Testing audiophile claims and myths

[castleofargh]

My vague memories are saying: power supply, whatever capacitive load for the "nominal" operation, and feedback loop for what remains.
Shouldn't people pick the op amp and develop around it, instead of randomly swapping the op amp in a fixed design?

 

[2leftears]

My vague memories are saying: power supply, whatever capacitive load for the "nominal" operation, and feedback loop for what remains.
Shouldn't people pick the op amp and develop around it, instead of randomly swapping the op amp in a fixed design?

In theory yes. In practice many modern opamps are "ideal enough" in their behaviour that the circuit design can be done based on a idealised opamp model. Many circuit designs hence allow for the substitution of a range of different opamp models.

But there are still some minor differences in characteristics to be considered, e.g. whether the opamp is unity-gain stable or needs compensation for different gain configurations. Input impedance and noise characteristics may matter for some circuits, and of course not all opamps have the same allowed supply voltage or permitted output power.

What matters in context of this thread, is that many low-power opamps are "ideal enough" that in the audio band (and quite a bit beyond it) the characteristics of the circuit are pretty much entirely determined by the circuit design around the opamp rather than the opamp itself, and switching a standard opamp for an 'audiophile' one will make no difference in the sound, let alone be measurable in some cases.

Perception is of course a different matter.

 

[PhonoPhi]

1. It is not the opamp in itself but the total amplifier design that determines the sound.
2. Audibly transparent (audibly perfect) amplifiers are easy and cheap to make for decades.
3. Audibly better than audibly transparent is not possible.
4. If changing the opamp audibly changes the sound of an originally audibly transparent amplifier then it is an objective downgrade. Why on earth would someone want that?
5. It is of the utmost importance to understand that human hearing perception is a construction by the brain depending on many things besides only the actual sound going into the ears and that because of that it is possible to "hear" (in fact perceive) (even night and day) differences even when there is actually no audible difference in sound.

Only people who are not aware of the "audibly transparent" concept and the fact that it is easily achievable in certain parts of an audio system (amps, dacs, cables, most electronics), and don't know that many reported audible differences are in fact non-existent but actually differences in sound perception that can be caused by many different things, could think there could be any gain in things like changing opamps. And of course it may feel 'flamboyant and arrogant' to such people when they are told the actual facts for the first time and they don't fully grasp them.

Using your great arguments, let's consider whiskey/whisky (W) vs. pure alcohol solution in water of the right concentration for efficient smooth consumption, aka purified vodka (PV).
In the spirit of this thread, we can call it bigshot's take on the Universe (all name similarities are purely coincidental; no creatures are subjected to any stress or, God forbid, A/B tests during these gedanken experiments).

1. It is just the total alcohol (ethanol) content that matters, not any additives or the colour of the bottle or the fancy labels.

2. The effect of the beverage is based entirely on its alcohol content, and it can't be possibly better than this in any way.

3. Nothing can be added to alcohol to make it any better for controlled intoxication.

4. If adding something to PV change some properties, why on Earth anyone wants to ever do it (or even think of it). Smoked moss by-products and feeble selection of single malts of W are just useless and futile marketing attempts to improve on PV.

5. Intoxication is purely alcohol-dependent, and let's keep it this way - pure and simple!

P. S. I can add that I never could understand those smoked peat preferences.... Yet, I am not going to Liquor stores to pedal my views to whiskey buyers, yet alone based on solid scientific views of intoxication...

P. P. S. Keeping cognac out of (exempt of) these considerations given the Castle moderator prowess/influence

P. P..P. S. No discussions, sorry, being there here - no more.

 

[gregorio]

Using your great arguments, let's consider whiskey/whisky (W) vs. pure alcohol solution in water of the right concentration for efficient smooth consumption, aka purified vodka (PV). …

I don’t understand your post. If it’s supposed to demonstrate something through analogy, it’s not analogous and therefore doesn’t demonstrate whatever it’s supposed to. If it’s not supposed to be analogous then its purpose also eludes me. Either way, I don’t get what you’re trying to say or demonstrate?

