General debate thread about audio measurements

[g29]

Don't misunderstand, one amp is clearly more noisy than the other. However the horns high sensitivity creates the problem. The noise is inaudible with normal speakers and not a problem.

So yes circumstance can create issues, but you can't say that particular noisier amp is less revealing when used in appropriate circumstances.

So would you agree what you call "high sensitivity" can also be considered "more revealing", like putting on the readers and actually being able to see what I am typing, or using TSA cameras to reveal ??? Just because I am accustomed to my 80 column CRT doesn't mean a 4K display caused a problem or isn't more revealing.

 

[LTig]

So would you agree what you call "high sensitivity" can also be considered "more revealing", like putting on the readers and actually being able to see what I am typing ???

Regarding noise yes, the high sensitivity of a horn speaker reveals noise a speaker with lower sensitivity may hide. You can put your hands around your ears (creating a horn) and listen to a speaker with lower sensitivity, this may also reveal noise of the power amp.

In theory a horn speaker also has lower distortion (distortion increases with the amplitude of the movement of the membrane which is lower in a horn) and may be able to reveal distortion from the feeding chain. However in reality almost all horn speakers are plagued by resonances which may again hide other short comings.

 

[amirm]

See that crystal at the top right? That's the master clock. The "turntable motor servo-amp" at the bottom right adjusts the rotational speed of the disc to provide the correct data rate. A FIFO buffer somewhere in the chain absorbs the inevitable variations in the raw signal rate.

That's not my read of it. The clock is needed to regulate the speed of the drive. Here is the picture again:

See the clock regen is prior to the sync, detection, and time with said crystal. The only input to the clock regeneration is the output of the optical drive, not the other way around.

In other words, they use a crystal controlled clock to regulate the motor speed and then assume post some filtering, what comes out of the disc must be the correct clock.

To be sure, the clock regen could use a PLL and reduce the optical jitter but the drive is still the master.

It was not until way later that manufacturers built asynchronous players. These required data drives which were more expensive.

 

[Blumlein 88]

So would you agree what you call "high sensitivity" can also be considered "more revealing", like putting on the readers and actually being able to see what I am typing, or using TSA cameras to reveal ??? Just because I am accustomed to my 80 column CRT doesn't mean a 4K display caused a problem or isn't more revealing.

What is happening is probably poor gain staging.

For instance, an amp might have an SNR of 100 db with 30 db of gain. Another might have an SNR of 100 db with 20 db of gain. You connect your DAC to each. Your DAC has a fixed noise floor. Let us say it is also - 100 db from full signal level. With the first amps the noise coming from your speakers is 10 db louder than the noise from the second one. Because the 1st amp is amplifying the existing noise floor more than the second one. With one you hear noise with the other you didn't. And that is assuming both amps have equal Signal to noise ratios. If one is higher gain and lower SNR, the problem would be even worse. Some variation of that was likely happening. So all that really means is not that your speakers are more revealing of anything other than noise levels due to their high sensitivity.

 

[andreasmaaan]

For example, Nelson Pass discovered a few years ago that the addition of third harmonic increases the sense of depth, but only if it's added out of phase. Now who the hell would have guess that? And even if you knew, how would you infer it from a measurement?

This is really interesting. Is this documented? If it's true then that technique should be used in the studio instead of the playback chain.

I read this at the time and was curious as to whether there was really something to it. Although my recollection was that his claim related to the second harmonic, not the third??

Anyway, I digitally added inverted second harmonic distortion to music to see if it sounded different/better than in-phase second harmonic and I wasn't sure I could hear any difference. Both just sounded like distortion, and needed to be quite high in level to be audible. I'll try to work out again how I did it and upload some files if I can...

 

[March Audio]

So would you agree what you call "high sensitivity" can also be considered "more revealing", like putting on the readers and actually being able to see what I am typing, or using TSA cameras to reveal ??? Just because I am accustomed to my 80 column CRT doesn't mean a 4K display caused a problem or isn't more revealing.

