different type caps sound different?

[R.G.]

... I tired several of the same ones

Forgive me for ruffling your feathers or even giving you the impression I doubted your word. I was too tired last night to be thinking straight.

I should have simply asked the pertinent question. Can you please tell me the value and brands of the capacitors in question? I'd like to obtain samples of each brand and find out what makes them different.

Even better, would you sell me two each pretested "good" and "bad" caps?

[R.G.]

Let's not completely write out the possibility that the "bad sounding" Silver Mica cap had failed...
...
And again: it tested just fine
Silver Mica are notoriously unreliable, the silver starts migrating through the mica under the electromagnetic forces. After a while, it creates small paths where tiny little arcs form randomly under high DC. It will test just fine, and it's not a completely open circuit either, but it just doesn't work the same anymore.
I haven't had any problem since switching to the Cornell Dubilier ones

Very interesting. I didn't know that about silver mica in particular, since I try never to use them. That may not be the mechanism at work here, but dang, it's a good plausible explanation. I should have asked some of my older DIY ham radio friends. I suspect they know this about silver mica backwards and forwards.

It satisfies the symptoms: brand is different, which is the external symptom; value is nominally the same; tests OK at low voltages, but turns leaky and noisy at high voltages. The DC leakage could well cause bias shifts which would cause ugly distortion.

But the underlying cause is not simply that it's a different brand. It's the high voltage leakage mechanism. And it neatly sidesteps the issue which the original poster asked about and which has been almost lost by now.

[daz]

Yes, similar to what someone suggested the area where it was used IS indeed a area thats very very sensitive to whatever i put there. But that said i tried the same thing at a second filter like this one further upstream and it did the exact same sort of thing except it didn't become horribly bright and harsh. It became much much brighter with one and much softer with the other. So they are still in fact very different, but the change from great to horrible in the amp was due to these components being used in a area thats been uber sensitive to any tweak i do there. but never the less the original question was about differences between caps, and there is a huge one here in any case.

Can you please tell me the value and brands of the capacitors in question? I'd like to obtain samples of each brand and find out what makes them different

No, i see no brand on them, at least as far as i can tell. But one was the typical ones you get from most any supplier online and one i got locally. heres a pic. the one on the right is the soft sounding one i got locally.

http://i24.photobucket.com/albums/c39/dazco/cap.jpg

[beefwhiff]

Why can't this be proven?

How 'bout building a test circuit that we can plug-in various capacitors and sweep through the audio spectrum for frequency response? Maybe a signal generator as the source and spectrum analyzer w/ peak hold function to view the frequency response? Or a sweep generator/comparator and an oscilloscope using two channels with XY function?

Seeing is believing.

[Steve Conner]

It's already been proven to my satisfaction, and I'm a practicing design engineer who has to specify different capacitors for different applications every day of the week.

A capacitor doesn't have a frequency response, it has an impedance curve represented by the equation

Xc = j/(2*pi*f*C)

and the frequency responses of circuits containing capacitors follow from that by well-known laws that don't need testing.

Capacitors have other parameters that make them deviate slightly from this ideal curve: ESR, ESL, tan delta, dielectric absorption, voltage coefficient of capacitance, and so on. But I believe that none of them has an audible effect in guitar or hi-fi amp circuits, except possibly the voltage coefficient of ceramics, and the ESR of 40 year old tired electrolytics.

Somewhere there's a website with XY scope plots showing the performance of various capacitor dielectrics. The electrostatic equivalent of the B-H loop of a transformer, so a D-E loop or whatever. The only one that shows non-linearity is ceramic.

Audiophiles have made test rigs that claim to show differences between capacitors, but I believe these show up ESR, ESL, tan delta and dielectric absorption as I mentioned above. These four are linear effects, which is to say that they just modify the frequency response of a capacitor-containing circuit very slightly, they don't actually distort anything. They open the D-E plot out from a straight line into an ellipse, but that isn't distortion. The ceramics warp it into an S shape.

I tend to agree with Hardtailed, it's likely that the bad sounding silver mica capacitor that Daz found is a dud.

[uvacom]

Better yet, why doesn't anybody record these substantial before/after differences? I think the best way to do it would be to get a good session player and have him/her play a few different licks right into a DI and then directly into a high-quality ADC stage.

Then, using a high-quality DAC, play that through a re-amp box into the amp, and record the output of that, mic'ed up.

Then, being very careful to change *only* a capacitor with a different type (but as close to nominal value as feasibly possible, probably by hand-matching from a larger batch) but leaving the amp/mic position exactly the same, probably by using a switch to change out the cap, record that output.

