If "Tube Sound" Is a Myth, Why Tubes?

[Julf]

Yeah, but and still, things can get downright hostile 'round these parts.

Sure. A lot of us are rather tired and frustrated at the same old "I have been musical director of a horn orchestra for 50 years. Are you saying I can't hear a difference?" stuff.

 

[Robin L]

Sure. A lot of us are rather tired and frustrated at the same old "I have been musical director of a horn orchestra for 50 years. Are you saying I can't hear a difference?" stuff.

I'd say a musical director of a horn orchestra is lucky to hear anything after 50 years. I was a recording engineer for 10 years and I know my upper registers are pretty much blown. Glad I don't have tinnitus [knock on wood].

 

[Blumlein 88]

I'm aware of all that.

My running hypothesis has been that the CJs do sound different than the typical SS amp, and it continues to be my presumption. But appropos of the very subject of this thread, I'm always aware of the possibility of sighted bias, and several members have given reasons to maintain those doubts (including levimax's blind test results with his Dynaco ST-70).

So I tried to balance my post in tune with the OP: both the question of "why tubes" (here's why I stick with my tube amps) and the question of the audibility of tube amp behavior.

If I did a blind shoot out of the CJ with the Bryston and couldn't tell the difference, on one hand I'd be shocked because the difference I hear seems so distinct (not "huge" but distinct). On the other hand I wouldn't be shocked because I'm acutely aware of the power of expectation effects etc.

So I think this thread stands a nice "open argument" questioning the assumptions regarding the audible qualities of tube amps.

On the other hand, for the rest of the discussions, I think it's reasonable to continue along discussing preferences for things like vinyl, tube amps and whatever speakers we choose, on the assumption of audible differences, which are plausible in such cases. If every time we wanted to mention a speaker we liked someone said "but that's out of bounds unless you have been to the Harman Kardon facilities to truly properly blind test speakers" then discussion would get impractical. I think, for general purposes, perhaps tube amps can go in to that category, with the caveat that while the audible differences can be plausible, we all know that it's always possible some bias is playing a part.

At least that's my take in trying to wend through these issues.

Be nice to do the Swedish style serial amp test.

Next best thing would be to make yourself a voltage divider, and record at the speaker terminals using a good ADC. Capture both amps playing the same music and let us hear for ourselves.

 

[SIY]

...the puzzler then remains why the sonic difference I hear with the CJ, it's "character," seems so consistent across different speaker loads, including "easy" to drive speakers.

That is an extremely pertinent observation and strongly points to the answer.

 

[ahofer]

We've been banned, herded into special corners where we're only allowed to discuss the rational side of things, and gang raped by the members for daring to question their religion. The Spanish Inquisition version of audio. That won't happen here.

That might be a *bit* much. Audiogon types (generally older and likely hearing impaired) are constantly claiming I must have tin ears and a crap system, and the nitwits on the Harbeth Facebook group just ignore me, but it isn't quite being put on the strappado or the rack until I confess.

 

[MakeMineVinyl]

.... "audiophile" sites where the magic dust believers have run us off with stakes and pitchforks whenever we tried to explain the facts to them about the snake-oil, pure BS things they gush over every day like digital cables, power cords, magic dots and all the rest. We've been banned, herded into special corners where we're only allowed to discuss the rational side of things, and gang raped by the members for daring to question their religion. The Spanish Inquisition version of audio. That won't happen here.

As much as I like this site and believe overwhelmingly in its scientific approach, to be honest I've seen "Spanish Inquisition version of audio" happen here too, and much too often. I don't like closed minds, whether on the subjective or objective side. I don't appreciate dogma of any flavor. It screams of propaganda, and its something which I try to avoid strenuously. Listening to music is by definition a subjective experience, although some people will dwell on the composition structure more and try to find some mathematical logic to it all. Electronic equipment qualification is largely an objective experience although some people will be convinced that there are senses being activated which transcend the state of the art of measurements. Speakers, phono cartridges and other electro-mechanical transducers (including microphones) have a foot in both realms. Let's live with those realities, because they ain't changing.

 

[mocenigo]

As much as I like this site and believe overwhelmingly in its scientific approach, to be honest I've seen "Spanish Inquisition version of audio" happen here too, and much too often.

