I previously shared the background behind why I began these circuit experiments. However, starting something for the first time is always a challenge, as the initial point of departure sets the direction for everything that follows.
My choice emerged naturally from experience. When I first started these experiments, I was at a level where I could build very basic oscillators. Consequently, it felt intuitive to begin with the Integrated Circuits (ICs) I already had on hand. At the time, I had a sort of circuit practice toolbox that my friend Satoshi had passed down to me, which contained various IC chips. My starting point was to organize them, figure out what each one did, and pick one to work with. The most abundant chip in the box was the 4049 CMOS. I discovered it was a NOT gate; seeing that it simply outputted 1s and 0s, I thought it would be a perfect place to start.
Inverter (NOT gate)
An inverter, or NOT gate, is a fundamental component in digital circuits, designed to control input and output logic level voltages representing binary bits of 0 and 1. These binary values are depicted through voltage signals in relation to ground within the circuit. The functionality of an inverter extends to its ability to manage currents in two primary ways: Sourcing and Sinking.
Sourcing current involves connecting the output terminal to the IC’s power source (usually called Vcc), effectively “pushing” the current out. Conversely, Sinking current entails connecting the output terminal to the ground (often labeled as Vss), completing the circuit by “pulling” the electricity in to enable logic operations. To put it simply, it’s a component that “pushes and pulls” electricity.
Another frequently used IC for inverters is the 40106, which contains six Schmitt triggers. A Schmitt trigger is an inverter with hysteresis. But what exactly is hysteresis?
It refers to a property where the output of a system depends not only on its current state but also on its past state. By setting different thresholds for “stepping up” and “stepping down,” the system avoids wavering in ambiguous middle zones. This allows the system to remain stable and unfazed by external fluctuations or minor noise. In short, it is a more stable, noise-filtered inverter, and it is preferred in oscillator design due to that very reliability. (See the diagram below)
The interesting part of choosing between these two lies right here. Usually, one would choose the Schmitt trigger for its robustness against noise, as the 4049 seems to require a lot of effort to produce a clean square wave. To investigate further, I decided to compare the two side-by-side. This kind of comparison is an experiment that can only be done out of “ignorance”—a lack of prior knowledge. I designed the oscillators as follows:
Theoretically, in the design above, both should oscillate properly. The formula for calculating frequency is $1 / (RC \times t)$. I also learned that the value of $t$ (propagation delay) can usually be found in the datasheet. Looking at the waveforms at the bottom of the diagram, you can see that the frequencies of the two inverters are completely different. Specifically, the 4049 oscillator practically runs wild because it allows even the most minute changes to pass through. It’s a noise hellgate!
Consequently, I realized that to build an oscillator with the 4049, one must mix multiple oscillators together. In other words, you have to pass through various inverter gates to self-correct. The design is as follows:
Through this experiment, my choice naturally gravitated toward the more “problematic” one. Starting with the 4049 and my first oscillator design, I built three oscillators into a single IC and began tinkering. I tried connecting different points, breaking connections, and replacing capacitors with different values or materials. The first circuit I created is shown below:
At this point, a question arose: where should I listen to the sound? When working with analog, the starting and ending points are often unclear. For someone like me, who values tinkering over the “orthodox” way, the sheer number of choices was a bit paralyzing. So, I initially used the output junction shown above. When connecting to other equipment (e.g., a mixer), both devices must share the same ground, and the audio cable should not interfere with the signal flow. A good way to prevent this is to build a simple pre-amp using an Op-amp. I’ll explain the Op-amp some other time—that story is quite long!
So, the final design came out like this:
Each point is numbered; these are the contact points where I can interfere with the circuit in various ways. This idea was actually inspired by the Crackle Box. I created multiple contact points in advance so that the character of the circuit changes depending on how each point meets. Unexpectedly, it works quite well.
The following is a video of the test. This attempt eventually became the catalyst for composing my 2024 work, Cross-wired Xylophone.
