Can you hear a difference in Sound between Audio Amplifiers?

[AcuDefTechGuy]

I don't think we should debate whether we can hear differences among amps, preamps, and DAC.

Why do we?

What's the point? Has it gotten anywhere in the last 30 years?

Let's just give them the benefit of the doubt and say that there are potential differences that could be significant enough for some people.

Instead, let's ask:
1. Are the differences night and day?

2. Do you care enough about these differences to justify spending an extra $1,000 on amps?

Doing the same thing repeatedly and expecting different results is just......well, insane.

Do 50% of all people believe the differences are night and day?

Are 50% of all people willing to pay even $500 more for that difference?

 

[mtrycrafts]

Where I work there are some real bona fide statisticians who earn a good living telling the rest of us doing cancer clinical trials how to get statistics right with the fewest number of patients. ...

Once I was talking to one of them about, of all things, audio, a subject he liked. ...

What a surprise.

He didn't know a good answer for how many repeated tests one individual should do. He hoped 12 repeats would be enough. But he was very clear about how many individuals are needed. If the likelihood of just guessing is 50/50, it will require about 50 individuals in the test. The test subjects should have enough opportunity to hear an A/B test (where the answer might be 'yes, I hear a difference') as well as an A/A or B/B test (where the answer should be 'no I cannot hear a difference').

And the conclusions from a test with 50 people (N=50) are only a simple yes or no. To have a statistically valid estimate of how many people might actually hear differences would require 300 to 500 people (N=300-500)!

Now you know why I roll my eyes when one person (N=1) claims to hear differences.

Well, if that 1 person had at least 12, 15-20 preferred, and a good dbt, maybe even a very good single blind, it would show to that person if he/she could hear a difference, not a population.
Why would I buy something that I could not properly differentiate enough to make it worthwhile. But then I am not hard core like some in Japan installing their own telephone poles and transformers.

Thanks.

 

[PENG]

Where I work there are some real bona fide statisticians who earn a good living telling the rest of us doing cancer clinical trials how to get statistics right with the fewest number of patients. They really understand how to design blind trials so the answers mean something.

Once I was talking to one of them about, of all things, audio, a subject he liked. So I asked him his take on how many trials and how many test subjects would he like to see in a statistically valid blind listening test.

He didn't know a good answer for how many repeated tests one individual should do. He hoped 12 repeats would be enough. But he was very clear about how many individuals are needed. If the likelihood of just guessing is 50/50, it will require about 50 individuals in the test. The test subjects should have enough opportunity to hear an A/B test (where the answer might be 'yes, I hear a difference') as well as an A/A or B/B test (where the answer should be 'no I cannot hear a difference').

And the conclusions from a test with 50 people (N=50) are only a simple yes or no. To have a statistically valid estimate of how many people might actually hear differences would require 300 to 500 people (N=300-500)!

Now you know why I roll my eyes when one person (N=1) claims to hear differences.

He did make the point that in audio tests there were skill involve for reasons he also explained, or tried to explain...

If we are talking about SQ differences between speakers, room effects, MP3 128 or even 320 to lossless, we wouldn't have beaten the horse for so long because in those cases no (or just very little) listening skills/training would be required.

 

[herbu]

What is THD? Amps all publish this number. How is it determined?

I can see taking a known input signal, measuring the amp output, and comparing them. Wouldn't this indicate "distortion"? And wouldn't careful measurement even reveal harmonics and at least some differences in "tone"?

So if "distortion" in each of 2 amps is clearly below human distinguishing, exactly how can one amp "sound" different than another? Exactly what is it that "sounds" different, and why can't it be measured?

 

[Swerd]

He did make the point that in audio tests there were skill involve for reasons he also explained, or tried to explain...

We did talk about that, but as a good statistician, his answer was "without any data testing skilled vs. unskilled listeners, I couldn't answer without guessing". He did point out that medical clinical trials ignore any issue about prior patient experience. Sometimes clinical results lab test results, clinical assessments by MDs, or patient self-assessments.

Did Floyd Toole and associates once examine speaker listening test results of trained vs. untrained listeners? I don't have a clear memory of his conclusion about this.

