As I learn about guitar tone, both in creating and using it with guitars and amplifiers as well as designing circuits to make good tone, I feel at a loss of words in describing what I'm doing. Descriptive words that I often come across when I read about products seem to fall quite short of really making me understand what they're talking about. I hear words like searing and ballsy and scooped and so on. I sort of understand what people mean, and I've certainly heard a lot of guitar tones on recordings and wonder what was used to make them. But those sort of words above seem to fall quite short of really describing the sound, or offering enough information to reproduce it either with equipment or by designing a circuit.
When reading specs for things like hi-fi amplifiers and non-guitar gear, a spec often used is THD, which is usually rated in percent. This means that the output of the amplifier is compared to the input (ignoring the change in amplitude) and however the signal differs because of slight changes to the signal added by the amp is measured. The percentage difference is recorded and used in the spec sheet for the device. This may or may not also included noise added by the device in addition to distortion. (I suppose that noise is a form of distortion, however it is often measured separately)
Although it seems perfectly nice to have a small list of parameters like distortion, frequency response, etc. to describe the performance of an amplifier or other device, this doesn't come close to describing the actual device will sound or perform in real life. Certainly extremely poor specs would be an indication that there are major problems with a design. But for devices with relatively good specs, what makes one sound better than another? Certainly marketing, price and physical aesthetics have a lot to do with how people perceive and enjoy things. But I'm trying to separate the technical and aesthetic parts of a design for the purpose of my investigation.
When people talk about distortion, especially in hi-fi amplifiers, the detail most often ignored is what the distortion is made of. No circuit is absolutely perfect, although some come really close. So we can assume that most circuits add distortion. But simply measuring a percentage of distortion tells us very little about what effect it will have on the sound. What we really need to know what the distortion is. Choosing a product with the lowest possible value of distortion might not be the best choice if the type of distortion produced is offensive to the ear. Certainly after learning about guitar circuits where various types of distortion are desirable, this became very clear. In a guitar amplifier the character and overall tone of the sound is created by various types of distortion interacting in subtle and sometimes not so subtle ways. Whether this was originally intended by amplifier designers is unclear. Early amplifiers had a lot of shortcomings that were inherent in the designs and parts that were available which have since gone on to define the sounds that defined entire generations of music. After studying a lot of classic amp designs, it's pretty clear to me that if designers had originally intended guitar amps to simply make a guitar sound louder, it probably didn't take very long for them to realize the appeal of deliberate distortion.
I never really thought about this before when I used to look at a lot of hi-fi and recording equipment specs. I just always thought that I wanted the lowest possible value. But after learning more about guitar circuits where certain kinds of distortion are often desirable (and other kinds of distortion are to be avoided) I've started to take more interest in learning not only how much something is distorted, but more importantly, how it is in fact distorting. It's hard to design a circuit to produce a certain sound if you don't know how to describe the sound. I know the basic characteristics of certain distortion sounds and can translate that into circuit ideas. But there are a lot that I can't understand, or subtle differences that are hard to describe.
Obviously it would excellent to have a language and notation for describing any kind of sound or sound processing, but audio is a complex thing with virtually unlimited possibilities. Perhaps someone with a bigger brain than me will come up with something like this. But for now I'm focusing on distortion and pondering whether it's possible to accurately notate or describe a distortion sound with human-readable parameters. Hopefully with few enough parameters that different circuits could be measured and compared easily. Some things that I currently (as of this writing) believe to be true about distortion are as follows:
Now that I've laid out some of the criteria and assumptions that I've made, hopefully we're all on the same page. If you disagree with my assessment of the situation please let me know. In the meantime I'll carry on with this report and see where we get.
We know from mathematics that sound (or many other signals for that matter) can be described in terms of samples in time (time domain) or by frequency amplitudes and phase in chunks of time. (frequency domain) Indeed digital samples can be converted into the frequency domain and back again without any loss. And let's agree that digital sampling does a good job of accurately capturing and playing back sound. This means that whatever analysis can be applied to digital audio signals is valid for our purposes. That's a good thing because the results of digital recordings are finite and easy to look at mathematically.
Considering that we now have defined that we have a way to record and analyze sound, let's discuss how we might apply that to a practical analysis of a distortion circuit. The goal of the analysis would be to produce enough data to be able to recreate the same distortion sound from the data alone. This would then present a language that could be used to describe the complex sound of a distortion circuit.
I have a few hypotheses about how a circuit could be analyzed. This might be an oversimplification... let me know if you think so. I would argue that there are two factors that affect how a distortion circuit changes an input signal. One is the amplitude of the input signal, and another is its frequency. Equalization before and after the distortion circuit (and even within it) seems to be a major factor in the sound of the distortion. So too does the amplitude of the input signal and how this signal is converted to an output signal. Think of a transfer curve for transistors or tubes. These are usually measured at various operating points, but do a pretty good job of showing how the device will behave in a real circuit.
Suppose that we were to map out a distortion circuit in some way to determine the distortion response, we could build some sort of distortion map or graph of the result. Obviously if we assume that the amplitude and frequency of the input signal are important, this would generate a lot of resulting data. For each frequency at each amplitude, an entire spectrum of output signals would result. This would be hard to graph as there are too many dimensions to the output. But what if we ignore frequency response? What if we just look at a single input frequency and measure at 1dB or even smaller intervals? We could measure perhaps 60dB of input signal range and measure the output spectral response at each input signal level. We could end up with a set of transfer curves of sorts that may give clues as to what happens to the distortion at various input levels. This could be graphed in two dimensions like a transfer curve with each output response stacked one on top of each other. By going back and measuring different frequencies, a three dimensional map could be visualized that could tell the entire story. A multi-layered and multi-dimensional set of transfer curves that tell a distortion story.
So, my major question is: does my analysis example tell the full story? Or would there be major pieces of the puzzle missing from such analysis? And even if there was useful information from such a test, would it give information that humans could use to understand how and why a certain circuit sounds like it does?
The practical part of me just wants to sit down and make some circuits, hoping to stumble upon a magical sound. But another part of me wants to analyze and know why something sounds the way it does.
Go to the Projects Page