Threads like this one show the limits of the tests currently carried out on PAs, because they are simply too magnifying and ignore one or the other parameter.
What is the goal of a PA? Quite simple really: it should make a weak input signal usable for a loudspeaker. Technically speaking, it is an impedance converter: energy is added to the input signal. Of course, the whole thing should be linear (= neutral at the end of the day). The impedance converter should not add anything under all real existing conditions and of course not remove anything either (here the input as well as the output interface has to be considered).
As simple as the requirement sounds - as complicated is the technical implementation.
From my point of view, the current tests only provide a statement as to whether a PA meets the minimum requirements with regard to linearity. Or to put it another way: Below these requirements, one must assume in any case that the signal is audibly altered. But they do not allow any final conclusion as to whether the PA is actually neutral. The existing measurements are the view with the magnifying glass on individual special parameters, which can be measured more or less easily today.
I miss the overall picture that results from the sum of several smaller errors / deviations, which can then finally lead to audibility under certain conditions (the conditions in the real world are highly diverse and complex).
From my point of view, the following tests would help much more to evaluate a PA in a resilient way:
The test should "simply" compare different complex input signals or even quite real signals with the output signal and show their difference (i.e., gain and phase are subtracted linearly). In this way it is made visible what the PA "adds". Ideally, the difference is 0. The whole thing of course at some different interface conditions (keyword capacitance, inductance and resistance). The whole naturally also with many different volumes.
For comparison of two PAs e.g.:
E2E test: the two real generated results at the output of the same loudspeaker (recorded via microphone) are superimposed and the difference is formed. If both PAs are "neutral", the difference must be 0 at this point. However, it is highly unlikely that this will be the case (But I would like to be convinced otherwise). Ideally, this E2E comparison is done with a few very different loudspeakers (to finally show potentially different behavior of the same PA at different loudspeakers).
What is the big advantage of an E2E test: Due to the additional influences of room and transducer (both of which are not error-free either), existing deviations, even small ones that are considered irrelevant, are further amplified under unfavorable circumstances, so that they do become audible at the end of the day.
Tests like this would help shed more light on the matter.
What is the goal of a PA? Quite simple really: it should make a weak input signal usable for a loudspeaker. Technically speaking, it is an impedance converter: energy is added to the input signal. Of course, the whole thing should be linear (= neutral at the end of the day). The impedance converter should not add anything under all real existing conditions and of course not remove anything either (here the input as well as the output interface has to be considered).
As simple as the requirement sounds - as complicated is the technical implementation.
From my point of view, the current tests only provide a statement as to whether a PA meets the minimum requirements with regard to linearity. Or to put it another way: Below these requirements, one must assume in any case that the signal is audibly altered. But they do not allow any final conclusion as to whether the PA is actually neutral. The existing measurements are the view with the magnifying glass on individual special parameters, which can be measured more or less easily today.
I miss the overall picture that results from the sum of several smaller errors / deviations, which can then finally lead to audibility under certain conditions (the conditions in the real world are highly diverse and complex).
From my point of view, the following tests would help much more to evaluate a PA in a resilient way:
The test should "simply" compare different complex input signals or even quite real signals with the output signal and show their difference (i.e., gain and phase are subtracted linearly). In this way it is made visible what the PA "adds". Ideally, the difference is 0. The whole thing of course at some different interface conditions (keyword capacitance, inductance and resistance). The whole naturally also with many different volumes.
For comparison of two PAs e.g.:
E2E test: the two real generated results at the output of the same loudspeaker (recorded via microphone) are superimposed and the difference is formed. If both PAs are "neutral", the difference must be 0 at this point. However, it is highly unlikely that this will be the case (But I would like to be convinced otherwise). Ideally, this E2E comparison is done with a few very different loudspeakers (to finally show potentially different behavior of the same PA at different loudspeakers).
What is the big advantage of an E2E test: Due to the additional influences of room and transducer (both of which are not error-free either), existing deviations, even small ones that are considered irrelevant, are further amplified under unfavorable circumstances, so that they do become audible at the end of the day.
Tests like this would help shed more light on the matter.
