Surely most tube amplifiers are flat when loaded with a 8 ohm resistor - but as soon as you connect them to a real loudspeaker thinks change due to the variable impedance of the loudspeaker versus frequency. I was addressing the reality, not the lab.
Why do audiophiles care about flat measuring gear when most rooms are not flat?
- Thread starter caesar
- Start date
Surely most tube amplifiers are flat when loaded with a 8 ohm resistor - but as soon as you connect them to a real loudspeaker thinks change due to the variable impedance of the loudspeaker versus frequency. I was addressing the reality, not the lab.
Do you consider a 10 dB variation in a response smoothed to 1/6th octave to be flat? Although the scale is very compressed the average level in the bass seems at less 3 dB higher than the treble.
Yeah, tube amp designs will have challenge to reach even just below 1 ohm for output impedance, with some exceptional ones hitting close to 0.5ohms I think.
But then Atmasphere does mention this with the context of how much feedback is used and loudspeaker curves but with you context needs focus on practical real world consideration, good example: http://www.stereophile.com/content/...erence-monoblock-power-amplifier-measurements
In general though, this topic makes it pretty impressive what the Devialet does even for solid state.
Cheers
Orb
What Mike was referring to was that 'first watt'. In SETs and Push-pull amps that don't have a dedicated phase splitter circuit, the distortion can decrease linearly to zero (unmeasurable) as power is decreased. Most push-pull amps (tube, solid state or class D) have a rising distortion curve as power is decreased below a certain point, typically around 5% of full power. So in many cases an SET can be **more** linear than a transistor amplifier with high feedback as long as you stay inside this window. This is one reason why choosing the right speaker to go with a particular amp is so important- if you are using a push-pull amp for example, you want to make sure that the amp is not too powerful such that you are frequently listening at the lower, nonlinear, portion of the power curve.
As far as the tubes themselves: triodes have been the most linear form of amplification for decades. There are now some low signal transistors that operate in the same region, but there are no power transistors that do. You might consider this to be the envelope to push in transistor design, but with the industry rapidly switching over to class D (if you will pardon the expression-
) such devices seem less likely to happen but we can always hope.So why are tube amps less linear if the devices used to make them are *more* linear?? Its all in the design. For example. If the amp is single ended, the primary distortion component will be the 2nd harmonic. If an amplifier is push-pull with a single-ended input, the distortion component will still feature the 2nd, with some 5th thrown in. If the amp is fully balanced from input to output, the primary distortion component will be the 3rd harmonic at a lower level than seen in an SET. This, mind you, is all assuming zero feedback. Add enough feedback and the distortion can be brought to very low points. Futterman built amplifiers in the early 1960s that had THD in the neighborhood of 0.005%. The question is really 'does lots of feedback actually sound better' and that is a question worthy of its own thread.
(the short answer being no- most feedback applications are not done correctly, resulting in audible problems caused by the feedback itself, the most common being the introduction of RF into the circuit...)
Many tube amps have a damping factor of 15:1 and even over 20:1. With such output impedances, the frequency response difference as compared to a solid state amplifier of 100:1 would be a difficult thing to measure or hear. *That* is the reality. This is accomplished for the most part through the application of feedback, which causes the amplifier to dynamically adjust its output power with respect to the impedance of the load so as to get predictable flat frequency response. This is the basis of the Voltage Paradigm which was championed by Electro Voice and MacIntosh in the late 1950s. You can read more about this at
http://www.atma-sphere.com/Resources/Paradigms_in_Amplifier_Design.php
The Voltage Paradigm is so effective that going beyond 20:1 in damping factor yields no further measurable improvement.
The point I was making was that in most cases, feedback is improperly applied, resulting in audible artifacts (brightness being the most common) due to added distortion and RF. This is something independent of whether the amp is tube or solid state. As a result, there is an argument that zero feedback amplifiers are actually more linear, not less.
For those interested, here is are two articles outlining what the problems are in the application of feedback:
http://www.normankoren.com/Audio/FeedbackFidelity.html
https://www.passdiy.com/project/articles/audio-distortion-and-feedback
Could I ask your thoughts on Bruno Putzey's argument that there is no such thing as too much feedback?
Sure. Take a look at the articles I linked, in particular the one by Mr. Koren. All the stuff I have seen from Bruno so far suggests that he is only operating on a theoretical level, and not dealing with real world issues. For example if you don't do it right, RF will appear at the input of the amp (the speaker cables being the antenna) which is never a good thing (if nothing else, causes the amp to make distortion due to intermodulations, which the ear will interpret as brightness)!
A simpler answer is found in Nelson's article, the more feedback you have, the more gain you need for it too, meaning that to kill the extra distortion of the added gain stages, you need more feedback; there is a carrot on a stick that you just can't reach. So I recommend reading the articles at the links and then you will see the flaws in Bruno's position. I am not opposed to feedback so much as I am opposed to its improper execution!
I had hopes for HEXFETs a long time ago too, as they were very linear and now also long gone.
I really don't like how the speed of semiconductor obsolescence seems to be accelerating!
I really don't like how the speed of semiconductor obsolescence seems to be accelerating!
That's like asking why a car needs proper front end alignment, when most roads are not straight.
Tim
Last edited:
Soulution seems to say the same thing in an old TAS article as I recall.
but Nelson's arguments make much more sense to me. That white paper on feedback is one of the all time greats.
Between the not flat speaker output, the not flat room, and ones not flat hearing,
how do you do this without going into the digital world? I'm not sure I want to go through an A to D converter to make my vinyl audio reproduction flat. I would like to hear such before attempting it.
how do you do this without going into the digital world? I'm not sure I want to go through an A to D converter to make my vinyl audio reproduction flat. I would like to hear such before attempting it.
Similar threads
