Sound Quality Vs. Measurements

[wahab]

Out of curiosity i made some listening tests and surprinsingly
when using a 1KHZ sine THD contents are clearly audibles at levels
higher than -60db , wich is 0.1%.

A screen of the THD distribution , quite tubish for a class AB
solid state...

[ThorstenL]

Hi,

Originally Posted by wahab

Out of curiosity i made some listening tests and surprinsingly when using a 1KHZ sine THD contents are clearly audibles at levels
higher than -60db , wich is 0.1%.

At what SPL was the test made and what is HD level (per harmonic - not THD) of the transducers used at that level.

Ciao T

[Wavebourn]

Originally Posted by wahab

A screen of the THD distribution , quite tubish for a class AB
solid state...

Do you mean "tubish" on onset of full power?

Tubish does not mean that. Tubish mean that when you decrease power higher order harmonics disappear, and at 1/10 of power only 2'nd and 3'rd are visible, and even they go down when you further decrease power.

[bcarso]

I was just reading the historic series of articles by Jung et al. and subsequent letters exchanges from TAA about slew-induced distortion. Quite fascinating from both a technical and sociological point of view. So far I haven't found anyone mentioning that the reason the two-close-spaced-tone IM distortion test is poorly correlated with other measures of slew limiting effects is that, if the slew rates are symmetrical, the distortion is odd-order --- and thus the IM products, for example for a cubic nonlinearity, are at 2f1 - f2 and 2f2 - f1, i.e. on either side of the two stimulus frequencies.

When the slew rates are asymmetrical, then you will indeed get some even order including the simple difference frequency f2 - f1. But if that's the only thing you are measuring, trying to diagnose slewing problems, you will understandably be getting some poor correlations with, among other things, high frequency THD.

But surely someone discusses this sooner or later --- I just haven't seen it yet in TAA. But I have plenty of reading left to do.

[wahab]

Originally Posted by ThorstenL

Hi,
At what SPL was the test made and what is HD level (per harmonic - not THD) of the transducers used at that level.

Ciao T

Level is 12V Pk on a 4.7R + (0.5R//8R) load , the speaker is the 8R load,
this allow to reduce the SPL level while preserving about 16 damping factor.

The speakers are studio monitors Quested H108 , wich in this case were
driven with 1.2V Pk , about 100mW.

Their efficency is 90.5 db/1W/1m.

On the screen , the fundamental is at -5.83dB , very close to the -6dB line ,
while the line graduations are spaced with 6dB , wich imply that H2 is 46dB below the fundamental.

Originally Posted by Wavebourn

Do you mean "tubish" on onset of full power?

Tubish does not mean that. Tubish mean that when you decrease power higher order harmonics disappear, and at 1/10 of power only 2'nd and 3'rd are visible, and even they go down when you further decrease power.

The amp is close to clipping and as output level is reduced
high order harmonics decrease more than the fundamental
level variation.

Low order only THD is reached well before 1/10 of Pmax
despite the amp having meager NFB.

[Wavebourn]

Originally Posted by bcarso

I was just reading the historic series of articles by Jung et al. and subsequent letters exchanges from TAA about slew-induced distortion. Quite fascinating from both a technical and sociological point of view. So far I haven't found anyone mentioning that the reason the two-close-spaced-tone IM distortion test is poorly correlated with other measures of slew limiting effects is that, if the slew rates are symmetrical, the distortion is odd-order --- and thus the IM products, for example for a cubic nonlinearity, are at 2f1 - f2 and 2f2 - f1, i.e. on either side of the two stimulus frequencies.

When the slew rates are asymmetrical, then you will indeed get some even order including the simple difference frequency f2 - f1. But if that's the only thing you are measuring, trying to diagnose slewing problems, you will understandably be getting some poor correlations with, among other things, high frequency THD.

But surely someone discusses this sooner or later --- I just haven't seen it yet in TAA. But I have plenty of reading left to do.

Yes. Also, when sidebands are of different highs that means phase intermodulation.

Signal generators and oscilloscope still are nice tools, if to know how to use them properly. Like, a flashlight is good tool when you can guess where you could drop that car keys on dark parking lot.

[Wavebourn]

Originally Posted by wahab

The amp is close to clipping and as output level is reduced
high order harmonics decrease more than the fundamental
level variation.

Low order only THD is reached well before 1/10 of Pmax
despite the amp having meager NFB.

So, on 1/10 of Pmax it is way below 0.1%, and low order.
I doubt it will be audible, even if on peaks it is as pictured.

Thorsten asked precisely right question, about level of loudness. Dynamics matters more than static numbers when they are so small.

[wahab]

Originally Posted by Wavebourn

So, on 1/10 of Pmax it is way below 0.1%, and low order.
I doubt it will be audible, even if on peaks it is as pictured.

Thorsten asked precisely right question, about level of loudness. Dynamics matters more than static numbers when they are so small.

Indeed , with a musical signal , even the screened high THD
would be inaudible given the huge amount of harmonics
produced by instruments.

I did the test at low SPL , agree though that perception
is different when reaching tympani compression....

[ThorstenL]

Hi,

Originally Posted by wahab

Level is 12V Pk on a 4.7R + (0.5R//8R) load , the speaker is the 8R load, this allow to reduce the SPL level while preserving about 16 damping factor.

The speakers are studio monitors Quested H108 , wich in this case were
driven with 1.2V Pk , about 100mW.

Their efficency is 90.5 db/1W/1m.

So the SPL is around 80dB at 1m and probably lower than that at listening distance. The human hearing becomes more sensitive at these fairly low levels.

Speaker HD would arguably be low.

Originally Posted by wahab

On the screen , the fundamental is at -5.83dB , very close to the -6dB line ,
while the line graduations are spaced with 6dB , wich imply that H2 is 46dB below the fundamental.

So it is not 0.1% THD, but over 0.5%, a difference of more than 14dB compared to your claim.

Can 0.5% THD with harmonic structure like you show be heard on an 80dB SPL pure tone? Yes, very likely.

Now can 0.1% almost pure 2nd HD be heard at 108dB SPL with pure tones?

Ciao T

[dvv]

Which slew rate?

Voltage, as V/uS?

Or current, as A/uS?

All too often, and very typical of particularly Japanese gear, while the voltage slew rate can even be outstanding, the current slew rate is something they know nothing about, and if even mentioned, you are expected to hara kiri at once.

Others, such as for example SAE, did quote it as 20 A/uS for their mammoth 500 WPC amp.