What Is an Analog Chip? Analog vs Digital ICs Explained
An analog chip is a semiconductor that handles smoothly changing quantities, such as sound, light and temperature, directly as voltages or currents. It makes small signals bigger, turns them into numbers and back, steadies supply voltages and creates timing signals. It links digital chips, which compute with 0 and 1, to the real world.
What you'll learn
- How "analog" differs from "digital," and why we need both
- How circuits make signals bigger and turn them into numbers, and what "16-bit, 44.1 kHz" means
- Why analog design is hard, and why analog chips are not made on the newest processes
- What the analog chip industry looks like in the news
Prerequisites: none. It helps to know that computers store numbers as 0s and 1s ("How Do Computers Use Binary?").
In everyday life: "analog chips" in the news
You may see the phrase "analog chips" in business news. It often appears in stories about the earnings of companies such as Texas Instruments (TI) in the US. Analog chips are less famous than CPUs or memory. Even so, they are inside almost every electronic device.
When you talk on a phone, the microphone turns your voice into an electrical wave. A thermometer measures a voltage that changes with temperature. In each case, the first thing to arrive is a smoothly changing signal. The chip that receives it is an analog chip.
Analog vs digital
Sound is a wobble in air pressure. Its strength changes smoothly and can take any value in between. Brightness and temperature behave the same way. A quantity that changes smoothly like this is called analog. So is a signal that copies it directly.
A computer, on the other hand, writes numbers as strings of 0s and 1s. It only uses separate, fixed values. This is called digital. With only two values, a small wobble in the voltage rarely causes a misread. That makes digital good for exact calculation and storage.
The real world, however, does not send information as 0s and 1s. What a sensor picks up first shows up as a smoothly changing voltage. A chip that handles this voltage in that form is an analog chip.
Note that industry statistics also count integrated circuits (ICs) that steady supply voltages as "analog." Most power semiconductors, which switch large amounts of power, are counted separately as single components ("What Is a Power Semiconductor?").
The figure below shows how a voice is recorded and played back. Analog circuits sit at the entrance and the exit of the digital circuits.
How it works 1: making small signals bigger
The voltage from a microphone or sensor is often very small. As it is, the circuit that turns it into numbers cannot read the fine differences. So the first step is to make the signal bigger while keeping its shape. This is called amplification, and the circuit that does it is an amplifier.
A widely used building block for amplifiers is an IC called an operational amplifier (op-amp). It takes the difference between the voltages at its two inputs and multiplies it by a huge factor. In one widely used product, the factor is about 200,000.
The factor is so large, though, that it is hard to use as it is. So part of the output is fed back to the input. Then the overall gain is set by the ratio of two resistors connected outside. Resistor values are easy to make accurate, so the gain becomes accurate too. This way of connecting is called negative feedback. You can try how resistor values set the gain with the Op-Amp Gain Calculator.
How it works 2: removing unwanted parts
Besides the part you want, a signal also picks up unwanted parts. Examples are the low "hum" that leaks in from the power supply, or fast ripples from other circuits. These parts are told apart by how many times per second the wave swings. That count is called the frequency, and its unit is the hertz (Hz).
A circuit that passes only a set range of frequencies and weakens the rest is called a filter. It is built from resistors, capacitors (parts that store charge for a short time) and op-amps.
How it works 3: turning voltage into numbers with an ADC
A circuit that turns the cleaned-up voltage into numbers a computer can use is called an analog-to-digital converter (ADC). As the name says, it converts analog into digital. An ADC turns a voltage into numbers in two steps.
The first step is to measure the voltage at fixed time intervals. It throws away the values between those moments and keeps only the measured ones. This step is called sampling.
The second step is to round each measured value to one of a fixed set of levels. This step is called quantization. The number of levels depends on how many bits (binary digits, each 0 or 1) the number uses. With 3 bits there are 8 levels; with 16 bits there are 65,536.
The dots in the figure are slightly off from the original curve. This small gap, created when values are rounded to a level, is called quantization error. The finer the levels, the smaller the gap.