G

 

[Androxylo]

If the feedback loop in opamp is necessary, than a slew rate becomes critical because the feedback may come too late when the signal has to change fast. Extracting from AI answer:
"
For audio signals, the slew rate matters most when you’ve got high-frequency components with significant amplitude. Take a 20 kHz tone (the upper limit of human hearing) with a 5 V peak amplitude—pretty reasonable for line-level audio circuits. The required slew rate is:

2π×20×10 ^3 ×5≈0.628V/μs

An old-school op-amp like the LM741, with a slew rate of 0.5 V/μs, falls short here. The output won’t track the input accurately; instead of a clean sine wave, you’ll get a distorted shape—think of a triangle wave creeping in where crisp highs should be. This softens transients (like drum hits) and smears harmonic detail, making the sound dull or harsh."
However most opamps are fast enough

 

[2leftears]

If the feedback loop in opamp is necessary, than a slew rate becomes critical because the feedback may come too late when the signal has to change fast. Extracting from AI answer:
"
For audio signals, the slew rate matters most when you’ve got high-frequency components with significant amplitude. Take a 20 kHz tone (the upper limit of human hearing) with a 5 V peak amplitude—pretty reasonable for line-level audio circuits. The required slew rate is:

2π×20×10 ^3 ×5≈0.628V/μs

An old-school op-amp like the LM741, with a slew rate of 0.5 V/μs, falls short here. The output won’t track the input accurately; instead of a clean sine wave, you’ll get a distorted shape—think of a triangle wave creeping in where crisp highs should be. This softens transients (like drum hits) and smears harmonic detail, making the sound dull or harsh."
However most opamps are fast enough

"The output won’t track the input accurately; instead of a clean sine wave, you’ll get a distorted shape—think of a triangle wave creeping in where crisp highs should be. This softens transients (like drum hits) and smears harmonic detail, making the sound dull or harsh." 10/10 for AI regurgitating meaningless audiophile drivel.

A 20 kHz tone with 5V peak amplitude has an energy content far exceeding that which you would expect a line-level audio signal to contain. 5V peak amplitude can be expected for the lower bass frequencies, but not for the high frequencies. The higher the frequency, the lower the amplitude will be for normal audio; 5V amplitude 20kHz at line level will blow up your tweeters if not severely damage your hearing (assuming a fairly nominal amplification chain and volume setting).

If you needed a clean 20 kHz 5V peak sine wave without any ultrasonic harmonic distortion components (which you cannot hear anyway), then you would indeed need to get a fast opamp, or supplement a slower opamp with an additional amplification stage.

So essentially, you have found a borderline theoretical situation that does not occur for normal audio electronics, in which one of the most pedestrian archaic opamps out there doesn't quite cut the mustard.

 

[VNandor]

5V is excessive even for low frequencies with most headphones. At least the AI is correct about the maximum slew rate of a 5V 20kHz sine if nothing else.

 

[XTF1]

An old-school op-amp like the LM741

Definitely old (1968, I believe…), and mostly known to teach when opamp shortcomings become an issue in electronic circuits. “Old + school” is a good qualifier.

But sure, if you dig deep enough, you’re going to find opamps that, if misapplied, do make a difference…

 

[gregorio]

An old-school op-amp like the LM741, with a slew rate of 0.5 V/μs, falls short here. The output won’t track the input accurately; instead of a clean sine wave, you’ll get a distorted shape—think of a triangle wave creeping in where crisp highs should be. This softens transients (like drum hits) and smears harmonic detail, making the sound dull or harsh."

“Instead of a clean sine wave, you’ll get” pretty much no “shape”, it won’t just “softens transients and smears harmonic detail …” it will pretty much remove them entirely, because 5V at 20kHz with headphones when playing music would pretty much guarantee at least a huge TTS, if not blowing out your ear drums. At which point, any sane person will have more to worry about than half a century old opamp specs, slew rates or “crisp highs”!

