Transistors
From manual switching to mechanical switching to relays to vacuum tubes to transistors and beyond
We understand that everything that goes on inside a computer is down to strings of zeros and ones—binary strings.
We call the individual digits of a binary string as “bits” (which is short for “binary digit”). This is the smallest amount of information a computer has to work with—one binary digit—and we know that it must be a zero or one. Those are the only possibilities.
We’ll get into more detail later, but for now, let’s just consider how one might change a bit. Surely anything interesting that goes on in a computer isn’t static. Things change. For example if we add two numbers, we need a way of performing the operation and representing the result. So there has to be some way to change a zero to a one and a one to a zero as demanded by various operations. But how might this happen?
If you’ve read Cixin Liu’s “The Three-Body Problem” (or see the Netflix series based on the book), you’ve encountered a human powered computer.
In this scene, a massive army of soldiers hold flags which are black on one side (I assume representing one) and white on the other (representing zero). The solders are highly regimented and follow precise instructions regarding when they should flip their flag from zero to one or vice versa. In this way, the human computer performs a complex computation.
This may seem absurd, but there’s absolutely nothing in principle that prevents one from performing computations in this manner. It might be slow, it might be huge, it would need regular feedings, but you could construct a computer this way. Keep in mind that all these soldiers are doing is flipping bits from zeros to ones and back according to some instructions.
How could we improve on this? We could use mechanical switches to replace the flags.
This may be a small improvement. With smaller mechanical switches, a human operator could manage a few dozen, but it’s not automated and it certainly wouldn’t scale well.
Electromechanical relays
In the early 1800s, a number of scientists and engineers developed electric-powered relays (priority of invention isn’t very clear).
Relays of this type were developed for early telegraph systems. A switch could be opened or closed by applying a voltage to an electromagnet. No one had to throw the switch manually—it could be done with electricity. Arguably, this is one of the most important inventions in the history of computers, and it was invented before computers existed!
However, as you can see, the relay above is rather large, clunky, and heavy. It works, but if we were to construct a computer out of such relays, the device would be immense and would require considerable energy. Indeed, some early computers were built with electromechanical relays.
Vacuum tubes
In order to improve reliability and speed, a new invention was needed: the vacuum tube. Vacuum tubes were invented in the early 1900s and they had the advantage of being able to perform switching operations without any moving parts! Mechanical parts wear, and to build a computer, you’d prefer not to spend more time servicing it than running it. Vacuum tubes don’t last forever, but they do last longer and are more reliable than electromechanical relays.
Like relays, vacuum tubes can switch signals on and off using electricity.
When the ENIAC (the first programmable digital computer) was built in the late 1940s, vacuum tubes were at the foundation.
Vacuum tubes were a big improvement over electromechanical relays—they were faster and more reliable. However, they’re not small. To significantly improve performance and reliability a new invention was needed: the transistor.
Transistors
The transistor was invented in 1947 by Bardeen, Brattain, and Shockley at Bell Laboratories. Like a relay or vacuum tube, the transistor functions as an electrically controlled switch, but they are much smaller and have no moving parts—indeed, they are entirely solid.
The first transistor looked like this:
This was, of course, a working protype. Soon they came to be mass produced. Here’s what early transistors look like (each is smaller than a dime).
Check in
Notice that these transistors have three leads each. Why is that?
Integrated circuits
With time, manufacturing processes improved and transistors could be miniaturized to the point where several could be included in a tiny package called an integrated circuit.
Here’s what one looked like.
Again, this is about the size of a dime, but there are multiple transistors in the same package. This particular integrated circuit was used in the Apollo guidance systems.
By 1974, Intel was producing the 8080 microprocessor. This package included approximately 6,000 transistors and could perform several hundred thousand operations per second.
By 1985, the ARM 1 was released, with 24,800 transistors.
Nowadays, integrated circuits have hundreds of billions of transistors. So, in the last 80 years or so, we’ve gone from computers built with relays and vacuum tubes—machines that filled an entire large room—to system-on-a-chip designs that easily fit in your pocket.
Copyright © 2025 Clayton Cafiero
No generative AI was used in producing this material. This was written the old-fashioned way.