Can you hear a difference in Sound between Audio Amplifiers?

[Art Vandelay]

If you read and understood my post, you would not be saying this. I used a specific example of an 80W/100W AVR versus a 200/300W 8/4 ohm rated amplifier using 800 diamond speaker to listen at 85 dB spl in a large room, the only thing I omit is stating the listening distance. So let me make it more explicit, say distance is 4 meter, that should be reasonable in a 15' wide 25' long room. That little 80W/100W 8/4 ohm rated AVR is going to clip quite often during movie/music peaks, don't you think, to the point that it would be audible? If not enough, say that AVR is rated only 50W/60W 8/4 ohm and listening distance is 15 ft. Note that I used 800 diamond deliberately to make an extreme case, their impedance dips down close to 3 ohms.

At least I mtrycraft agrees with me, that's good enough for me to make my case.

How did you arrive at those figures?

A pair of 800 Diamonds will produce about 87-88dB spl at 4m with only 2.87vrms input, so even your little AV amplifier will be able to produce well over 100dB sound pressure without clipping. Add to that the fact that pretty much all music these days is pre clipped / limited during production, so you could probably get away with a 25W AV amplifier. In my own listening room which is about 20 x 15 I rarely exceed 20W peak during any listening session.

 

[AcuDefTechGuy]

You should know by now I meant 20 ft because you could see how desperate I was to make my point that amps could sound different under certain different test conditions, i.e. one is over stressed while the other is not.

I know. I just didn't want someone else to get a chance to even rebut.

 

[AcuDefTechGuy]

How did you arrive at those figures?

A pair of 800 Diamonds will produce about 87-88dB spl at 4m with only 2.87vrms input, so even your little AV amplifier will be able to produce well over 100dB sound pressure without clipping. Add to that the fact that pretty much all music these days is pre clipped / limited during production, so you could probably get away with a 25W AV amplifier. In my own listening room which is about 20 x 15 I rarely exceed 20W peak during any listening session.

I can definitely attest to being able to easily power the B&W 802D2 with just the Denon 3312 AVR to very high volume from 15FT listening distance and sounded pretty great.

I laugh when I hear guys say things like, "My B&W speakers need a lot of power to sound good".

 

[PENG]

How did you arrive at those figures?

A pair of 800 Diamonds will produce about 87-88dB spl at 4m with only 2.87vrms input, so even your little AV amplifier will be able to produce well over 100dB sound pressure without clipping. Add to that the fact that pretty much all music these days is pre clipped / limited during production, so you could probably get away with a 25W AV amplifier. In my own listening room which is about 20 x 15 I rarely exceed 20W peak during any listening session.

Right, but that would be peak. That means to get say 75 dB average, the peaks will be 6 to 25 dB higher and the resulting distortions may provide enough clue to someone with very discerning hearing if the right music is used for the test. Note that the 800 diamonds have impedance dips to almost 3 ohm so your 87-88 dB at 2.83V will have to be de-rated a little more in terms of W/1m. It also has some significant phase angel that aligned with the low impedance points. That means a lesser amp may run hot if pushed, and that would affect it's performance. IIRC, ADTG had used his AVR-331X with the 802 D2 but at least that's a respectable mid range AVR.

In any case, you don't have to convince me on how little power one may actually need. I have amps from 50-300/500 WPC and have taken measurements myself, as well as watching the peak power meters on the power amp. I know most of the time my amps output fractional watts on average. Let me repeat again, for the last time, I am only trying to make my point on why I don't think the poll question is specific enough for meaningful results. I enjoyed the discussion but am ready to move on, thank you.

 

[Art Vandelay]

Right, but that's would be peak. That means to get say 75 dB average, the peaks will be 6 to 25 dB higher and the resulting distortions may provide enough clue to someone with very discerning hearing if the right music is used for the test. Note that the 800 diamonds have impedance dips to almost 3 ohm so your 87-88 dB at 2.83V will have to be de-rated a little more in terms of W/1m. It also has some significant phase angel that aligned with the low impedance points. That means a lesser amp may run hot if pushed, and that would affect it's performance. IIRC, ADTG had used his AVR-331X with the 802 D2 but at least that's a respectable mid range AVR.

