Capacitor Sound

[Jazz P Bass]

Here is a link to a very informative group of articles written by a capacitor designer.
He covers the distortion that capacitors create, the test equipment that he "hand built" to facilitate the tests and an overall coverage of each type of capacitor.
Link: Capacitor Sound
Enjoy.

[Steve Conner]

I can't load it for some reason, but I'm guessing that it's the series of articles by Cyril Bateman. Those were great

[pdf64]

Yes, that's the guy, from Electronics World July 2002 - Jan 2003.

[Jazz P Bass]

Capacitor Sound PDF

My, My aren't we well read, Steve.
Yup. Cyril Bateman.
Attached PDF file.
So he states that "distortion" in a capacitor is not something that is even checked for by the capacitor manufacturer.
Huh.
He goes on to say that equipment builders should inspect for this.
And if they do or do not, that should be stated in the specifications.
(and if they do not, he would not trust the stated distortion figures of the equipment)
Imagine trying to repair something that has "sorted" capacitors in it.

[Steve Conner]

Yeah, the science is sound, and the implications are worrying. I read those articles a few years ago, but I kept quiet about it, because on a casual reading they play right into the camp of the golden-eared corksniffing brigade.

From what he says, the non-linear effects in film capacitors come from the end terminations. This is possibly quite relevant to our work: there could be some interesting distortion effects in those ancient paper-in-oil caps where the leads were simply shoved into the rolled-up electrodes, and maybe that explains some of the "mojo" when they're used as guitar tone caps.

Hopefully good quality film caps such as the Wima ones won't suffer from serious distortion. These caps usually have a specified ESR and pulse rating, and they have to control the quality of the terminations to meet that spec.

I have a distortion analyser, but I've yet to see any distortion on it that I could trace to a capacitor. (To be fair, it's not as sensitive as Bateman's.) Right now the biggest distortion source in my prototype hi-fi amp is the speaker relay contacts. Brand new relay, never been arced, and already they're starting to cause trouble.

So there's another hypothesis for the tube sound pot: maybe tube amps sound better because they don't have speaker relays.

[diagrammatiks]

there's a lot going on...

scientifically there are differences...you can plainly see it on a scope...

doesn't always translate to sonic differences though...

but i think the key is trying to determine how much of a difference something has to make in order to be important.

[km6xz]

There is a safe bet than it matters not a bit in a guitar amp since the noise floor is 30 db higher than any but the worst(defective?) offenders. His work however vindicates to a large extent the use of electrolytic caps where such values are needed, showing much better fidelity than claimed by the golden ear crowd.
As has been mentioned before, none of this really matters in music production amplifiers but does in music re-production amplifiers since the former has no reference and the latter has to conform to a readily available reference, where there is something to be faithful to, as in fidelity.

[diagrammatiks]

i think there's a lot to be said for the notion of fidelity in guitar amps.

we're far from the era where you just turned an amp all the way up with a wish and a prayer.

[Steve Conner]

Yep, nowadays we turn it all the way up with an evil cackle instead.

[km6xz]

How is that?
The whole purpose of the a guitar amp is to have low fidelity because the source is not a desired final result. Magnetic pickup, cables, playing styles and the low fidelity characteristics of the amplifier combine to make a musical sounding tone. A highly faithful amp will sound anything but musical since most of the transducer signal is the fundamental by a wide margin, with few of the brain expected harmonics that offer the clues needed to perceive it as music.
It there a new fad of only fundamental frequencies in guitar amps that I missed?
That is why the hi-fi purist myths are so dumb sounding when repeated by guitar amp builders, the most revered classic amps were particularly lo-fi, which in this case is good.
Once the guitar, player, amp, speaker cab, acoustic space combination producing a listenable musical tone, fidelity certainly does become important in the re-producing the listenable sound.

[Rhodesplyr]

scientifically there are differences...you can plainly see it on a scope...

doesn't always translate to sonic differences though...

but i think the key is trying to determine how much of a difference something has to make in order to be important.

