. And if so, why does this nonlinearity always manifest itself as n-order harmonics? Great question. To appreciate it, math is involved, and that math involved functions that have squared terms in them, among other things. In the time domain, if you have a component whose transfer function introduces And how does clipping come into the picture? Take an input of sine(t). The ideal output would G*sine(t) where G is a linear multiplier representing the gain of the amplifier stage. Clipping results when the the amplifier runs into the supply rail. In the extremest case of clipping your sine wave looks like a square wave. If you do a Fourier transform on a square wave you get a very long equation that shows the square wave as a summation of sine waves all harmonically related t othe original. If the original input is at say 100Hz, the frequency components of the square wave will be weighted sum of 100Hz 200Hz 300Hz 400Hz, ad infinitum. I forget the specifics of the math, but mentally envision an equation that takes the original sine wave, and adds sine waves and successive harmonics. That's where you begin to appreciatiate how clipping introduces new frequenies in the signal that are multiples of the original. And hence, the term harmonic distortion. Thanks Todd for that reply. So let me rephrase so ensure I'm along the right path: - unwanted distortion is due to nonlinearities - some nonlinearities exist even within the "linear" operation range of the amplifier - some nonlinearities exist when approaching the real-world limits (power rails) of the amplifier - It is not that the amplifier does anything special to add distortion in terms of fundamental multiples, but rather when mathematically transformed into the frequency domain (FFT) the distortion is manifested that way. Another question that comes to mind, then, is that if a wave is made up of fundamental pure sines with different phases and frequencies (that makes sense, I knew that one already), I guess I'm wondering why amplifying, say a 1 kHz sine, doesn't introduce some 1.05 kHz sine as some distortion coefficient. It would seem to me that if distortion is caused by nonlinearities, then there must be an infininte collection of possible nonlinearities that could incur the creation of a harmonic of some decimal-multiple instead of whole-multiple of the fundamental. Your thoughts? No, a pure sine wave can only have integer harmonics. With a 1KHz pure sinewave input, 2KHz is the next lowest frequency distortion component you will ever see, no matter what. |
#7
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Karl Uppiano wrote: No, a pure sine wave can only have integer harmonics. With a 1KHz pure sinewave input, 2KHz is the next lowest frequency distortion component you will ever see, no matter what. Playing the child's voice... "why?" I do understand the definition of a harmonic is an integer multiple, but throwing away terminology for a moment, I do not understand why distortion is always manifested in integer harmonics. Is it just a mathematial fact, or a physics fact, or something else... ? And thanks for your Wiki link, I'll look into that and follow its links, also. ...dane |
#8
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
wrote in message oups.com... Karl Uppiano wrote: No, a pure sine wave can only have integer harmonics. With a 1KHz pure sinewave input, 2KHz is the next lowest frequency distortion component you will ever see, no matter what. Playing the child's voice... "why?" I do understand the definition of a harmonic is an integer multiple, but throwing away terminology for a moment, I do not understand why distortion is always manifested in integer harmonics. Is it just a mathematial fact, or a physics fact, or something else... ? Ecch. You would have to ask that :-) I could take the easy way out, and say it is just a physical fact (or a mathematical one, since the mathematics describes the physics) - which it is. But I should explain it better than that. If you feed a 1KHz sinewave into a nonlinear amplifier (that is to say, any amplifier), the non-linearity will bend and distort the original sinewave into a new shape. But the bends and distortions can only occur along the original sinewave, not along something else that was not there to begin with. Let's say that you *did* detect a 1.05KHz signal distortion product. That signal would be in phase at some point, but it would drift in and out of phase, resulting in a slowly changing amplitude at the output. The only way that can happen is if you put in two signals, or if the *nonlinearity* is constantly changing. But that contradicts our given conditions, that we put in only a single frequency, and the unspoken condition that the amplifier's linearity is a constant (it probably isn't exactly, but that's a different problem). On the other hand, any integer multiple of the 1KHz input sinewave will have a fixed phase relationship with the input, and the output waveshape will be constant. This is not a rigorous answer, but I think it provides an intuitive grasp of the big issues. |
#9
