Classic Marantz and Citation vs modern

[atmasphere]

But of course the GedLee metric shows nearly any engineered amp to be innocuous. And Earl uses cheap solid state electronics in his system.

Your audibility speculations are still unsupported with evidence.


Did you see what the level of those higher harmonics is? The claim that they are audible is (I'm sorry to use this term but it's accurate) ludicrous.

To the first, I was not talking about the metric beyond the idea that harmonics should be weighted.

Why do you think they would be inaudible? Do you think they are masked?

 

[SIY]

Why do you think they would be inaudible?

Because they are ridiculously low. Like four orders of magnitude lower than the very best transducers.

 

[atmasphere]

They are quite low!

So you are saying that even though they modify the signal, that the tonality they add is inaudible, simply because the modification to the signal is so slight. Do I have that right?

BTW I'm not disagreeing that they might not be audible. Its just that I've found that distortion is a lot more audible than most people realize.

 

[SIY]

So you are saying that even though they modify the signal, that the tonality they add is inaudible, simply because the modification to the signal is so slight. Do I have that right?

Not just slight, absolutely buried under the rest of the system distortion (specifically transducers). I mean, you can't listen to an amplifier's electrical signal, you have to hook up speakers or headphones. And, at least for real instruments, you need a microphone.

Its just that I've found that distortion is a lot more audible than most people realize.

A few minutes playing with @pkane 's terrific Distort software will disabuse people of that notion. We are highly sensitive to level, frequency response, and localization. Our sensitivity to distortion is relatively mediocre.

 

[atmasphere]

Not just slight, absolutely buried under the rest of the system distortion (specifically transducers). I mean, you can't listen to an amplifier's electrical signal, you have to hook up speakers or headphones. And, at least for real instruments, you need a microphone.

A few minutes playing with @pkane 's terrific Distort software will disabuse people of that notion. We are highly sensitive to level, frequency response, and localization. Our sensitivity to distortion is relatively mediocre.

The microphones I own are Neumann U67s. I bought them for $1500/pair in the early 1980s and it seemed like a foolish thing to do at the time, but they really have paid off.

I don't see how the initial statement here holds water. The distortion of other parts of the system simply adds up and changes sound. Since you have tube equipment, you know that even though your speakers have more distortion than the amps, you can easily hear the difference between your tube amps and most solid state amps. Sometimes lower orders can mask higher orders, but the signal itself cannot- its modified by whatever distortions are present and since the ear uses harmonics to sense tonality, the tonality of the system changes. That's the same reason the differences you hear between amps are literally the distortion signatures of those amplifiers. None of this should be controversial and IMO its a shame that it is.

When the 'sound' of something is discounted as anecdotal, the same sin is committed as the subjective guys do when they discount measurements. If all the measurements are there and are accurate, you should be able to tell how an amp sounds if you understand the significance of the measurements (that latter bit is rare, even amongst the people doing the measurements). Put another way "If it measures good and sounds bad, -- it is bad. If it sounds good and measures bad, -- you've measured the wrong thing." (Daniel von Recklinghausen)

Distort is a great bit of software, but according to the designer, while he wrote a bit of software to allow for distortion rise with frequency, he didn't include it in the release because he felt it too arcane. So what Distort really shows is how distortion becomes inaudible if it does not rise with frequency (something rare in amplifiers that employ feedback). So its good for that but not very useful if you want to simulate the sound of a certain amp and you know the turnover frequency where distortion begins to rise, along with its slope.

My contention is that if distortion rises with frequency, it can become unmasked and the resulting (and very slight) distortion above the turnover frequency can be interpreted as brightness. That is why I celebrate newer solid state amplifier designs where the designer solved this problem: self-oscillating class D amps and a few others like the Topping B100 and Benchmark.

 

[SIY]

you can easily hear the difference between your tube amps and most solid state amps.

No, I can't. My tube amps are well designed, with the idea of making small signals larger.

When the 'sound' of something is discounted as anecdotal

Only when there's no controls. Do an actual ears-only listening test and maybe there's something to discuss.