The position I've arrived at after years spent as a hobbyist, experimenting with various capacitors, is that a critical part of this discussion is the interaction of a particular capacitor with the surrounding circuitry. There are certainly circuits where you can hear a difference by swapping in different types of capacitors, and there are also circuits where it seems not to make the slightest difference whether you use Z5U ceramics or film & foil polypropylene.

As an example, when you have a capacitor that functions as a high-pass filter in a crossover in series with a high-frequency driver, you can almost always hear differences between capacitor types.

[J M Fahey]

Well, *I* canīt.
Iīm not sensitive to 0.00074% distortion, sorry.
Check Fig 1 , (printed) page 48.
And thatīs a "bad" one, a vile Class 2 ceramic. (ugh !)
So bad, in fact, that its distortion rises 24X (yes , 24 times, amazing) when biased with 18V DC, which is a reasonable DC voltage it may easily find in the real world.
And how much does this despicable monster distort now?
An unbearable 0.0205% !!!! [Fig. 2]
Please, call the paramedics !! Iīm having a heart attack !!!
Pleeeeease *read* Batemanīs actual measurements.
As in [Fig 5] , *still* a vile ceramic, although now Class 1.
Even with 18V bias it produces 0.00009% (0.00006% without).
Oh well.

[Wombaticus]

Well, *I* canīt.
Iīm not sensitive to 0.00074% distortion, sorry.
Check Fig 1 , (printed) page 48.
And thatīs a "bad" one, a vile Class 2 ceramic. (ugh !)
So bad, in fact, that its distortion rises 24X (yes , 24 times, amazing) when biased with 18V DC, which is a reasonable DC voltage it may easily find in the real world.
And how much does this despicable monster distort now?
An unbearable 0.0205% !!!! [Fig. 2]
Please, call the paramedics !! Iīm having a heart attack !!!
Pleeeeease *read* Batemanīs actual measurements.
As in [Fig 5] , *still* a vile ceramic, although now Class 1.
Even with 18V bias it produces 0.00009% (0.00006% without).
Oh well.

Yes, I'm with you J M. There's a worthwhile followup to Bateman's results in Doug Self's (highly recommended) book "small signal audio design". The bottom line for us in the MI amp world is that unless you are going for super- exacting hifi, you can pretty much forget about caps as a significant distortion source. It's orders of magnitude less than other sources in any MI amp. Small Signal Audio Design - Douglas Self - Google Books

[J M Fahey]

When I started studying Engineering, (a loooong time ago), the *first* thing we were taught in "Física 1" (similar to Physics 101 elsewhere) was "Error Theory", which in a nutshell helped us decide what *realistic* precision we should aim for when measuring, and also that we should not cheat ourselves with Math results full of decimal places, when the original data was not *that* precise.
The first experiment we made, measuring the Electron charge by aligning the Cathode Ray Tube (Oscilloscope Tube) first along and then perpendicular to the Earthīs North/South magnetic field, and then measuring how far the bright dot position changed, was limited by the smallest division on the plastic ruler we used to measure it (about 1 mm).
A real humbling experience.
We had carefully calculated constants with lots of decimal places, but our measurements had a built-in unsurmountable error of, say, 1 mm in a 12mm measurement, about 8%.
Yet the experiment was valid , just that we had to realize that on real world everyday problem solving, we had to be practical.
I guess thatīs the difference between a good Engineer and a good Physic.

[diagrammatiks]

most of those results are referenced to hi-fi because that's where the bulk of research/hype/heresy in this matter is.

fidelity does not imply high fidelity nor is low fidelity simply the absence of a fidelity standard.

and of course you are right...how well the circuit reproduces a fundmental is not relevant to us.

as such neither are those results. Look at the scopes of plate bypass capacitors on a triode stage expected to pass wide band harmonics with split second transient response changes.

look at cathode bypass capacitors connected to triode stages expected to undergo heavy asymmetrical clipping. compare the differences in capacitors between saturation and cut-off.

the circuit and the activity of the circuit is crucial for making any sorts of worthwhile judgments.