People that believe they can assess the quality of a DAC, say, from the SINAD value, are as clueless as those that claim that only subjective listening counts.

Let us take a DAC. SINAD is measured at 4V output. Who ever does output everything always close to 4V? Most of the cases, unless you are listening to some heavily compressed disco music, or some heavy metal, even at -0db you are outputting 0.5V-1V AT MOST. Often even less. Of course for those values both distortion and noise would be different. And, here's the rub, the behaviour of noise and distortion at attenuated values is not always clear: with a SMD or a simple R2R DAC the distortion will increase as the output decreases, roughly in a linear way. With a sign-magnitude R2R DAC much less so – in fact the original signal will still be detectable at very high attenuations (even 100Db!!!) whereas with a SDM DAC you can forget it. Of course this is as much as extreme as measuring everything at 4V (measurements at both 0.25V and 1V would complement that nicely, and give a better picture) but it tells us that there is a cutoff somewhere. So a sign-magnitude R2R DAC may perform much better than a "ASR 4V output TOTL" DAC at realistic levels, with a higher noise floor but with less distortion – or maybe not, but we do not have conclusive evidence of that from the information available here. But sometimes, pointing out at these basic facts led to attacks.

I love my D90 and it sounds simply divine. It happens it is also one of the "best measuring DACs ever". Indeed, I *do* believe that there is a correlation between Amir's and Wolf's measurements and sound quality – I just do not believe that there is an equivalence. Also, the choice of OPA1612 in the IV conversion circuit is clearly driven by optimising noise and distortion. Performance under difficult loads may be better served by OPamps that can handle higher currents, for instance. May it have sounded even better if the design had not been driven by the need to win the SINAD race here? Some Japanese DIYers and modders find other opamps there sounding better even if you lose a couple of Db in the SINAD.

Please do not misunderstand me – this site is very useful. But the extreme oversimplification of the way measurement results are read is not helping the cause of higher quality audio reproduction as it could. I also understand that the masses (even small masses as the lovers of quality audio) need simple explanations (as in politics), as the full story is always too difficult to grasp.

 

[ahofer]

And I am not correlating engineering quality with listening impressions at all...

Indeed, I *do* believe that there is a correlationbetween Amir's and Wolf's measurements and sound quality – I just do not believe that there is an equivalence.

I agree SINAD is not one measurement to rule them all, but I'm not clear on what you are saying. Did you mean non-signal-related "engineering" in the first statement?

I suppose the pertinent question in all of this is whether there is an aspect of sound quality that isn't measurable in any step prior to the speaker?

Evidence suggests not, to me.

 

[andreasmaaan]

With a sign-magnitude R2R DAC much less so – in fact the original signal will still be detectable at very high attenuations (even 100Db!!!) whereas with a SDM DAC you can forget it.

I tend to agree with the gist of your post, notwithstanding that I think a number of members here put a lot of work into explaining the subtle details of psychoacoustics and subjective-objective correlation (although I agree that not everyone does).

But this particular comment above seems a stretch to me. Are you saying that no SDM DAC is capable of reproducing an even detectable signal at -100dB?

 

[DSJR]

That might be a *bit* much. Audiogon types (generally older and likely hearing impaired) are constantly claiming I must have tin ears and a crap system, and the nitwits on the Harbeth Facebook group just ignore me, but it isn't quite being put on the strappado or the rack until I confess.

Which Harbeth facebook group as I believe there are two? One is a dealer based one I think and I was quietly deleted from it for going against the valve-vinyl vibe it had. The other one ain't so bad and the people there do seem to listen once evidence is submitted, at least in my experience

 

[StefaanE]

Some Japanese DIYers and modders find other opamps there sounding better even if you lose a couple of Db in the SINAD.

With the performance of current DACs, a couple of dB in the SINAD will be inaudible, so how do you define “sounding better”? A device they modded and which has now measurably worse performance “sounds better” to them. How did they come to this conclusion? Do they use flowery audiophool language to describe an “improvement” based on their memory of the pre-modded sound?
Or are you suggesting that there are other things to measure than the Signal-to-noise and Distortion to determine that one device ”sounds better” than another? If so, what are they?