My contention is that if distortion rises with frequency, it can become unmasked and the resulting (and very slight) distortion above the turnover frequency can be interpreted as brightness.

OK, let's see some evidence. Contentions are cheap.

 

[SIY]

The microphones I own are Neumann U67s. I bought them for $1500/pair in the early 1980s and it seemed like a foolish thing to do at the time, but they really have paid off.

They're good mics. But still have MASSIVELY more distortion than an engineered amplifier. I have a diverse collection of mics (condenser and ribbon) which include some very low distortion models, and even those absolutely bury the distortion of any decent amp.

And then there's speakers. Sorry, if you think that having 0.1000% and 0.1001% high order harmonics is audibly different, you have a lot of work to demonstrate that.

 

[atmasphere]

No, I can't. My tube amps are well designed, with the idea of making small signals larger.

Only when there's no controls. Do an actual ears-only listening test and maybe there's something to discuss.

OK, let's see some evidence. Contentions are cheap.

They are!
I'd be interested in seeing the circuit you used- I'm assuming the tube amp you use is one you built.

They're good mics. But still have MASSIVELY more distortion than an engineered amplifier. I have a diverse collection of mics (condenser and ribbon) which include some very low distortion models, and even those absolutely bury the distortion of any decent amp.

And then there's speakers. Sorry, if you think that having 0.1000% and 0.1001% high order harmonics is audibly different, you have a lot of work to demonstrate that.

I never said 'having 0.1000% and 0.1001% high order harmonics is audibly different'; if you are thinking that I meant that your reaction is very reasonable.

To clarify, I don't think you can hear the difference between 0.1 and 0.001% unless distortion rises with frequency. If it does so, the distortion will be considerably higher. I think the THD figure tends to hide this fact.

 

[SIY]

unless distortion rises with frequency. If it does so, the distortion will be considerably higher.

It does, usually, but if the distortion is tiny, who cares?

My own circuits usually use the SYclotron voltage amp/phase splitter, but I've also used somewhat modified versions of the Mullard 5-20 and Williamson, as well as the Red Light District amp. And I have a pair of late '50s Macs.

 

[atmasphere]

It does, usually, but if the distortion is tiny, who cares?

My own circuits usually use the SYclotron voltage amp/phase splitter, but I've also used somewhat modified versions of the Mullard 5-20 and Williamson, as well as the Red Light District amp. And I have a pair of late '50s Macs.

What is meant by 'tiny' (keeping in mind how sensitive the ear is to higher ordered harmonics)? If DvsF takes off at 1KHz, the 9th harmonic could be 18dB higher in level; that could move something that is 0.001% to the neighborhood of 0.02%- it all depends on what frequency the distortion is measured. I remember my SWTP Tigers to be pretty painful so I relegated them to driving woofers. They were supposed to be 0.01%. But at 9KHz I suspect their distortion was much higher.

My point here is only how important DvsF is as a measurement- far more so than simple THD. If you don't consider how the ear is so sensitive to higher orders then low THD looks very attractive. But if DvsF is a ruler flat line THD has almost no meaning. I don't have citations on this stuff as its simply something I noticed over the decades of amplifier work. But apparently Bruno Putzeys feels its important too if you read his stuff.

Interesting- we use a differential cascode Voltage amplifier with CCS in our tube stuff...

 

[SIY]

What is meant by 'tiny' (keeping in mind how sensitive the ear is to higher ordered harmonics)

Orders of magnitude below any transducer. And "the ear" (to use your synecdoche) is not terribly sensitive to harmonics, period. Dial up Distort, put in, say, 0.001% 7th (or whatever harmonic you like), and play music. That higher order harmonic is not "masked" (to use your trope). Can you detect the difference blind? If so, please post the files and your ABX log.

 

[SIY]

we use a differential cascode Voltage amplifier with CCS in our tube stuff...

Note that my circuit is not a cascode.

They were supposed to be 0.01%. But at 9KHz I suspect their distortion was much higher.

So what? The harmonics are above the audible range. It takes special pleading and reference to discredited "research" (can you say "Oohashi?") to claim any audibility.