"16-bit, 44.1 kHz" on a music CD
The music CD standard measures the sound voltage 44,100 times per second. The number of measurements per second is called the sampling rate, and for CDs it is 44.1 kHz (kilohertz). Each value is stored with 16 bits, which means 65,536 levels. There are two channels, left and right. This format is set out in the international standard IEC 60908.
How fast do you need to measure?
To record a fast-swinging wave, you have to measure just as fast. The rule of thumb is to measure more than twice as often as the highest frequency you want to record. Then, in theory, you can rebuild the original wave from the measured points. This rule is called the sampling theorem.
Put another way, half the sampling rate is the highest frequency you can record. This limit is called the Nyquist frequency. For CDs, it is half of 44,100, or 22,050 Hz. The highest pitch the human ear can hear is roughly 20,000 Hz (20 kHz). CDs are designed to record a little above that.
Suppose swings faster than this limit are still mixed in when you measure. When you join the measured points, a slow wave appears that is not really there. This effect is called aliasing. That is why a filter that removes components above the limit is usually placed just before the ADC.
There are several kinds of ADC, some built for speed and some for fine detail. A table of the differences is in "SAR vs Delta-Sigma vs Pipeline vs Flash ADC."
How it works 4: turning numbers back into voltage with a DAC
Going the other way, a circuit that turns numbers back into a voltage is called a digital-to-analog converter (DAC). The voltage a DAC puts out looks like a staircase, with one step for each number. A filter smooths it, and an amplifier strengthens it to drive the speaker.
How it works 5: steadying the power supply
The circuits inside a chip need electricity at a set voltage, delivered steadily. A battery's voltage, however, changes as it runs down. So a circuit makes the needed voltage from the incoming one and holds it constant. This circuit is called a voltage regulator.
There are two main types. One type burns off the extra voltage as heat. A common kind is called an LDO (low-dropout regulator). It wastes more energy, but its output voltage has very little ripple. The other type is the switching regulator, which changes the voltage by turning the current on and off very fast. It is more efficient, but it adds some ripple. When to use each is shown in a table in "LDO vs Switching Regulator."
How do regulators differ from power semiconductors? The difference is how much power they handle, and where they sit. Power semiconductors switch the large power that flows through motors and chargers. Most regulators are ICs on the board that holds the parts, and they tidy up the smaller power sent to chips. A phone uses a power management IC (PMIC), which combines many regulators in one chip.
How it works 6: radio signals and clock signals
The parts that send and receive a phone's radio signals are also analog circuits. One example is an amplifier that boosts a very weak incoming signal while adding as little noise as possible ("What Chips Are Inside a Smartphone?").
Digital circuits move one step at a time, in time with a signal that arrives at regular intervals. This signal is called the clock.
The circuit that makes the steady swinging signal behind the clock is an oscillator. Many devices use a quartz crystal, which vibrates at a fixed rate when a voltage is applied. It works the same way as a quartz watch.
The frequency a crystal can make directly is far lower than the speed a CPU runs at. So another circuit takes the slow, steady signal and makes one tens to hundreds of times faster. This circuit is called a PLL (phase-locked loop).
Why analog design is hard
In a digital circuit, a small wobble in voltage changes nothing, as long as a 0 is not misread as a 1. In an analog circuit, even a tiny wobble becomes an error in the signal. Here are four main sources of trouble.
- Noise: Heat makes the electrons inside every part jiggle at random. This motion creates small voltage wobbles that cannot be avoided. This is called thermal noise. Noise also leaks in from other circuits and from the power supply.
- Temperature changes: The behavior of parts shifts slightly with temperature. Designers need tricks so the circuit gives the same value outdoors in winter and in a hot car in summer.
- Variation between parts: Two transistors made to the same shape still behave slightly differently. In an op-amp, which amplifies a difference, this mismatch turns directly into error. By a rule of thumb that Pelgrom and colleagues published in 1989, making a transistor four times larger in area halves the variation.
- Layout: Performance changes with how parts are placed. Examples are placing paired parts symmetrically, and keeping them away from digital circuits.
For digital circuits, design software decides much of the layout automatically. For analog circuits, layout changes performance a lot, so more is left to the designer's judgment. A 1991 book on analog design, edited by Jim Williams, shows this. Its title includes the word "Art" next to "Science." In this field, skill built on experience matters a great deal.