If we are talking about SQ differences between speakers, room effects, MP3 128 or even 320 to lossless, we wouldn't have beaten the horse for so long because in those cases no (or just very little) listening skills/training would be required.

Yes. If there are likely to be more obvious sound quality differences, as with speakers, the number of individuals needed to achieve statistical significance drops. When comparisons may make it hard to tell if there is a difference is when statistics requires a larger number of people.

 

[Irvrobinson]

He did point out that medical clinical trials ignore any issue about prior patient experience. Sometimes clinical results lab test results, clinical assessments by MDs, or patient self-assessments.

That's irrelevant. Whether or not a rash clears up is not a matter of subtlety or guessing, like the vast majority of audio trials are. Even subjective medical results, like pain reduction, doesn't focus on subtle differences.

The problem is that audio comparisons, blinded or sighted, produce mostly useless results in audio when the differences between the two sources being compared are slight. Anyone who has ever participated in a comparison test knows this. Guessing in one of these tests is just GIGO. And, no, I don't know of a better way.

 

[Swerd]

That's irrelevant. Whether or not a rash clears up is not a matter of subtlety or guessing, like the vast majority of audio trials are. Even subjective medical results, like pain reduction, doesn't focus on subtle differences.

There are trials of long-term use of pain medications where some or all of the primary data is patient self-assessment. I personally prefer to see hard measurable data like lab test results, image scans, or at least clinical assessments from the MDs.

The problem is that audio comparisons, blinded or sighted, produce mostly useless results in audio when the differences between the two sources being compared are slight. Anyone who has ever participated in a comparison test knows this. Guessing in one of these tests is just GIGO. And, no, I don't know of a better way.

Exactly.

When differences are slight, or none at all, is when a large sample population (N) allows you to safely conclude that no differences were detectable, or are likely to be detectable among a larger sample population. I don't think I've ever seen that done properly in audio listening tests, except by Toole & Olive.

 

[RichB]

What is THD? Amps all publish this number. How is it determined?

I can see taking a known input signal, measuring the amp output, and comparing them. Wouldn't this indicate "distortion"? And wouldn't careful measurement even reveal harmonics and at least some differences in "tone"?

So if "distortion" in each of 2 amps is clearly below human distinguishing, exactly how can one amp "sound" different than another? Exactly what is it that "sounds" different, and why can't it be measured?

The statement that two similarly powered amps with very low distortion will sound indistiquishable when operation within their limits.

The subject interests me so my thoughts on the matter are presented below. Some of this information may are may not be facts

Resistive versus Reactive Loads
For the most part, amps a measured with simple tones into resistive loads. Speakers are reactive loads and music are complex tones. Some amps may current limit when the resistance and phase change. Speakers and amps can also change their characteristics: linearity and distortion levels with temperature.

THD Measurement
I read a review of the Parasound A31 on HomeTheaterHifi that stated that THD includes harmponics +/- 250Hz. If an amp has intermodulation outside that range it is not included. The general consensus distortion further from the initial tone are easier to detect and less desirable. For this reason, HTHFI includes all harmonics to characterize the full range of harmonics. Audioholics also shows the full set of harmonics.

Linearity
I would expect that most SS amps remain linear into reactive loads but some seem to do so more than others. They can also drop in measurements of quality amps up to .5 DB into 4 and 2 ohm loads. This is not likely audible but the performance is related to the speaker load. Linearity is not considered distortion. For example, soft-clipping amps limit distortion by limiting the output. In this case distortion is kept low, but the signal is not properly amplified. This can be a good thing in that it prevents damaging the amp and speakers.

Anecdotal Stuff
My brother-in-law had an ICE amp based Pioneer SC-07 (I think) driving B&W CM10's and it sounded flat when driven. It was reviewed by many sites and measured great. Audioholics found that it had issues driving low impedance loads:

http://www.audioholics.com/av-receiver-reviews/pioneer-sc-07

In this case, a better measurement found an issue, which has since been corrected in the new models.

Different measurement methodologies
When comparing AVR to separate amps, they are not measured to the same levels. Audioholics does but others don't. HTHFI has measured mono-blocks while driving Carver ribbon tweeters but not an AVR.