You started with “How to finish this discussion once and forever with hard facts, not just listening?”. The hard facts are: This is not the 1960’s and I presume you still have ear drums! As is so often the case, it seems some audiophile marketing guy is going through ancient magazines, correspondence and discussions in the scientific/engineering audio world looking for issues that may have existed many decades ago but were solved three or more decades ago, in order to “lie by omission” (not mention that the issues were solved) and market their world leading design engineers and cutting edge products that can solve those issues. Of course, world leading audiophile design engineers, cutting edge technology and audiophile grade components are never cheap, so expect to see an extra couple of zeroes on the price compared to cheap mass produced units (that solved the issue for peanuts 20-50 years ago)!

G

 

[71 dB]

People can learn (with or without the help of AI) to calculate things, but it is another thing to know how to interpret the numerical results.

Yes, an old opamp may struggle with 20 kHz tones at 5 V, but dropping the signal level to 4 V solves the problem. That's only about 2 dB drop in level. That alone renders this problem almost insignificant, but there is so much more to this than that, the things addresses by others.

This is a total non-problem which would be easy to fix (by using more suitable opamps with bigger slew rates) if it was a problem. Audiophile community creates these non-problems to confuse people...

 

[2leftears]

People can learn (with or without the help of AI) to calculate things, but it is another thing to know how to interpret the numerical results.

Learning with the help of AI (or just asking AI) will probably become the norm for these kind of questions.

There are plenty of good academic books on this subject matter, but they tend to be a bit more expensive. On the other hand, one of those books can give someone a much better and more in-depth understanding of audio/amp design, within a much shorter period of time.

 

[castleofargh]

I'm using Gemini all the time when I need nothing. (it sucks so bad at replacing Google Assistant, I've lost hand free functionality because I must unlock the phone to record or access anything, uselesssssssss. IDK how I'm going to do because the only time I need it is when I can't put what I'm doing down to grab the phone, it's a real issue right now and google plan to completely phase out the Assistant this year... halp!!!!! I need a fully hand free memo, chrono, and ideally calculator sometimes when I'm outside).
But to find a song, discuss fusion reactors, and to make me laugh by telling me jokes that make no sense. It's amazing. Probably a perk of the French version, but every 3 or so jokes are nonsense and IDK if it's a literal translation from jokes in another language that failed to carrying a play on word, or some deeper hallucination? What I know is that her failures sometimes make me laugh more than the best stand-up comedian right now. Priceless.
But yes, the directive to answer can bias the replies quite a lot(same with gpt). Add to that how it's referring to the internet with most audiophile data coming from overconfident mostly unqualified audiophiles, and talking to AI about audio seems like a recipe for failure. More so with loaded question(same issue as when using google or inquiring about anything in general, there's a reason why the scientific method tells us to look for why our ideas are wrong instead of validation. Seeking validation almost always returns validation, no matter how nonsensical an idea can be.

I do use https://scispace.com/ on occasion as it relies on papers instead of audiophile forums, and it gives me the name or even a link to the various papers it used to make its answer, so I then just go read some of the papers that seem to concern my initial inquiry. It's not what it's for, but I use it like a Google for nerds from time to time, with some success(and sometimes not, just like Google). But for codes, to find who's got cancer, and to deal with astronomical amounts of data, LLMs are cool.

 

[Androxylo]

Oh, finally I found someone measuring opamps the way to expose their measured differences! And before anyone complains, yes, this person did not hear any audible differences, only measured ones - which is fine, because if the power amp or speakers are not resolving enough there should be no differences. Also, the discrete opamp tested against IC NE5532 is some $18 no-name from China, it's not Sparkos or Burson. So, the measurement difference with such a small step up is impressive:


So, what's different? The NE5532 square wave at 10 kHz:

and 20 kHz:

Now Chinese noname opamp at 10 kHz:

20 kHz:

and even 50 kHz:

and mind you, a square wave is still a very simple waveform to capture. This is a single grand piano key struck spectrogram:

The top chart is a strong key hit, the bottom is the same key pressed softly. The difference between those spectrograms is the whole idea of the music - when a pianist wants to express a violent fortissimo or a soft mood ambient melody. If you volume match those sounds because they obviously have different volume, the difference of sound to a human ear and brain are not just different, they are completely different. We are capable of capturing all transient variants between those two extremes.

