If this is the case, why is often much more feedback applied in solid state circuits? I would have thought that the more linear the device, the less feedback needed.
If you still spin CDs, this may be the best option.
- Thread starter thomask
- Start date
If this is the case, why is often much more feedback applied in solid state circuits? I would have thought that the more linear the device, the less feedback needed.
That's a very solid argument. The truth is not very many people are good at getting very low distortion numbers (like, below 0.001%). I think people are just use to using transistors in certain ways, they're a bit vain about the distortion numbers, and people like to read low distortion numbers even when they don't necessarily like the sound. Some manufacturers purposely use less transistors in a design than others; and some use A LOT because it's so easy to push numbers that way.
You do see some designs contrary, but it's just so easy to design for more feedback and it's *correct* in most engineers' minds. It's also competitive between people/manufacturers. With tubes, everyone knows you'll never get silly low distortion so the competition doesn't exist.
And a lot of people want that feedback because of the speaker type they have. More feedback = more damping (lower output impedance).
It would be fun to see some very tube like SS amps. Although, tubes do lend themselves to very high voltage rather easily. Not all transistors do.
Good subject for discussion, but IMHO this was true up to the 70s - then experts studied the mechanisms of feedback and found what was needed to avoid these poorly sounding artifacts. Some high feedback designs sound poor because of other poor design options, not because of the feedback itself.
Yes, we agree that two poorly sounding pieces of equipment can't sound good when paired. But sound signatures are too subjective to get a general agreement on what sounds good and poor, and their perception depends a lot on system. Although I prefer avoiding mixing different technologies, I do it for peace of mind and economy of time, not for audio knowledge or audio religion - I have listened to too many great systems with mixed technologies.
Both tube and transistor gear have audible signatures...the tubes often get much of their signature from their output transformers (talking mainly about amps) and/or feedback. Transistors mainly from feedback and their inherently lower linearity (compared to triodes at least). The latter seems to be less natural to a growing number of audiophiles...this would fit with what is known from the distortion they produce and studies on perception of distortions. Tubes don't have to sound "soft" and transistors, if very carefully handled, don't have to sound analytical but more often than not that is how they sound and thus the stereotypes. I have gone one step further and banished push/pull, which also alters the distortion pattern, IMO for the worse. I would be more inclined to have a single ended transistor amp (like a Pass SIT) than I would a push/pull anything. I go back from time to time...just to be sure. The last three PP amps were the Einstein "The Absolute Tune", VAC 30/30 MKIII and the PureSound A30. The last big SS I owned was a while ago (Moon W5). I had one really nice sounding PP hybrid (Sphinx Project 14 MKIII) and OTLs (Silvaweld reference monos), which were shocking in clarity but still somewhat sharp and ultimately less natural sounding. I still want an OTL SET to complete the survey.
Nice to know of your own evolution and preference, but where do you get the idea that transistors seem to be less natural to a growing number of audiophiles? From a small sample of audiophiles who share your preferences?
IMHO we must banish the stereotypes from discussions and debate the specific best of each technology to learn anything new about the technical and sound evolution of current products. Otherwise generalization without proper valid statistics is inappropriate, each of us favoring his current preferred topology.
Last edited:
Curiously purity in this sentence only means simplicity - although purity looks much better in the sentence. From an objective point of view SETs are not pure at all!
Well said. Some like to be in their own bubble.
Everybody perceives differently, and nobody can claim an 'absolute truth' in that regard. Be happy with what works for you, and don't be surprised if others disagree. If you found your truth (or preferences, based on individual perception and priorities), don't think others must share it. And don't think if others don't share it, they are "wrong".
Audio is not like scientific facts.
It is known that tube amplifier tend to show higher distortion No. but with benign second harmonics.
SET tend to have less crossover distortion due to class A operation.
But important thing is how it sounds with your speaker.
I am happy with my current combination.
In Korea, most shops use SS amplifiers for auditioning product.
The reason is to reduce warmup time and heat in small soace.
I like the sound from Lampi Pacific coupled with SS amp.
But I get the feeling that it could be too full with tube amp.
Audio is personal choice.
I prefer transparent soundstage with lot of details but not lean or analytical.
With combination of Lansce 4.1, Line Magnetic 508 SET, Chord Dave and Mscaler, I am happy to get the sound that I want.
Please also note that I am using Spiritual Audio power conditioner and 5 separate isolation transformer for each components and LI battery pack for Mscaler.
With wrong combination, Chord Dave and Mscaler could sound analytical and lean.
My 9 years old EMMLab Dac2 is no slouch in details, dynamics, wide and deep soundstage.
But Chord Dave and Mscaler is better in more details and transparent and deeper soundstage.
Thus I am happy with this combination.
I am playing with several digital cables that I have.
R2R Dac s are very popular with organic sound.
But some of them has shortage of bass slamm (not MSB Select II and Pacific).
Chord Dave and Mscaler, could give excellent result with proper system matching.
If I could stretch my budget to 100k$, then I would have wandered between MSB Select II and Metronome Kalista.
