While working on a trace recently, we came across this quote from the circuit’s designer:

Another key parameter I worked is parts selection. I spent much time comparing the qualities of different types of capacitors and resistors in each position in the circuit and selected the types that gave the best possible tone and response. Many pedals are designed with no regard to the type of parts used, only considering the nominal value (e.g. 2.2uf or 1kOhm). It makes a difference! Sometimes subtle, sometimes epic! I don’t like to leave stones unturned so I investigate this area with every pedal I design in order to achieve the best sound and response I can.

If you check out the photos from the tracing journal, you’ll notice two carbon comp resistors and several different types of capacitors: orange drop, polyester film, MLCC (ceramic), and one that appears to be polypropylene, alongside standard electrolytics.

In the analog audio world, particularly guitar electronics, there’s a lot of mythology around the physical composition of components. You’ll hear people describe certain types of capacitors as having “clear highs”, or certain resistors as being “warmer”.

This is usually subjective and almost always anecdotal. What they really mean is: when I switched to this type, it changed the sound. Let’s dig into this.

Dielectric differences

Time and time again, countless scientific tests have shown that the dielectric or physical substance usually doesn’t make a difference except in specific, measurable areas of performance. Here are several:

• Some types of resistors have lower thermal noise (e.g. metal film)
• Some types of resistors maintain a more stable resistance across changes in ambient temperature changes, called “ppm/°C”
• Certain types of capacitors may have more measurable harmonic distortion (e.g. tantalum or poor-quality ceramic)
• Some dielectrics or compositions allow for lower tolerances (e.g. polystyrene) or better equivalent series resistance (e.g. tantalum)
• More expensive ceramic capacitors (e.g. NP0/C0G) are less susceptible to microphonics
• Poor-quality ceramic capacitors may have nonlinear characteristics (different capacitance under different voltages, temperatures, or other conditions)
• Some types of components degrade over time (e.g. electrolytic capacitors or carbon composition resistors)

There are certainly pros and cons of each type of material, otherwise there wouldn’t be so many different types—but these properties can all be described in datasheet terms.

Beyond these specific characteristics, generally speaking: in low-voltage audio-frequency circuits, if good-quality resistors or capacitors have the same actual (measured, not nominal) value, electrons will pass through them in the same way.^[1]

Confounding variables and controls

One of the most basic principles of scientific experimentation is the concept of the confounding variable. When setting up an experiment, you must identify any external factors that could influence the results and include a control for each of these variables, which means a method of negating their influence. Any conclusion that does not control for a confounding variable is considered useless.

So, if you’re evaluating different types of resistors or capacitors—or just reading an account from someone who did—here is the pass/fail question: Was each individual component measured before comparing them? If the answer is anything but “yes”, then we’re beyond the realm of science and strictly in anecdote. The actual values of the components are a massive confounding variable because they offer a very compelling alternative explanation for why two components sound different.

Actual vs. nominal values

Let’s go back to the use of carbon comp resistors in the pedal we traced. The two carbon comp resistors were nominally 15k and 150k, with a 5% tolerance. But the 15k actually measured 17.1k (+14%) and the 150k measured 159.8% (+6.5%).

If these were auditioned alongside 5% carbon film resistors of the same nominal value, they would not resist the flow of electrons by the same amount. The carbon film resistor is guaranteed to be between 14.25k and 15.75k and does not exhibit any drift over time. But the carbon comp resistor is already well outside the spec, and in another 20 years it will likely have drifted even farther, as high as 25k.

It’s an obvious point, but one that doesn’t seem to get near enough attention from people who should know better: The electrons don’t care about the color code painted on the side, only the actual value in raw ohms. If you like the way the carbon comp resistor sounds in your circuit, the next step is to measure it and find the closest precision metal film type available, and then perform the comparison between those. You’ll most likely find that you were hearing the ohms, not the material.

Real-world component tolerances

There’s also another angle on this: if you read our article on component tolerances, you’ll remember that in practice, the tolerance doesn’t usually have an even distribution.

If, for example, you tested a batch of 1,000 polyester film capacitors from a certain brand, nominally 100n with a +/-5% tolerance, you may find that most of them are between 102n and 105n and only a few that test below 100n. If you tested the same number of a different type such as polypropylene, you may find that they’re mostly 96n to 98n. Both are within 5% of 100n, but the specific types often have a much narrower range than you would expect.

So while it’s mostly a myth that capacitor dielectrics sound different, it can still be true that two types of capacitors consistently produce different results, and this conclusion could certainly be supported by a full double-blind listening test or via spectral analysis using expensive test equipment.

If you had a breadboard and were swapping back and forth between different types of capacitors, you might conclude that you preferred the polyester film. You might say the polystyrene is a bit too hi-fi and sterile sounding and doesn’t have enough warmth. But are you absolutely certain that what you’re hearing is the dielectric (polystyrene vs. polyester) and not the capacitance?