Now, if there is a true sonic difference then that should be measurable through a simple null test. Also, if the person taking the audio recordings doesn't know which cap was used with each recording (say they were only allowed to operate the switch but did not wire the switch and cannot tell what is being switched), we have a decent start for a true double-blind test.

[Hardtailed]

No, i see no brand on them, at least as far as i can tell. But one was the typical ones you get from most any supplier online and one i got locally. heres a pic. the one on the right is the soft sounding one i got locally.

http://i24.photobucket.com/albums/c39/dazco/cap.jpg

That's exactly the same that went bad in my amp. It's what you get from AES or Tube Depot.

In my experience though, it sounded exactly the same as the Cornell Dubilier part, until it went leaky. I still have the no-names as bright caps and have no issues with them (they don't block high-voltage DC there).

The replacements look almost identical to the no-names, but it says "CD" on them. I also have some brown no-names, which have been reliable so far.

[Steve Conner]

Fine, send me your capacitors folks, and I'll do it, and you can have 24-bit FLAC files to download and do your ABX tests.

I already know what the answer's going to be, though

Golden ear answer: The imperfections of the digital equipment* swamp all differences in the capacitors. The only valid test is one where I know the expected outcome so I can cheat.

Engineer's answer: There's no point in doing this, it wastes time that could be used for playing guitar too loud and drinking.


*in my case a Focusrite Twintrak with the 24/96 ADC option board

[Hardtailed]

Very interesting. I didn't know that about silver mica in particular, since I try never to use them. That may not be the mechanism at work here, but dang, it's a good plausible explanation. I should have asked some of my older DIY ham radio friends. I suspect they know this about silver mica backwards and forwards.

To be honest, I read that on the internet, so maybe I shouldn't post it as true science.
No matter what the mechanism, the result was there and the proof was found by measuring the treble pot top lug and finding over 100VDC on it.

I try to avoid them too, but the amp I'm working on right now is AC30 inspired, and that has a 47pF cap in the tone stack. Not being able to find exactly what I wanted, I ended up paralleling in a 36pF "good" Silver Mica with a 20pF ceramic. Curious as to how that will work out. The "good" SM might very well go bad in this amp too.

[daz]

it's likely that the bad sounding silver mica capacitor that Daz found is a dud

That wouldn't explain the fact i have a few of these and they all exhibit the same phenomenon. You guys can argue this all you want and to be honest, theory about caps and what they do means nothing here because the fact is, the original post concerned whether caps can sound very different. I don't care what the reason.....if they do for WHATEVER reasons, they CAN sound different ! Do you see what i'm saying here? Maybe there was a manufacturing defect, but the caps still went out and were for sale. so whether or not there is some reason like that, all that matters is the end result. And the end result has been proven to me. that would be that while it may be rare, they can indeed be very different. Doesn't matter what the reason is ! If a cap for sale can cause this without some external reason being at fault, and it's recreatable in every one of them and not just one defective cap from the batch, then the answer to the original question is YES, they *can*. likely or not, rare or common, they can. It's wierd, probably rare, but it can and could be the reason for a small percentage of the people who have heard large differences in caps.

the facts as i know them are that whatever is causing it, it's in the cap itself, not the circuit, and it's not a defective cap in and of itself. maybe a run of bad or poorly made caps is a possibility. But then whether or not thats the case, doesn't still point to the answer being yes, caps CAN be very different? see what i'm saying here? If the end result is yes, and it's not a single cap's fault, then yes IS the answer to the question regardless of why.

Now, you can take the debate from this point on and argue about exactly what it is about the brand/model of cap that caused it, but that doesn't alter the answer to the original question. If you buy THAT brand and model, it WILL do this, and if there are others like that they too will. rare or not.

[Steve Conner]

But then whether or not thats the case, doesn't still point to the answer being yes, caps CAN be very different? see what i'm saying here?

Yes, but they're not supposed to sound different, so if two caps sound different, one of them is broken. It's not a philosophical issue, it's a quality control issue.

It's not unknown for bad batches of components that were rejected by a large-scale buyer to get dumped on hobbyists. Case in point, "NOS" germanium transistors. If you buy on Ebay you get the stinkers that couldn't be sold out the front door because they failed the leakage spec back in 1960-whatever. All the good ones are gone because they got sold, to be built into equipment that became obsolete and was dumped.

[uvacom]

It's already been proven to my satisfaction, and I'm a practicing design engineer who has to specify different capacitors for different applications every day of the week.

