HIFI Diary: What is Audio Clock? The Starting Point of All Sound

For a long time, audiophiles—including the BLOG owner—have frequently tossed around terms like “clock,” “jitter,” and “synchronization” when discussing digital audio systems. This is hardly a new concept; clocks emerged alongside digital audio long before the advent of the CD. Fast forward to today, debates surrounding clocks are still raging among novice and seasoned audiophiles alike. In an era where even a budget digital audio interface costing a few hundred yuan boasts a “femtosecond clock,” why do we still often hear that a system’s “clock” isn’t up to par? To truly make sense of this, we need to dive deep into what a “clock” actually is in digital audio, and whether the “clock specifications” on hardware spec sheets truly align with the “clock sound impressions” described by audiophiles.

I. The Core Role of Clocks in Digital Audio: The Ruler of Time

The essence of digital audio is slicing a continuous sound waveform into discrete points (samples) along the time axis. To reconstruct these points, they must be measured against a precise “time ruler”—and that ruler is the clock.

If the time ruler used during recording and playback is perfectly identical, the sound can be reproduced exactly as it was. However, if the ruler fluctuates in speed or interval during playback, the originally evenly spaced sample points will be “skewed”—at best resulting in muddy sound and a disorganized soundstage, and at worst causing audible distortion. Therefore, clock precision determines the accuracy of this time ruler.

II. Physical Definitions of Clocks: Frequency, Phase Noise, and Jitter

To quantify the quality of a clock, we primarily look at its precision.

2.1 Ideal Clocks vs. Real-World Clocks

• Ideal Clock: A perfect square wave or sine wave where the time interval of every cycle is completely identical, like a perfectly straight road.
• Real-World Clock: Constrained by crystal cut precision, circuit noise, temperature fluctuations, and other factors, the duration of each cycle exhibits subtle, random deviations. It is like a potholed road where cars bump along as they drive.

2.2 Two Core Metrics for Measuring Clock Precision

Metric	Definition	Analogy
Frequency Accuracy (ppm)	The long-term deviation between the clock’s actual frequency and nominal frequency	How many seconds a wristwatch gains or loses per day
Phase Noise (dBc/Hz)	The purity of the clock signal in the frequency domain, describing the distribution of noise power over frequency offsets	The instantaneous speed fluctuation with every tick of a watch
Jitter (fs/ps)	The instantaneous timing deviation of clock cycles in the time domain (usually in seconds), which is the integral of phase noise over the frequency domain	Whether the watch ticks uniformly second by second—the cumulative degree of speed fluctuations

For audio playback, short-term jitter (phase noise) has a far greater impact than long-term frequency deviation (frequency accuracy). A long-term deviation (such as a 10 ppm offset, or 10 parts per million) merely causes the music to play slightly faster or slower overall, which is virtually imperceptible to the human ear. In contrast, short-term jitter directly undermines the temporal alignment accuracy of sample points, resulting in smeared sound and a flat soundstage.

III. Frequency Points

In the context of crystal oscillators and clocks, a frequency point refers to the oscillator’s nominal center frequency—the specific frequency value it is designed to stably generate. For example, 22.5792MHz, 24.576MHz, 48MHz, and 100MHz are all common frequency points.

3.1 Where Do Frequency Points Come From?

Frequency points are determined by the physical cut dimensions of the quartz crystal and the circuit design, fixed once manufactured. It represents the frequency the oscillator is “supposed” to output, though in actual operation it undergoes slight shifts due to temperature, aging, and other factors (as reflected in the aforementioned frequency accuracy in ppm).

3.2 Common Frequency Points in Audio Equipment and Their Uses

Frequency Point	Typical Application	Relationship with Audio Sample Rates
22.5792MHz	Audio master clock (MCLK) for the 44.1kHz family	44.1k × 512 = 22.5792MHz 88.2k × 256 = 22.5792MHz 176.4k × 128 = 22.5792MHz
24.576MHz	Audio master clock (MCLK) for the 48kHz family	48k × 512 = 24.576MHz 96k × 256 = 24.576MHz 192k × 128 = 24.576MHz
48MHz	Reference clock for USB 2.0/3.0 controllers (e.g., TUSB7320 default)	No direct integer multiple relationship with sample rates; primarily used for data transmission
100MHz	PCIe bus reference clock, direct reference for USB 3.0 controllers	Unrelated to audio sample rates
10MHz	Reference frequency for external high-precision clock sources (OCXO)	Serves as a unified system clock baseline, synthesizing other frequency points via PLL

3.3 Frequency Point ≠ Sound Quality

Many beginners mistakenly assume that “a higher frequency means better sound,” which is a misconception. The frequency point itself is merely an operating parameter of digital circuits; what actually determines sound quality is the phase noise (dBc/Hz) and jitter (fs) at that frequency point.

