Pneumonic

Yes, my use of thew word complex implied varying combinations of all. Music consists of thousands of frequencies but, the fact is, they all combine together to form a single pressure at a given point in time. Of course, this pressure will vary rapidly (up to 20,000 times per second at the highest audio frequency) as the pressure waves in the air pass by the microphone and are captured as a single output voltage.

What's more, there is no such thing as a perfect square wave because a perfect square wave would have infinitely fast rise and fall times. No such thing is possible in nature. So we can only approximate a true square wave. Using this post to expand on my previous point about square wave testing ...... As a general rule, to produce a reasonably good square wave, the DUT must have linear frequency response 10x higher than the square wave frequency. In other words, to produce a good square wave (one that has almost vertical rise and fall times) at 1 KHz, the amplifier must have linear frequency response to at least 10 KHz. In practice, this means that you cannot use square waves at the higher end of a component's frequency range as a square wave will look like a sine if it produced near the upper limit of the frequency range.

The reference signal is a voltage at that instant in time the mic captures it; no matter how complex the sines inputted into the mic. The audio device tasked with processing this signal has no perspective of the signal arriving on its input; it simply processes things as it comes. As for relationship to hearing ..... as stated before ..... we reference the distortion composition and levels to established hearing limits. That determines if the distortion that is added is @ audible levels or not. If the added distortion is below audibility then we don't hear the distortion that is added. If the added distortion is @ audible levels then someone might be able to hear it.

The mic captures all the complex waveforms it sees and combines them all to form 1 single, instantaneous, voltage. This voltage has a rate of change that is set by the frequency response and amplitude capability of the device that is tasked with processing the signal.

Indeed, and even if they did having practical meaning (at the design level of gear being bandied about in forums such as these)*, there is no accepted engineering standard that could be published that would show that one component was better than another based on square wave testing. * It is true that square wave performance will be reflected in an amplifier's slew rate. But the slew rate is simply a function of the combination of an amplifier's high frequency response and voltage output. An amplifier simply needs to have sufficient high frequency response and power to meet the demands of the music. Its slew rate will be determined by those factors.

Any piece of audio electronics only ever has to reproduce 1 frequency at a specific voltage, no matter how complex the music. Do you agree? If so, it then follows that if said device only deals with one voltage at any point in time, it will reveal its flaws if you test it with just one frequency. In practice, a 1 kHz signal is used because it allows for subsequent harmonics (and random noise) to be easily seen with a spectrum analyzer so that they can be analyzed for audibility. If one wanted to test for intermodulation distortion, we’d simultaneously use 19kHz and 20kHz test signals in order to produce IMD that can then be measured. However, from a practical standpoint, it is easily observed that IMD and THD will be essentially the same in any properly designed gear. Therefore, it’s not necessary to produce IMD #’s as it can be deduced from the THD measurement.

OK, I am navel-gazing and you are speculating on some future scientific revelation that may or may not transpire.

If you want to load an amp's PSU differently then sine waves then use square waves. But, overall, it is a very specific test that is crude and ltd and, in the case of non egregious designs, only relevant for go/no go testing.

Not meant to sound patronizing but I suppose that sometimes writing an explanation in a forum such as this may come across as such. I am not an audio expert though I am an EE with 20+ years experience in the >20k world. Oh, and I did own a pair of Stratus Silvers for a few years. ;-)

This is another reason why, if accurate results is the objective, one should test with objective instrumentation rather than rely on ones ears. ;-)

Re: instantaneous switching. It is needed to avoid the several second echoic memory limitation and to not offer a clue that a switchover occurred. Until such time as you have properly identified the section of playback that allows to to identify between A-B-X you can take as long as you desire to figure out the aspects of the sound that allows you to make this differentiation.

Teresa/mmerrill. Re: real world signal testing. Ah, I see where the confusion lies. No piece of audio electronics ever sees what you refer to as real world signal. While music consists of thousands of frequency tones keep in mind that that they all combine together to form 1 single voltage at a given period of time. In this case, at the time the microphone captures the soundwave and converts it to an electrical signal (ie a voltage). As a result, no matter how complex the music and no matter how many frequencies are involved, the voltage at any given time will be a single value. It doesn’t matter if it’s asked to reproduce 1, or 2 or 1,000 or 1 million tones; the audio signal the piece of audio electronics deals with will only ever consist of 1 voltage at any given time. It is this single voltage that our pieces of audio electronics then process.

I have stated, several times now, that I am open to the possibility that the generally accepted limits of human hearing are not carved in stone. But, until such time as I see documented proofs that alter them, I will continue to go with what science has told us for eons. As for the spectra, I have no way of knowing what procedures and processes were used to produce the result so have no way of replicating it to test for external validity. Have the audibility results ever been offered up for peer review?

I answered in post #1293. Here it is again. "If these measurements reveal distortion errors that are below the generally accepted hearing limits then there will be no sound to hear. If these measurements reveal errors that are above accepted hearing limits then someone might be able to hear the sound of these measured errors." It fits just as I stated. Listen for hours, days, months even if that is what is required to readily identify the discernible differences in DUT's. Once those part differences are noted, then one would want to switch up within the echoic memory limitation to properly identify them in a listening test.

I am asking the questions in order to better understand your thought processes so that I can better respond. I don't think I am asking them in an aggressive way. Echoic memory is as long as 3 seconds so is not really instantaneous. But, yes, ideally one should switch quickly in order to compare within the echoic memory confines.