Hi,
My take on this. A sine starting a t isn't a sine. Its a sine modulated with a step, at lest around t.
A modulated sine does/can look different than a sine.
/örjan
My take on this. A sine starting a t isn't a sine. Its a sine modulated with a step, at lest around t.
A modulated sine does/can look different than a sine.
/örjan
Guess what, petr, odds are you are in that suit put on crooked the entire time yourself. That you came to this place trying so hard telling people THAT alone would be the right way of wearing it is quite a comic. I wished you had been doing an April-fool trick in a wrong month.
Perhaps you would try one more time going over that Bonsai post without having in mind a "right way" of putting on a suit or things like that?
Very nice, yes that is correct and valid way of looking at it. That also fully explains the amp response to such a signal.
Jan
The core issue is that claims have been made based on empirical investigation. We have learned that proof methodologies must be employed that can be peer reviewed. Empirical investigation may yield a new global truth, but without reference it becomes hearsay.
Peter Baxandall an acknowledged genius of the audio world, made empirical investigations about feedback and even produced some nice graphs. He did not however provide mathematical proof. His teachings became widely accepted, due to his stature as an engineer. Those teachings have been shown to be invalid. Geniuses make mistakes.
Graham was a good engineer and his GEM circuit goes against Baxandall's teachings. He also understood the requirements for a fast amplifier and it is easy to see where he enhanced the phase margins in his circuit. The circuit submission by petr_2009 has no phase margin enhancement. Graham like Baxandall, made claims based on empirical investigation without proof methodologies. He did design a nice amplifier.
Peter Baxandall an acknowledged genius of the audio world, made empirical investigations about feedback and even produced some nice graphs. He did not however provide mathematical proof. His teachings became widely accepted, due to his stature as an engineer. Those teachings have been shown to be invalid. Geniuses make mistakes.
Graham was a good engineer and his GEM circuit goes against Baxandall's teachings. He also understood the requirements for a fast amplifier and it is easy to see where he enhanced the phase margins in his circuit. The circuit submission by petr_2009 has no phase margin enhancement. Graham like Baxandall, made claims based on empirical investigation without proof methodologies. He did design a nice amplifier.
@spladski
I think you have to work quite a bit more to convince us that Baxandell "have been shown to be invalid".
Please don't forget to add provide mathematical proof.
/örjan
I think you have to work quite a bit more to convince us that Baxandell "have been shown to be invalid".
Please don't forget to add provide mathematical proof.
/örjan
Rules for feedback have not changed. Device parasitics determine phase shift, which determines viable feedback. The devices of today are very different from when Baxandall made his observations.
We really should stop attacking each other specially with personal tones.
We should seek for a common language and understanding and try to create
a model that maybe even has both aspects without any contradiction.
It's not clear to me why anyone is harrassed/bothered if FCD can also
be described with other methods as well: why is it a problem at all?
IMV FCD just focuses/concentrates on a specific aspect from the whole spectrum.
Even on a loudspeaker datasheet there are sometimes results about step responses.
Why? We could say that it's non-sense as music is never like that.
For me it's similar (and sorry again for the metaphorical examples) as
acceleration vs speed tests (drag racing cars vs F1). I'm not saying that the
one and only and above all aspect is the step (sine wave) response but likely
it can have some info about an amplifier. Or is the debate about that it is
completely redundant and unnecassary as a simple Bode also tells the same?
Or if we go back for my theoretical example: can anyone imagine a case
where we start to measure the THDistortion with a 1kHz sine wave and if we
could start it from 0 and measure it on every period it would be THD wise:
0.2%, 0.1%, 0.01% for the first 3 periods? Then how we should/could deal with it?
What would this even mean? Wouldn't it tell us something important about the
transient behaviour that we usually miss when measuring a continouos/steady
and averaged sine wave result? Or is such a test out there already?
What if amplifier "A" would be better at such a test vs "B"..?
Anyways regarding distortion I guess there are still lot of potentials as we
probably shouldn't run just a THD test for 1 frequency but also check the distortion
coherency/stability against period, frequency, amplitude, time, load, back EMF, etc.
We should seek for a common language and understanding and try to create
a model that maybe even has both aspects without any contradiction.
It's not clear to me why anyone is harrassed/bothered if FCD can also
be described with other methods as well: why is it a problem at all?
IMV FCD just focuses/concentrates on a specific aspect from the whole spectrum.
Even on a loudspeaker datasheet there are sometimes results about step responses.
Why? We could say that it's non-sense as music is never like that.