 

[Blumlein 88]

People that believe they can assess the quality of a DAC, say, from the SINAD value, are as clueless as those that claim that only subjective listening counts. And I am not correlating engineering quality with listening impressions at all.

Let us take a DAC. SINAD is measured at 4V output. Who ever does output everything always close to 4V? Most of the cases, unless you are listening to some heavily compressed disco music, or some heavy metal, even at -0db you are outputting 0.5V-1V AT MOST. Often even less. Of course for those values both distortion and noise would be different. And, here's the rub, the behaviour of noise and distortion at attenuated values is not always clear: with a SMD or a simple R2R DAC the distortion will increase as the output decreases, roughly in a linear way. With a sign-magnitude R2R DAC much less so – in fact the original signal will still be detectable at very high attenuations (even 100Db!!!) whereas with a SDM DAC you can forget it. Of course this is as much as extreme as measuring everything at 4V (measurements at both 0.25V and 1V would complement that nicely, and give a better picture) but it tells us that there is a cutoff somewhere. So a sign-magnitude R2R DAC may perform much better than a "ASR 4V output TOTL" DAC at realistic levels, with a higher noise floor but with less distortion – or maybe not, but we do not have conclusive evidence of that from the information available here. But sometimes, pointing out at these basic facts led to attacks.

I love my D90 and it sounds simply divine. It happens it is also one of the "best measuring DACs ever". Indeed, I *do* believe that there is a correlation between Amir's and Wolf's measurements and sound quality – I just do not believe that there is an equivalence. Also, the choice of OPA1612 in the IV conversion circuit is clearly driven by optimising noise and distortion. Performance under difficult loads may be better served by OPamps that can handle higher currents, for instance. May it have sounded even better if the design had not been driven by the need to win the SINAD race here? Some Japanese DIYers and modders find other opamps there sounding better even if you lose a couple of Db in the SINAD.

Please do not misunderstand me – this site is very useful. But the extreme oversimplification of the way measurement results are read is not helping the cause of higher quality audio reproduction as it could. I also understand that the masses (even small masses as the lovers of quality audio) need simple explanations (as in politics), as the full story is always too difficult to grasp.

We can measure distortion at any level we choose. The results don't match your description. Sdm dacs can have signal well below -100 db.

Just for example here is a measured 20th bit from a Focusrite Forte SDM based DAC. Measured with another ADC. 128 k FFT. The signal sits between -120.4 and -120.5 dbFS.

 

[ahofer]

Which Harbeth facebook group as I believe there are two? One is a dealer based one I think and I was quietly deleted from it for going against the valve-vinyl vibe it had. The other one ain't so bad and the people there do seem to listen once evidence is submitted, at least in my experience

https://www.facebook.com/groups/176400719064958/

Currently called "Harbeth Speaker Enthusiasts"

 

[mocenigo]

I agree SINAD is not one measurement to rule them all, but I'm not clear on what you are saying. Did you mean non-signal-related "engineering" in the first statement?

I suppose the pertinent question in all of this is whether there is an aspect of sound quality that isn't measurable in any step prior to the speaker?

Evidence suggests not, to me.

I must admit that the first sentence you quoted must have been the by product of sloppy editing and I have removed it. It is NOT what I mean and I was not able to reconstruct what I wanted to write (cough).

I meant that there is correlation between engineering quality and this value, but it is a number that does not tell the whole story, as I explained later. If that number is bad, then it is unlikely that the DAC is ok at other output levels - but it could happen if 4V is a level past clipping.

 

[mocenigo]

We can measure distortion at any level we choose. The results don't match your description. Sdm dacs can have signal well below -100 db.

Just for example here is a measured 20th bit from a Focusrite Forte SDM based DAC. Measured with another ADC. 128 k FFT. The signal sits between -120.4 and -120.5 dbFS.

View attachment 98104

thank you, I stand corrected. Also SDM has improved here. Good to know.

 

[Sal1950]

As much as I like this site and believe overwhelmingly in its scientific approach, to be honest I've seen "Spanish Inquisition version of audio" happen here too, and much too often. I don't like closed minds, whether on the subjective or objective side.

