Audibility thresholds of amp and DAC measurements

[peng]

Have you seen the prices of some Classe Audio’s new gear?! Ouch!

I’m going to push the envelope to see how many specs I can get. Not that big of a deal because I really can’t hear any distortion, but I would just like to know.

Not recently, but when I was shopping for power amps in that price range, they were priced about the same as the comparable Bryston power amps (another Canadian make brand). I eventually bought Bryston's for the look that I preferred. So I can imagine what their prices would be like today with high inflation and supply chain issues unsolved..

 

[j_j]

I agree it was true in the past, but all newer devices measure THD harmonics simply at multiples of fundamental frequency, as a software action.

That strikes me as allowing inaccuracies. Especially if any time propagation or jitter is involved.

 

[Blumlein 88]

That strikes me as allowing inaccuracies. Especially if any time propagation or jitter is involved.

The software also adjusts for the time propagation for fixed signals so that part is taken care of to prevent issues. I suppose jitter could be an issue if it is large enough, but usually isn't.

 

[RayDunzl]

Opinion:

Audibility of distortion:

With a sine wave or other contrived signal - very low values of distortion may be audible.


With Music...

Musical tones are full of harmonics - equivalent of wild levels of distortion of a sine.

It would take a large dose of added distortion to do more than slightly change the tonality.

A single bass guitar note (no added distortion, just the signal fed to an ADC and REW):

22.7% "distortion" baked in as "timbre".

@pkane has a distortion tool to play with with music

 

[Blumlein 88]

Here is pkane's Distort free software. Lets you add distortion to music in the amounts and types you choose.

Distort audibility tester | Distort documentation

distortaudio.org

 

[j_j]

The software also adjusts for the time propagation for fixed signals so that part is taken care of to prevent issues. I suppose jitter could be an issue if it is large enough, but usually isn't.

Obviously it has to adjust for phase of the fundamental.

For the rest, I really do wonder. I've seen some very strange stuff in new amps, especially some of the lesser switching ones.

 

[restorer-john]

For the rest, I really do wonder. I've seen some very strange stuff in new amps, especially some of the lesser switching ones.

Probably the bandwidth of the old skool THD meters is half the problem. My trusty old Leader is 200kHz. Can't get it near a cheap class D.

 

[j_j]

Probably the bandwidth of the old skool THD meters is half the problem. My trusty old Leader is 200kHz. Can't get it near a cheap class D.

No, I'm using good ADC's and software. And, yes, we do keep our stuff quiet. Back in the good old days, we had our AA501 residual connected to a Nicolet Spectrum analyzer.

One finds some funny things there, once one figures out that one MUST bypass the crap out of the Nicolet power line, and make sure you have a rock-solid ground.

 

[rdenney]

Opinion:

Audibility of distortion:

With a sine wave or other contrived signal - very low values of distortion may be audible.


With Music...

Musical tones are full of harmonics - equivalent of wild levels of distortion of a sine.

It would take a large dose of added distortion to do more than slightly change the tonality.

A single bass guitar note (no added distortion, just the signal fed to an ADC and REW):

22.7% "distortion" baked in as "timbre".

@pkane has a distortion tool to play with with music

Here's the spectrum of a tuba (a 58-Hz low Bb), measured by me at least 25 years ago:

I've heard systems that make a tuba sound like a euphonium, or a horn sound like a trombone. In the former case, a lot of odd-order harmonics de-emphasize the spacing of the harmonics, and emphasize the overtones that would normally be related to the note an octave higher. Yes, it takes a lot of distortion to make this happen, but maybe not as much as we think.

A horn sounding like a trombone can simply be plain old distortion of the usual kind: an emphasis of high harmonics in the mix, particularly as a result of clipping distortion. Trombones have edge in the sound not present in horns (unless played fully stopped, but that's not likely to be mistaken for a trombone). That edge is what? Non-harmonic high frequencies and sibilance. IMD might product something like that if severe. But a spectral tile that depresses frequencies can do it, too. An F horn has the same length tubing as an F tuba (half again longer than at trombone), and though the player is usually buzzing higher frequencies, there will be some excitation and resonance at the fundamental (though the fundamental of a Bb tuba in the FFT above is at 26 Hz, but my microphone probably couldn't pick that up--I think I recorded that on a Sony Minidisc recorder).

