Choosing a DAC: A Guide to the ‘Chipset’ That Defines Your Sound
The exceptional specifications mentioned in Part 2, such as an SNR of 120 dB and a THD+N of 0.0001%, did not come out of nowhere. They are the result of the physical computing architecture of the chipset (IC) located at the heart of the motherboard and its surrounding circuitry.
Most audio devices we use are designed based on components made by a small number of global chipset manufacturers. In this chapter, we will analyze the architectural characteristics of the major chipsets dominating the Hi-Fi audio market, along with interesting behind-the-scenes stories and the sonic profiles associated with each manufacturer.
Understanding DACs for Beginners: A 3-Part Series
- Part 1: What is an Audio DAC? The first thing to buy when your earphones sound dull
- Part 2: How to read DAC spec sheets in 1 minute: 24-bit? SNR?
- Part 3: Which DAC should I buy? A guide to the ‘chipset’ that defines your sound
ESS vs AKM vs Cirrus Logic: Architectural Differences of the Top 3 Delta-Sigma Chipsets
More than 90% of modern DAC chipsets use the ‘Delta-Sigma’ conversion method, which increases data resolution through high-frequency oversampling. However, the decoding algorithms and filter designs vary by manufacturer, resulting in subtle differences in the final output waveform. Let’s examine the characteristics of the three leading brands that dominate the market.
ESS Technology (Sabre DAC)
– Source: DANAWA –
This brand holds the most overwhelming market share in the global Hi-Fi market today. The numbering convention seen on spec sheets is divided into the ES90xx series (e.g., ES9038PRO, ES9039PRO) for flagship desktop units, and the ES92xx (e.g., ES9219) series for portable dongles and DAPs.
In the past, measurements by famous overseas audio communities revealed a widespread phenomenon known as the ‘ESS Hump’ (IMD Hump), where distortion (IMD) spiked at specific mid-volume levels in devices using ESS chips. Ironically, this incident forced a level-up in the amplifier circuit design standards across the entire audio market.
The signature ESS Sabre sound relies on their proprietary HyperStream algorithm to keep THD extremely low. It is characterized by distinct sound contours, clean decay, and precise separation that accurately pinpoints the location of each instrument. If you want analytical, monitoring-style gear that delivers the original sound without embellishment, a device with an ESS chip is a great choice.
AKM (Asahi Kasei Microdevices)
– Source: DANAWA –
A Japanese brand focused on smoothing out the cold, sharp sound often associated with digital audio. AKM’s DAC lineup uses the AK44xx (e.g., AK4490, AK4493, AK4499) series numbering.
AKM faced its greatest crisis during the massive fire at its Nobeoka semiconductor plant in October 2020. The plant was destroyed, cutting off the global supply of AKM chips, forcing many audio manufacturers to switch to ESS chips for their new products. While this caused a significant drop in market share, AKM made a spectacular comeback with an innovative flagship architecture that physically splits the roles of the digital computing unit (AK4191) and the analog output unit (AK4499EX) into two separate chips.
Their sound profile aligns with their “Velvet Sound” slogan. By suppressing digital ringing noise that occurs when sounds cut off, they create a smooth texture and rich mid-to-low frequency harmonics reminiscent of analog reel-to-reel tapes. Known for excellent vocal density and a non-fatiguing listening experience, they are highly favored in high-end portable lines, including Astell&Kern.
Cirrus Logic (CS)
– Source: DANAWA –
A US semiconductor company that designs high-efficiency micro-architectures optimized for mobile environments and low-power operation. Their DAC lineup numbering follows the CS43xx (e.g., CS43131, CS43198) series.
After acquiring the UK audio chipset company Wolfson in 2014, they absorbed their warm sonic DNA and audio engineering expertise.
The CS43131 chipset integrates amplifier circuitry inside the chip very efficiently, minimizing heat and battery consumption. They dominate the mobile ecosystem in the dongle DAC and mid-range DAC/DAP markets, where preserving smartphone battery life is critical. Their sound is flat and clean, avoiding emphasis on specific bands, while incorporating a touch of the warm, natural signature inherited from Wolfson.
R-2R Ladder Network: Ditching the Chip for a Resistor Ladder
Inside the Gustard R26 R2R + 1Bit DSD DAC
Beyond popular Delta-Sigma chipsets, the other pillar of high-end audio is the R-2R ladder network method.
Instead of relying on a single chipset for calculation, this method involves arranging hundreds of physical resistor components on a circuit board in a ladder pattern. It is an intuitive, analog-style structure where switches open and close based on the input 0 and 1 bit values, distributing voltage instantly.
Implementing 24-bit audio requires numerous resistors per channel, and even a 0.01% error in a single resistor can ruin the sound image. While design difficulty is extremely high, production costs are steep, and device volume is large, it remains the most primitive conversion method without digital filter processing, resulting in very natural harmonic characteristics. It is highly supported by enthusiasts who prefer an analog texture.
Parallel Chipset Design in High-End Devices: “Throwing Hardware at the Problem”
TOPPING D90 III
Even in equipment that doesn’t cost millions, you will often find two flagship chips (e.g., ES9039PRO) that support 8-channel processing inside a single unit. Since the music we listen to is 2-channel stereo, why include hardware capable of processing 16 audio channels (8 channels x 2)?
This is a method of inputting the same audio signal into multiple channels simultaneously for distributed processing and then merging them. While the original sound signal adds up and becomes larger, the random mechanical noise generated in each channel cancels out as they are out of phase with one another.
By distributing the 8 paths in a single chipset to 4 channels each for left and right and merging them into one line, noise is canceled, theoretically yielding an SNR gain of approximately 6.02dB. Manufacturers opt for this parallel structure to suppress noise below the measurement threshold at the chipset stage before the signal even reaches the amplifier.
The Chipset is Just the Beginning; Circuit Design Defines the Sound
We have explored various chipsets and design methods. However, once you spend some time in the hobby, you will often hear that “the chipset doesn’t determine the sound.” No matter how good the flagship chip is, it cannot perform to its full potential if the board’s power design is inadequate.
High-quality DAC equipment uses LDO (Low Drop-Out) regulators on the board to block power noise from USB or adapters, supplying only pure DC power to each component. The final sound profile is ultimately shaped by how the amplifier stage amplifies the delicate analog signal extracted from the chipset.
Ultimately, choosing a good audio device is not just about finding one with an expensive chip, but evaluating the ‘overall engineering design’—how well the system controls digital noise and preserves the purity of the analog signal.
I hope this three-part guide helps you select equipment that fits your taste and budget, and allows you to enjoy your audio life at the intersection of digital and analog.
🔗 Original Post :
어떤 DAC를 사야 할까? 나만의 소리를 찾아주는 핵심 ‘칩셋’ 가이드
https://blog.naver.com/PostView.naver?blogId=k5kun&logNo=224325183262