In any case, you don't have to convince me on how little power one may actually need. I have amps from 50-300/500 WPC and have taken measurements myself, as well as watching the peak power meters on the power amp. I know most of the time my amps output fractional watts on average. Let me repeat again, for the last time, I am only trying to make my point on why I don't think the poll question is specific enough for meaningful results. I enjoyed the discussion but am ready to move on, thank you.

i was actually specifying peak (instantaneous) SPL. fwiw, most recorded music has <10dB dynamic range these days and typically >6db for popular music, thanks mostly to the overuse of peak limiting during recording and mastering. It mostly doesn't sound very pleasant but it does mean that domestic amplifiers are more likely to be operating within dynamic limits, although that's not why it's done of course.

Probably the single most audible parameter (for most people) is damping factor, and particularly so with speakers such as B&W 800 Diamonds. Damping factor is the ratio of load to output impedance, which includes the connecting cable. All of my power amplifiers sound different at low frequencies due mostly to differences in DF. That said, most AV amplifiers sold these days for mass consumption are based on generic designs with a high DF. Ironically perhaps, "audiophile" designs tend to have lower damping factors and there's a greater variation between various audiophile brands too.

For people who care though, and with good ears, it's what goes on at high frequencies that's most important. From my experience, amplifiers can and do sound different at high frequencies due to a combination of factors, but mostly it boils down to a combination of dynamic and non-linear distortions. Some of the former are difficult to measure but it's generally accepted that small amounts of high order harmonic distortion is more likely to be audible than than low order HD up to as much as 1%. Similarly, people can be sensitive to modest amounts of high frequency inter-modulation distortion.

Manufacturers specifications will provide us with a distortion number of course, but it's essentially meaningless.

 

[mtrycrafts]

...
Probably the single most audible parameter (for most people) is damping factor, and particularly so with speakers such as B&W 800 Diamonds. Damping factor is the ratio of load to output impedance, which includes the connecting cable. All of my power amplifiers sound different at low frequencies due mostly to differences in DF. That said, most AV amplifiers sold these days for mass consumption are based on generic designs with a high DF. Ironically perhaps, "audiophile" designs tend to have lower damping factors and there's a greater variation between various audiophile brands too.

....

Where did you get this nonsense?
Where is you evidence?

 

[Art Vandelay]

Where did you get this nonsense?
Where is you evidence?

Please do tell what you think is "nonsense" and why.

All conclusions I've arrived at are based on & supported by sound engineering principles and confirmed by listening tests.

 

[mtrycrafts]

Please do tell what you think is "nonsense" and why.

All conclusions I've arrived at are based on & supported by sound engineering principles and confirmed by listening tests.

DF? Sound engineering and listening tests? Really? How was that tested and conducted?

DF add below threshold of detection loss, period. Please do tell us how you arrived at the measured losses due to DF, numbers please, in dB would be great. Listening protocol specifics as well.

 

[Art Vandelay]

DF? Sound engineering and listening tests? Really? How was that tested and conducted?

DF add below threshold of detection loss, period. Please do tell us how you arrived at the measured losses due to DF, numbers please, in dB would be great. Listening protocol specifics as well.

Measured losses are usually small enough to be ignored. A poor DF will result in a change of the designed system Q, which will manifest itself as audibly less defined or boomy bass. Certainly a low DF is audible to me when I'm testing a TC tube amp, but some solid state amplifiers that omit use of negative feedback can also have a DF<40 and the bass can suffer as a result. Whist it's audible to me it may not be audible to others, though most if not all of my audiophile friends are equally sensitive to poor DF.