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Karl Uppiano wrote: No, a pure sine wave can only have integer harmonics. With a 1KHz pure Playing the child's voice... "why?" Ecch. You would have to ask that :-) Well, I am an engineer, and the son of an engineer... It is a phenomina that both my mother and my wife roll their eyes at, but accept as fact nevertheless. :-) Karl Uppiano wrote: I could take the easy way out, and say it is just a physical fact (or a mathematical one, since the mathematics describes the physics) - which it is. But I should explain it better than that. If you feed a 1KHz sinewave into a nonlinear amplifier (that is to say, any amplifier), the non-linearity will bend and distort the original sinewave into a new shape. But the bends and distortions can only occur along the original sinewave, not along something else that was not there to begin with. Let's say that you *did* detect a 1.05KHz signal distortion product. That signal would be in phase at some point, but it would drift in and out of phase, resulting in a slowly changing amplitude at the output. The only way that can happen is if you put in two signals, or if the *nonlinearity* is constantly changing. But that contradicts our given conditions, that we put in only a single frequency, and the unspoken condition that the amplifier's linearity is a constant (it probably isn't exactly, but that's a different problem). On the other hand, any integer multiple of the 1KHz input sinewave will have a fixed phase relationship with the input, and the output waveshape will be constant. This is not a rigorous answer, but I think it provides an intuitive grasp of the big issues. W-O-W. (big light blinds on above head) That's about the best explanation I could have asked for. You're right-- since the phase of the 1 kHz and 1.05 kHz signals would have to drift, one could not have been created "from" the other. Unless, as you theorized, the nonlinearity itself drifted (which can be assumed false for any sort of modern electronics, esp. over such a short timeperiod as the harmonic under discussion would exist in the system). I didn't figure on getting all my questions so thouroughly answered in one day. Now I'll have to let my brain work on it for a few weeks to ensure I retain it all. Thanks for all the help and great answers. ...dane |
#10
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Todd -- yours and Karl's replies have been extremely helpful in my comprehension of the issues around distortion, and fundamentally why it is created in the first place. I applaud you both and would buy you both steak dinners if I was able. thanks so much to you and everyone else. ...dane |
#11
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Todd H. wrote: Ah ha--Karl said the magic word here. I think a _time varying_ system is what's needed to generate frequencies other than harmonics. So basically you'd need to be playing with circuitry involving a 2nd signal source, delay lines, and/or an oscillator and such to create these non-harmonic frequencies. Which sounds like based on yours and Karl's posts, that if two signals are used, or a time-varying transfer function is used, either way we are beginning to transition into inter-modulation distortion aspects rather than harmonic distortion... If the input is stable but the reference changes (transfer function varying with time), or if the input is stable along with a secondary input, both situations will cause modulation distortion (beat noises, I think they are also called) in the output. Chorus effect explained (delay+pitch modulation mixed with original signal): http://www.harmony-central.com/Effects/Articles/Chorus/ A nifty bucket brigade device I've never seen befo http://www.doepfer.de/A1881.htm You would have to give me MORE references to look up, wouldn't you?? :-) Okay, on to my next question, but I'll post a new reply on that one. ...dane |
#12
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
wrote in message oups.com... (It was suggested that I bring my question over here....) While so many people run around with their hands in the air talking about this amp and that amp, their distortion measurements, tube-vs-transistor, yadda-yadda-yadda, I'm concerned with something much more fundamental. If the phrase is true "all amplifiers cause distortion," which I believe is the case, my question is simple... How is it created? Please see http://www.pcavtech.com/soundcards/techtalk/nlinear/ |
#13
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
|
#14
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Since you two (Todd and Karl) are being so responsive to my questions,
I want to pull another one in. I originally posted on rec.audio.opinion, but was suggested to move it over here so I did. But I did get a response over there that I wanted to quote here and ask about. Trevor Wilson wrote: Even if that is assumed to be true, what causes the tube to have a steeper harmonic rolloff? **Output transformer. Is that due solely to the frequency response (higher freq. rolloff) of the output transformer? IOW, if you used an "ideal transformer" with a flat FR from 0-infinity, would the tube amp's harmonic content be the same as a transistor amp (with no xfmr) of the same topology? Of course that would not make sense if the input frequency was relatively low in the audio band (say, 1 kHz), because surely the transformer would still exhibit fairly nice frequency response out to at least 10 kHz if not higher.... I suppose now that I understand "WHY" harmonics are created as whole-order multiples of the fundamental, now I'd like to consider topologies of audio amplifiers and their harmonic content differences. And here I'm not quite sure the questions to ask, so I'll just put back out there again that "I read somewhere" that a class-A tube amp (with output xfrmr) has a steeper harmonic content rolloff than a transistor-based push-pull amp with no xfmr. Granted there are at least three variables here (tube/transistor, class A/class AB, and xfmr/no-xfmr) if not more, but I'd like to dive in and learn what I can learn... ...dane |