Now compare this waveform to a 1 kHz square wave.

Ignore the noise, just the main harmonics. There are 12 of them. A single grand piano key has 60. A continuous grand piano music will have like 300 or even 500 of them. So if the NE5532 was unable to reproduce a square wave a grand piano will be much worse.

 

[gregorio]

and mind you, a square wave is still a very simple waveform to capture.

The exact opposite of the truth, a square wave is an impossible waveform to capture, it’s also impossible to reproduce.

This is a single grand piano key struck spectrogram:

That is not a spectrogram, a spectrogram indicates frequency, amplitude and time. What you’ve posted only indicates frequency and amplitude. It’s also rather bizarre as neither represents the harmonic series of a single note, maybe the note was played with the sustain pedal engaged and they mic’ed it in such a way to pickup the sympathetic ringing of the other strings.

The difference between those spectrograms is the whole idea of the music - when a pianist wants to express a violent fortissimo or a soft mood ambient melody.

The difference between those graphs has little to do with music at all. Again, they are not spectrograms and do not indicate time, virtually “the whole idea of music” is based on one or a combination of melody, rhythm and harmony but without any time none of those are possible. So again, complete BS!

If you volume match those sounds because they obviously have different volume, the difference of sound to a human ear and brain are not just different, they are completely different.

That too is complete nonsense. If the sound were “completely different” then you would be incapable of recognising that a piano played fortissimo is still a piano.

Ignore the noise, just the main harmonics. There are 12 of them. A single grand piano key has 60. A continuous grand piano music will have like 300 or even 500 of them.

The 1kHz square wave has 12 harmonics because all rest have been filtered out! A square wave has an infinite number of harmonics while a single grand piano key will typically have about 4-5 or up to 8 or so if you really bash it, although you might get 60 freqs with the sustain pedal down from all the other strings “sympathetic” ringing. “A continuous grand piano music” would commonly have no more than 10-20 or so harmonics at the same time as many of notes will be harmonically related and therefore have overlapping harmonics. A very loud tone cluster with the sustain pedal down might give you as many as 100 harmonics/freqs but even if it were 500, thats still a great deal less than the infinite number of harmonics required for a square wave.

So if the NE5532 was unable to reproduce a square wave a grand piano will be much worse.

The NE5532 is unable to reproduce a square wave, your own oscilloscope pictures demonstrate only an approximation of a square wave because a real square wave is impossible to capture or reproduce and a grand piano is literally INFINITELY easier to capture and reproduce. While a grand piano is challenging to mic and mix, from a digital audio point of view it’s trivially easy. It has a relatively limited dynamic range and a quite limited frequency range, it’s entirely possible to have a good grand piano recording with little or nothing above about 12kHz.

Again, pretty much every assertion you made is completely false! Posting false assertions to a science discussion forum is rude/insulting, when are you going to stop being so rude and ignorant?

G

 

[VNandor]

So, what's different? The NE5532 square wave at 10 kHz:

If you heard about the fourier series, you wou would know that the difference between a 10kHz sine wave and a 10kHz square wave is the harmonics of the square wave.

A 10kHz square wave will sound the same as a 10kHz sine wave because the first non-zero harmonic is at 30kHz for the square series. It doesn't really matter what the opamp does as long as the fundamental of the square wave is amplified correctly, although a gentle rolloff above 20kHz is what's probably the best for most cases so you don't accidentally fry components down the line with ultrasonic noise.

The difference between the two scope displays are the harmonic content way above 20kHz which is why it does not matter for listening. It's super easy to measure differences between opamps at or above 100kHz or 1MHz, it doesn't mean that their audio performance are not the same.