Listening tests of high end capacitors...

[JGundry]

If you are going to do this test properly make sure the foil caps have the outer foil oriented the same to ground. The outer foil end used to be marked on foil caps but no longer is. Orient it properly and the signal flow will be shielded better and will reduce noise. Also there is a slight difference in resistance with orientation which in this test could affect treble bleed.

Here is a simple way to test it.

[Ronsonic]

I have no idea whether voltage rating means anything at all in the case of passive treble cut capacitors, but the only way to be fair in a test such as I am proposing would be to attempt to use the same rating for every cap and test the caps so they are within + or - 2% or better.

I don't think we need to be "fair." What's fair to an inanimate object.

I would think the objective would be to determine if there are indeed differences in the sound produced using different caps outside what we would expect from the standard measured characteristics and whether any of those differences could be reasonably considered an improvement.

Just saying that different physical sizes and voltage ratings should be fair game in such a test, instead of controlling just for construction. We just care if there's an improvement to be had and if any greater cost is necessary to achieve it if so.

[Dave Kerr]

While I'm very interested in the results of your testing, I'm equally interested in whether you or your client sees a risk in some reviewer noticing the cap swap and goes on a diatribe along the lines of "Turner's doing to Henman what CBS did to Fender". Even if your trial is well designed and the guy who can tell the difference between battery brands powering a drive pedal can't perceive a difference between the caps, that won't stop a guitar mag from performing a sloppy test and declaring that you're sacrificing tone for a few bucks/unit.

[Rick Turner]

I'm well used to unfair comparisons, for instance between my acoustic electric basses and those of Rob Allen (whose work I do admire). In a good eight on-line comparisons, what I see is that people are actually comparing the strings that I use (Thomastik Infeld Acousticores) with those that Rob uses (LaBella black nylon tape wound), and they ascribe the tonal difference to the basses, not the strings. Obviously the only proper way to compare the basses would be with the same type of strings on each.

There's a lot of sloppy "testing" going on both on-line (really bad) and in magazines (I think better but not perfect). The hearsay and mythology that I see on forums is just mindblowing, and that includes both electric and acoustic forums. The consistently best lutherie information I see on-line is over at the Mandolin Cafe where practically every top mandolin maker and repair tech in the US and beyond participates. There is a much higher rate of professional involvement there than on any other forum devoted to instruments that I visit. Maybe there's the equivalent in the bowed instrument world, but not in fretted. If you're at all interested in acoustic instruments, check it out.

BTW, JGundry, I appreciate the tip on the foil orientation test. Shall do when it comes time. Thanks.

[dcoyle]

First, I am not trying to make any point about what the performance of any cap is, I am trying to explore the underlying thinking of what has an effect and what doesn't.

I question the designation of being in the "signal path" , or not, as being a somewhat arbitrary designation, compared to being in the "circuit". My concern is that the idea that a device is "out of the signal path" might lead one to discount something that matters.

The term has a history of being misapplied, IMO. Back in the day, there was a lot of advertising copy about this or that component, then out of favor, being used, but "not in the signal path". I find that unless a device has an input impedance many, many times higher than the Z of the load, it is affecting the signal.

If I have a signal and do something to it, by putting it through a device and passing on the signal to the load, or by using a device to bleed off some of the signal and send on the rest to the load, I have affected the signal. If the device is in series or in shunt, its sonic signature is there. Shunts are in the signal path.

If you define distortion as deviation from ideal behavior, then a cap which has inductance, resistance and dielectric absorption, as all real world caps have, does something other than present the simple first order rising impedance of an ideal cap.

There are two types of distortion, things added and things taken away. If a device is in series, the distortions it adds are delivered to the next stage, along with the signal as modified by the input impedance of the device. If the device is in shunt, depending on the input impedance of the next stage and the input impedance of the shunt, signal passes on, but the distortion products of the device are not added, they are shunted to ground. Either way, what ever effect the device has is there. Whether or not it is perceptible or matters is what you are trying to discern, it seems to me.

The distinction of "active" and "passive" circuits may be thought of as speaking about how zesty the signal strength (source impedance) is when facing the input impedance of the load, it doesn't really change what the effects of the load are. I wonder if your observation of the effects of caps being less obvious in passive circuits than active ones does not speak to the qualities of the cap, but to the nature of the circuit.

As a side note, I have also spent time with dc coupled circuits, and find them to be great for HiFi, but not in musical instrument amps, because guitarists don't want to go that low and the ability to go low in anything but a recording situation for a bass quickly gets out of hand with over excursion.

Dan

[salvarsan]

First, I am not trying to make any point about what the performance of any cap is, I am trying to explore the underlying thinking of what has an effect and what doesn't.

I question the designation of being in the "signal path" , or not, as being a somewhat arbitrary designation, compared to being in the "circuit". My concern is that the idea that a device is "out of the signal path" might lead one to discount something that matters.

The treble bleed cap is a shunt impedance across the pickup coil and the signal that it passes contributes to the electrical ground of the signal path.

Even if the guitar output were implemented as a differential signal, the shunted treble would still be subtracted out at the first input stage.

For a collection of near equal-valued capacitors, the audible differences will be insignificant provided that the caps are not pathologically microphonic (viz., high-K discs) or leaky (bruised or NP electrolytic).

Any listening test of capacitors in treble cut circuits will be ones of listening for information absence -- people will be looking for something that isn't there.

Personally, I would find that difficult.