• A 22.5792MHz femtosecond crystal oscillator, though much lower in frequency than 100MHz, delivers extremely low time-base error when used in a digital interface, yielding pure sound.
• Conversely, a generic 100MHz SMD crystal oscillator may have a high frequency, but with large jitter and poor close-in phase noise, it can actually degrade sound quality.

3.4 Why Does Audio Feature These Two “Odd” Frequencies: 22.5792MHz and 24.576MHz?

Because they maintain an integer multiple relationship with standard audio sample rates (44.1kHz / 48kHz).

• 22.5792MHz = 44.1kHz × 512
• 24.576MHz = 48kHz × 512

By using such integer multiple frequencies, the internal DAC circuitry only needs simple frequency division to generate the required sampling clock without relying on complex fractional-N PLLs, thereby minimizing added jitter. This is why high-end DACs often feature dual oscillators—one dedicated to the 44.1kHz family and the other to the 48kHz family.

IV. Phase-Locked Loop (PLL)

In the clock chain of digital audio systems, the PLL (Phase-Locked Loop) is an ubiquitous yet frequently overlooked core building block. Simply put, it acts like a “frequency translator,” translating the frequency of a reference clock into another required frequency while simultaneously cleaning up jitter in the clock signal.

4.1 Fundamental Working Principle of a PLL

A typical PLL consists of four basic modules forming a closed-loop feedback system:

Module	Function	Analogy
Phase Detector (PD)	Compares the phase difference between the reference clock and the feedback clock	A conductor listening to the orchestra’s tempo and comparing it against their baton
Loop Filter (LF)	Filters out high-frequency noise and outputs a smoothed control voltage	The conductor adjusting the tempo of their gestures based on the perceived difference
Voltage-Controlled Oscillator (VCO)	Outputs the corresponding frequency based on the control voltage	The orchestra adjusting its playing speed according to the conductor’s motions
Frequency Divider	Divides the VCO output frequency and feeds it back to the phase detector	Translating the orchestra’s overall tempo into a beat-by-beat rhythm fed back to the conductor

Workflow: The PLL continuously compares the reference clock with the divided VCO clock. When their phases align, the loop achieves “lock.” At this point, the VCO output frequency = reference frequency × division ratio.

4.2 Core Roles of PLL in Audio Systems

Frequency Synthesis: Generating Required Frequencies

Audio systems are filled with diverse frequency requirements, and PLLs enable them to share a single reference source:

• 10MHz external clock input → PLL → 22.5792MHz (as master clock for the DAC)
• 10MHz external clock input → PLL → 100MHz (for the USB controller)
• 48MHz on-board crystal → PLL → 5GHz (USB 3.0 physical layer serial clock)

Jitter Attenuation: Purifying the Clock Signal

The loop filter in a high-performance PLL acts as a low-pass filter:

• High-frequency jitter (far-out phase noise): Filtered out, making the VCO output cleaner
• Low-frequency variations (close-in phase noise / frequency drift): Retained, with the VCO tracking the reference clock

This explains why an ordinary 10MHz Oven-Controlled Crystal Oscillator (OCXO), paired with a high-quality PLL, can synthesize an ultra-high-quality 22.5792MHz audio clock—the PLL accomplishes both frequency conversion and signal purification.

4.3 Integer-N PLL vs. Fractional-N PLL

Type	Working Principle	Advantages	Disadvantages
Integer-N PLL	The multiplier/divider ratio is an integer (e.g., ×4, ×10)	Simple design, introduces very low phase noise	Output frequency can only be integer multiples of the reference frequency, lacking flexibility
Fractional-N PLL	The multiplier/divider ratio is a fractional value (e.g., ×4.1667, ×221.5)	Can synthesize arbitrary frequencies with high flexibility	May introduce fractional spurs, demanding stringent circuit design

In audio applications:

• Integer-N PLLs are generally preferred because they are theoretically “cleaner.” For instance, 25MHz × 4 = 100MHz, or 10MHz × 100 = 1GHz followed by frequency division.
• Fractional-N PLLs are inevitable when multiple unrelated frequencies must be synthesized from a single reference; high-end PLL chips (such as the Si5381) suppress fractional spurs to extremely low levels through optimized design.