No, its just higher sensitivity . As @Blumlein 88 alludes to above the gain staging needs to be appropriate for the chosen equipment. The amps could still have the same SNR and same distortion levels. Equally it could be worse or actually better! Its just that hearing noise with those horns is not the correct way to judge the overall transparency or how revealing an amp is.

High sensitivity speakers simply need very quiet amps.

 

[RayDunzl]

Nelson Pass discovered a few years ago that the addition of third harmonic increases the sense of depth, but only if it's added out of phase.

Anyway, I digitally added inverted second harmonic distortion to music to see if it sounded different/better than in-phase second harmonic and I wasn't sure I could hear any difference.

Experiment with harmonic phase in-room:


If I create a high level 2nd harmonic at 0 degrees, it's phase angle relative to the fundamental is all over the place at the listening position as the frequency of the fundamental (and the harmonic) changes.

Speakers: JBL LSR 308

Fundamental frequency and phase reported of the 2nd harmonic generated at 0 degrees for a few piano fundamental frequencies at the listening position.

440 .....111
466.16 ..... -106
493.8 ..... 127
523.25 ..... -175
554.4 ..... -52

and so on

Measuring the 3rd harmonic (no second added) generated by the speaker itself, yields a similar range of phase angles.

As "proof" that the problem is at least partly rooted in room reflection, I moved my torso by leaning back in my chair at the desk, and the reported phase angle changed.

 

[josh358]

Experiment with harmonic phase in-room:


If I create a high level 2nd harmonic at 0 degrees, it's phase angle relative to the fundamental is all over the place at the listening position as the frequency of the fundamental (and the harmonic) changes.

Speakers: JBL LSR 308

Fundamental frequency and phase reported of the 2nd harmonic generated at 0 degrees for a few piano fundamental frequencies at the listening position.

440 .....111
466.16 ..... -106
493.8 ..... 127
523.25 ..... -175
554.4 ..... -52

and so on

Measuring the 3rd harmonic (no second added) generated by the speaker itself, yields a similar range of phase angles.

As "proof" that the problem is at least partly rooted in room reflection, I moved my torso by leaning back in my chair at the desk, and the reported phase angle changed.

For those who are curious, here's Nelson Pass's account of his experiment.

"So why is the phase important? Well, it's a subtle thing. I don't suppose everyone can hear it, and fewer particularly care, but from listening tests we learn that there is a tendency to interpret negative phase 2nd as giving a deeper soundstage and improved localization than otherwise. Positive phase seems to put the instruments and vocals closer and a little more in-your-face with enhanced detail.

"Your results may vary, but when I first explored this with the SIT-1 amplifier at First Watt, I had a knob on the front of the amplifier which varied the amount and phase of the 2nd harmonic. It was easy enough to lend the amplifiers to listeners who didn't know what the knob did and gather their comments. Roughly speaking, they tended to prefer about 1% negative phase 2nd harmonic, so it became my standard setting for that knob."

https://positive-feedback.com/audio-discourse/the-pass-h2-harmonic-generator/

 

[andreasmaaan]

Experiment with harmonic phase in-room:


If I create a high level 2nd harmonic at 0 degrees, it's phase angle relative to the fundamental is all over the place at the listening position as the frequency of the fundamental (and the harmonic) changes.

Speakers: JBL LSR 308

Fundamental frequency and phase reported of the 2nd harmonic generated at 0 degrees for a few piano fundamental frequencies at the listening position.

440 .....111
466.16 ..... -106
493.8 ..... 127
523.25 ..... -175
554.4 ..... -52

and so on

Measuring the 3rd harmonic (no second added) generated by the speaker itself, yields a similar range of phase angles.

As "proof" that the problem is at least partly rooted in room reflection, I moved my torso by leaning back in my chair at the desk, and the reported phase angle changed.

That makes sense, yeh. What I understood of Nelson Pass' claim was that the harmonic should be inverted relative to itself, not the fundamental (as you point out it can't be, because the relative phase of the two is always changing).

