I don't want to beat a dead horse but I'm bugged.

It seems many amplifiers have lots of negative feedback and they have low distortion at 20Khz.

I don't think it is back in the day any more. There are lots of amplifiers on Stereophile and ASR that have very low distortion at 20Khz. Lower than this Rogue at almost any frequency.

It needs to be more than just low distortion at 20KHz although that certainly does not hurt. What is important is four things; if these four things are met, the amp will be easy to listen to, regardless of its technology:

1) the distortion will be consistent at all frequencies- the same at 100Hz, 1KHz and 10KHz.

2) the distortion spectra at 1 Watt will be such that the higher ordered harmonics will not be audible- for whatever reason.

3) The distortion spectra at 6dB below full power will also prevent higher ordered harmonics being audible.

4) IMD is well controlled.

These things sound simple and easy but they can be quite tricky! Most amps fail them outright- and sound bright and harsh as a result, even though they 'measure well' otherwise. The real problem isn't that they measure well, its that the proper measurements aren't made at all!

Many amplifiers on tested on Stereophile and ASR meet those criteria @atmasphere .  By low, I don't mean only low at 20KHz, but low at all frequencies, <<0.01% at all frequencies. I don't think distortion has any particular meaning at >10KHz as the distortion is >20KHz. I assume the measurements are using an extended bandwidth beyond human hearing.  ASR shows distortion from low to high power at a range of frequencies in their reviews.

@atmasphere 

Most amps fail them outright- and sound bright and harsh as a result, even though they ’measure well’ otherwise. The real problem isn’t that they measure well, its that the proper measurements aren’t made at all!

The absolute truth in my opinion. Current test bench measurements are inadequate. It makes no sense to measure good and yet sound poorly. What good are measurements such as these?

There are numerous audio components that don’t measure particularly well yet sound fabulous. Ralph is correct in that what really matters in determining good sound quality is not being done currently with measuring instruments. I do believe at some future point this dilemma will be properly solved. It sure is not the case now.

I guess for some there’s a degree of reassurance or peace of mind with current methods of audio component bench testing. As it stands presently the ears are far more reliable and trustworthy in judging audio product sound quality.

Charles

I assume the measurements are using an extended bandwidth beyond human hearing.

The problem if i may give my grain of salt, is that we dont exactly know what are human hearing abilities...

Human hearing limit in decibel scale and in the frequency scale yes we know but what about the way the brain/ears used the information and decipher it with the non linear tool kit of the cochlea/brain ?

In the words of these mathematical physicists turned acoustician:

Jacob N. Oppenheim and Marcelo Magnasco of the Laboratory of Mathematical Physics at Rockefeller University have conducted experiments indicating that the human brain does not use the Fourier transform when resolving a cacophony of noise into individual sounds and voices.

While the Gabor limit associated with the Fourier transform stipulates that you can't simultaneously determine a sound's frequency and duration, the 12 musicians subjected to Oppenheim and Magnasco's battery of tests beat the limit by as much as a factor of 13.

The Fourier transform cannot, therefore, fully explain the machinations of the human brain. "The actual algorithm employed by our brains is still shrouded in mystery," says Magnasco.

«You see, physicists tend to think hearing is spectrum. But spectrum is time-independent, and hearing is about rapid transients. We were just told, by the data, that our brains care a great deal about timing.»

«For the first time, physicists have found that humans can discriminate a sound’s frequency (related to a note’s pitch) and timing (whether a note comes before or after another note) more than 10 times better than the limit imposed by the Fourier uncertainty principle.»

«

The results have implications for how we understand the way that the brain processes sound, a question that has interested scientists for a long time. In the early 1970s, scientists found hints that human hearing could violate the uncertainty principle, but the scientific understanding and technical capabilities were not advanced enough to enable a thorough investigation. As a result, most of today’s sound analysis models are based on old theories that may now be revisited in order to capture the precision of human hearing.»

This article is particularly illuminating if someone take the time to read it and his imnplication to understand hearing and why our measurements are not the good one...

Many amplifiers on tested on Stereophile and ASR meet those criteria

This statement is misleading IMO. Most fail requirement 1) of my prior post. If you look at the distortion amount and its spectra, you’ll find that most amps have distortion rising with frequency and the spectra changing along with it.

This is because (as I’ve mentioned before) that they lack sufficient feedback, because they also lack sufficient Gain Bandwidth Product. There are exactly two ways around the problem of rising distortion and changing spectra:

1) build the circuit with zero feedback and wide bandwidth

or

2) build the circuit with really enough Gain Bandwidth Product to support over 30dB of feedback (such that feedback is actually the same at 15KHz as it is at 100Hz).

Most amps spec’ed to have ’30dB of feedback’ might have that at 1KHz but they will be less than that at 10KHz since the feedback is falling off due to insufficient GBP. This causes distortion to rise with frequency and it also means that the distortion caused by the application of feedback can’t be corrected by the circuit. That translates to ’harsh and bright’ and we’ve all heard amps with this problem; we’ve been hearing them for the last 60-70 years! There are plenty of amps made like that today and only a very few that actually meet the requirements stated in 2) of this post above if they have feedback.

I do believe at some future point this dilemma will be properly solved. It sure is not the case now.

Its not that we as an industry can’t do it, its that we lack the will, plain and simple. You do also need to know enough to understand how something will sound by looking at the distortion spectra... I’ve no doubt that some of these measurements don’t occur because there isn’t the education out there to really understand the data! Plus- heaven forbid- you know how something sounds by looking at the measurements!