4.4 Real-World PLL Applications in Audio Equipment

Device Type	Role of PLL	Typical Chips
USB 3.0 PCIe Card	Multiplies the on-board 48MHz crystal to the high frequency required by the USB physical layer	Integrated inside the TUSB7320 controller
Digital Audio Interface (DDC)	Synthesizes clocks required for USB reception from 22.5792M/24.576M audio crystals	Integrated inside XMOS or custom DDC controllers
High-End DAC	Receives external 10MHz reference and synthesizes audio master clocks and system clocks	Si5381, CS2100
Master Clock Generator	Outputs 10MHz reference, or directly outputs word clocks (multiples of 44.1k/48k)	Custom OCXO + distribution circuitry

4.5 The Double-Edged Nature of PLLs

While PLLs are an indispensable link in audio clock paths, they are not without trade-offs:

• A well-designed PLL can substantially enhance clock quality: Using a high-grade OCXO as a reference, a PLL can synthesize various frequencies that closely match the purity of the reference source.
• A poor PLL can bottleneck a great crystal: Even if a high-grade femtosecond oscillator precedes it, poor PLL design (subpar loop filter, noisy VCO) can actually degrade the jitter performance of the output clock.

This is also why certain high-end devices emphasize a “direct clock path”—minimizing PLL stages or driving the audio master clock directly from the oscillator without passing through a PLL.

V. Overview of Crystal Oscillator Types

5.1 SPXO: Simple Packaged Crystal Oscillator

• Operating Principle: The simplest type of oscillator, consisting solely of a quartz crystal and an oscillation circuit. Its output frequency drifts with ambient temperature changes.
• Performance Characteristics:
  • Temperature Stability: Relatively poor, typically ranging from ±20ppm to ±50ppm (ppm = parts per million).
  • Jitter: Typically between tens to hundreds of picoseconds (ps).

5.2 TCXO: Temperature-Compensated Crystal Oscillator

• Operating Principle: Builds upon the SPXO by adding a temperature compensation circuit. By sensing ambient temperature, it automatically adjusts voltage to offset frequency drift caused by thermal variations.
• Performance Characteristics:
  • Temperature Stability: Substantially improved, typically reaching ±0.5ppm to ±2ppm.
  • Jitter: Usually lower than SPXO, achieving several hundred femtoseconds (fs) to a few picoseconds (ps).
  • Advantages: Compact form factor, low power consumption, instant start-up (no warm-up required).

5.3 OCXO: Oven-Controlled Crystal Oscillator (Defaulting to AT-Cut)

• Operating Principle: Houses the quartz crystal inside a miniature temperature-controlled chamber (oven). A heating circuit keeps the crystal temperature constant near the inflection temperature on its frequency-temperature curve (usually where the temperature coefficient is zero or near zero, typically 75°C to 95°C). Regardless of external environmental fluctuations, the crystal always operates at this “golden temperature point.”
• Performance Characteristics:
  • Temperature Stability: Extremely high, typically reaching ±0.005ppm to ±0.05ppm.
  • Phase Noise / Jitter: Extremely low. Top-tier OCXOs can achieve phase noise of -100 dBc/Hz or even lower at 10Hz offset, with superb short-term stability.
  • Trade-offs:
    • High Power Consumption: Typically requires 1–3 watts during initial warm-up heating, and several hundred milliwatts to maintain steady temperature thereafter.
    • Warm-up Time: Requires 5 to 30 minutes or even longer of warm-up time after power-on to reach a stable state.
    • Large Size: Due to the oven chamber structure, physical dimensions are usually relatively bulky.

5.4 SC-OCXO: Stress-Compensated Cut Oven-Controlled Crystal Oscillator (“SC” denotes the crystal cut type, not a model prefix)

• Comparison of Cut Types:
  • AT-Cut: The most widespread crystal cut. Mature manufacturing process and low cost, but its frequency-temperature inflection curve is relatively steep.
  • SC-Cut: An advanced stress-compensated cut featuring a double-rotated quartz geometry.
• Performance Characteristics:
  • Stability: Compared to AT-Cut OCXOs, SC-Cut improves short-term stability (Allan deviation) by approximately one order of magnitude.
  • Phase Noise: Exceptional close-in phase noise performance (1Hz–10Hz offset), which is paramount to “black background” and “imaging stability” in audio listening impressions.
  • Immunity to Interference: Far lower sensitivity to gravitational orientation and mechanical vibration/acceleration compared to AT-Cut.
  • Aging Rate: Extremely low annual aging rate (below ±0.05ppm/year), maintaining precision over long service periods.