Fortunately, most HT amplifiers these days are solid state class AB or D with DF>>50, so unless the speaker cable is very long and HR the DF shouldn't be a problem

 

[Johnny2Bad]

" ... - No, you are describing a music lover, not an audiophile. One can be both, of course, but they are not synonyms. ..."

Bullcrap. If you're not listening to music, you're not going to spend money on a sound system.


" ... - What "history"? We are talking the science of sound. Many disciplines go into it, and I am not aware of any "history" of perplexed audio fans desperately trying to discover this science so they can measure it. ..."

Wow. Just wow.

-" ... Because you have a number of psychological, and acoustical, factors you are not even acknowledging. Until you do, you will not be informed on this subject. It's rather like the patient who takes a
homeopathic remedy and starts to feel better, even though it's only placebo effect. Billions are spent by fools who would know better if they cared to learn. Instead, they buy the hogwash that a substance diluted in water to the point there are no molecules of the substance left is still effective because the water "remembers" the molecules. I know audiophiles who laugh at the suckers, then claim that, a la Synergistic Research, that cables improve in sound quality as the electrons "learn" the best path through them. ..."

What in the world are you talking about?

" ... - You keep saying this. Your source for this? ..."

Aaah, the troll's response that the OP should go about perform a bunch of work for his benefit. Surely you can find the information yourself, it's not a secret.

" ... - No, you wouldn't, because you misunderstand the fundamental of scientific progress. And human nature. Look how many pointless "advances" have been made in digital conversion that are for all intents inaudible. There is, maybe unfortunately, no shortage of audio designers who believe they can make a better mousetrap. ..."

You speak of "scientific progress" after denying that there is a history composed of those who design audio circuits who came to their innovations because something didn't sound right to them. Whatever.

" ... - How would that possibly be so? Are the ones more "oneish" and the zeros more "zeroish"? And I note how quickly you dispose of the need for DBT. Without DBT, you cannot support your claim. Is that why you resist it? ..."

You seem to be under the impression that digital storage and digital processing somehow exists in a digital domain; that electrical ones and zeros somehow exist. They do not.

Everything digital ... from the storage medium, to the CPUs in computers, to the transition from one device to another through a cable, exists in the analog domain. Digital chips contain resistors, transistors, and capacitors, not some digital-domain pixie dust.

The zero bit is not at zero volts and the one bit at [some positive value] volts; they are (for example) +2 volts and +4 volts with a zero crossing point of +3 volts. Errors are common; all digital systems specify an error rate (error vs time) and typically employ error correction which may, or may not, work on individual bits of data.

Double Blind Testing is useful to a point. It is not how we actually listen to systems, so it does not replicate fully the listening experience, the very job we expect our sound systems to perform. I do not dismiss it as useless, but to suggest it is the definitive measure is naive and narrow minded. *Properly Performed* DBTs can provide useful information, but many DBT's I see are flawed and do not follow the basic protocol they purport to represent.

Apologies for delving into this "history" ... excuse me ... "scientific progress" but Dr Harry F Olsen (RCA) showed in the 1940's one result of DBTesting; he set up a system consisting of the best equipment available at the time, added a selectable distortion generating circuit, and applied frequency-limiting to the reproduced sound (3K, 5K, 7K, 10K and 15K) with an untrained audience and a true DBT regimen. The result was the higher the frequency limit, the less distortion the audience could tolerate.

Audio Precision recently repeated a version of this experiment at the Rocky Mountain Audio Fest without the frequency limiting but with the distortion generation; the audience of show goers (apparently, in your world "audiophiles" who were not necessarily "music lovers") were asked to put their hands up when the distortion became unpleasant. The entire audience didn't put their hands up until 2Hd of 30% was reached, while everyone's hand was up by 3% of 7Hd and the majority before 1% 7Hd.

In 1945, Howard A Chinn and Phillip Eisenberg of CBS performed another true DBT; this time limiting the frequency response to a set of bands (40~10K; 80~7K and 180~4K. The audience preferred the 180~4K limited music to the 80~7K and the fewest preferred the 40~10K. The conclusion was that people preferred the sound quality they were used to.