But mostly systems make tubas sound like euphoniums because they can't put enough power in the overtones under 200 Hz.

I'm not particularly sensitive to harmonic distortion, but I am particularly sensitive to tone color and spectral tilt.

Rick "thinking it unlikely that this is an amp problem unless overdriven" Denney

 

[j_j]

No, I'm using good ADC's and software. And, yes, we do keep our stuff quiet.

Here's the spectrum of a tuba (a 58-Hz low Bb), measured by me at least 25 years ago:

I'm not particularly sensitive to harmonic distortion, but I am particularly sensitive to tone color and spectral tilt.

Rick "thinking it unlikely that this is an amp problem unless overdriven" Denney

Well, I have to say that while I am sensitive to distortion, pure harmonic isn't a big problem unless it's berserk, because your auditory system expects harmonics. But the same transfer function can create some terrible IMD (if you have harmonic distortion you have by definition also IMD) and that bugs the (O*(*& out of me.

So does timbre change. So you have my sympathy on that front!

 

[dlaloum]

Here's the spectrum of a tuba (a 58-Hz low Bb), measured by me at least 25 years ago:

I've heard systems that make a tuba sound like a euphonium, or a horn sound like a trombone. In the former case, a lot of odd-order harmonics de-emphasize the spacing of the harmonics, and emphasize the overtones that would normally be related to the note an octave higher. Yes, it takes a lot of distortion to make this happen, but maybe not as much as we think.

A horn sounding like a trombone can simply be plain old distortion of the usual kind: an emphasis of high harmonics in the mix, particularly as a result of clipping distortion. Trombones have edge in the sound not present in horns (unless played fully stopped, but that's not likely to be mistaken for a trombone). That edge is what? Non-harmonic high frequencies and sibilance. IMD might product something like that if severe. But a spectral tile that depresses frequencies can do it, too. An F horn has the same length tubing as an F tuba (half again longer than at trombone), and though the player is usually buzzing higher frequencies, there will be some excitation and resonance at the fundamental (though the fundamental of a Bb tuba in the FFT above is at 26 Hz, but my microphone probably couldn't pick that up--I think I recorded that on a Sony Minidisc recorder).

But mostly systems make tubas sound like euphoniums because they can't put enough power in the overtones under 200 Hz.

I'm not particularly sensitive to harmonic distortion, but I am particularly sensitive to tone color and spectral tilt.

Rick "thinking it unlikely that this is an amp problem unless overdriven" Denney

I had a somewhat similar experience, where the big TOTL Martin Logan hybrid Electrostatic (with the big dynamic bass woofer) - made bass singers sound timbrally like baritones - whereas the all stat CLS - which had a lot less bass - the bass singers sounded like bass singers, and were timbrally different from baritones.... This may relate to the much lower distortion on the ESL panel vs the Magnet/Cone Dynamic speaker - don't know for sure - but I do look on the increase LF distortion with some suspicion.... I am pretty sure it is altering timbre / tone color

 

[j_j]

I had a somewhat similar experience, where the big TOTL Martin Logan hybrid Electrostatic (with the big dynamic bass woofer) - made bass singers sound timbrally like baritones - whereas the all stat CLS - which had a lot less bass - the bass singers sounded like bass singers, and were timbrally different from baritones.... This may relate to the much lower distortion on the ESL panel vs the Magnet/Cone Dynamic speaker - don't know for sure - but I do look on the increase LF distortion with some suspicion.... I am pretty sure it is altering timbre / tone color

I would suspect time coherence for this particular issue. I also must quibble with the idea that ESL's are necessarily lower distortion in general, although I am not aware of the ML hybrid's performance.

 

[dlaloum]

I would suspect time coherence for this particular issue. I also must quibble with the idea that ESL's are necessarily lower distortion in general, although I am not aware of the ML hybrid's performance.

Typically 0.5% THD max on stats... most dynamics never even get close to 1% THD (3% is typical - particularly in the bass - for many Dynamics - and that is well above the audible threshold) - the most difficult area is of course the bass - and the measured results pretty much reflect that.