Unless the two capacitors both measure the same actual value, then you can’t know this for sure. Depending on where it’s used in the circuit, the 102n cap might allow more midrange and bass frequencies than a 98n, and this is generally associated with warmer tone. As a result, you may conclude that polyester film sounds different when it’s really just a matter of raw capacitance.^[2]

Therefore, you should avoid inferences like “I prefer polyester capacitors” or the more universal “polyester capacitors have these characteristics” when what you really mean is “the polyester had a more preferable capacitance range in my application”. It’s not a waste of money to use more expensive parts, at least within reason. They probably do sound different due to the distribution of actual capacitance values within the batch. Just don’t fall into the trap of assigning non-scientific characteristics to components that behave in a very scientific way.

Conclusions

If you take away one thing from this article, make it this: any comparison between components that does not control for the measured value of the components should be ignored. In our experience, this is nearly all of them. We are naturally susceptible to absorbing bad conclusions from forums, YouTube videos, or even major guitar magazines^[3] and then mistaking this for aphorism or data.

A few final words as a postscript: in this article, we’re speaking primarily of passive components like resistors and capacitors. Once you get into semiconductors like transistors or diodes, the question of physical materials (e.g. silicon vs. germanium) does become very important and makes a huge difference—particularly in how they handle signals outside their limits, which produces clipping.

We’re also specifically talking about pedals and other low-voltage applications—not amplifiers or hi-fi equipment, and not fully passive circuits like guitars. Most of the above still applies across these different fields, but there are some exceptions. For example, carbon-comp resistors do exhibit some measurable distortion effects at high voltages, and at certain positions in a 400V tube amplifier they may not in fact pass electrons the same as a different type of resistor of the same value. Whether this produces a noticeable difference in sound is up for debate, but it is real.

Addendum: Echoic memory

Measured values are not the only confounding variable when comparing component types—in our experience, they’re just the most overlooked, and so it’s what we focused on for the majority of this article. But another significant one is called echoic memory, which is our ability to accurately remember sounds after hearing them.

Numerous studies have shown that actual sound is only retained in our brains for about 3 or 4 seconds on average, after which it gets processed into memory—pitch, timbre, and feelings associated with it. As with most inner workings of the brain, we don’t realize this is happening. But in practice, it means we can’t accurately compare the subtleties of two different sounds unless we hear them less than four seconds apart.

So when you read accounts of component comparisons, aside from the earlier question about the measured vs. nominal value, you should also ask: How long does it take to change out a capacitor in a Stratocaster? What about in an amp or a pedal? Even with sockets or a breadboard, it’s hard to make a swap in four seconds.

But if you’re aware of how echoic memory works, you can easily compensate for it. The simplest way is to put each component on a toggle switch and toggle between them with a continuous signal playing—either from another person playing the guitar or from a pre-recorded phrase, e.g. using a looper. It’s a bit of work to set up, but it defeats the time-based limitation of echoic memory and allows for reliable (though still subjective) listening comparisons.

The more involved but far more accurate way is to use test equipment for frequency analysis. Generally, a swept-frequency test signal is output by the equipment itself, and then it measures the differential between input and output so that you can see what change is being made to the signal across the spectrum. By capturing the frequency plots under various conditions, you can actually visualize the differences.

The limitation of this method is that you can only compare one snapshot in time to another. You can see changes in frequency response, but you can’t easily capture time-based qualities like sustain or attack. With passive components like resistors and capacitors, this isn’t generally necessary, but once you get into semiconductors or tweaking full circuits, things get more complicated.

Further reading

Cyril Bateman’s Capacitor Sound articles: Several different types of tests, disproving many myths about all different capacitor types. However, he is mostly concerned with hi-fi and measuring harmonic distortion, which are for the most part undetectable in guitar audio due to the amount of intentional distortion that we use.

Rod Elliott’s Capacitor Characteristics article: Again, not guitar-specific, but very in-depth. The section on ceramic capacitors is particularly interesting and explanatory if you want to know why there’s so much derision toward low-quality ceramics like X7R.

1. We tried to word this very carefully. It doesn’t cover every single fringe case, but it’s much closer to the truth than the pervasive belief that dielectrics have different audio characteristics. ↩
2. This makes it especially difficult to compare capacitors, because you may not be able to find two 100n polyester and polystyrene capacitors that measure the same value if their distributions. ↩
3. We came across this article awhile back which has the following quote from Dirk Wacker, speaking about guitar wiring: “Some customers report they get the best results with NOS Allen-Bradley carbon composition (CC) resistors. These resistors have less background noise and were the ‘gold standard’ back in the ’50s and ’60s for all Fender amps—not a shabby point of reference!” CC resistors absolutely do not have less background noise than modern carbon film or metal film and were only used because there was no other practical option available. Manufacturers switched over to film resistors as soon as they could. ↩