In audio, as in many other fields of electronics, accuracy and consistency are critical. If a circuit is well-designed, it will have predictable performance from unit to unit, and so the designer needs to be aware of how the individual components will vary across a production quantity of the circuit.

“Tolerance” is defined as the allowable variation in a given specification of an electronic component. It’s almost always expressed as an above/below percentage value, such as +/-10% (plus or minus 10 percent) of the nominal value of the component, e.g. 0.01uF. Since the plus/minus is assumed, it’s often shortened to just “10% tolerance”.

This means that a 100k ohm resistor with a 1% tolerance may measure as high as 101k and as low as 99k. If it measures 101.1k, then it’s considered off-spec and it shouldn’t pass quality control at the factory.

Here are some common tolerances in the type of components we use in pedals:

• Metal film resistors: +/-1%
• Carbon film resistors: +/-5%
• Film capacitors: +/-5% or +/-10%
• Electrolytic capacitors: +/-20%

In general, lower (tighter) tolerance is more expensive because the processes and materials are more expensive.

Note that the over/under percentages don’t have to be the same. For example, many general-purpose ceramic disc capacitors are +80%/-20%. This means that these types are best used in applications where the value needs to be at least a certain value, but higher is OK, such as power filtering.

Tolerances in the real world

That’s how tolerance is defined, but it doesn’t generally pan out this way in practice. The tolerance the maximum allowable deviation beyond which the part would be considered off spec, but with modern manufacturing, often the actual parts are grouped much more closely together. It is not by any means a bell curve with standard distributions.

For example, we once measured a couple hundred 10k carbon film resistors, specified as +/-5% tolerance, and found that every single one of them was within a range of 9.75k to 9.85k. So if the nominal value had been 9.8k instead of 10k, then the true tolerance across the whole group would have been 0.5%, which is better than most precision metal film resistors.

Likewise, if you measured 10,000 0.22uF/10% capacitors, you might not find a single one that is 210n, even though that’s only -4.5%. But you may find 1,000 of them that measure 230n (+4.5%). It’s typically more cost-effective for a factory to develop processes and source materials that always result in a certain tolerance than it is to sort parts into tolerance groups afterwards.

Carbon comp resistors

Carbon composition resistors are a special case. You can read more in our article about audio qualities of different component types, but briefly, this type of resistor is unique in that it exhibits a great deal of drift in value over time. Many of them are specced at 10% tolerance, but this is only at the time of manufacture. They start changing before long—and not by small amounts. We’ve pulled some from 1960s tube amps that measured as much as 70% off of the nominal value. For these, the tolerance rating is essentially worthless in the long term.

Tolerance in active components

Active components such as transistors and JFETs don’t have a tolerance rating in the strictest sense—percentage of deviation from a nominal value—but they do have mininum and maximum ranges for important specifications such as h_FE (gain) or Vgs_OFF. As with passive components, they often measure inside of a much tighter range in practice.^[1]

In fact, some semiconductor manufacturers will make a single physical part and then assign different part numbers afterward depending on how they measure. This is why we have certain transistors with very similar designations, such as the 2N3903/2N3904 and BC549A/B/C. The specs are usually the same except for the gain range.^[2]

Operational amplifiers

Op-amps are a whole other story. Inside the plastic package is a full circuit consisting of transistors, resistors and capacitors (hence the name “integrated circuit” or IC). They’re designed in such a way that the performance of the IC as a whole is very consistent from unit to unit despite the variance in the individual parts inside.

There are still several factors such as noise level or input offset current that can vary. However, the major distinction with op-amps is that most datasheet specs have a “typical” value, which is what you can expect most of them to test at or near. This is very different than just giving a range without any guidance as to where in the range you can expect it to be.

Plus, with op-amps, the types of specs that do vary piece to piece are typically not important for low-level audio work. The actual performance of a modern op-amp is predictable and precise without any biasing or calibration. If you put a 100k resistor in the feedback loop and a 10k from inverting input to Vbias, it’s going to give you a gain of 10. The tolerance of those two external resistors will have a much bigger impact on the circuit than anything going on inside the chip.

Tolerance in circuit design

A circuit designer’s goal is not just to make something that sounds great. It must also be repeatable—and ideally efficient, though inefficiency can be offset by increased cost, as is the case with many hand-made pedals and amplifiers.

Therefore, a well-designed circuit will take all of the above into account. If a circuit only sounds right with selected or audited parts, this makes it very difficult to scale.^[3] It’s much more ideal to include trimmers in critical positions, e.g. transistor bias or clock calibration, so that the circuit can be dialed in after assembly. Still better, though not always possible depending on the circuit, is to design it in such a way that the component variance does not matter.

For example, the Big Muff is an all-transistor circuit, but the transistors are used in a way that the individual gain (h_FE) of each transistor does not impact the level of amplification of each of the four stages. The circuit works with no biasing or resistor selection and pretty much always sounds like a Big Muff no matter what parts EHX was using at the time.

Now, a Big Muff is not a Fuzz Face and is not per se a better circuit than more complex designs. But a big part of its success was in how cheap it was to manufacture. It outsold most of its competitors, and it wasn’t long before Sola Sound, Maxon/Ibanez, and other major manufacturers took notice and released their own versions of the circuit.

It’s in the manufacturer’s best interest to design a circuit that does not require any selection or auditing of parts, having a consistent sound with a broad range of parts. A Big Muff won’t sound much different with transistors in the 150 range or the 400 range for h_FE, because the gain of each stage is set externally. And it won’t sound much different if a production batch of resistors is +2% or -1% of the nominal value.

If precision is absolutely critical, better tolerances can of course be bought. Film capacitors with 5% tolerance are more expensive than those with 10%, and you can find certain types at 2.5% or 1%. If standard 1% metal film resistors aren’t enough, you can buy 0.5% or 0.1%.^[4] But good circuit designers are judicious about where they use more expensive parts. It’s very rare to have an application in guitar audio where that level of precision will truly make a difference, and otherwise it’s a waste of money.

1. For example, the 2N3904 has a maximum h_FE of 300, but we’ve tested hundreds and found that almost all of them are in the 170 to 200 range. We’ve never seen a single one above 220. ↩
2. The semiconductor manufacturing process is much more accurate today than in the 1970s and ’80s when most of these parts were originally developed. They usually don’t leave it up to chance in this way anymore. ↩
3. For many types of vintage effects such as germanium fuzzes, there’s no getting around the scalability problem. It’s the reason that manufacturers like Sola Sound switched to silicon in the early 1970s. ↩
4. Fortunately, metal film resistors have come down significantly in price over the past 20 years and they now cost about the same as 5% carbon film. ↩