#15
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Ron Capik wrote: Ah but don't let your brain rest too much yet. ;-) of course not! Non harmonic distortion can be created by microphonics in a vacuum tube. Forces can be created by the signal that cause the elements of the tube to move and thus create modulation effects linked to mechanical resonances in the tube ...not a likely effect for solid state devices. Can you provide any measured data about the magnitude of such modulation distortion? Don't assume I doubt you, I have heard the term microphonics with tubes and while I didn't know its definition, I can understand the principals you describe. I am curious though as to how much they really play a part (assuming of course the tubes are *physically* isolated from external mechanical forces like the rumbling of the speakers). Hmmm... that being said, I wonder if microphonics are more commonly caused from air coupling of the speaker outputs shaking the tubes (similar to a feedback equation with delay), and how much is due to internal system issues in-time with the signal being amplified? ...dane |
#16
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
wrote in message oups.com... Ron Capik wrote: Ah but don't let your brain rest too much yet. ;-) of course not! Non harmonic distortion can be created by microphonics in a vacuum tube. Forces can be created by the signal that cause the elements of the tube to move and thus create modulation effects linked to mechanical resonances in the tube ...not a likely effect for solid state devices. Actually, there are two effects (1) Modulation distortion in the form of modulation of an audio signal being amplified at the same time. (2) Noise caused by modulation of the bias voltages present in the tube. Can you provide any measured data about the magnitude of such modulation distortion? IME (1) is pretty small, and (2) is quite a bit larger. Don't assume I doubt you, I have heard the term microphonics with tubes and while I didn't know its definition, I can understand the principals you describe. I am curious though as to how much they really play a part (assuming of course the tubes are *physically* isolated from external mechanical forces like the rumbling of the speakers). Hmmm... that being said, I wonder if microphonics are more commonly caused from air coupling of the speaker outputs shaking the tubes (similar to a feedback equation with delay), Or mechanical vibration propigated through furniture and room structure. and how much is due to internal system issues in-time with the signal being amplified? Now we're back to the more common form of distortion due to the nonlinearity of the tube. |
#17
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Is it just a mathematial fact, or a physics fact, or something else...
? It's a mathematical fact, _if_ the transfer function of the device is a function of the input value, and only the input value. You can prove this via a simple method: that of contradiction. That is, assume that it's not true, follow this assumption logically, and show that the conclusions of this line of thought contradict one of your starting assumptions. So, let's assume the following three things: [1] The output value (i.e. the amp's output voltage) is a deterministic function of the input value (i.e. the input voltage), and nothing else. Vout = f(Vin) [2] Vin is a sinusoidal function of time, Vin(t) = sin(rate * t) where "rate" is a constant [3] Vout, the output, contains one or more sinusoidal frequency components which are not members of the set sin(rate * t) sin(2*rate * t) sin(3*rate * t) That is, Vout(t) contains some nonzero (and of course not necessarily integer) amount of sin(X*rate * t) where "X" is not an integer. Now, let's follow these assumptions forwards. Let's evaluate Vout at various times. Let's look at times t=Q and t=Q+2*pi, where Q is any starting time value you want to choose. Vout(Q) = f(sin(Q)) Vout(Q+2*pi) = f(sin(Q+2*pi)) Fair enough? We're simply evaluating the output of the amp's transfer equation, precisely one cycle apart in the input sinewave. Now, we can make use of the fact that sin(X) = sin(X * N*2*pi) for any integer N - that is, any sinusoidal function repeats its value once per cycle. This means that sin(Q) == sin(Q*2*pi) which further means that Vout(Q) == Vout(Q+2*pi) That is to say, since we're feeding the same value into the function at these two points in time, and since the function's output depends only on this one input (our assumption #1), we know that the function's output must be the same at these two moments. Here's where the fun kicks in. We've also assumed that the output function has some frequency component which is not an integral harmonic of "rate". In other words, we've assumed that Vout(t) = a*sin(rate*t) + b*sin(2*rate*t) + c*sin(3*rate*t) + ... + w*sin(X*rate*t)) where "w" is the relative amount of this particular non-harmonic. But, our knowledge of the behavior of the sin(x) function leads us to an uncomfortable conclusion. If X is not an integer, then sin(X*rate * t) != sin(X*rate * (t+2*pi)) except on special occasions. If there's any nonzero amount of such a