A capacitor doesn't have a frequency response, it has an impedance curve represented by the equation

Xc = j/(2*pi*f*C)

and the frequency responses of circuits containing capacitors follow from that by well-known laws that don't need testing.

Capacitors have other parameters that make them deviate slightly from this ideal curve: ESR, ESL, tan delta, dielectric absorption, voltage coefficient of capacitance, and so on. But I believe that none of them has an audible effect in guitar or hi-fi amp circuits, except possibly the voltage coefficient of ceramics, and the ESR of 40 year old tired electrolytics.

Somewhere there's a website with XY scope plots showing the performance of various capacitor dielectrics. The electrostatic equivalent of the B-H loop of a transformer, so a D-E loop or whatever. The only one that shows non-linearity is ceramic.

Audiophiles have made test rigs that claim to show differences between capacitors, but I believe these show up ESR, ESL, tan delta and dielectric absorption as I mentioned above. These four are linear effects, which is to say that they just modify the frequency response of a capacitor-containing circuit very slightly, they don't actually distort anything. They open the D-E plot out from a straight line into an ellipse, but that isn't distortion.

I tend to agree with Hardtailed, it's likely that the bad sounding silver mica capacitor that Daz found is a dud.

I agree with all of this, *but* when we talk about guitar amps for the most part we are using well-defined subcircuits which have specific characteristics, i.e. baxandall tone stacks, long-tailed inverters, etc.

That being the case, there could be in certain common amplifier configurations some tonal differences due to these other secondary characteristics of capacitors other than their capacitance.

What I'm saying is that the typical tube guitar amp designer isn't doing their design from the ground up with a rigid EE approach, taking into consideration a priori the effects of ESR, etc. They're building some permutation of a proven design element and tweaking the coefficients through the values (and sometimes the type/brand) of the various elements often without any particular regard for the pure electrical behavior of the circuit, but rather the change in predefined output parameters of that circuit.

So, enough people claim to hear a difference that it does make sense to perform tests to first confirm whether or not this is the case, and if so, then to perform analysis to find out why. To do the analysis first and conclude there should not be a difference and so write it off is in my mind a fallacy.

On a personal level I think that people are easily fooled, I've fooled myself on a few occasions and I actually think I do have very good ears when it comes to absolute/relative frequency detection. I think most of what people think about capacitors is basically hocus pocus. But I also think that it would be astonishing if we happened to live during that precise threshold where humanity has finally erased all of the serious misconceptions we have about the nature of the physical world and everything we think we know is correct. That is essentially what every generation prior has thought and they have all been proven wrong. That's why it is worth it to challenge the well-proven. You just never know when somebody will discover something contrary to what we all previously thought, which in hindsight was so obvious.

[Steve Conner]

OK, so go look up some capacitor datasheets to get a feeling for typical values of ESR, ESL and tan delta.

Then try to think of some place in some amplifier where those extra milliohms (OK, so some small electrolytics can have a few ohms) and nanohenries could make a few dB of difference to the frequency response.

BTW: Does anyone know these figures for old paper-in-oil caps? I've heard that some of them can develop ESR upwards of 1k as they age, and that would make a difference for sure. They could have lousy dissipation factor too.

[daz]

Yes, but they're not supposed to sound different, so if two caps sound different, one of them is broken. It's not a philosophical issue, it's a quality control issue.

It's not unknown for bad batches of components that were rejected by a large-scale buyer to get dumped on hobbyists. Case in point, "NOS" germanium transistors. If you buy on Ebay you get the stinkers that couldn't be sold out the front door because they failed the leakage spec back in 1960-whatever. All the good ones are gone because they got sold, to be built into equipment that became obsolete and was dumped.

But how about this....say that assuming you believe me and that this CAN happen, what if you were given the opportunity to answer the original question and you were only allowed to say yes or no and not another word beyond that. This is the point i'm making. That for all intents and purposes, if a person wants to know the answer to this question the only correct one is "yes", again assuming you believe me. And I don't think most here truly do because for one they don't know me and for all they know i may be tone deaf, an idiot, susceptible to placebo effect, etc etc etc. But thats no longer the point anyways, and i'm done addressing that because I know the answer and thats all that really matters to me anyways.

[Steve Conner]

what if you were given the opportunity to answer the original question and you were only allowed to say yes or no and not another word beyond that.

I'd answer no, because the question concerns different types of capacitors, and a faulty capacitor isn't a capacitor, it's just a useless blob of something. It goes in the trash can along with the cratered MOSFETs and the dark-emitting diodes.

If "broken" counted as a type of capacitor, then I'd have to answer yes.