[Rick Turner]

Well, the opinions are going about as I'd expect! Equal measures of "no difference" and "of course I hear a difference." And frankly, the difference between the two camps seems to be a function of the rigor of the tests and listening trials performed.

It may be a few months before I can gather my test results, but I shall report in when I can on all of this.

[David Schwab]

Any listening test of capacitors in treble cut circuits will be ones of listening for information absence -- people will be looking for something that isn't there.

Well said! I can hear it now "the absence of treble with cap "A" is more vacant than with cap "B". Surely the absence is more pure."

Or how black is black?

Now how about ESR? Does that matter at all in guitar tone circuits? (I think ESL only affects very high frequencies)

[Mike Sulzer]

I question the designation of being in the "signal path" , or not, as being a somewhat arbitrary designation, compared to being in the "circuit". My concern is that the idea that a device is "out of the signal path" might lead one to discount something that matters.

Any differences in sound have to be due to non-linear effects. Effects due to loss in the capacitor are very small compared to other losses in the low Q guitar circuit. These capacitors do not have significant inductance. You have a capacitor; if you match capacitances among the different types under test, you are then looking for the sound of harmonics and intermodulation distortion.

Yes, "in the signal path" is a somewhat vague way of indicating how a capacitor affects the sound. Circuit elements are characterized by their V/I curves. For example, Bateman (capsound3.pdf, the most useful of files 1 to 6) puts a sine wave of about 3 volts across a capacitor (or two sine waves) and effectively looks at the current as a response, including the harmonics (or intermodulation products) that are generated. In a similar way, one could put a sinusoidal current through the capacitor and look at harmonics in the voltage across it. In general, a capacitor in a circuit is driven from a finite impedance, and you expect harmonics in both the voltage across it and the current through it.

Then one can use the laws of circuit analysis to see how big the non-linearity is and how much the distortion affects the output of the circuit. For example, consider a 500K resistor in series with a .02 microfarad capacitor with a signal at 1 KHz. The impedance of the capacitor at 1 KHz is about 8K, and so the voltage across the capacitor is a tiny fraction fraction of the three volts that Bateman used in order to see significant distortion in some capacitors. Remember, distortion is non-linear; if you reduce the signal to the capacitor by a factor of ten, you expect the relative distortion to drop by at least a factor of one hundred.

The harmonics from the capacitor get back into the guitar output through the 500K resistor. The impedance of the pickup at 1KHz is a lot less than 500K, and so you expect the very small harmonics to be further reduced.

As you reduce the 500K resistor, the capacitor is driven harder and produces more harmonics. But as the resistor approaches zero, the guitar output has that bassy sound which is not a good test signal, lacking the midrange-to-low-high content where human hearing is most sensitive.

[12Bass]

FWIW, I've reliably detected differences between capacitor types when used in the signal path. Polypropylene seems to sound a little bit clearer than some of the other types, as if they are a little more transparent to high frequency detail. So, if it turns out that polypropylene caps are slightly more "clear" in their path to ground, this difference may be audible, though admittedly subtle. By this reasoning, one might expect the "clearer" capacitor types to sound a bit "darker" when used to shunt high frequencies to ground.

[salvarsan]

Now how about ESR? Does that matter at all in guitar tone circuits? (I think ESL only affects very high frequencies)

The ESR's for film caps at the frequencies of interest are similar and interesting if you rigorously model the circuit.

Empirically, ESR is only an issue in very lossy caps such as NP electrolytics...which would appear in a guitar treble cut circuit only by mistake since they aren't available in sub-microfarad values.

My opinion from long experience is that, unless the capacitor has large non-capacitive properties, you won't hear a difference in a passive subtractive circuit.

If the cap is part of an active feedback loop and typically sees voltage excursions larger than your average pickup is capable of, the cap quality may be important.

In simple words, unless the cap is an utter piece of shit, it will sound fine. For the purposes of guitar builders, film caps behave consistently well at the low voltages they experience in the treble cut circuit.

In simpler words:
NP electolytics: BAD
high-K disc ceramics: BAD
Plastic film caps: GOOD

[David Schwab]

In simpler words:

"NP electolytics BAD, smoke good"

[RoadToNever]

Picking Capacitors - Walter G. Jung and Richard Marsh

[salvarsan]

Capacitor coda

The Jung-Marsh article on capacitor testing, despite its 30 year vintage, is germane.

It made explicit the fact that capacitors are not purely capacitative. Further, more insightful tests demonstrated not only resistance, inductance, and a dissipation factor separate from dielectric loss, but (horrors!) piezoelectricity, an overall hysteresis, and simple microphonics dependent on the dielectric's elasticity or pliability.

You shouldn't ignore that capacitor manufacture has also improved in the intervening 30 years. Once manufacturors saw a market for caps built for a particular use, they tested and built accordingly.

By 2000, ordinary film caps were generally much better than what you could get for a premium in 1980. Specific exceptions like boutique high purity films (teflon, polypropylene) between tin or copper foil conductor have always been available to the monied delusional types.

If you're worried about capacitor quality today, briefly reflect that you never had it so good as today.

Sincerely,
Dr. Pangloss

[Rick Turner]

David, ESL is only a problem if you're trying to read something above grade level in English...

OK, bad pun...

And so far, what I'm reading from those most knowledgeable in electronic theory here would seem to indicate that nobody is likely to hear the difference between a half way decent capacitor and an over the top cap IN THIS APPLICATION...a passive treble bleed low pass RC circuit.