non-harmonic in the signal, then its presence would cause the output voltage to be different at two points in the output cycle which are precisely one input period apart... points at which we know that the inputs to the transfer function are identical. Or, in other words, the presence of a nonharmonic component in the output means that Vout(Q) != Vout(Q+2*pi) So, combining our set of initial assumptions in different ways have led us to two conclusions: that the output voltages at two specific points in time must be identical, and must not be identical. We've got a logical contradiction. The initial assumptions, taken as a group, are incompatible with one another. The only way to resolve this contradiction, and create a consistent set of conclusions, is to eliminate one of the initial assumptions. We can either give up the requirement that the system be able to generate non-integral harmonics, or we can give up the requirement that the output be a function of only the input value, or we can give up the requirement that the input value be a strict sinusoidal function of time. -- Dave Platt AE6EO Hosting the Jade Warrior home page: http://www.radagast.org/jade-warrior I do _not_ wish to receive unsolicited commercial email, and I will boycott any company which has the gall to send me such ads! |
#18
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
wrote in message ups.com... Since you two (Todd and Karl) are being so responsive to my questions, I want to pull another one in. I originally posted on rec.audio.opinion, but was suggested to move it over here so I did. But I did get a response over there that I wanted to quote here and ask about. Trevor Wilson wrote: Even if that is assumed to be true, what causes the tube to have a steeper harmonic rolloff? **Output transformer. Is that due solely to the frequency response (higher freq. rolloff) of the output transformer? IOW, if you used an "ideal transformer" with a flat FR from 0-infinity, would the tube amp's harmonic content be the same as a transistor amp (with no xfmr) of the same topology? Of course that would not make sense if the input frequency was relatively low in the audio band (say, 1 kHz), because surely the transformer would still exhibit fairly nice frequency response out to at least 10 kHz if not higher.... I suppose now that I understand "WHY" harmonics are created as whole-order multiples of the fundamental, now I'd like to consider topologies of audio amplifiers and their harmonic content differences. And here I'm not quite sure the questions to ask, so I'll just put back out there again that "I read somewhere" that a class-A tube amp (with output xfrmr) has a steeper harmonic content rolloff than a transistor-based push-pull amp with no xfmr. Granted there are at least three variables here (tube/transistor, class A/class AB, and xfmr/no-xfmr) if not more, but I'd like to dive in and learn what I can learn... This starts to get into areas where I do not have specific knowledge about specific devices. Triode tubes and transistors have remarkably different transfer characteristics (bipolar junction transistors and field effect transistor curves are shaped a bit more like tetrodes, but you can't even take that to the bank). The differences in transfer characteristics alone are enough to give rise to different harmonic content. That is a reasonable starting point, anyway. My personal opinion (flame bait) is that both amplifier types (tubes or transistors), when linearized using properly-designed negative feedback should have such low distortion characteristics as to be indistinguishable. Perhaps theoretical or measurable, but my opinion (again, flame bait) is that if you can actually *hear* a difference between two amplifiers of similar power and bandwidth, then one of the amplifiers is poorly designed. :-) |
#19
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Arny Krueger wrote: Actually, there are two effects (1) Modulation distortion in the form of modulation of an audio signal being amplified at the same time. (2) Noise caused by modulation of the bias voltages present in the tube. Wouldn't a low-impedance bias driver help reduce the levels of (2) above? (Granted I don't know amplifier design theory, so I'm not sure that the bais point CAN be low impedance. But it would seem to me if the current flowing through the bias point was substantial (10x) compared to that flowing(leaking) into the tube bias, that the bias point voltage would therefore not change much as a function of the leakage through the tube grid. Or mechanical vibration propigated through furniture and room structure. of course; I assumed physical structure isolation, but did not mention it. and how much is due to internal system issues in-time with the signal being amplified? Now we're back to the more common form of distortion due to the nonlinearity of the tube. Ah. Very good. yes. ...dane |
#20
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Dave Platt wrote: ....[a lot!]... Wow, Dave. Excellent response. Well-done. Not much to reply to there, seems fairly straightforward (as most math proofs are, I suppose) ...dane |
#21
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
writes:
Thanks Todd for that reply. So let me rephrase so ensure I'm along the right path: - unwanted distortion is due to nonlinearities Yup. - some nonlinearities exist even within the "linear" operation range of the amplifier Yes. Because no matter how brilliant the design, how carefully biased the transistors are to work in the linear portion of their transfer curves, we're strill trying ot make linear output using non-linear devices. - some nonlinearities exist when approaching the real-world limits (power rails) of the amplifier Absolutely. - It is not that the amplifier does anything special to add distortion in terms of fundamental multiples, but rather when mathematically transformed into the frequency domain (FFT) the distortion is manifested that way. Or, alternatively, it's just the nature of the beast, that you're using nonlinear devices (transistors/tubes) doing everything you can to confine their state to the linear portion of their transfer curves, but necessarily you're going to have nonlinearities when the input pushes outside of the limits of "linear" operation. I put linear in quotes because it's not entirely linear. Here's a paper I found that talks in terms of the non-linearities introduced by the components and talks about things in the terms you're grappling with http://www.passlabs.com/downloads/articles/cascode.pdf Another question that comes to mind, then, is that if a wave is made up of fundamental pure sines with different phases and frequencies (that makes sense, I knew that one already), I guess I'm wondering why amplifying, say a 1 kHz sine, doesn't introduce some 1.05 kHz sine as some distortion coefficient. It would seem to me that if distortion is caused by nonlinearities, then there must be an infininte collection of possible nonlinearities that could incur the creation of a harmonic of some decimal-multiple instead of whole-multiple of the fundamental. Your thoughts? The problem is that to create arbitrary non-harmonic frequencies like that, you need a transfer function that is not readily available in the transfer characteristics of the components involved. I'd need to back to my notes of my third analog electronics classes and my communication engineering class to see the math again, but I think it's basically because the transfer function of transistors is polynomial, you're simply not able to create non-harmonic frequencies without really going out of your way to do so (e.g. synthesis). -- /"\ ASCII Ribbon Campaign | Todd H \ / | http://www.toddh.net/ X Promoting good netiquette | http://triplethreatband.com/ / \ http://www.toddh.net/netiquette/ | http://myspace.com/mytriplethreatband |
#22
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Todd H. wrote: IOW, if you used an "ideal transformer" with a flat FR from 0-infinity, would the tube amp's harmonic content be the same as a transistor amp (with no xfmr) of the same topology? No, I don't think so. #include the even'odd harmonics stuff. If it were this simple, you'd see transistor amps transformer coupled and sold as botique. :-) I've been convinced through my research thus far that the even/odd harmonic stuff between tubes and transistor amps is hogwash, that it's much () more related to the topology (class A, push/pull, feedback/no-feedback, etc) than the technology (tube vs transistor). Your last statement is what I'm getting at... Albeit added expense and unnecessary for sales, I wonder if a transformer-coupled push-pull transistor amp would sound more like a class-A tube amp. This of course assumes that a good bit of signal shaping is done by the transformer... which may itself be hogwash. ...dane on a never-ending quest |
#23
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Karl Uppiano wrote: This starts to get into areas where I do not have specific knowledge about specific devices. Triode tubes and transistors have remarkably different transfer characteristics (bipolar junction transistors and field effect transistor curves are shaped a bit more like tetrodes, but you can't even take that to the bank). The differences in transfer characteristics alone are enough to give rise to different harmonic content. That is a reasonable starting point, anyway. I guess we're close to me having to start digging up some datasheets.. :-) My personal opinion (flame bait) is that both amplifier types (tubes or transistors), when linearized using properly-designed negative feedback should have such low distortion characteristics as to be indistinguishable. Perhaps theoretical or measurable, but my opinion (again, flame bait) is that if you can actually *hear* a difference between two amplifiers of similar power and bandwidth, then one of the amplifiers is poorly designed. :-) I would agree with your assessment, with the only caveat that when the output approaches clipping, differences will occur due to nonlinearity differences outside of the feedback's area of operation. But that would be expected. What's interesting (e.g. yet another tangent road I'd like to travel down further) is that traditional tube designs, or so I hear, don't use feedback, but to design a transistor-based method without feedback would be dangerous (unstable). Again, haven't gone down that road yet, so that comment could be hogwash also. :-) I guess I gotta get home to let the dog out. Thanks for all the feedback today; back here on Monday. Have a great weekend. ...dane |
#24
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
"Karl Uppiano" writes:
Let's say that you *did* detect a 1.05KHz signal distortion product. That signal would be in phase at some point, but it would drift in and out of phase, resulting in a slowly changing amplitude at the output. The only way that can happen is if you put in two signals, or if the *nonlinearity* is constantly changing. But that contradicts our given conditions, that we put in only a single frequency, and the unspoken condition that the amplifier's linearity is a constant (it probably isn't exactly, but that's a different problem). Ah ha--Karl said the magic word here. I think a _time varying_ system is what's needed to generate frequencies other than harmonics. So basically you'd need to be playing with circuitry involving a 2nd signal source, delay lines, and/or an oscillator and such to create these non-harmonic frequencies. Electric guitarists familiar with teh "chorus" effect are familiar with these shimmering introduction of frequencies around an original frequency, and its done by adding analog delay lines (bucket brigade devices) that are clocked by an oscillator to perform the magic. Chorus effect explained (delay+pitch modulation mixed with original signal): http://www.harmony-central.com/Effects/Articles/Chorus/ A nifty bucket brigade device I've never seen befo http://www.doepfer.de/A1881.htm Best Rgards, -- /"\ ASCII Ribbon Campaign | Todd H \ / | http://www.toddh.net/ X Promoting good netiquette | http://triplethreatband.com/ / \ http://www.toddh.net/netiquette/ | http://myspace.com/mytriplethreatband |
#26
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
wrote in message ups.com... Karl Uppiano wrote: My personal opinion (flame bait) is that both amplifier types (tubes or transistors), when linearized using properly-designed negative feedback should have such low distortion characteristics as to be indistinguishable. Perhaps theoretical or measurable, but my opinion (again, flame bait) is that if you can actually *hear* a difference between two amplifiers of similar power and bandwidth, then one of the amplifiers is poorly designed. :-) I would agree with your assessment, with the only caveat that when the output approaches clipping, differences will occur due to nonlinearity differences outside of the feedback's area of operation. But that would be expected. There are tons of caveats and provisos in my assertion, which is why I labeled it "flame bait". It might be a design goal to produce a certain amound of distortion. Musical instrument amplifiers tend to be designed this way, or at least, they don't try to design out all the distortion. Also, while most amplifiers used for playback don't clip much, many professional applications do, because the sound level is not known in advance. So designers must take clipping into account. Soft clipping vs. hard clipping, and what happens if an amplifier doesn't come out of clipping cleanly are all things we have to think about. What's interesting (e.g. yet another tangent road I'd like to travel down further) is that traditional tube designs, or so I hear, don't use feedback, but to design a transistor-based method without feedback would be dangerous (unstable). Again, haven't gone down that road yet, so that comment could be hogwash also. :-) Tube amps use feedback, starting in the 1920's I think. The phone company used tube amplifiers as repeaters to increase long distance service, and they found that by the time the audio made it through several of these devices, it was almost unitelligible. They started using feedback. Problem solved. In my first engineering job, I had to maintain lots of tube amps as well as solid state amps, and I can guarantee that they *all* had feedback. Degenerative cathode feedback, overall feedback, RIAA equalization using feedback, you name it. I guess I gotta get home to let the dog out. Thanks for all the feedback today; back here on Monday. Have a great weekend. Thanks :-) ..dane |
#27
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
wrote in message ups.com... What's interesting (e.g. yet another tangent road I'd like to travel down further) is that traditional tube designs, or so I hear, don't use feedback, False. Feedback has been part of tubed amplifier designs going back to the 1930s. but to design a transistor-based method without feedback would be dangerous (unstable). False, sort of. Audio amplfiers are actually designed in two domains - AC and DC. The DC design of an audio has traditionally related to stabilizing operating points for the active components. The AC design relates of course to how the amplifier amplifies audio. In some cases the AC and DC designs converge and the amplfier operates the same at all frequencies including zero frequency or DC. If you don't stabilize the operating points of either a tubed or SS circuit it will be prone to either drift to a high distortion operating point, or drift to saturation and perhaps overheat, or drift to cutoff and not amplify at all. |
#28
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
writes:
Since you two (Todd and Karl) are being so responsive to my questions, I want to pull another one in. I originally posted on rec.audio.opinion, but was suggested to move it over here so I did. But I did get a response over there that I wanted to quote here and ask about. Trevor Wilson wrote: Even if that is assumed to be true, what causes the tube to have a steeper harmonic rolloff? **Output transformer. Is that due solely to the frequency response (higher freq. rolloff) of the output transformer? Yes. The output transformer has all the nifty properties we've come to love from ferromagnetic materials. Yeah, there's the rolloff among them, soft compression/saturtion in tape is another nice property. Some studios still track drums to tape because there's no good digital equivalent to nice tape compression. IOW, if you used an "ideal transformer" with a flat FR from 0-infinity, would the tube amp's harmonic content be the same as a transistor amp (with no xfmr) of the same topology? No, I don't think so. #include the even'odd harmonics stuff. If it were this simple, you'd see transistor amps transformer coupled and sold as botique. :-) And here I'm not quite sure the questions to ask, so I'll just put back out there again that "I read somewhere" that a class-A tube amp (with output xfrmr) has a steeper harmonic content rolloff than a transistor-based push-pull amp with no xfmr. Granted there are at least three variables here (tube/transistor, class A/class AB, and xfmr/no-xfmr) if not more, but I'd like to dive in and learn what I can learn... Start lurking rec.audio.tubes and alt.guitar.amps And google tube solid state harmonics for a lot of stuff on that. -- /"\ ASCII Ribbon Campaign | Todd H \ / | http://www.toddh.net/ X Promoting good netiquette | http://triplethreatband.com/ / \ http://www.toddh.net/netiquette/ | http://myspace.com/mytriplethreatband |
#29
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
"Karl Uppiano" wrote in message news:Ir85h.53$xD.38@trndny08... In my first engineering job, I had to maintain lots of tube amps as well as solid state amps, and I can guarantee that they *all* had feedback. Degenerative cathode feedback, overall feedback, RIAA equalization using feedback, you name it. Agreed. In the Army I maintained 3 Hawk CW radars that were pretty much entirely tubed*. They had audio-like circuits for handling the Doppler and the modulation of the transmitter, and they had DC-coupled circuits for antenna servos, and numerous other things. The largest had about 400 tubes. As you say, these tubed circuits were full of every kind of local and loop feedback one could imagine. We had one feedback loop that essentially traversed the entire radar from transmitter to receiver with numerous subloops that stabilized segments of the overall loop. IOW this feedback loop passed through about 350 tubes. The other 50 or so tubes were in regulated power supplies and other service circuity. MTBF was about 1 day. * there were some solid state tubed replacements that were packaged as tubes and were socketed. Most were in the regualted power supplies. They were 300B replacements. There was also some SS-based built-in test equipment. Ironically the SS built-in test equipment was a bit marginal in places and would not work when it was really cold, because the beta of some critical transitors would fall off too much. |
#30
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
wrote in message oups.com... Todd H. wrote: IOW, if you used an "ideal transformer" with a flat FR from 0-infinity, would the tube amp's harmonic content be the same as a transistor amp (with no xfmr) of the same topology? No, I don't think so. #include the even'odd harmonics stuff. If it were this simple, you'd see transistor amps transformer coupled and sold as botique. :-) I've been convinced through my research thus far that the even/odd harmonic stuff between tubes and transistor amps is hogwash, that it's much () more related to the topology (class A, push/pull, feedback/no-feedback, etc) than the technology (tube vs transistor). Even/odd harmonic differences are generally due to the use of push-pull versus single-ended circuitry. Push-pull or balanced circuits can reduce even harmonics very dramatically, depending primarily on how well the halves are matched. Your last statement is what I'm getting at... Albeit added expense and unnecessary for sales, I wonder if a transformer-coupled push-pull transistor amp would sound more like a class-A tube amp. No, but as Nelson Pass tries to show, a single-ended SS circuit can produce harmonic structure that is similar to a single-ended tube circuit. This of course assumes that a good bit of signal shaping is done by the transformer... which may itself be hogwash. Transformers are for impedance matching, They can shape the signal by adding nonlinear distortion and also by adding phase shift and frequency response variations. Most tube amp output transformers are marginal and start losing efficiency at the low end, but still inside the audio band. ..dane on a never-ending quest |
#31
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
wrote in message oups.com... Arny Krueger wrote: Actually, there are two effects (1) Modulation distortion in the form of modulation of an audio signal being amplified at the same time. (2) Noise caused by modulation of the bias voltages present in the tube. Wouldn't a low-impedance bias driver help reduce the levels of (2) above? (Granted I don't know amplifier design theory, so I'm not sure that the bais point CAN be low impedance. But it would seem to me if the current flowing through the bias point was substantial (10x) compared to that flowing(leaking) into the tube bias, that the bias point voltage would therefore not change much as a function of the leakage through the tube grid. By definition, tube grids in linear amplifiers draw essentially no current and are therefore always high impedance. From practical experience I can tell you that a tube with a grounded grid (hard to imagine a lower impedance source!) can be microphonic. |
#32
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
