Driving Modern Society: Semiconductors (2)

Written byKalanKalan
💡

If you have any questions or feedback, pleasefill out this form

This post is translated by ChatGPT and originally written in Mandarin, so there may be some inaccuracies or mistakes.

Yesterday, we introduced the functions and basic operating principles of semiconductors (in a simplified version) and included this image at the end:

20103565Hn1Trf1bqI

This transistor is called a Bipolar Junction Transistor (BJT). However, in practical applications, another type of transistor is more commonly used, known as the Field Effect Transistor, abbreviated as MOSFET.

Field Effect Transistor (MOSFET)

Like the Bipolar Junction Transistor, the MOSFET has three terminals: the gate, the drain, and the source, along with a substrate known as the body.

When voltage is applied to the gate, an electron channel can be created between the source and gate, thereby allowing current to flow through the circuit. Since we can control whether the circuit conducts by applying voltage, it effectively functions as a switch.

There are several reasons why MOSFETs are used more frequently than BJTs:

  • There is a Si oxide film between the gate and the P-type substrate, which means that even when voltage is applied, no current flows (high impedance). As a result, the power consumption is relatively lower than that of BJTs.
  • The switching speed (frequency) of MOSFETs is significantly higher than that of BJTs.
  • Due to their structural design, MOSFETs are generally smaller than BJTs.

Because of these advantages, MOSFETs are typically used in semiconductor manufacturing processes.

Manufacturing Methods

For CPUs or other circuits, the most fundamental component is the transistor; however, the positioning and connection of the transistors are crucial. To achieve this in a sufficiently small size, the most critical method used today is photolithography. The main principle involves creating a photomask of the circuit and exposing the circuit diagram onto the wafer.

As semiconductor manufacturing has advanced to the scale of a few nanometers, extremely precise operations are required. Recently, EUV (Extreme Ultraviolet) technology, which uses even shorter wavelengths, has been frequently discussed; exposure machines are key to semiconductor manufacturing, costing billions in NT dollars.

Packaging

The etched circuits on the wafers need to undergo packaging before they can be used by others because the circuits inside are relatively fragile and require a protective outer layer. Additionally, the contact points on these wafers are too small and need to be connected by packaging factories to be functional. The same chip may have different packaging methods, such as DIP, PGA, SMD, etc.

Writing Methods

A CPU can have billions of transistors, a scale that cannot be completed by just a few people manually wiring and drawing. Typically, computer assistance is required. Currently, logic circuits are mainly designed at the RTL (Register Transfer Level), with Verilog being one of the more famous languages used.

module half_adder(A, B, sum, c);
  input A, B;
  output sum, c;
  xor (sum, A, B);
  and (c, A, B);
endmodule

It looks very much like the code we usually write; however, because the underlying operation is directly converted into physical circuits, it differs from typical programming. In physical circuits, they execute simultaneously. This means that the half adder computes the sum and carry in one cycle simultaneously. After writing the logic circuit, it is first simulated and tested to ensure it aligns with expectations before layout begins.

Various Applications

Transistors have a wide range of applications, making it impossible to list them all. Here, we will share two common applications.

CPU / MCU

Both CPUs and MCUs are composed of countless transistors. But how can a multitude of transistors combine to form a CPU or MCU?

The circuits within a CPU can be broadly simplified into several types: control units, arithmetic units, and storage units. The control unit determines which branch to execute based on instructions, the arithmetic unit is responsible for data calculations, and the storage unit can store data.

For more information on how each unit is simulated with circuits, you can refer to [Day8] Games Related to Programming (2) – TIS-100 and Turing Complete. Essentially, all functions can be realized using logic gates, which in turn can be represented by transistors.

But computers are not just 0s and 1s, right? For example, how can a screen display so many colors? Indeed, the technologies involved extend beyond electrical signals; however, if we imagine a screen as composed of many tiny LEDs, we can combine different voltages to produce various brightness levels and colors. Thus, the display of a screen can be abstracted into "calculating the voltage needed for this color and where to display it."

There are also chips behind the screen controlling it; by changing the values in the screen's memory addresses, they can alter the colors. Therefore, the entire process can be further simplified to "calculating which memory address value needs to be changed." We can see that many problems, when simplified, ultimately become computable issues, and as long as something can be computed, a CPU can be utilized.

Boost Converter

Suppose there is an IC that requires a power supply of 5V or 12V, but a single dry cell only provides 1.5V, and two cells only yield 3V. What should we do? As mentioned earlier, transistors serve a function as "switches." The principle behind a boost converter involves inductors (coils). Inductors have a fascinating property: when the current changes, the inductor generates an electromotive force to counteract this change.

By continuously switching, we can allow the inductor to accumulate energy and release it to charge a capacitor, resulting in a voltage higher than the input voltage. This achieves the boost effect. The first time I learned about this circuit, I found it fascinating—boosting is simply about continuously switching!

Of course, this isn't magical; while theoretically, the voltage can be boosted indefinitely, this would require the MOSFET and inductor to withstand a very high power level, almost akin to a direct short circuit. Even if it were possible, the energy from two 1.5V dry cells is limited, so boosting too much would deplete the energy quickly.

If you found this article helpful, please consider buying me a coffee ☕ It'll make my ordinary day shine ✨

Buy me a coffee