For me it's similar (and sorry again for the metaphorical examples) as
acceleration vs speed tests (drag racing cars vs F1). I'm not saying that the
one and only and above all aspect is the step (sine wave) response but likely
it can have some info about an amplifier. Or is the debate about that it is
completely redundant and unnecassary as a simple Bode also tells the same?
Or if we go back for my theoretical example: can anyone imagine a case
where we start to measure the THDistortion with a 1kHz sine wave and if we
could start it from 0 and measure it on every period it would be THD wise:
0.2%, 0.1%, 0.01% for the first 3 periods? Then how we should/could deal with it?
What would this even mean? Wouldn't it tell us something important about the
transient behaviour that we usually miss when measuring a continouos/steady
and averaged sine wave result? Or is such a test out there already?
What if amplifier "A" would be better at such a test vs "B"..?
Anyways regarding distortion I guess there are still lot of potentials as we
probably shouldn't run just a THD test for 1 frequency but also check the distortion
coherency/stability against period, frequency, amplitude, time, load, back EMF, etc.
What makes you think there is a problem that you need to deal with? THD by definition is a measurement of residue against a base tone, when you have a suddenly emerged 1KHz tone, it is not a sine. It is the product of a step function and a 1KHz sine, just like fellow member orjan pointed out a few posts ago. At the beginning of it, because of the step function, it manifests a nasty spread in frequency domain with peaks all over the place in the spectrum towards infinity, drastically different from what a 1KHz tone is supposed to look like, your THD reading of that signal itself would go through the roof. It would take a while before the spectrum settles down and diminishes into a single peak at 1KHz, and THD reading approaches zero. (Do you know how long that "a while" is, by the way?) All of these are about the property of that signal itself alone, THD reading declining over time, before your amplifier even comes into picture.
Now let's take a look at your amp. Does it have an infinite bandwidth? No? Then you will have trouble reproducing that nasty spectrum at the beginning, cuz an infinite bandwidth is required to do that. Any finite bandwidth amp would have trouble doing it.
So what would you do about it? Slowing down the rise of that step fiction so that your limited bandwidth amp is able to handle it appears to be the only way to go. Naturally, you would then ask yourself how fast a slope of that step signal you yet still need your amp to be able to handle, and why you need that fast for an audio amp.
Where to look for answers to those questions then? You may have noticed by now it is that step signal that matters, the sine played no part at all in this made up FCD drama. We could then get rid of the sine, replace the step signal with a bandwidth limited square wave, and arrive at Bonsai's analysis at post #561.
FCD, what more does it do than confusing people?
I have an original Bode book from 1952 where this all is being explained exactly as done here.
It was sooo much easier pre-internet to find factual info ;-)
Jan
It was sooo much easier pre-internet to find factual info ;-)
Jan
I have to admit I am a believer. I strongly believe that physics are not debatable. And ignoring them does not change anything.
Yes, it's a comfort, isn't it. The irony is that both camps will express this with a cosy feeling...
//
//
Professional physicists sometimes do debate physics. Its just that they are much more qualified to do it than some of the participants here.
Last edited:
Let's leave the very 1st one. The 2nd period is already ok to use it as a 100% normal sine wave?
What if the 2nd period has a THD of 0.3%, the 3rd 0.2% and the 4th 0.01%?
That's probably not okay.
What exactly is a "100% normal sine wave"? If it is a single peak in frequency spectrum at f that is what your suggested, it requires a time domain function sin(2*pi*f*t) where t extents from -∞ to +∞, if you would agree with Mr. Fourier. And if you do agree with Mr. Fourier, then no part in your signal is a "100% normal sine wave".
You may consider abandoning the sine. The FCD illusion comes from a signal that has a start in time, what that signal itself is or how long it sustains does not matter.
As Nattawa said (but may have been missed):
At the beginning of it, because of the step function, it manifests a nasty spread in frequency domain with peaks all over the place in the spectrum towards infinity, drastically different from what a 1KHz tone is supposed to look like, your THD reading of that signal itself would go through the roof. It would take a while before the spectrum settles down and diminishes into a single peak at 1KHz, and THD reading approaches zero. (Do you know how long that "a while" is, by the way?) All of these are about the property of that signal itself alone, THD reading declining over time, before your amplifier even comes into picture.
The subtlety is that the high initial 'distortions' are not distortions of the 1kHz signal at all; it is the harmonics of the step input, which will die out over time, and then what remains if the 1kHz and its (low) harmonics.
Jan
Last edited:
At this point you have to take into account how many cycles the brain needs to work out what is going on vs how often this occurs. If it's on every note there is a problem. If it's when you hit an amplifier with no input LPF with a DC-daylight impulse that cannot exist in music that's just telling you to add that LPF for some piece of mind. But real music doesn't suddenly start from 0.