Why would capacitors have an audible characteristic that is not measurable?... I believe there are contributing factors to an audible difference between capacitors that are simply not revealed by electrical measurements.
Differences between capacitors are very easy to measure, as my small series has repeatedly shown - for example, the increased resistive active component (ESR) of electrolytic capacitors, compared to film capacitors.
The equivalent circuit diagram of a capacitor shows us immediately that in reality a capacitor unfortunately not only has capacity, but there are also undesirable effects.
The crucial question is whether these differences play a role in the frequency range of human hearing when using undamaged capacitors in "normal" quality compared to, for example, a high-end capacitor - see part 1
Among other things, in order to exclude possible speculations about undiscovered physical phenomena or "things" not revealed by electrical measurements, the experimental setup was deliberately chosen with a 0.75'' tweeter and a measuring microphone (and the capacitors connected in series).
No one in their right mind would claim that there is an "undiscovered way of air molecule movement" (back and forth from tweeter to the measuring microphone) which comes into play when high-end capacitors are used, but not with normal film capacitors.
Thus, the question is reduced to the sensitivity of the experimental set-up
and whether all parameters/aspects decisive for hearing are covered
by the measurements.
a) sensitivity of the experimental set-up
The sensitivity of the test setup was truly surprising. I would never have thought that such small deviations in the capacitance of the capacitors could be detected.
See the section "Capacitors - low capacitance deviations" in part 1
. Even 0.3% capacitance deviation in the capacitors was detected by the test setup as sound pressure deviation.
And the measured deviation in sound pressure was about 0.02dB. This is much better than the human ear is able to distinguish (at least according to all sources I know).
The measurements of the multitone distortions showed that the differences there were at -60dB or less. The measuring microphone used for the measurements shows even extremely low THD values up to 128dB sound pressure.
So also in this area a high sensitivity of the test setup can be assumed.
b) measured parameters/aspects
The investigations on "normal film capacitor versus high-end capacitor" compared
1) the frequency response curves
2) the impulse responses (the temporal behaviour)
3) the CSD diagrams (decay behaviour of the tweeter)
4) the multitone distortions (which capture HD and IMD)
With capacity differences <<0.5% (in part 1 the capacity deviation was about 0.2%) all measured parameters showed no significant difference.
From the results of the section "Capacitors - low capacitance deviations" in part one, we know that even a 1% capacitance deviation can lead to 0.1 dB sound pressure difference and that this is a deviation that is within the range of what can be perceived by human hearing under certain circumstances.
Further down I will discuss whether these measurements are sufficient to explain possible auditory impressions after a capacitor exchange.
c) psychological effects
had already given an example of this. I can report something similar.
When tuning the crossover, it happens again and again with dozens of tunings that I forget to really implement the changes (whether passive or active) and when switching between the two versions I still think I hear a difference in sound.
However, this aspect of human hearing should not be part of this three part series.
My modifications were fortunate to start with the designers hand drawn crossover schematic. He hand matched drivers and crossover components were selected to achieve specific values. I followed that specific value process during modifications. Every capacitor removed was checked. All were within tolerance and displayed no physical damage. My musically trained ear told me audible differences existed at each step. The Engineer knew how to conduct a valid experiment.
The decisive factor is not whether the capacitors were within their tolerance range, but the relative capacitance deviation of the capacitors among themselves. In order to determine the relative deviations of the exchanged components a cheap LCR-meter or software like Arta-Limp (available in a free version) is sufficient.
In order to rule out changes in sound caused by different capacitance values, the deviation of the replaced capacitors should be less than 1%.
This is damn hard to realize, but anything else quickly leads to 0.2dB and more deviation in the overall frequency response of the speaker.
Here is a small example. If the deviation of a capacitor is 5% (5.3µF replaced by 5.57µF) this leads to a deviation of up to 0.25dB over three octaves for an [email protected]
This deviation will be audible - by direct comparison.
If an electrolytic capacitor is replaced by a film capacitor, the ESR difference is very likely to be significant. This effect also very quickly leads to a 0.2dB difference in sound pressure during replacement, as shown in part two in the section "Comparison with 8.2µF capacity"
Are the measurements used today sufficient to explain auditory impressions?
This question arises not only in connection with an alleged sound change caused by high-end capacitors, but also frequently here in the forum with the sound impression of speakers that show very similar spinorama, angular frequency responses and distortion and yet sound different.
I think many people don't realize how much even the smallest changes in the frequency response of a loudspeaker can affect the sound.
Here again a small example. Below are the horizontal angular frequency responses of a speaker with slightly different crossover tuning.
I'm pretty sure that everyone here in the forum would immediately and without difficulty notice a difference in sound when comparing them directly.
Even significantly smaller changes are easily audible in direct comparison.
The measurements made in the ASR are therefore perfectly sufficient to explain the different auditory impressions of almost identically measuring loudspeakers - to predict the sound only on the basis of the measurements in detail is however very difficult for us humans, a self-learning AI is certainly an advantage.
Therefore, when assessing the reasons for supposedly heard sound differences when replacing capacitors in the crossover of loudspeakers, it is crucial that the capacitance values of the capacitors involved are precisely determined in order to rule out that they are responsible for the different sound impressions (aside from the other influences, such as non-blind testing....)