Later, Roger E Kirk of Ohio State University repeated the Chinn-Eisenberg experiment with untrained college students chosen at random. After confirming the Chinn-Eisenberg results (they were the same) he divided the students into three groups; one which was exposed to a series of music performances via an unfiltered system, one exposed to the same music with a moderately filtered system, and one whom were not exposed to any music at all.

Now those who were trained, so to speak, on the HiFi system now preferred that system; those exposed to the MidFi system now preferred it, and those who were not exposed retained their initial preference to LoFi.

Just one example where DBTs can give misleading results ... after all, the untrained audiences preferred the lowest quality sound while the trained audience preferred the highest .

A proper DBT is not particularly easy to perform; although it is certainly possible. It is imperative that the experimenters watch for unintended consequences and incorrect interpretation of the test result data. THEN the results are valid, within the limit mentioned previously (that we don't expect to listen to music that way).

Is that enough "history" ... oops, did it again ... "scientific progress" for you? It shouldn't be ... you should be interested in doing more research into how audio has evolved to the point were it is today, but I suspect your mind is already made up, and that you "know everything".


- How have we improved upon "perfection"? Where are your DBTs showing listeners can reliably differentiate between 16/44 and 24/192? ..."

What has 16/44 and 24/192 have to do with it? Have you even listened to a 1985-era CD player lately? My God, if you can't hear a difference vs a modern 16/44 file and DAC set to 16/44 (no upsampling), well, that explains a lot.

* If you want some real fun, tell a young woman today that she is a "feminist"; there's a good chance she will become offended, despite the fact that her beliefs are EXACTLY the same as the feminists of yore.

" ... - Actually, no there isn't such a "good chance." And your quip is offensive. ..."

You obviously don't know many young women. You should get out more.

 

[Johnny2Bad]

Hmmm....
Have to debate this point in the last 25 years there have been tremendous strides in amplifier measurements... In the old days we used the HP 221C then moved up to the Sound Technology 1700 but the entire landscape changed when Audio Precision entered.. The problem with the early measuring stuff was that their residual THD was often higher than the component being measured... Then the Cascade 2 & 586 became available so we can now measure components down to 0.0001%..

But as pappy always said "It if measures good, but sounds bad, it is bad.."

Just my $0.02..

I meant that we still use the same metrics (frequency response, power bandwidth, s/n, THD, IMD) as in the past, and that the measuring equipment was capable of precision, to the point where it could measure below the residuals of much of today's audio equipment.

Check out the December 1970 'High Fidelity' magazine test of a SONY TA-1144 integrated amplifier (page 39). THD measured at 0.057% 15w 20~20K L channel and 0.068% 30w 20~20K, and long before the Sound Tech 170x series was available a few years later.

http://www.americanradiohistory.com/Archive-High-Fidelity/70s/High-Fidelity-1970-12.pdf

Also note the breadth of measurement criteria as specified by the IHF-202 standard (1978):

http://www.gammaelectronics.xyz/audio_06-1978_ihf.html

And finally a test report from Audio comparing two flagship receivers; note THD measurement capability below 0.01% and full spectrum analyzer photo showing the harmonic structure (and interestingly, that spectrum seems indicitave of high levels of Global Negative Feedback, typical of the components of the era when a "spec war" was ongoing amongst the Japanese manufacturers).

The practise was mercifully eventually abandoned as complaints began being voiced that certain gear, as you say, tested good but sounded bad. Which, apparently according to some here (not you), never happened.

http://www.gammaelectronics.xyz/audio_01-1977_titians.html

 

[TechHDS]

Yes anyone should be able to hear a difference between amps. Just don't adjust tone control's on the pre-amp first..

 

[Johnny2Bad]

Damping Factor is not, in my opinion, a measure of sound quality, or more precisely, may be an indication perhaps of poor sound quality.