I don't see that happening here at all. Our minds are wide open to investigating any claims that can be supported by any scientific method. We just don't swallow the "I hear it, so it is real" claims. If you insist you hear some abnormality, just back it up with bias controlled DBTs or measurements. I just don't want to hear any more magic dust claims, I can go anywhere else and read that crap till my ears bleed. Hell I already do with subscriptions to TAS and Stereophile going back to near issues 1, (like the Playboy readers, I do it for the stories, (music reviews). )

As to SINAD and all that, Amir does fantastic investigations into which companies designs offer the best engineering and are producing SOTA design and numbers. But like amplifiers, the problems that make a DAC circuit have audible issues have been solved for decades and fairly major design failures have to be made to introduce audible differences when listened to under controlled conditions. Like the subjectives, we are all also influenced by bias. If you believe getting the best measuring component with give you the best sound, it will. OTOH, The fact is that it is a guarantee that SOTA measurements will give you SOTA audible performance, how far degraded the numbers have to get before SQ is audibly effected is highly debatable.

 

[ahofer]

I don't see that happening here at all. Our minds are wide open to investigating any claims that can be supported by any scientific method. We just don't swallow the "I hear it, so it is real" claims.

I do. When someone chimes in with "I hear a difference...", people sort of assume they are going to get the usual accompaniments (my ears are better, my equipment is better, the market proves my point, etc.) and jump in pretty aggressively, as if they had said all those things.

It's like bad relationships with family or spouse, you repeat the pattern, and assume the provocation even when it isn't (yet) in evidence, and then get outraged when it comes in response to your aggression. I'm not paranoid, they are out to get me.

The Isolda guy, for instance, was willing to mix it up in this forum, which I applaud. He wanted to convince objectivists. I assumed we would obtain one more data point of wire testing, but everyone jumped on him (I understand his test design may have been misleading, possibly by intention). I was surprised he stuck around. I wish more of those types would come and duke it out on something as substantial as a measured test.

Overall, people read way too much intent into statements on the internet. More than they do in person. Hard habit to moderate, let alone break.

btw, I wish someone would produce a mysterious blind test result. It would mean a new design frontier was opening.

 

[andreasmaaan]

btw, I wish someone would produce a mysterious blind test result. It would mean a new design frontier was opening.

Or that a control hadn’t been properly implemented...

(Generally agree with your post though.)

 

[ahofer]

Fremer has supposedly had some blind success. He picked out amps in the original challenge - https://www.stereophile.com/features/113/index.html

"As The Abso!ute Sound's Michael Fremer reports in this month's "Letters" section, no only did he identify correctly five times out of five whether one amplifier was the same as the next or different, he also volunteered identifications of the amplifiers under test—Crown, Threshold, and VTL—and got that right four times out of five."

and then cables here, allegedly -
https://www.wsj.com/articles/SB120044692027492991

Anyone know the back story? Is he having us on?

 

[StefaanE]

The Isolda guy, for instance, was willing to mix it up in this forum, which I applaud. He wanted to convince objectivists. I assumed we would obtain one more data point of wire testing, but everyone jumped on him (I understand his test design may have been misleading, possibly by intention). I was surprised he stuck around. I wish more of those types would come and duke it out on something as substantial as a measured test.

I read the whole thread until the pause, and felt in a sense sorry for him. It was a good story without a conclusion, as I remember posting at the time. Max then reappeared and doubled down on his claims.
As a result, I had a gander at his website, and it became abundantly clear he is a real snake-oil salesman. A “pre-amp” that’s nothing more than a passive six-way switch with an attenuator for £, to be used with his special “Fractal” cable at £2200 per meter, with the following “qualities”:

• Unrivalled sound quality with crystal clear treble, open mid-band and subterranean bass
• Ultra high end performance in a small package
• Purely passive Autotransformer design wired throughout with Fractal-Wire™ to perfectly match any source component with any power amplifier
• No mains noise, no electronic hash and no requirement for esoteric mains conditioners or power cord

The rest of his products are similar scams. The man is a shameless huckster who will abuse the semblance of seriousness the discussion of his meaningless test on this forum has given him.
I am the last one to complain when gullible people with more money than sense buy overpriced products, but this guy is a swindler.
The line has to be drawn somewhere...