Jim G wrote:
I don't know the crossover details either. I do remember assembling several brands of film capacitors @ 4.7uf. None actually measured 4.7uf however. There were easily heard differnces between the capacitors with slightly different values. Only 2 brands measured close enough to double blind test. I think they were approximately 4.4uf and within 1% of each other. No one was able to correctly identify the difference. I imagine few actually measure the competing capacitor brands they are comparing in this type of application to less than 1% and most of the differences heard are the varying values.
To be fair, the system was set up pretty hastily and not set up for optimum performance before starting. Hell, that could take weeks/months and require significant room treatment. I believe it may be possible there could be a difference between capacitors in spacial or atmospheric characteristics had the system (including the room) been capable of showing those characteristics. We really were only able to compare basic, but critical characteristics like toneality, and timbre.
Jim,
The tolerance of capacitance does no not account for audible differences.
Assuming a 2nd order Butterworth network was used, 4.7uF capacitor equates to 3000Hz crossover frequency. With a crossover frequency that high, it is most likely a 3 way system.
Let us look at few cap values within +/-5% tolerance
4.7uF gives 3000Hz crossover at -3dB
4.5uF gives 3072Hz crossover at -3.2dB (-5%)
4.9uF gives 2922Hz crossover at -2.78dB (+5%)
Suppose, we have two crossovers the caps of same make and mode at extreme ends of the tolerance, 4.5uF and 4.9uF, the difference in crossover is only 150Hz and .45dB at 3000Hz. I doubt any human ear can consistently detect that small difference of the signal crossover assuming neither of the drivers are rolling off near the crossover frequency. This is very important to understand because we are talking of same model and same make caps, so any differences we expect to hear should be strictly due to the crossover frequency and slope, both of which are fairly constant.
Secondly,
Not all capacitors sound drastically different, irrespective of their cost. If the caps in the test are "similar" sounding, it would be hard to detect the difference in double blind testing (thats another subject which I will try to address shortly).
Moreover we need to know exact crossover topology and the caps changed. In our example, if only the HF capacitor is changed, frequencies below 3kHz remain fairly unaltered which makes it extremely difficult to detect the sonic differences of the capacitor of interest.
IOW, there are more questions about this test than information to postulate any inferences.