Bonsai was about the only guy here selling amps (atmasphere and AKSA also do).
Others are so clever to stay out of this nonsense.
Others are so clever to stay out of this nonsense.
Bonsai sells nfb amps, but he may still be a source of wisdom and perspective. AKSA is great, no mater what he sells. And Atmasphere has experienced a most amusing religious conversion, perhaps backed up by a solid business plan 😛
Well, the early development of the hypothetico-deductive-probabilistico-approximative Formal Axiomatic Method and its pillar, the Symbolic Logic was initiated back in the seventeenth century by some mathematicians imbued with the lofty romantic-idealistic-deterministic vision of establishing an analogy between numbers and Nature through many-layered abstractions. It`s a pretty much all-round method, its application is limited only by our imagination.
Practical feedback amplifiers can have a positive or a negative output impedance. But not zero ?
An argument to explain why an amplifier could have not have zero impedance (it does not need to have infinite negative feedback for that) would be more interesting than your favorite leitmotiv.
This is an excellent univocal definition which, slightly modified, could be extended to differential inputs.
The real old school, that means Georg Ohm, teaches us I = V/R.
Where I is the effect and V is the cause. Voltage makes the law.
It's time to remember it.
It depend what goal of the amplifier specification.
If the goal is best linearity, then choose high loop gain at 20 Hz to 20 kHz.
If the goal is sound pleasing, you must define what make sound pleasing or not pleasing. Many people have their own idea about pleasing sound.
The problem is many people do not understand the correlation between sound pleasing with measurement.
The other problem is people who do not understand engineering talk about engineering, blame the measurement and blame the technique.
Sound has a "highest purpose" arithmetically determined, just like human does have. Therefore it can have characteristics that are positive on a scale of purposefulness. In these modern days music is not always about harmony. Nothing wrong about it. The normalization process may be applied to difference between individuals. To create amplifier that does the job - "amplifying" - not only a single recipe should suit. There could be different approaches. Instead of global feedback, one could perhaps use certain patterns of numbers and create local "spins" around amplifying devices, by defining desired order of pattern. Negative multiplied with negative is positive number. Not fully true, if the process is within boundaries of negative amounts always being in return. "But supply voltage is free from negative form..unless it is off". Sure, go that way. And record is free from soul too.
Unless you paid for it more and more. Just like buying woman "does not reduce feminine essence, because they are such negative kind of kind".
Be careful with numbers. You might kill your reason. Mortal treason.
Mortal Treason - Khampa Nomads.
Be careful with numbers. You might kill your reason. Mortal treason.
Mortal Treason - Khampa Nomads.
Haha, ego driven engineers don't have any sense of humor...
Latest is new-age explanation of feedback 😀
Latest is new-age explanation of feedback 😀
This is on a humorous note.
About 30 years ago, someone described feedback as:
When you deficate and eat it or part of it.
This person now is almost retired and possesses a Higher Technician Diploma. I still remember the transformer phasor diagram on his study table.
About 30 years ago, someone described feedback as:
When you deficate and eat it or part of it.
This person now is almost retired and possesses a Higher Technician Diploma. I still remember the transformer phasor diagram on his study table.
Last edited:
Nah, what are you talking about....? That's NOT new-age, but decomposing-old-age!
Originally Posted by jan.didden 
This can be and has been technically demonstrated as fact. See the various writings of Self, Cordell, and earlier Baxandall.
First I'd like to say that the paper you linked is great, and it's surprising to me that this was published in 1938. Global feedback in audio power amplifiers wasn't seen much until the 1950s. and did not become de riguer until the late 1960s. The fact of the matter is that it's difficult to apply global feedback to a transformer coupled amplifier, although the Williamson and Ultralinear topologies became somewhat popular (tubes were losing their popularity and were considered old fashioned) and showed that it could be done.
Second, I'd like to point out that you're both right. Global feedback does change the harmonic profile and it does skew it towards higher order harmonics. But the design theory (which I realize isn't always achieved) is that if there's enough loop gain in the passband, and the higher order harmonics are below the threshold of audibility (even mediocre modern amplifiers achieve -80 dB which is pretty good), then the distortion is negligible. I've seen specs that claimed the higher order harmonics were -110 to -120 dB which is almost certainly negligible.
Do amplifiers achieve their published specifications? There's a lot of reasons why they might not while in service, but that's another rabbit hole.
This can be and has been technically demonstrated as fact. See the various writings of Self, Cordell, and earlier Baxandall.
First I'd like to say that the paper you linked is great, and it's surprising to me that this was published in 1938. Global feedback in audio power amplifiers wasn't seen much until the 1950s. and did not become de riguer until the late 1960s. The fact of the matter is that it's difficult to apply global feedback to a transformer coupled amplifier, although the Williamson and Ultralinear topologies became somewhat popular (tubes were losing their popularity and were considered old fashioned) and showed that it could be done.
Second, I'd like to point out that you're both right. Global feedback does change the harmonic profile and it does skew it towards higher order harmonics. But the design theory (which I realize isn't always achieved) is that if there's enough loop gain in the passband, and the higher order harmonics are below the threshold of audibility (even mediocre modern amplifiers achieve -80 dB which is pretty good), then the distortion is negligible. I've seen specs that claimed the higher order harmonics were -110 to -120 dB which is almost certainly negligible.
Do amplifiers achieve their published specifications? There's a lot of reasons why they might not while in service, but that's another rabbit hole.
It is not true. It depend on all design. Example: https://www.diyaudio.com/forums/solid-state/338815-anistardi-peletuk-4.html#post5835050
This amplifier have high loop gain and have H2 dominant.
Derivation of the foundation formula for Negative Feedback, the economical magic, which makes cheap amplifiers behave like prohibitively expensive ones.
Suppose an amplifier has a transfer function:
V = G.i
Where V = output voltage, G is a constant and i is the input voltage.
We will apply feedback to the amplifier to change its transfer function into a line with a lower gradient.
Let us say the amount of feedback is f where f < 1 but not negative.
Applying the transfer function, we get:
V = G.(i - fV)
=> V = Gi - GfV
=> V(1 + Gf) = Gi
Rewriting:
Gi = V(1 + Gf)
The new gain is:
V/i = G/(1 + Gf)
If G is very large, 1 + Gf >> 1, implying we can eliminate 1.
Therefore:
V/i = 1/f
This result concludes that when G is large but not necessarily stable and constant, a better behaving amplifier can be made by applying a feedback fraction of f.
Suppose an amplifier has a transfer function:
V = G.i
Where V = output voltage, G is a constant and i is the input voltage.
We will apply feedback to the amplifier to change its transfer function into a line with a lower gradient.
Let us say the amount of feedback is f where f < 1 but not negative.
Applying the transfer function, we get:
V = G.(i - fV)
=> V = Gi - GfV
=> V(1 + Gf) = Gi
Rewriting:
Gi = V(1 + Gf)
The new gain is:
V/i = G/(1 + Gf)
If G is very large, 1 + Gf >> 1, implying we can eliminate 1.
Therefore:
V/i = 1/f
This result concludes that when G is large but not necessarily stable and constant, a better behaving amplifier can be made by applying a feedback fraction of f.