Overview: Oladra Design

Oladra digital audio sources (DAS) incorporate the serving of stored files, re-serving of internet streams and providing low-noise and precisely timed digital audio outputs. Oladra takes a total system approach so that it can ensure a clean digital audio signal is received by your DAC. The cleaner the signal a DAC receives, the better the DAC will perform, because any noise entering the DAC will degrade the timing accuracy of sample conversions, and can even directly interfere with the analogue signal circuitry in the DAC.

Oladra’s design philosophy is set out in the articles listed below. The articles are an attempt to explain how we approach this complex field, in a way that is complete and yet understandable for the average audiophile. It is not our intention to present these articles as if it is our original thinking. It does express Oladra’s philosophy and approach, but it also draws extensively from the accumulated experience and insights of many practitioners in this field, including Dunn’s AES work on jitter, XMOS technical documentation, and the work of Jung and Putzeys.

We realise that there are some people that cling to the fallacy that playing digital audio is similar to digital transmission of a file from one storage device to another. Hopefully our articles will explain why this is a misconception, and describe the magnitude of the real design challenge.

Articles

1. Music, Audiophiles & The High-End

This article sets out how we see the purpose of music, how audiophiles are defined by the depth to which they seek to engage with recorded music, and what the goal of high-end audio is.

2. The Digital Audio Source (DAS)

This article proposes and explains the term DAS, to describe whatever method and collection of devices is used to feed a DAC with a digital audio signal. The complexity and confusion that persists in digital audio has created a plethora of partial methods and ‘silver bullet’ products, alongside total system products that complete both the music server and streamer roles. This article seeks to clarify the reasons behind the total system approach.

3. DAS Total System Design

This article provides an overview of the key areas and key objectives for the design of a Total System DAS. The key areas are further expanded on in the following sub-articles.

Glossary

Our glossary defines what is meant by the technical terms used in the articles.

1. Music, Audiophiles & The High-End

The Common Goal

The purpose of music is to create an emotional connection between composers, performers and audiences. Like any artform, how you connect with it and the emotions you derive from it are personal to you. But the common goal of Oladra, audiophiles and the high-end audio industry is to create profound emotional connections through high fidelity reproduction of recorded performances.

At first glance, high-end audio might seem frivolous and even decadent. After all, most audio equipment today delivers measurably accurate playback, is affordable, and integrates easily into modern life. But high-end audio isn’t about convenience or measured accuracy (ie. flat frequency response and low distortion), it’s about deepening the listener’s emotional connection with music. It exists for those who care about experiencing music in a more authentic and engaging way.

The detailed resolution of a system is not to serve something as banal as sound. If that was all that was achieved, then it would not hold the attention of a listener for long. The purpose is to serve the communication of the emotional content in the music, which requires a degree of naturalness and a form of authenticity that enables us to feel and immerse ourselves in the message. It is not something you can objectively measure, but it is something you recognise when you feel it.

Music as Human Expression

Music is fundamental to human experience. Across cultures and history, music has served as ritual, story, comfort, entertainment, and inspiration. It shapes memory, alters mood, and strengthens identity. Music connects us to others and to ourselves. But it is important to appreciate that different people engage with music in quite different ways.

Different Ways of Engaging with Music

Some people enjoy making music, by composing, singing, or playing instruments. Others love recording and producing it. Many people use music for ambience, mood or motivation. Each of these relationships is valid and personal. But music lovers seek to get closer to the recorded performance and to enjoy something more intense and immersive.

The Audiophile Perspective

For any music lover, the goal is not just to hear the music. It’s to feel it. The audiophile goes even further, wanting to be drawn into the recorded performance as if they were there. This requires equipment that doesn’t just reproduce the notes, but gets out of the way, preserving spatial cues, fine nuance and inflexion, separating harmonics and resonances. The listener may never know what being at the recording would have sounded like, but like every other music lover, they can recognise what real voices and instruments sound like in real spaces. It is fundamental that the system preserves the emotional message, through revealing fine details that reflect the artistic expression in the performance.