But within budget of 20K, I am happy to get the sound that I want.
SET tend to have less crossover distortion due to class A operation.
But important thing is how it sounds with your speaker.
I am happy with my current combination.
In Korea, most shops use SS amplifiers for auditioning product.
The reason is to reduce warmup time and heat in small soace.
I like the sound from Lampi Pacific coupled with SS amp.
But I get the feeling that it could be too full with tube amp.
Audio is personal choice.
I prefer transparent soundstage with lot of details but not lean or analytical.
With combination of Lansce 4.1, Line Magnetic 508 SET, Chord Dave and Mscaler, I am happy to get the sound that I want.
Please also note that I am using Spiritual Audio power conditioner and 5 separate isolation transformer for each components and LI battery pack for Mscaler.
With wrong combination, Chord Dave and Mscaler could sound analytical and lean.
My 9 years old EMMLab Dac2 is no slouch in details, dynamics, wide and deep soundstage.
But Chord Dave and Mscaler is better in more details and transparent and deeper soundstage.
Thus I am happy with this combination.
I am playing with several digital cables that I have.
R2R Dac s are very popular with organic sound.
But some of them has shortage of bass slamm (not MSB Select II and Pacific).
Chord Dave and Mscaler, could give excellent result with proper system matching.
If I could stretch my budget to 100k$, then I would have wandered between MSB Select II and Metronome Kalista.
But within budget of 20K, I am happy to get the sound that I want.
Last edited:
It clearly never occurred to you that correlating sound quality with measured data indicates a lot of testable hypotheses regarding designs and the concept of linearly.
Not to mention that it was already known a long time ago that it is not amount of distortion but the order of the distortion. When I say harmonics I am not talking about harmonics in music , I am talking about the additional harmonics generated by a non-linear system. This is clear with single tones and only gets much worse with real music and negative feedback. Pass wrote an interesting paper on the subject as did Cheever, Crowhurst and others. Boys and Sussmann showed numerical the clear issues with the non-linearity of transistors...they are significantly less linear devices than triodes. This is simply physics. Cleaning it up with feedback creates other issues.
You think high order distortion is a thing of the past?? Go look at measurements from many modern amps...this only a part of the story but it points the direction.
No, before feedback is applied the bipolar transistor is the least linear and the triode the most linear with the FETs somewhere in between. Their transfer functions are fairly well known...see Boyk and Sussmann. This is why simple, no feedback amps are even possible with reasonable distortion using triodes...good luck with bjts. It seems Pass did it with special SITs or some FETs.
And yet one can clearly still hear those signatures. A Tubey sounding SS amp only sounds that way in a very superficial manner. Compared to a good triode amp (PP, SET or OTL) it will be less transparent and usually with a darkish character rather than an open one like the good tube amp. Heard this with many brands over the years. ALL amps have signatures regardless of how low their distortion is supposed to be and to my ears there are telltale signs of transistors and also some for tubes...Good subject for discussion, but IMHO this was true up to the 70s - then experts studied the mechanisms of feedback and found what was needed to avoid these poorly sounding artifacts. Some high feedback designs sound poor because of other poor design options, not because of the feedback itself.
Yes, we agree that two poorly sounding pieces of equipment can't sound good when paired. But sound signatures are too subjective to get a general agreement on what sounds good and poor, and their perception depends a lot on system. Although I prefer avoiding mixing different technologies, I do it for peace of mind and economy of time, not for audio knowledge or audio religion - I have listened to too many great systems with mixed technologies.
Nice to know of your own evolution and preference, but where do you get the idea that transistors seem to be less natural to a growing number of audiophiles? From a small sample of audiophiles who share your preferences?
IMHO we must banish the stereotypes from discussions and debate the specific best of each technology to learn anything new about the technical and sound evolution of current products. Otherwise generalization without proper valid statistics is inappropriate, each of us favoring his current preferred topology.
I tried to get SET OTL sound from one of the worlds only SET hybrids (NAT Symbiosis SE) but even though it had only a single MOSFET on the output stage the signature of a transistor was there and never achieved the transparency of s true OTL. In some ways it sounded amazing but ultimately slightly unnatural.
At Munich there were several rooms with a particular type of SS amp of the ultra modern persuasion...all these rooms had a similar signature despite different speakers, sources, cables, power , room treatment etc. I was struck by this sameness that must be a coloration.
Just go to Munich in the last 10 years or Frankfurt before and see the industry evolution it is clear that SETS , which were once an odd preoccupation of the Japanese are now mainstream in the audiophile world ( and for sound quality this is the world that matters). It was literally an explosion in choices and owners.
Preferred sound signature can be possibly be determined scientifically but like all psychological endeavors it will have many exceptions. The way we hear though and masking effects impacting perception can be determined...what you like though may depend a lot on prior conditioning of expected norms.
If you run a transistor in saturation or completely open loop without anything in the circuit, sure. The tube has a minor benefit running open but it’s still stuck within an operating range because it will kill itself completely open, rather easily. It’s perfectly possible to build feedbackless transistor designs that don’t use open loop for gain. And yes, they can have very low distortion.
