Sound differences between cap brands? Test results?

[Steve Conner]

I'm going to wash out my ears with C37 lacquer and try again tonight!

[Enzo]

Politics aside, daz, I won't argue that you do or do not hear/feel some difference in caps. If you think you do, then you do. But when you talk of mounting mallory caps, then sozo caps and back and forth, I have to ask, did you MEASURE each cap to be sure the Sozo and Mallory caps at each position had the EXACT same capacitance? I don't care if the 0.022 wound up being 0.019 as long as BOTH brands were 0.019. Because if the Sozo cap measures 0.025 and the Mallory measures 0.019, that is going to affect what you hear a lot more than brands.

[Jazz P Bass]

Also, I'm sure that even in the CERN Large Hadron Collider, there are capacitors in the signal path of some of their instrumentation. I wonder what they think about this mysterious, immeasurable mojo that some capacitors have?

CERN:"We think we have located the Higgs Boson.
But to be sure we have to change some capacitors in the detector"

[g1]

I have to ask, did you MEASURE each cap to be sure the Sozo and Mallory caps at each position had the EXACT same capacitance? I don't care if the 0.022 wound up being 0.019 as long as BOTH brands were 0.019. Because if the Sozo cap measures 0.025 and the Mallory measures 0.019, that is going to affect what you hear a lot more than brands.

I'd like to know how much this is discounted by those who have much better ears than I. What tolerance differences are they allowing when the caps are said to be "of equal value"?
Daz: sorry if you covered this earlier but do you check capacitance and ESR, and if so, what tolerance do you allow?
It would make an interesting twist to the experiments to also include caps of the same brand/type but with 10 or 20% difference in value, not letting the subject know when the change is type and when the change is value only.

[kleuck]

When i compare, caps are under 2%, and when talking coupling caps, the difference in BP is only on the lows, where you have to use very different caps to hear a difference, so the differences in sound do not lie in the tolerance (why would you hear difference in brightness in this case ?).
BTW, modern caps (i include the NOS Soviet PIO i use in my axes) do have very low tolerance, not 10 or 20 %, but most of the time under 2, and actually under 1,5 %

[Steve Conner]

I agree, changes in capacitance shouldn't cause changes in the upper midrange and treble. So the question is, what does cause these changes? How can the capacitor dielectric affect the sound at high frequencies where the capacitor is basically a short circuit and passes the signal unchanged?

I can't think of any scientific explanation, and I haven't seen anyone else come up with one either.

[markr14850]

There's some interesting stuff here: The "Sound" of Capacitors

[defaced]

I agree, changes in capacitance shouldn't cause changes in the upper midrange and treble. So the question is, what does cause these changes? How can the capacitor dielectric affect the sound at high frequencies where the capacitor is basically a short circuit and passes the signal unchanged?

I can't think of any scientific explanation, and I haven't seen anyone else come up with one either.

I have no explanation, but I will propose a test:

I think an interesting test would be to do something similar to how they proved the Higgs Boson was discovered. The best I understand the method, you take many many readings and look for statically significant differences (that would probably otherwise be considered outliers) and compile them. If the readings consistently appear at the same point, the artifact is real. Conceptually it seems pretty straightforward, hell it might even have a proper name (I stay far away from statistics), but I haven't the foggiest idea of exactly how to perform the experiment.

Ref: CERN announces Higgs boson discovery: 'I think we have it' | DVICE

Thanks to exclamationmark for mentioning CERN.

[Enzo]

That might work, but you have to bombard the caps with protons moving at 0.999 the speed of light. Buried under the Alps.

[Steve Conner]

Mike, that's exactly what I'm trying to set up with my blind test. If someone gets 3 out of the 5 clips right, that is a fairly compelling demonstration, there are odds of 60:1 that they got it right by chance. The ABX comparator plugin I referred to earlier will keep track of your score and calculate the statistical significance as you go along.

Unfortunately every time one of these tests is proposed, the people who claim to be able to hear capacitor sound vanish in a squid-ink-like cloud of excuses and insults. The same seems to have happened this time.

Maybe the caps are too hard, but surely someone can identify the NOS preamp tubes?

PS: At the Hydrogen Audio forum they ban any subjective discussion that isn't backed up by ABX test results.

[exclamationmark]

The only reason I could think of that some capacitors might sound different in the higher registers is harmonic distortion. The distortion of most capacitors (espescially high k ceramics) increases when a bias voltage is applied. I would also imagine that distortion increases with increasing signal amplitude. As far as I know, no one has actually measured distortion numbers when you have a bias voltage of 150v+, with a signal amplitude of a hundred volts or more. Usually higher voltage capacitors have lower distortion numbers compared to ones rated at lower voltages, so it might turn out that the increase isn't that great (all the tests I have seen done only test up to about 3v RMS on 50v and lower rated capacitors) . Regardless of this, the THD of your average film capacitor tested at 3 volts is somewhere in the range of 0.001% and lower. Put another way, that's one part per million. Though as I said, who knows what happens when you stick them in a guitar amp.

C. Bateman's "capacitor sound" article series is a good read if you want actual data on these sorts of things. Also just for fun, my guesses for the ABX test are: (highlight to see) 1. NOS mustard, 2. Sozo vintage, 3. Sozo Standard, 4. Mallory, 5. Nos tubes. My curiosity got the better of me, so I opened up the files to try and match their waveform - However, they all looked slightly different to the original, so either it's been slammed by some sort of audio codec or there is some funny business going on

[Jazz P Bass]

Also, there is a point that seems to be missing in this discussion, exactly what is happening in the brain when it is said that we 'like' something.
There is a chemical flood going on constantly & it changes in a certain way when we percieve "likeing' something or not.
And that in itself has to be different from individual to individual.

[Steve Conner]

Mark, the segments that I cut out weren't processed in any way. I made them with Sound Forge and I'm pretty sure this is bit-perfect from previous experience. Also, being WAV files they are uncompressed.

If they look different, it's probably because of how your audio software displays the waveform. When the display is zoomed out, it has to map several samples to one pixel and it won't necessarily always choose the same samples.

Since it is a two-terminal device, a capacitor can only distort the voltage that appears across it, not the voltage to ground. So coupling capacitors shouldn't generate distortion at high frequencies, where the signal voltage across them is negligible. Every graph I ever saw of capacitor distortion is worse at low frequencies, and falls off at least 6dB/octave. So, my suspects for capacitor distortion are undersized coupling caps, undersized cathode bypass caps, and the caps in the tone stack, particularly the treble one as it sees a lot of signal voltage in the midrange.

[Jazz P Bass]

There is an awful lot of technical as well as historical info in this Wiki article as to the differences in film caps.
Link: Film capacitor - Wikipedia, the free encyclopedia.

[Steve Conner]

Thanks to those of you who have had a go at identifying the clips. I'm not going to say anything about the results just yet because it would spoil the test for everyone else.