Why analog chips are not made on the newest processes
The main computing chip in a phone, the SoC (system on chip), is made on a leading-edge process such as "3 nm." (What these process names mean is explained in "What Does 2nm or 3nm Mean?") Most analog chips, by contrast, are made on proven processes several generations older. There are three main reasons this works fine.
- They handle higher voltages: Leading-edge transistors are built to run at low voltages, around 1 volt. Analog ICs often take higher voltages, straight from a battery or a power adapter. They need sturdy transistors that can withstand those voltages.
- Shrinking gains little: As the last section showed, accuracy needs area. TSMC gave figures in 2020. Its 3 nm process fits about 1.7 times as much digital circuitry in the same area as its 5 nm process. For analog circuits, the gain is only about 1.1 times.
- Cost: The newer the process, the more both manufacturing and design cost. For circuits that do not shrink, it does not pay off.
One way to cut costs instead is to use bigger wafers, the round slices of silicon that chips are made on. According to TI's annual report for 2024, making chips on 300 mm wafers beats 200 mm wafers. The cost per chip, before it goes into a package, is about 40% lower.
The analog chip industry
The industry group WSTS tracks integrated circuits by type, such as analog and memory. By its count, the analog market in 2025 was about $86.5 billion. That was about a tenth of the whole semiconductor market (about $795.6 billion).
- A huge number of products: Each mix of what is measured, voltage, speed and accuracy calls for a different product. According to TI's website, the company sells more than 80,000 products.
- They sell for a long time: Factory machines and cars use the same parts for many years. In its annual report for 2009, TI said that high-performance analog products often keep selling for more than 10 years.
The big companies focused on analog chips include TI and Analog Devices (ADI), both in the US. In August 2021, ADI acquired Maxim Integrated, another major analog company. Infineon and STMicroelectronics in Europe, and Renesas and ROHM in Japan, also make them.
The first widely used analog IC was the μA702 op-amp, released in 1964. It was made by Fairchild in the US (according to the Computer History Museum).
Common myths
Myth: Analog is old technology that digital has replaced
In fact, what changed was mainly how information is stored and carried. Picking up sound, driving speakers, sending radio waves and steadying power are still jobs for analog circuits. The more digital devices there are, the more analog circuits are needed at their inputs and outputs. The myth spreads because "analog" is also used for old systems, as in "analog TV."
Myth: An ADC with more bits always gives better results
In fact, the number of bits only sets the upper limit on how fine the levels are. If noise or distortion (damage to the wave's shape) is bigger than a level, the lowest digits just wobble and carry no meaning. A figure that counts the usable bits, including noise and distortion, is called ENOB (effective number of bits). The ENOB in a part's datasheet is lower than its nominal bit count.
Myth: Analog chips also get better on newer "nm" processes
In fact, many analog ICs use proven processes, because handling high voltage and accuracy come first. Even so, leading-edge SoCs also contain analog circuits, such as PLLs. Those are made on the newest process to match the digital circuits around them, and careful design keeps them accurate. The myth arises because chip news is often told as a race to shrink.
By the numbers
| Item | Number | Note |
|---|---|---|
| Levels with 16 bits | 65,536 | Music CD |
| CD sampling rate | 44.1 kHz | Highest recordable frequency: 22.05 kHz |
| CD data per second | 1,411,200 bits | 44,100 × 16 bits × 2 channels (our calculation) |
| Analog market (WSTS) | About $86.5 billion | 2025; about a tenth of all semiconductors |
| TI products | More than 80,000 | Company website, as of 2026 |
You can try data size calculations with the Image, Audio and Video Data Size Calculator.
Learn more
- Compare in a table: Op-Amp vs Comparator, Class A vs AB vs D Amplifiers, PLL vs DLL, Single-Ended vs Differential Signaling
- Terms for the next step: successive-approximation ADC, oversampling
- Try the numbers: ADC SNR and ENOB Calculator (how bits relate to noise), dB and dBm Converter
- Related articles: "What Is a Microcontroller?" (microcontrollers with a built-in ADC), "How Do Image Sensors Work?" (from light to numbers)