 

[jon_s]

No, you can't really tell that by eye. And Stereophile helpfully doesn't show the lower power spectrum nor the spectrum for higher frequency fundamentals.

Fourier by eye, the only thing in there that you might hear in the Figure 5 2W spectrum is 2nd and a whiff of 3rd. It is nothing like the near-clipping spectrum shown.

 

[SIY]

Fourier by eye, the only thing in there that you might hear in the Figure 5 2W spectrum is 2nd and a whiff of 3rd. It is nothing like the near-clipping spectrum shown.

Don't let the log scale fool you. If something is 20dB down from the overall residual, it's damn hard to see in the linear time domain waveform. 40dB down, forget it.

 

[jon_s]

Don't let the log scale fool you. If something is 20dB down from the overall residual, it's damn hard to see in the linear time domain waveform. 40dB down, forget it.

View attachment 396024View attachment 396025

I'm talking about what I can hear, in contrast to probably the 2nd, 3rd, and 4th being audible in the 67W spectrum for a 50Hz signal in Figure 6. In the Figure 5 of that review, the visible and audible harmonic in the distortion waveform are clearly 2nd. That's at 2W, so this is already < 0.1% based on the prior figures (ignoring that they don't note the scale in this plot), and I can't hear 20-40dB below that. Even the author agrees: "The waveform of the distortion at 2W into 4 ohms is shown in fig.5. It is heavily second-harmonic, with limited higher-order components." https://www.stereophile.com/content/audio-research-vt100-power-amplifier-measurements

My own circuits usually use the SYclotron voltage amp/phase splitter, but I've also used somewhat modified versions of the Mullard 5-20 and Williamson, as well as the Red Light District amp. And I have a pair of late '50s Macs.

(Also, I finally realized that you're the same SY from forum days of yore ... it has been a while since I've spent any time on audio discussions, haha.)

 

[atmasphere]

Orders of magnitude below any transducer. And "the ear" (to use your synecdoche) is not terribly sensitive to harmonics, period. Dial up Distort, put in, say, 0.001% 7th (or whatever harmonic you like), and play music. That higher order harmonic is not "masked" (to use your trope). Can you detect the difference blind? If so, please post the files and your ABX log.

Again, Distort does not include distortion rise with frequency (if it did you would have to set the turnover frequency and slope parameters). I agree that you won't hear any difference (or it would be very difficult to detect) if distortion is a ruler flat line across the audio band; Distort demonstrates this in spades so we're on the same page here as far as Distort is concerned.

Note that my circuit is not a cascode.


So what? The harmonics are above the audible range. It takes special pleading and reference to discredited "research" (can you say "Oohashi?") to claim any audibility.

I think you misunderstood. 9KHz is the 9th harmonic of 1KHz and I think even I can hear that still. I was not referring to any harmonics above 20KHz!

 

[SIY]

I think you misunderstood. 9KHz is the 9th harmonic of 1KHz and I think even I can hear that still.

No, I did not. You continually refer to the rise in distortion with frequency in feedback amps, which almost invariably happens well above 1kHz. Your specific statement was "I remember my SWTP Tigers to be pretty painful so I relegated them to driving woofers. They were supposed to be 0.01%. But at 9KHz I suspect their distortion was much higher."

You also keep talking about masking as the reason that upper harmonics can't be heard with tube amps. So my proposed experiment completely removes all masking which, if your hypothesis is correct, would be audible to you. So let's see if your hypothesis can withstand a basic test. I suspect you'll get a null result, but I'm willing to be surprised.

 

[atmasphere]

No, I did not. You continually refer to the rise in distortion with frequency in feedback amps, which almost invariably happens well above 1kHz. Your specific statement was "I remember my SWTP Tigers to be pretty painful so I relegated them to driving woofers. They were supposed to be 0.01%. But at 9KHz I suspect their distortion was much higher."

You also keep talking about masking as the reason that upper harmonics can't be heard with tube amps. So my proposed experiment completely removes all masking which, if your hypothesis is correct, would be audible to you. So let's see if your hypothesis can withstand a basic test. I suspect you'll get a null result, but I'm willing to be surprised.

