Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Tubes are objectively best! Feedback is bad! (Read 31630 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Tubes are objectively best! Feedback is bad!

Reply #75 –
Quote
That makes me wonder what a tube amp without any feedback would sound or evel look like. I guess you'd have to use a pentode instead of a triode? And it would behave more like a single-ended transistor, with a highly quadratic response?
[a href="index.php?act=findpost&pid=311669"][{POST_SNAPBACK}][/a]


A pentode is not actually less linear than a triode, in a practical sense. If you look at the characteristics you'll see that, beyond the "knee" they're actually quite linear, but the slope is very different from that of a triode, hence the very high output impedance and high mutual conductance. A pentode without feedback would be very difficult to control and prevent instability and oscillation. In any case, the output impedance is so high that such a circuit would need a ridiculously high impedance output transformer, leading to even more likelihood of oscillation.

Interestingly, the supporters of the current fad for valve (tube) amps often decry solid-state amplifiers universally, yet many of those are based on FET devices that have essentially the same characteristics as triodes, minus the disadvantage of needing output transformers. Single-ended MOSFET amp anyone?

Back to your point, I'd say that a working and usable (in a practical, audio sense) amp without any feedback is impossible, at least at present (and probably ever). It would need perfect components with zero drift. Someone tell me how, if they know different...

Tubes are objectively best! Feedback is bad!

Reply #76 –
Quote
A pentode is not actually less linear than a triode, in a practical sense. If you look at the characteristics you'll see that, beyond the "knee" they're actually quite linear, but the slope is very different from that of a triode, hence the very high output impedance and high mutual conductance.[a href="index.php?act=findpost&pid=311700"][{POST_SNAPBACK}][/a]



We certainly don't agree, and the high output impedence is why we don't agree.

There is a voltage feedback mechanism in a triode, there isn't in a pentode, unless you use the screen-grid "triode" connection.

Pentodes have (or had) their uses, indeed, that's not what I'm arguing, I'm saying that a pentode lacks the internal feedback mechanisms of the triode.
-----
J. D. (jj) Johnston

Tubes are objectively best! Feedback is bad!

Reply #77 –
Quote
Quote
A pentode is not actually less linear than a triode, in a practical sense. If you look at the characteristics you'll see that, beyond the "knee" they're actually quite linear, but the slope is very different from that of a triode, hence the very high output impedance and high mutual conductance.[a href="index.php?act=findpost&pid=311700"][{POST_SNAPBACK}][/a]



We certainly don't agree, and the high output impedence is why we don't agree.

There is a voltage feedback mechanism in a triode, there isn't in a pentode, unless you use the screen-grid "triode" connection.

Pentodes have (or had) their uses, indeed, that's not what I'm arguing, I'm saying that a pentode lacks the internal feedback mechanisms of the triode.
[a href="index.php?act=findpost&pid=311855"][{POST_SNAPBACK}][/a]


I don't see how we disagree then. I suggest you go look at the cascade curves. Pentodes have a longer "knee" and different slopes, but they are just as linear along the main part of the charcteristic, if not more so. I'm not disputing the "feedback" inherent in triodes, the screen grids of a pentode were introduced exactly to counteract this feedback. As a result. they are useful HF amplifiers, which triodes are not. The consequence of the screen grids is the high output impedance, but I don't see how that affects the linearity or otherwise. Pentodes deviate in a different manner, which leads to odd-order harmonics which triodes are less prone to, but neither is absolutely linear in any case.

Tubes are objectively best! Feedback is bad!

Reply #78 –
Quote
Quote
Quote
If exactly the same signal reaches a loudspeaker from two different amplifiers (as best as we can measure), they'll sound the same.
Wrong. The amps damping factor has direct influence on the speakers behaviour at resonance frequency. Dont underestimate the direct corellation speaker <--> amp in this discussion.[a href="index.php?act=findpost&pid=311267"][{POST_SNAPBACK}][/a]


I don't. In the hypothetical comparison you describe, the signal at the loudspeaker terminals is measurably different for the two different amps.


Right, i overlooked the obvious  . But even if you compensate for this effect by varying the input signal two both of the amps, so that the output signal reaching the speakers is exactly the same, sound will be different as the amp with the lower damping factor will result in a longer 'Nachhallzeit' at the bass speakers frequency ....

Christian

Tubes are objectively best! Feedback is bad!

Reply #79 –
But Christian,

The damping (or lack of) is mediated through an electrical signal, and is therefore measurable at the output.

So, if the two outputs are forced to be identical by changing the input, then this implies you are implementing some kind of active feedback to cause the damping to be accommodated earlier in the circuit.

In reality, I don't think you could do this (so the outputs would be different), but if (and only if) you could, then the outputs would be the same. Thus the sound would be the same.

In effect, if you were successful, you would have made the two different amplifiers into electrically equivalent circuits by actively pre-distorting the input to one amplifier.

Cheers,
David.

Tubes are objectively best! Feedback is bad!

Reply #80 –
Quote
But Christian, The damping (or lack of) is mediated through an electrical signal, and is therefore measurable at the output.


Right again  ... /me hides ......