VI. The Clock Path in Digital Audio Systems

Having understood the clocks themselves, let us examine their signal routing through the entire system. A typical PC-HiFi audio setup undergoes at least three stages of “clock domain transitions.”

6.1 Stage 1: PC / Digital Transport — USB Port (Transmitter Side)

The USB host card plugged into the PC motherboard has one primary mission: transmitting audio data from the computer out via the USB cable.

The USB card is equipped with its own clock oscillator, typically:

• USB 3.0 Controller: Typically uses a 48MHz crystal oscillator (dictated by chipset architecture)
• USB 2.0 Controller: Commonly uses 24MHz, 25MHz, or 48MHz

This clock ensures that the USB controller pushes data onto the USB bus at the correct pace. However, it is solely responsible for “data transmission,” not “sound quality.” As long as data packets are transmitted without error, the jitter of this clock has a relatively minor direct impact on sound quality—because far more critical stages follow downstream.

6.2 Stage 2: Digital Audio Interface (DDC) — USB Port (Receiver Side)

After receiving data over the USB cable, the digital audio interface must perform two tasks:

1. Accurately receive and decode data packets from the USB bus
2. Transmit the audio stream to the DAC timed precisely to the audio clock

Here lies the fundamental conflict: The timing pace of USB data transfer and the timing pace of audio playback belong to two entirely independent clock systems.

The USB receiver chip has its own reference clock (e.g., 24MHz, 25MHz, or 48MHz) driving USB physical layer communication. However, audio playback requires an audio master clock (MCLK) running at 22.5792MHz or 24.576MHz—the former for the 44.1kHz sample rate family, and the latter for the 48kHz family.

The introduction of Asynchronous Mode resolved this contradiction elegantly:

• The digital interface uses its own audio crystal oscillators (22M/24M) as the master timebase
• Via the USB protocol feedback endpoint, it instructs the computer: “Here is my current clock rate; please adjust your packet transmission rate to match my pace”
• The USB host controller on the PC adjusts its transmission rate accordingly

This means: What truly dictates audio performance is the audio crystal oscillator on the digital interface, not the upstream USB card oscillator. The front-end USB card only needs to ensure bit-perfect data delivery, while the downstream audio clock serves as the true “master conductor” of the sound.

6.3 Stage 3: DAC (Digital-to-Analog Conversion)

Once the digital interface formats the data stream, it feeds it to the DAC chip alongside the audio master clock (MCLK). Guided by this master clock, the DAC converts discrete digital sample points back into continuous analog waveforms.

If this clock suffers from jitter, the DAC exhibits timing jitter (JITTER) during reconstruction—voltage values that should have occurred at precise instants are output at slightly displaced points in time. This error manifests audibly as:

• Smeared transients and loss of micro-detail
• Compressed soundstage and inaccurate imaging localization
• Harsh, grainy treble that induces listening fatigue

Therefore, the single most critical clock in the entire audio chain is the audio master clock arriving at the final DAC stage.

VII. Unified Master Clock Synchronization

While the asynchronous mode described above allows the system to be “slave to the receiver,” pursuing ultimate performance opens up an even more thorough approach: using a single external high-precision master clock to supply a unified reference to all devices across the entire system.

7.1 Why Synchronize to a Unified Master Clock?

In a standard setup, the USB card has its own oscillator, the digital interface has its audio oscillators, and the DAC may also house its own oscillator. These separate oscillators operate independently; although coordinated via asynchronous protocol, they share no common “time baseline.”

The concept of unified clocking is: use an ultra-high-precision, ultra-low phase noise 10MHz Oven-Controlled Crystal Oscillator (OCXO), routed through high-performance PLL chips (such as the Si5381), to synchronously synthesize the required clock frequencies for all system components:

Device	Required Clock	Generation Method
USB PCIe Card	100MHz (or 48MHz)	10M → PLL → Required frequency
Digital Audio Interface (DDC)	22.5792M / 24.576M	10M → PLL → Audio master clock
DAC	Same 22M / 24M	Directly uses clock from DDC, or independently synthesizes from 10MHz

In this manner, the entire chain from USB data transmission to DAC conversion shares a singular time baseline, eliminating “relative jitter” between intermediate clock domains.