"Ron Capik" wrote in message ... wrote: Ron Capik wrote: Ah but don't let your brain rest too much yet. ;-) of course not! Non harmonic distortion can be created by microphonics in a vacuum tube. Forces can be created by the signal that cause the elements of the tube to move and thus create modulation effects linked to mechanical resonances in the tube ...not a likely effect for solid state devices. Can you provide any measured data about the magnitude of such modulation distortion? Sorry, that's way out of my realm of expertise. Most of my experience with microphonic tubes goes back to my high school days as a TV repair person. I have heard tubes "sing" and thus extrapolated that observation. As tubes age they can become vibration sensitive and forces in the tubes can make the sensitized tubes "sing." Microphonic tubes are most apparent in low level circuits with flat frequency response. RIAA phono preamps would qualify except for the fact that they are integral with low pass filtering. Microphonic tubes were more common in PA system mic preamps. |
#33
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Arny Krueger wrote:
"Ron Capik" wrote in message ....snip.. Sorry, that's way out of my realm of expertise. Most of my experience with microphonic tubes goes back to my high school days as a TV repair person. I have heard tubes "sing" and thus extrapolated that observation. As tubes age they can become vibration sensitive and forces in the tubes can make the sensitized tubes "sing." Microphonic tubes are most apparent in low level circuits with flat frequency response. RIAA phono preamps would qualify except for the fact that they are integral with low pass filtering. Microphonic tubes were more common in PA system mic preamps. Ummm, gain structure. Later... Ron Capik -- |
#34
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Karl Uppiano wrote:
.....snip... Tube amps use feedback, starting in the 1920's I think. The phone company used tube amplifiers as repeaters to increase long distance service, and they found that by the time the audio made it through several of these devices, it was almost unitelligible. They started using feedback. Problem solved. Close, very close. Bell Lab's Harold Black invented/discovered the negative feedback stabilization technique in 1927. Scribbled his first notes on the subject on his New York subway ride to the West Street lab. Later... Ron Capik -- |
#35
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
See, Dave, unlike me, has not forgotten all the math. :-) Kudos. -- /"\ ASCII Ribbon Campaign | Todd H \ / | http://www.toddh.net/ X Promoting good netiquette | http://triplethreatband.com/ / \ http://www.toddh.net/netiquette/ | http://myspace.com/mytriplethreatband |
#36
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
On Nov 10, 12:26 pm, wrote: (It was suggested that I bring my question over here....) While so many people run around with their hands in the air talking about this amp and that amp, their distortion measurements, tube-vs-transistor, yadda-yadda-yadda, I'm concerned with something much more fundamental. If the phrase is true "all amplifiers cause distortion," which I believe is the case, my question is simple... How is it created? For instance, let's take the scenario of an all-analog, all pure class-A staged amplifier.. My *assumption* is that in an ideal model, this scenario would generate no distortion, but in using real-world components, distortion is still generated. I understand that there exist what are called "nonlinearities" in the amplifier, where at some input levels, a change of the input voltage causes a particular change in the output voltage, but at some other input level voltage, the same change in voltage (just offset from the original) would cause a different amount of change in the output. So is distortion's root this nonlinearity? And if so, why does this nonlinearity always manifest itself as n-order harmonics? You asked why one can't put a 100 Hz sinewave into a some kind of nonlinear device and get out 100 Hz + 105 Hz. A steady (reoccuring) wave is composed of a fundamental (sinewave) and various harmonics. (In a reoccuring wave, every cycle looks exactly like the previous cycle.) If there was a non-harmonic frequency component, such as a 105 Hz , added to 100 Hz, the sum of the two waves would vary in amplititude as the two components went in and out of phase. That would be a non-reoccuring waveform. Puting a sinewave into a any nonlinear amplifier can only produce a reoccuring output waveform. Such a wave will always be composed of a fundamental and harmonics. No matter how much an amplifier bends and distorts the sinewave, the output wave will always be a constant waveform. Thus, it must be composed of a fundamental and various harmonics, only. Bob Stanton |
#37
Posted to rec.audio.tech
|
|||
|
|||
Distortion... why/how is it created?
Wow -- thanks to everyone for so much good discussion. I think I'm good for a while on this topic, I'll come back and revisit as more questions arise. thanks again to all! ...dane |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
Distortion... why/how is it created? | Audio Opinions | |||
From RAHE article on distortion | Audio Opinions | |||
Re KISS 123 by Andre Jute | Vacuum Tubes | |||
Setting up a two amp system | Tech | |||
Pioneer Clipping and Distortion was:DEH-P840MP, infinity kappa 693.5i and kappa 50.5cs component. | Car Audio |