Edmond,
Transform only.
Information carried by what?
It does, in the whole universe and multiverse.
janneman said:
I believe Cherry's analysis would not show a disproportionate reduction due to the op stage local feedback as long as the fundamental of the feedback signal is much larger than the distortion components
there could be some "extra" distortion reduction help from the local feedback increasing input impedance of the op stage, increasing gain of the VAS, hence increasing global loop gain
Lumba,
I think most participants in this thread are quite aware by now that feedback can't "fix" a high order nonlinearity - to the point that I find it insulting when people think they have to "remind" us of this "error" by rolling out this mouldy straw man
and we're pretty sure that the easiest way to avoid high order nonlinearity/dead zones is heavy op bias, well into Class AB - Bob points to >150mA/fet in MOSFET op, I think it was also established that Bipolar should also be biased hotter than Self's "optimum Class B" to avoid dynamic underbias with thermal transient, we also have Pass' "first watt" principle pointing in the same direction
the need to avoid dead zones extends to dynamic response with enough driver current to handle op gate/base charge demands
only when the amp can drive the worst case audio load/signal combination with only lower order distortion/lack of dead zones can it then be profitable to apply high feedback
and with modern op devices “high feedback” can be as high as 60 dB loop gain at 50KHz - well above the frequency content of the vast majority of recordings ( look up http://www.earthworksaudio.com/3.html - advocates of very wide band microphones and electronics and you will find their highest frequency recording mic is 50KHz and the bulk of their recording mic offerings are =< 30KHz )
"For higher audio frequencies, say above 8kHz or so, you don't need 'shaping' as the harmonics are above the audible bandwidth anyway. But for low-mid frequencies, you would like the first couple of harmonics to decrease a lot, even if they pop up as much higher harmonics above 20kHz. The amount of 'shaping' apparently depends on the OL harmonic contents, the OLG and the amount of feedback/ULG, but maybe there is something to be gained by carefully shaping the OL gain, distortion and feedback to get ALL harmonics, independently of frequency, above 20kHz."
A truly excellent reason for very high OLG (lets have more than 90dB here) with a single dominant pole at 100Hz.
A truly excellent reason for very high OLG (lets have more than 90dB here) with a single dominant pole at 100Hz.
Lumba Ogir said:
Come on Lumba. They are almost the same on different currents/voltages. "Very"/"not very" is not the technical term.
Wavebourn,
OK, what about saying that it`s sonic achievement is confined within narrow limits?
OK, what about saying that it`s sonic achievement is confined within narrow limits?
I'm not sure that "harmonics" is the best way to describe the
shape and smoothness of any non-linear transfer function.
Lets face it, we don't listen to pure sine waves with zero offset.
We usually listen to multiple waves, each offsetting the other to
another place on the curve. I highly doubt the "harmonics" of
most transfer functions are the same when so offset... Even if
that offset were only simple DC!
A high "harmonic" noted with a pure sinewave might be an
indication of a glitch in the curve, such as an AB crossing...
But all harmonics, even high ones (like the infamous 7th)
are musically relevant to a pure sinewave... IMD of such a
glitch in the presence of two or more tones is the real killer.
Especially so when such tones were not already musically
related as the strings and overtones of a single instrument.
Why we don't play Hi-fi through the typical guitar amp, but
the guitar gets away with it fine if not better.... Sideband
products of oh say drum + vocal IMD, would not as likely fall
by pure accident upon musically relevant sum and difference
frequencies. Yes, its all about the IMD ladies...
IMD can occur in feedback loops as well.
When we look at slew rate clipping (TIM or whatever) in
such loops, does the current domain also present us a
non-linear transfer curve well before clipping? And should
we speak of its distortions also in terms of "harmonics"?
Where does it all end?
shape and smoothness of any non-linear transfer function.
Lets face it, we don't listen to pure sine waves with zero offset.
We usually listen to multiple waves, each offsetting the other to
another place on the curve. I highly doubt the "harmonics" of
most transfer functions are the same when so offset... Even if
that offset were only simple DC!
A high "harmonic" noted with a pure sinewave might be an
indication of a glitch in the curve, such as an AB crossing...
But all harmonics, even high ones (like the infamous 7th)
are musically relevant to a pure sinewave... IMD of such a
glitch in the presence of two or more tones is the real killer.
Especially so when such tones were not already musically
related as the strings and overtones of a single instrument.
Why we don't play Hi-fi through the typical guitar amp, but
the guitar gets away with it fine if not better.... Sideband
products of oh say drum + vocal IMD, would not as likely fall
by pure accident upon musically relevant sum and difference
frequencies. Yes, its all about the IMD ladies...
IMD can occur in feedback loops as well.
When we look at slew rate clipping (TIM or whatever) in
such loops, does the current domain also present us a
non-linear transfer curve well before clipping? And should
we speak of its distortions also in terms of "harmonics"?
Where does it all end?
kenpeter said:
The story goes that a cop encounters a man crawling on the ground
beneath a street lamp.
"I'm looking for my lost keys," explains the (inebriated) man.
"You lost them here?" asks the cop.
"No, I lost them down the street, but the light's much better here."
😎
kenpeter said:
These are good points, although I don't usually get enthusiastic about the argument that we don't listen to sinewaves, suggesting that the value of sinewave testing is very limited. It is actually very powerful when properly interpreted and when more than one sinewave is employed.
The key point you make that is right on the money is that the likely biggest offender is IM, and that the nature of the sound of IM products can depend on the frequencies involved.
It is also very true that the harmonic distortion can be a strong function of the operating point where it is measured. One could, I suppose, measure modest-amplitude THD-20 that is mixed with a high-amplitude LF signal that is then taken out by the THD analyzer. One could then look at how the THD-20 residual changes as a function of the LF signal. This would be not unlike applying a SMPTE IM approach to looking at the distortion residual.
Cheers,
Bob
I'd even extend it to 4-quadrant LF bias, by means of backdriving the amp with another one through a resistor (sort of a current source, hence), fed with the LF componenent at a different phase angle and proper amplitude. Similar to the kind of stuff one would use for stability tests, with a square wave as the HF signal. Or to get small signal Zout-plots vs. freq and 4-quadrant bias, etcBob Cordell said:
- Klaus
Speaking of measurements, I used to observe on a 2- ray oscilloscope 10 KHz triangles of very small amplitude filtering out 10 Hz sine of a big amplitude. It is a very good test for transistor amps, not just to measure specs, but to see what is weak and to improve gradually. The hardest thing was to build a filter, it needed to be well screened and no iron cores allowed. It consisted of combination of passive then active HPF.
Wavebourn,
under any conditions. Necessarily, high mu tubes distort more. I would use another type.
under any conditions. Necessarily, high mu tubes distort more. I would use another type.
Lumba Ogir said:
I put one, and only one, stage with extremely high dynamic load where you will need more than one, as the result you will get more distortions, and their specter will be wider.
As you see, it is the matter of conditions. For example, it is optimal if you use a high mu tube to drive a grid of a low mu one that drives a loudspeaker.
Wavebourn,
OK, that`s nice, I still would try 12AT7 or even a medium-mu triode.
Right, but it`s not the case here.