It is actually fairly simple to connect an actual speaker system to an actual amplifier, and using a signal generator as a source, measure the frequency response and distortion of the amplifier with that load over the audio band with an oscilloscope and Distortion Analyzer.
John Atkinson does a distortion analysis for every amplifier that is reviewed in the magazine Stereophile and publishes the detailed results. Since a good distortion analyzer costs $30K and more, and requires considerable expertise and experience to use properly, it is not surprising that magazines do not usually do this. He points out the shortcomings of each amplifier, and the wise will do well to heed his analysis before purchasing an amplifier.
You would probably learn a lot if you were to read 10 or 20 of his amplifier test reports from past issues COMPLETELY AND CAREFULLY and let the information sink in a bit. It is obvious that you have never done so in the past. Educate yourself a bit and your questions and statements will possibly make more sense. I am always amused when people criticize Stereophile as being a purely subjective magazine, when they do much more objective scientific performance testing of products than any other. They obviously are not reading the whole article.
This is how amplifier design engineers do it, and if you were to do so you would find that the differences between amplifiers are quite significant and measureable.
Using the complex load of an actual speaker to test the amplifier will reveal the shortcomings of the amplifier very quickly, while a test with a resistive load will only give some information.
The amplifier performance will be different with every speaker, since some speakers have a greater degree of impedance variation and reactances than others.
Better-designed amplifiers, however, will perform better, maintaining more linear gain and producing less distortion, even with those speakers that are a more difficult load to drive.
Edited by commsysman - 7/4/13 at 7:22am
Originally Posted by beaveav
As a retired EE professor, what measurements would you suggest one do in order to best capture these sound differences?