7.2 The Hurdles of Unified Master Clocking

While ideal on paper, implementing a unified master clock carries considerable barriers:

• Devices Must Support 10MHz External Clock Input: Not all USB cards, interfaces, or DACs feature a 10MHz BNC input
• Requires a High-Grade Multi-Output 10MHz Master Clock: Premium OCXO master clock generators command substantial price tags
• Requires High-Performance PLL Chips: E.g., the Si5381, capable of synthesizing integer or fractional multiples from 10MHz into diverse frequencies with minimal phase noise
• Impedance Matching: A frequently overlooked point—10MHz reference clock connections typically use 50Ω coaxial cables, whereas digital audio word clocks usually use 75Ω coaxial cables. Strict attention must be paid to impedance matching when connecting external clocks.

VIII. Subjective Listening Impressions of Clocks

Having covered the technical fundamentals, let us talk about listening impressions. The audible impact of clock quality has long ceased to be audio “snake oil”—it has been thoroughly validated both in engineering measurements and subjective listening. However, with oscillator costs dropping over the past decade and HiFi hardware designs maturing significantly, many stock products now ship with quite respectable built-in clocks (such as the widespread Crystek CCHD-957 and Accusilicon AS318/338). Stepping up to an even higher-end 10MHz master clock may not necessarily deliver a “jaw-dropping, night-and-day” difference for every audiophile.

8.1 Perceptible Dimensions of Clock Upgrades

When system clock quality is upgraded (e.g., from generic SPXO to femtosecond crystal oscillators, or by adding an external 10MHz OCXO master clock), audiophiles commonly observe the following improvements:

Listening Impression	Technical Correlation
Blacker background	Lower noise floor; micro-details and low-level signals are not masked
Sharper imaging	Reduced jitter; more precise phase and timing alignment between left and right channels
More structured soundstage	More pinpoint instrument localization and more authentic acoustic spatiality
Smoother treble	Reduced high-frequency glare and sibilance, with more natural air and extension
Superior micro-dynamics	Enhanced reproduction of delicate, low-amplitude nuances

The underlying logic is: Lower clock jitter → Reduced time-base error → More accurate sample reconstruction → More authentic and well-defined sonic “body and shape.”

8.2 Macro-Dynamics vs. Micro-Dynamics

Referencing our previous framework when discussing dynamic range, the sonic impact of clock performance primarily affects low-level dynamics (micro-dynamics):

• Micro-Dynamics (Low-Level Dynamics): Refers to nuance reproduction and delicate contrast at quiet levels. Reduced clock jitter directly enhances the resolution of extremely faint signals, bringing out the subtle decay of a triangle and the authentic acoustic “air” of the recording studio.
• Macro-Dynamics (High-Level Dynamics): Refers to explosive impact and large-scale transient slam at loud volumes. This aspect depends primarily on the amplifier’s current delivery capability and power supply headroom, with clock improvements playing a much more indirect role.

Consequently, the gains brought by clock upgrades are rarely characterized by “punchier, more explosive slam,” but rather by “quieter background, greater precision, and effortless fatigue-free listening.”

8.3 Diminishing Returns of Clock Upgrades

Upgrading clock quality from “poor” to “decent” yields massive gains; moving from “decent” to “very good” brings noticeable improvements; but pushing from “very good” to “state-of-the-art” offers diminishing returns. For the vast majority of audiophiles, the high-grade femtosecond oscillators built into modern digital interfaces are already plenty capable, obviating the need for an external 10MHz master clock. Only when every other link in your audio chain (where main components typically run five figures apiece) has reached resolving transparency will clock performance become the true bottleneck.

IX. Conclusion

The clock is the most easily overlooked yet fundamentally essential pillar of digital audio systems. It does not boast glamorous headline metrics like flagship DAC chips, nor does it directly drive headphones like an amplifier, but it silently provides the crucial “temporal order” for the entire system. From the 48MHz crystal on a USB card, to the 22M/24M femtosecond oscillators in a digital interface, to the 10MHz OCXO synchronizing a unified master clock—every upgrade along the clock chain strives to reconstruct a more precise “time axis.”

And ultimately, every single sample point along this time axis translates into the physical texture, spatial positioning, atmospheric air, and emotional resonance we hear. Clock improvements rarely mean “changing the sonic signature,” but rather “letting music sound as it was truly meant to be.”

In the first half of 2026, the BLOG owner’s writing pace slowed noticeably due to the dual squeeze of work and gaming. This comprehensive guide to audio clocks is likely the most detailed and thorough breakdown available across the web. Consider it the BLOG owner’s earnest delivery to all our readers for the first half of the year =W=