In other words, generate harmonics and then invert their phase. Are we on the same page here? I'm not even 100% sure I've interpreted Pass correctly lol.

 

[Cosmik]

It was not until way later that manufacturers built asynchronous players. These required data drives which were more expensive.

The claim is not that the original CD players were asynchronous; they were 'loosely synchronous'. They used feedback (based on FIFO fill level) to keep the *average* rate of the motor at the correct speed and the FIFO to absorb the short term jitter, clocking the data out of the FIFO using the same crystal clock that provided the sample rate of the DAC. This from a 1985 SAA7210 data sheet:

In other words, a conceptually perfect system, completely immune to transport 'quality' as long as it gets the data into the FIFO before it's needed.

 

[Blumlein 88]

So the question is why did early and not so early CD players have rather variable J-test results? It is possible I still have the results on an old hard drive, but I measured some very early players and a couple DVD and a couple bluray players at one time. I still have the ADC used to measure and with current equipment know it had low jitter of its own as an ADC with a free running clock. I still have the goofy Pioneer DVD player (assuming it still works, did two years ago). Goofy because if you change tracks jitter goes sky high dropping to rather low levels over something like 12 seconds.

 

[March Audio]

The claim is not that the original CD players were asynchronous; they were 'loosely synchronous'. They used feedback (based on FIFO fill level) to keep the *average* rate of the motor at the correct speed and the FIFO to absorb the short term jitter, clocking the data out of the FIFO using the same crystal clock that provided the sample rate of the DAC. This from a 1985 SAA7210 data sheet:
View attachment 26741

In other words, a conceptually perfect system, completely immune to transport 'quality' as long as it gets the data into the FIFO before it's needed.

This was always my understanding and why two box players (dac and transport spdif linked) weren't always the good solution that was marketed. They lost this inherent dac clock stability.

 

[g29]

What is happening is probably poor gain staging.

For instance, an amp might have an SNR of 100 db with 30 db of gain. Another might have an SNR of 100 db with 20 db of gain. You connect your DAC to each. Your DAC has a fixed noise floor. Let us say it is also - 100 db from full signal level. With the first amps the noise coming from your speakers is 10 db louder than the noise from the second one. Because the 1st amp is amplifying the existing noise floor more than the second one. With one you hear noise with the other you didn't. And that is assuming both amps have equal Signal to noise ratios. If one is higher gain and lower SNR, the problem would be even worse. Some variation of that was likely happening. So all that really means is not that your speakers are more revealing of anything other than noise levels due to their high sensitivity.

Thanks for the insights. I tried to look up the specs for these 2 old units but could only find gain specs for 1.

- 70 watts continuous Power Output (20Hz - 20kHz, 8 ohms, 0.03% THD)
- Signal-to-Noise Ratio - 97dB
- Damping Factor .....(50 Hz, 8 Ohm) 60

- 60 watts continuous average power into 8 ohms at any frequency between 20Hz and 20kHz with both channels driven at less than 0.06% THD.
- Signal-to-Noise Ratio, A-weighted - 60 watts into 8 ohms.....< 100dB
- Damping Factor .....20Hz to 20kHz >400
- Gain .....27dB

Horns were 8 Ohms.

 

[Cosmik]

So the question is why did early and not so early CD players have rather variable J-test results? It is possible I still have the results on an old hard drive, but I measured some very early players and a couple DVD and a couple bluray players at one time. I still have the ADC used to measure and with current equipment know it had low jitter of its own as an ADC with a free running clock. I still have the goofy Pioneer DVD player (assuming it still works, did two years ago). Goofy because if you change tracks jitter goes sky high dropping to rather low levels over something like 12 seconds.