High DF is easily achieved by increasing the Global Negative Feedback, a practice that is depreciated generally amongst modern amplifier designers. GNF is useful but in small doses, as it negatively affects the harmonic structure (high order and odd order harmonic distortions) of THD.

Also, DF is usually much less than amplifier specifications suggest; the total impedance of the loudspeaker cable (both ways, so 20ft worth in a 10 ft cable) plus the crossover, plus the loudspeaker voice coil is summed and that value is compared to the output impedance of the amplifier in question. DF specs from an amplifier spec sheet ignore the complex impedance calculation and specify it only into a non-inductive resistor.

Some may counter that the resistance of 20 feet of speaker cable is negligible, and that would be correct, except that the output impedance of a Solid State amplifier is also negligible, so the ratio (which is what the DF spec is) becomes significant.

Many amplifiers specify DF at 1KHz, where it is essentially of no Earthy Use, rather than at 20 or 100 Hz where it's purported advantages should lie. DF at low frequencies is inevitably lower than at 1 KHz, so there is nothing to be learned from the specification as normally provided by manufacturers (note that the test frequency is rarely included).

The importance of DF was introduced in the early days of Solid State, as it was lower in that circuit than through a transformer-coupled Vacuum State amplifier. It is, basically, a "feature" dreamed up by the marketing department. Virtually all well designed amplifiers have adequate DF and when it is unusually high, it's a red flag rather than an indication of superior performance.

 

[PENG]

It seems to me this negative talks about negative feedback has been exaggerated a little too much. Nelson Pass surely is a respected audio amplifier designer but on certain things we still have to take him with a grain of salt. The thing is, if THD+N is near 0, say 0.002% to 0.01%, then regardless of the harmonic structure (in the distortions), you are not going to hear the difference between them. You can't hear something that is not there, regardless of what caused the disappearance. You may hear the byproducts, if such byproducts were created by the NFB but not the vanished distortions.

As for DF, Mr. Art Vandelay apparently referred to <40 being a potential issue. That's sounds reasonable, I would have problem if he meant to say DF needs to be as high as Crown Audio used to claim it needed to be.

 

[Art Vandelay]

High DF is easily achieved by increasing the Global Negative Feedback, a practice that is depreciated generally amongst modern amplifier designers. GNF is useful but in small doses, as it negatively affects the harmonic structure (high order and odd order harmonic distortions) of THD.

In a well designed amplifier it's very possible to use lots of GNFB without generating high order harmonics, but it's true that odd order distortion which is typical of amplifiers that use differential input stages and push-pull output stages, can result in a sterile sounding amplifier, even if the distortion is mostly below 0.05%. High NFB amplifiers need to use dominant pole compensation to prevent HF stability problems, but this measure itself has the effect of loading the input stage and generating odd order distortion at 12db / octave above a few kHz.

Also, DF is usually much less than amplifier specifications suggest; the total impedance of the loudspeaker cable (both ways, so 20ft worth in a 10 ft cable) plus the crossover, plus the loudspeaker voice coil is summed and that value is compared to the output impedance of the amplifier in question. DF specs from an amplifier spec sheet ignore the complex impedance calculation and specify it only into a non-inductive resistor.

That's not the correct way to calculate actual DF though. Actual DF is the ratio of the coil impedance to the amplifier output impedance - where the output impedance is the sum of all resistances (impedances) in between (Thevenin eq). It's why people will tell you that an amplifier DF>40 or 50 is mostly academic.

But a really high DF isn't necessarily wasted, because it's usually a measure of some level of immunity against speaker induced distortions that can plague some low feedback designs and affect LF performance. With GNFB it's generally best to use either none at all or lots of it.

 

[mtrycrafts]

Measured losses are usually small enough to be ignored. A poor DF will result in a change of the designed system Q, which will manifest itself as audibly less defined or boomy bass. Certainly a low DF is audible to me when I'm testing a TC tube amp, but some solid state amplifiers that omit use of negative feedback can also have a DF<40 and the bass can suffer as a result. Whist it's audible to me it may not be audible to others, though most if not all of my audiophile friends are equally sensitive to poor DF.