The best decisions are made with data and we do have good amp measurements and speaker measurements with resistance and phase.

Dealers
Harman does DBT to test their speakers and uses good science in all designs. On AVS, a dealer has stated that the Mark Levinson $15K each mono-blocks are specifically designed to drive the difficult load presented Salon2's. I don't believe that anyone needs to buy $30K amps to drive $22K speakers and this is the kind of statement that makes folks head explode

- Rich

 

[mtrycrafts]

He did make the point that in audio tests there were skill involve for reasons he also explained, or tried to explain...

...

Yes, but we are still talking about individual audiophiles/golden ears who make all those endless claims about their ability, not really asking those trained and honed listeners.

 

[mtrycrafts]

...
Did Floyd Toole and associates once examine speaker listening test results of trained vs. untrained listeners? ...

Yes, he has. They also included other folks I believe such as recording engineers, musicians, etc.

Another paper a good number of years ago, not the quality of Toole and associates, tested conductors and musicians and others if they could even tell if parts of their performance that was deleted afterwards due to some reason, was in their final cut or not. Been a while since I read it. No, they didn't do well at all.
The title is something like The Grass is Always Greener in the Outtakes by Glen Gould, 1975. Send me a private msg.

 

[mtrycrafts]

...

http://www.audioholics.com/av-receiver-reviews/pioneer-sc-07

In this case, a better measurement found an issue, which has since been corrected in the new models.

...
- Rich

There you go. An amp that was not well designed on the first run.
In that case, it was a broken amp.

 

[Swerd]

The statement that two similarly powered amps with very low distortion will sound indistiquishable when operation within their limits.

The subject interests me so my thoughts on the matter are presented below. Some of this information may are may not be facts

We really aren't so far apart on this.

Every so often I cite this interview with an electrical engineer, named Bob Cordell, who once had real experience with audio amplifier design. He later earned a living in the telecommunication world. In my opinion, his word about audio amplifiers was the gospel, and probably still is today.

http://www.helarc.com/guru/guruscordell.htm

Edit: I've now learned Bob Cordell has a web site: http://www.cordellaudio.com/

 

[herbu]

http://www.helarc.com/guru/guruscordell.htm

Great link! Thanks!

 

[jinjuku]

I am not arguing against blind testing but I find Iverson's observations interesting and sensible.

- Rich

As well you shouldn't. Blind testing is valid. I got rid of my Parasound HCA 1000's and replaced them both with a $179 Crown XLS 402D because they had a higher noise floor.

And I was able to go 10 for 10 blind in telling which amp was connected.

 

[auronihilist99]

I got rid of my Parasound HCA 1000's and replaced them both with a $179 Crown XLS 402D because they had a higher noise floor.

Great, I've always liked the newer Crown XLS Series II amps. This confirms that you really can hear differences in SS amps. Thanks, jinjuku.

 

[jinjuku]

I got rid of my Parasound HCA 1000's and replaced them both with a $179 Crown XLS 402D because they had a higher noise floor.

Great, I've always liked the newer Crown XLS Series II amps. This confirms that you really can hear differences in SS amps. Thanks, jinjuku.

Specifically what I heard was gain stage hiss with the Parasounds.

 

[auronihilist99]

Interesting, thank you for your clarification. It proves that just because a component is considered "audiophile", it doesn't always mean it is superior. Long live blind testing.

 

[jinjuku]

Interesting, thank you for your clarification. It proves that just because a component is considered "audiophile", it doesn't always mean it is superior. Long live blind testing.

What I found interesting about the Parasounds is that there was a lot of protection circuitry, which according to audiophiles is a no no, there was, I believe no fewer than 5 fuses in the amp.

 

[Diesel57]

This is great, which this thread provided me with how opinions differ on one's blind listening test as well as 1 amp vs another amp and a avr vs a amp on which sounds better and while it may have some value to what company or group or people listening experiences is, still makes it opinionated, while no one is wrong or right, what's left is the response and answers of one's listening ears...and what I have read here has been educational & informative=knowledge...science @ it's best...Thank You All

 

[mtrycrafts]

... Long live blind testing.

Let's hope so.