How we keep saying in Germany :

'Und ist der Ruf erst ruiniert, lebt sichs gänzlich ungeniert ....'

or in a more comprehensive language

' .. si tacuisses, philiophus mansisses ...'  !!

Seems like i am in sales for a too long time now  .....

Tubes are objectively best! Feedback is bad!

Reply #81 –
Quote
If one went on measurements alone, then almost all 'properly' designed SS ampilfiers should sound identical (THD figures down in the 2nd and 3rd decimal places etc), but they certainly do not


And you know this *how*?  Assume the usual caveats about not overloading the amp, level matching channels, etc.

Tubes are objectively best! Feedback is bad!

Reply #82 –
Quote
Quote
I must admit the analysis is extremely persuasive.


If you are interested in hearing from someone that has conducted many real-world listening tests, under blinded protocols, please refer to and contact Richard Clark at caraudioforum.com. He has conducted many blinded listening tests on audiophiles, music engineers and others; comparing vastly different grades(prices and topologies) of amplifiers. No one has yet demonstrated, conclusively, that they could tell a difference when the frequency response was the same and when no audible noise(s) were present. You can contact Clark directly from that forum. Tom Nousaine(you can find him on the rec.audio.high-end newsgroup) has also conducted blinded tests on amplifiers, using various audiophiles as subjects.

-Chris
[a href="index.php?act=findpost&pid=307906"][{POST_SNAPBACK}][/a]



Tom hasn't posted to RAHE, or any other Usenet group AFAIK, for some months now, since AOL dropped newsgroups.  But I've still been able to contact him at the AOL address he used when he was posting  (and I got to meet him at the HE 2005 convention a few months ago).

I'd like a scan of that article  if you would be so kind --

Tubes are objectively best! Feedback is bad!

Reply #83 –
Quote
I'd like a scan of that article  if you would be so kind --


E-mail me at wmax@linaeum.com

-Chris

Tubes are objectively best! Feedback is bad!

Reply #84 –
Quote
Quote
If one went on measurements alone, then almost all 'properly' designed SS ampilfiers should sound identical (THD figures down in the 2nd and 3rd decimal places etc), but they certainly do not


And you know this *how*?  Assume the usual caveats about not overloading the amp, level matching channels, etc.
[a href="index.php?act=findpost&pid=312633"][{POST_SNAPBACK}][/a]


Well, I do have to butt in here to point out that SNR (which is what THD really measures in the real world) is not the best information one can get about the distortions in an amplifier.

If the distortions, with real-world broadband signals, in-situ, are down 100dB, it is most likely that they are below the absolute threshold of hearing.

If they are 60dB down, it depends hugely on what the distortion characteristics are.

If they are 60dB down from full scale, and are from center clipping, they may be atrocious-sounding when one puts in a piano solo, for instance.

One can try this very easily in simulation and confirm it, of course.
-----
J. D. (jj) Johnston

Tubes are objectively best! Feedback is bad!

Reply #85 –
I don't think that we want to get into USENET discussions here. I can well understand why any competent scientist or sane person would simply walk.

Having said that, I miss quite some of the contributors there.
-----
J. D. (jj) Johnston

Tubes are objectively best! Feedback is bad!

Reply #86 –
BTW, this topic has been in the back of my head for a while, but I haven't looked into it deeply for a while.

The internal feedback of the triode was something I was completely unaware of originally, but I'm having trouble comprehending how that can be compared to the negative feedback in an amplifier circuit. The literature shows that simply adding/removing a bias resistor in a basic MOSFET amplifier circuit (!) implies negative feedback, which I believe is roughly equivalent to what you guys are talking about with triodes. Still, that doesn't completely explain how SS amplifiers with negative feedback contain no measurable harmonics, while the simulations and derivations provided by Cheever (and later, Boyk and Sussman, a followup of some specific claims of amplifier feedback performance) show massive harmonic distortion. My best guess about the situation is that Cheever etal are bashing strawman circuits that have far too little negative feedback, and production SS amps use much higher amounts of feedback, combined with much more open-loop gain than those simplistic circuits can handle at high feedback. Regardless, the internal feedback effect of the triode does nothing to help Cheever's credibility.

Chris and I were hashing out ways to simulate THD on arbitrary signals given given a harmonic spectrum chart; I was going down a rather interesting path that is probably over my head technically. Basically what I'd like to do is take the harmonics and reconstruct the nonlinear transfer curve, by performing an inverse fourier series and mapping from that distorted sinusoid to the reference sine wave, then applying that transfer curve to whatever I want. This could then be used to test Cheever's hypothesis about high harmonics being arbitrarily more audible than low harmonics, applied to real music instead of sine waves.

In theory, that should accurately simulate frequency-independent THD and static (but not dynamic) IMD. In practice, the system itself is underdetermined (you'd need the phase plots to actually do the inverse transform properly, although that may not be relevant for the purposes of audibility) and there's a bug floating around somewhere which makes the resulting harmonic distortion look nothing like what we started with.

I guess it would be far easier, and potentially more accurate, to just create a 64k lookup table mapping from samples in a wav to recorded output. But then you might have to deal with subsample delays from input to output. Suggestions?