Stat's don't do "loud" bass - but they do very clean, timbrally correct bass

 

[j_j]

Typically 0.5% THD max on stats... most dynamics never even get close to 1% THD (3% is typical - particularly in the bass - for many Dynamics - and that is well above the audible threshold) - the most difficult area is of course the bass - and the measured results pretty much reflect that.

Stat's don't do "loud" bass - but they do very clean, timbrally correct bass

Again, I suspect you've got a time-domain element off also in your listening comparison. As you said, not loud. You need to measure both devices at the same level, yes?

 

[dlaloum]

Again, I suspect you've got a time-domain element off also in your listening comparison. As you said, not loud. You need to measure both devices at the same level, yes?

Both items were never in the same room at the same time - but the experience has been consisten over decades of listening, and hearing those same components in various environments.

An observation is not invalid per se, because it is not in a measurably identical situation ....

It is investigation of such observations, under "laboratory conditions" that can then shed light on why the observation is experienced as it is.

My conclusions are my own hypothesis on the issues.

I am quite certain that the issue is likely to be time domain related, even if it is merely due to the fact that Woofers have mass (and more importantly inertia) several orders of magnitude greater than a sheet of mylar film - a mass that needs to be precisely accelerated and deccelerated to achieve optimal audio.... and mass does not like change! (Inertia) - the physics tend to advantage ultra low mass, more than they do ultra high power. (Lotus vs Aston Martin.... you don't need high power if you have low mass!)

 

[j_j]

I am quite certain that the issue is likely to be time domain related, even if it is merely due to the fact that Woofers have mass (and more importantly inertia) several orders of magnitude greater than a sheet of mylar film - a mass that needs to be precisely accelerated and deccelerated to achieve optimal audio.... and mass does not like change! (Inertia) - the physics tend to advantage ultra low mass, more than they do ultra high power. (Lotus vs Aston Martin.... you don't need high power if you have low mass!)

Well, for a valid measurement you do need the same conditions, including level, etc.

As to mass, the contribution of mass grows as the square of frequency. So at low frequencies it's not very big. A bigger problem in analog systems is going to be the crossover, to put it mildly, both for phase/time issues and frankly for distortion in anything involving an inductor that is not air-core.

I do measure rather a lot of things these days, and while I have seen loudspeakers much worse than you mention, I've also seen some pretty-much mundane bass drivers that anyone can buy inexpensively do much better, at least in a proper enclosure. I can't really speak directly to the driver, etc. We do not use analog crossovers, I can tell you that.

 

[dlaloum]

So at low frequencies it's not very big. A bigger problem in analog systems is going to be the crossover, to put it mildly, both for phase/time issues and frankly for distortion in anything involving an inductor that is not air-core.

My long term, unmeasured, subjective, comment on this, is I have had a long term preference for speakers that eschewed crossovers - particularly in the mid/high transition.

Given I don't have a speaker lab in my home, my observations will remain theoretical suppositions based on direct aural observation - and not measurements

 

[Jimster480]

Nice gear, and it’s interesting to see the mix of gear you guys are using. I also have a couple pieces of tube gear, one is a Fisher integrated tube amp from 1964, I believe this was the last full tube amplifier that they did, and in 1965 they went all solid state.

Another is a Tubes4hifi SP-14 preamp that I had built recently, and that is just great sounding and enjoyable to use.

I read your description, but wanted to ask you, do you feel that any of your tube gear sounds any different that it’s solid state counterpart?

Well I don't own tubes, but my friends did and they definitely sounded different than SS.
Lost details are the number one thing, also a warmer sound in some cases. Likely due to roll off in the highs.

 

[Peluvius]

I love this thread and am eternally grateful to Flipflop for initiating. The more I learn about this space that is a hobby of mine, the more I realise how little I know and I find the information here to be very grounding and accessible for the most part. I dream of a day when audio products come with a star rating that can be easily understood by anyone indicating how accurately they reproduce music, the discussions here make that more tangible.

 

[Newman]

Typically 0.5% THD max on stats... most dynamics never even get close to 1% THD (3% is typical - particularly in the bass - for many Dynamics - and that is well above the audible threshold) - the most difficult area is of course the bass - and the measured results pretty much reflect that.

Stat's don't do "loud" bass - but they do very clean, timbrally correct bass

…while delivering what SPL to the listening position at 80 Hz? I’m confident it would completely fall apart much below that.