I don't know that this patent is exactly relevant, but clearly the issue of rapid track selection has been an issue:

...in case the rough actuator is driven at a high speed for a fast access, the focus lens held by the fine actuator is deflected due to the acceleration caused at the start or halt of the rough actuator or by the vibration induced during the travel thereof, so that the fine actuator is kept in vibration for a while even after the halt of the pickup-base in the vicinity of the target, thereby raising a problem that the fine actuator cannot be brought into servo control with stability and rapidity.

https://patents.google.com/patent/US5117410

Maybe my claim that the system was immune to transport quality might not be absolutely true for systems that allow increased errors following a track change in order to speed up track access...?

It could also be as simple as power supply instability affecting the clock oscillator..?

It's difficult to find actual evidence that the basic system has always been the same and (conceptually) perfect hence the grubbing around in data sheets and patents. But the book "Understanding and Servicing CD Players" by Ken Clements includes the following extracts:

Putting those together, I think, makes the case that audio players have always worked the same basic way.

I am keen to quash the idea that CD players are, or have been, like a glorified record deck: an electronically regulated motor speed (using a crystal don't you know) with subsequent electronics that adapt to any instability in that speed. This is, and would always have been, the wrong way round. The motor is ultimately regulated by a crystal, but not for perfect short term stability; merely adequate stability to maintain the correct average rate as demanded by the central crystal oscillator.

P.S. Apologies for that green colour - a random selection and, I now realise, rather unpleasant

 

[Cosmik]

Re. CD motor control.

There might be a better way of putting the last bit of my previous post:

The motor isn't regulated against a crystal-controlled reference frequency; it is regulated against a crystal-controlled reference count.

The former allows for cumulative blips, errors and offsets over time, and the myth (I claim it is anyway) is that the subsequent electronics smooths out the speed variations.

The latter doesn't allow for any such errors over time *and* makes the system immune to transport quality. It uses almost the same basic hardware, but with a slight re-arrangement achieves a fundamentally better result - which is what elegant engineering is all about .

 

[josh358]

Do you know at which levels of SPL of ultrasonic sounds the mix products within the audible frequency range become audible?

Based on what I know of directional ultrasonic loudspeakers, it would likely be fairly high, because as I understand it in that application the demodulated envelope has on the order of 60 dB of attenuation. But that's only a rough guess since I don't know of any studies that apply to near audio frequencies, e.g., 22.05 kHz. I'm guessing that distortion products in the tweeter would be more significant here than the nonlinearity of the air, but I'm not aware of any studies. It's an interesting topic for research.

 

[mansr]

It's difficult to find actual evidence that the basic system has always been the same and (conceptually) perfect hence the grubbing around in data sheets and patents.

The schematics for the earliest CD players are readily available and show that they always worked fundamentally the same way.

The clock recovery circuit on the signal from the pickup is needed since the instantaneous rate can deviate considerably from the fixed crystal rate. To decode the bit values from the raw HF signal, a closely tracking clock, separate from the crystal reference, is required. Once the bits have been recovered, they can be buffered and reclocked, decoupling the timing from the rotation of the disc.

 

[March Audio]

Do you know at which levels of SPL of ultrasonic sounds the mix products within the audible frequency range become audible?

This is another total red herring. Any potential ultrasonic filter ringing is at very low level. You have to remember that music content in the transition band might be -70 dB down. Its low pass anti alias filtered. There is very little content to make a reconstruction filter ring at nyquist. Any ringing is therefore much further down in level and above your hearing range.

 

[Cosmik]

The schematics for the earliest CD players are readily available and show that they always worked fundamentally the same way.

It's just that it's hard to convince someone with a schematic. It's nice to see the actual words written down as 'a concept' - something that could possibly disappear into the mists of time unless written down somewhere by someone.

 

[andreasmaaan]

That would be true in a perfectly linear system. Neither amplifiers nor loudspeakers nor ears nor the air are perfectly linear and it's well known that the nonlinearity of air alone can make ultrasonic sounds audible.

Do you know at which levels of SPL of ultrasonic sounds the mix products within the audible frequency range become audible?

This very question was discussed previously here.