Fortunately, most HT amplifiers these days are solid state class AB or D with DF>>50, so unless the speaker cable is very long and HR the DF shouldn't be a problem

Interestingly AH has an interesting article on this:

http://www.audioholics.com/audio-amplifier/damping-factor-effects-on-system-response

 

[M Code]

And finally a test report from Audio comparing two flagship receivers; note THD measurement capability below 0.01% and full spectrum analyzer photo showing the harmonic structure (and interestingly, that spectrum seems indicitave of high levels of Global Negative Feedback, typical of the components of the era when a "spec war" was ongoing amongst the Japanese manufacturers).

The practise was mercifully eventually abandoned as complaints began being voiced that certain gear, as you say, tested good but sounded bad. Which, apparently according to some here (not you), never happened.

http://www.gammaelectronics.xyz/audio_01-1977_titians.html

The (2) mentioned receivers from Panasonic and Pioneer both failed the FTC preconditioning requirement, if the (1) hour preconditioning was done their published power output specs would be decreased 30-50%. The power face-off in the titan receivers was won by the Marantz 2500/2600. Whose amplifiers were designed in the USA by their pro-audio R&D team, based upon their highly sucessful pro amplifier 510/510M. The key to the excellent sonics of the Marantz amplifiers is that they used a minimum amount of negative feed-back with a full-complimentary output toplogy so they were very stable even under high driving transient conditions... Back in the late 70s', Marantz didn't play the same max spec game, they were truthful about their specs thats why in most brochure comparisons they had less power than the competition. However into a 4 ohm load they would generally show an increase of 25-35% in power output with a doubling of their THD specs..

Just my $0.02...

 

[PENG]

In a well designed amplifier it's very possible to use lots of GNFB without generating high order harmonics, but it's true that odd order distortion which is typical of amplifiers that use differential input stages and push-pull output stages, can result in a sterile sounding amplifier, even if the distortion is mostly below 0.05%. High NFB amplifiers need to use dominant pole compensation to prevent HF stability problems, but this measure itself has the effect of loading the input stage and generating odd order distortion at 12db / octave above a few kHz.



That's not the correct way to calculate actual DF though. Actual DF is the ratio of the coil impedance to the amplifier output impedance - where the output impedance is the sum of all resistances (impedances) in between (Thevenin eq). It's why people will tell you that an amplifier DF>40 or 50 is mostly academic.

But a really high DF isn't necessarily wasted, because it's usually a measure of some level of immunity against speaker induced distortions that can plague some low feedback designs and affect LF performance. With GNFB it's generally best to use either none at all or lots of it.

Simply put, DF is just another practically useless specs that tell less than 1/2 or even 1/4 of the story. Manufacturers would have told us more if they simply provide the output impedance range for the FR of say 10-20,000 Hz, and let us try to figure out the damping part if we really want to.

 

[PENG]

Why would they? To them, it sounds different under all sighted conditions.
And, they don't accept DBT so they exclude that as a given.

You meant to them, something like the ATI 6002 and 4002 will sound different when used at 30 dB below the rated output of the 4002, or likewise the MC1kW and MC2kW and so on and on....? I wish them luck in finding one that is best sounding to them and that they can own it for their best sounding speakers in their best sounding rooms.

 

[Art Vandelay]

Simply put, DF is just another practically useless specs that tell less than 1/2 or even 1/4 of the story. Manufacturers would have told us more if they simply provide the output impedance range for the FR of say 10-20,000 Hz, and let us try to figure out the damping part if we really want to.

There are a few manufacturers that do specify the output impedance at a specific frequency or range of frequencies, but it's a pretty simple calculation to convert from Z to DF, so I'm fine with either as long as the figure is reasonably accurate.

If I decide to design my own active sub-woofer with Qtc = 0.707 then I definitely do need to know the output impedance of my amplifier.