The whole distortion profile is so... decades ago.
We should be talking about nested feedback, error correction, crossover compensated plate amps, etc... But it’s like you grab the world’s dustiest violin every time you think you can impress with a technical post.
The whole distortion profile is so... decades ago.
We should be talking about nested feedback, error correction, crossover compensated plate amps, etc... But it’s like you grab the world’s dustiest violin every time you think you can impress with a technical post.
Bruno Putseys showed in a white paper that nested local feedback is mathematically equivalent to global feedback. What do you mean by “error correction” that isn’t feedback? Crossover compensated plate amps? Have you seen the measurements on something like that which indicates there is no zero crossing distortion? Even in 2019 you can see class AB amps that commit this sin and I am sure it’s audible.
Mathematically is not subjectively. Also some engineers that do nested feedback disagree about actual behavior under varied loads.
Error correction is a form of comparing, and doesn’t inherently change gain. Feedback only works by changing open loop gain. Error correction can be used in feedback, but can also be feedbackless.
The only crossover compensated amps I’ve seen so far are out sims. And yes, the whole point is they are immune to crossover problems.
Not sure what you mean by last sentence.
Error correction is a form of comparing, and doesn’t inherently change gain. Feedback only works by changing open loop gain. Error correction can be used in feedback, but can also be feedbackless.
The only crossover compensated amps I’ve seen so far are out sims. And yes, the whole point is they are immune to crossover problems.
Not sure what you mean by last sentence.
From what I have read about error correction it sounds like a mix of nested positive and negative feedback.
So, you don't know of any actual crossover compensated amps...?
I meant that in 2019 there are plenty of Class AB amps that commit the sin of zero crossing distortion.
https://www.stereophile.com/content/verity-audio-monsalvat-amp-60-power-amplifier-measurements (check Figures 7-10)
https://www.stereophile.com/content/bricasti-design-m15-power-amplifier-measurements (no crossover distortion but quite a bit of high order harmonics and a sharp rise in distortion at high frequencies)
https://www.stereophile.com/content...n-mono-monoblock-power-amplifier-measurements
https://www.stereophile.com/content...twenty-monoblock-power-amplifier-measurements
Curious that you refer to Bruno Putseys - he wrote a complete series about negative feedback showing mathematically that negative feedback is a good thing and the more the better, and that higher gain structures are faster than the lower gain structures preferred by many audiophiles.
IMHO high-end audiophiles should avoid science and mathematics - although superficially they can find some support in it, once we plunge in depth we find it usually turns against our view of the hobby.
Curious that you refer to Bruno Putseys - he wrote a complete series about negative feedback showing mathematically that negative feedback is a good thing and the more the better, and that higher gain structures are faster than the lower gain structures preferred by many audiophiles.
IMHO high-end audiophiles should avoid science and mathematics - although superficially they can find some support in it, once we plunge in depth we find it usually turns against our view of the hobby.
Seeing as I do science professionally (analytical chemist...I love measurements...but I love much more correlation of measurements to actual sonic characteristics. A test should reflect some critical quality parameter or it is a pointless test), I don't think I can really avoid delving into it in my hobby
I have read this from Bruno but given the sonic outcome I am not convinced he is right. Mathematically, negative feedback was generally considered a good thing by engineers (as Bruno is) who think the absolute amount of distortion is paramount and not the type (others have a differing opinion). For Folsom, I was merely pointing out that mathematically it can be shown that nested local feedback and global feedback behave the same (since he brought up I should be more focused on these more "modern" methods of signal "correction")...whatever you might want to make of that sonically...to me that means that people who claim to only use local feedback may not be gaining a sonic advantage (as they are clearly implying by stating this detail) that they think they are. Measurement wise, it matters very little...sonically it might.
Error correction IS FEEDBACKLESS. It increases the linearity of operation, as opposed to using feedback, to make a more linear output.
It is funny you quote Bruno, all things considered. He's a really smart guy, but you could say he's developed philosophy because research is not sales.
It is funny you quote Bruno, all things considered. He's a really smart guy, but you could say he's developed philosophy because research is not sales.
http://www.cordellaudio.com/papers/MOSFET_Power_Amp.pdf
And I quote:
"The technique of Fig. 1 1 also tends to become less effective at very high frequencies because, being a feedback loop (albeit not a traditional negative feedback loop), it requires some amount of compensation for stability, detracting from the phase and amplitude matching."
So, you can call it what you want, but part of the signal is still fed back into the original signal flow. Now, it seems that they are subtracting out, to some degree, the signal from the distortion signal and then feeding this (mostly) distortion signal back into the output stage.
When they say very high frequency they mean well past the audio band. And all circuits require something for some stability.
I don't follow your statement. It compares the original signal to the output and you're left with the original. While you can draw a line on the circuit that makes you think feedback, they're very different. It has no effect on gain, which is exactly how feedback works.
I don't follow your statement. It compares the original signal to the output and you're left with the original. While you can draw a line on the circuit that makes you think feedback, they're very different. It has no effect on gain, which is exactly how feedback works.