I feel like my words are being twisted a bit. I've have said that 1KHz is a common turnover frequency; in older amps it can happen at a lower frequency and 1KHz is still a common turnover frequency today. IOW not 'invariably'. Do you need examples? There are plenty of them on this site.

WRT the Tiger, because its distortion rose significantly above 500Hz, at 9Kz harmonics of lower frequencies manifested as brightness and harshness since the lower orders were insufficient to mask them at that frequency. In effect its like cranking up the treble at that frequency but of course when the amp was on the bench there was no sensitivity to load at that frequency nor any change in FR. This was an early indication to me that something else was afoot.

You mentioned

"the ear" (to use your synecdoche) is not terribly sensitive to harmonics, period"

which is false. Were it true, how would we tell the difference between a trumpet and a clarinet? if you view their waveforms you'll see the harmonic spectrum of the two instruments is what's different.

You can also demonstrate the ear's sensitivity to higher ordered harmonics to yourself with simple test equipment I'm sure you have on hand. If you need the procedure let me know.

Tube amps in general have more higher ordered harmonics than solid state which are masked by their H2 and H3 which tend to be more prodigious than most solid state amps; hence the higher orders are masked. So I would be very interested in a different explanation for their sound character, which in most cases anyone can hear. How do you propose to remove the masking principle, a function of the human ear? How would that be different from masking created by a 3rd harmonic in loudspeakers?

 

[SIY]

Were it true, how would we tell the difference between a trumpet and a clarinet?

The overtones are several orders of magnitude higher than the distortion of an amplifier, so the comparison is inapt.

Again, I have suggested an experiment to test your "masking" hypothesis. Just put in the higher harmonic at a level seen in solid state amps you consider "harsh" or "bright." For example, in my measurements of an Adcom 555, the 9th harmonic of 1kHz was at under -100dB. So we can round up.

No masking whatsoever, so if your hypothesis is correct, you should be able to demonstrate audibility. Files and ABX logs.

 

[atmasphere]

The overtones are several orders of magnitude higher than the distortion of an amplifier, so the comparison is inapt.

Again, I have suggested an experiment to test your "masking" hypothesis. Just put in the higher harmonic at a level seen in solid state amps you consider "harsh" or "bright." For example, in my measurements of an Adcom 555, the 9th harmonic of 1kHz was at under -100dB. So we can round up.

No masking whatsoever, so if your hypothesis is correct, you should be able to demonstrate audibility. Files and ABX logs.

I think you're missing something. The harmonics in question are higher order. If you look at a clarinet waveform you'll see the spread of harmonics. Now add a bunch of higher orders to that. You are saying they are inaudible because they are tiny; I'm saying 'not quite'- that's the crux of this. I think it might be important for you to understand how the ear senses sound pressure. It does so by sensing higher ordered harmonics. For this reason, most acoustic instruments do not have a lot of higher orders, like the trumpet at that same link.

I concede that hearing a single harmonic of a simple sine wave is a much harder task for the ear but we're talking about much more complex waveforms so the spectrum of higher ordered harmonics has to be considered.

The ear has to be keenly sensitive to higher orders because it has such a wide range; around 120dB. Add to that Fletcher Munson. So in the area audiophiles refer to as brightness (~3KHz and up) is also a region where the ear is the most sensitive combined with the fact that the ear has to be far more sensitive to higher ordered harmonics than lower orders, as a matter of survival (if you can't tell how loud sounds are, you could be dead- the fact that you are here says your ancestors shared this aspect of human hearing with you).

Here is a demo that shows how the ear reacts to higher orders. Run a low distortion sine generator into an amp and speaker, use a VU meter to set to 0VU. This need not be a loud demo. Then cover the meter, turn down the volume, set to square wave, turn up the volume until it sounds as loud as before. Uncover the meter. The purer your sine wave is, the better the result to demonstrate that the ear is using higher ordered harmonics to sense sound pressure. IMO/IME its lot more sensitive in this regard than most people give it credit.