This level of engagement is not about luxury or status. It’s about an audiophile’s personal priorities, which value the emotional experience of music enough to invest in equipment that supports it. Audiophiles learn to listen differently, to notice when something interferes with the experience. Over time, their systems become tuned to serve emotional transparency rather than just objective sound quality. An audiophile’s music system can capture your full attention for hours and is much more than just a backdrop to other activities.

Why Standard Equipment Falls Short

High-end audio exists to serve a deeper form of engagement. Its designers aim to preserve the emotional and spatial cues in music, which help the listener feel the intent of the performance. This goal influences every design choice, from component layout to power regulation to software control. But because the goal of conveying nuance, spatial realism and emotional content cannot be met by deducing it from the accepted science, great high-end audio comes from many years of experience and experiments that lead to accumulated design insights.

The key design decisions are not based on measurements. They are based on listening. The best designers tune by ear, balancing technical excellence with emotional response. They test not just for sound, but for emotional fidelity. That’s the real benchmark in high-end audio.

The Role of the Listener

This approach demands more from the listener too. To benefit from high-end gear, you must spend time listening, comparing, and refining your system. That journey is part of what makes the audiophile experience challenging. It is not about chasing perfection. It is about learning what moves you and building a system that gets out of the way of the music.

Conclusion

High-end audio isn’t for everyone, and it doesn’t need to be. But for those who seek a deeper connection with music, it offers something profoundly rewarding. It respects the power of music to move us, and it is built on the belief that sound quality matters because it enables emotion, presence and meaning.

In the next article, we look at the design principles behind the Digital Audio Source (DAS), which is the digital audio system’s source component, that sets the stage for everything the DAC will do.

2. The Digital Audio Source (DAS)

Every DAC needs to receive a digital audio signal so that it can produce music. This might be from a streaming service, a NAS, a laptop, a smartphone, a dedicated music server, or a streaming bridge. Whatever you use to deliver digital audio files to your DAC is your DAS (digital audio source).

For some people, a DAS solution is a cobbled together collection of devices, such as reclockers, isolators, and high-end network switches. But the best approach is a single device DAS because it enables a total system design approach. There are so many potential sources of noise and so many pathways that only a total system approach can deal with them comprehensively.

Proposing The Term DAS

Regardless of the solution you choose, we propose that an appropriate term for it is the DAS. Every DAC needs a DAS to feed it with a digital audio signal for it to decode and a DAS plays a foundational role in modern digital audio.

Unfortunately, a proliferation of opinions and devices makes the category confusing for many audiophiles. By grouping everything that delivers the digital signal to the DAC under the term DAS, we create a framework to understand the options you have in order to feed your DAC with a cleaner signal.

What A DAS Does

The DAS doesn’t convert the digital signal into an analogue signal. That is the DAC’s job. But the DAS is a major determinant of the environment in which the DAC operates. By ‘environment’, we mean the electrical, timing, and computational conditions at the DAC chip, including noise, jitter, voltage stability, and interface behavior. The DAS influences the DAC’s electronic environment in crucial ways, including:

• Electrical noise coupled through the digital interfaces
• Timing irregularities (jitter) in the arrival of the data
• Physical interface behaviour and impedance matching
• System noise passed through shared power or ground planes
• Latency and buffering strategies and implementation
• Network activity and packet scheduling (for Ethernet-based sources)

Each of these factors can degrade the DAC’s clock stability and introduce noise that interferes with the analogue conversion process.

When A DAS Is Poor

A poorly designed DAS, or a DAS that only focuses on reducing one type or pathway for noise, may still deliver all the right bits to the DAC. But it will pollute the DAC environment. It will introduce unpredictable noise patterns, fluctuating jitter, and power line interference. When the DAC environment is noisy, it performs less accurately in the time dimension. That makes the timbral and rhythmical patterns less natural to our brains, and the music starts to fail to engage us emotionally. When your system fails to make an emotional connection, you will stop listening to it.

Even excellent DACs underperform in that situation, not because the bits are wrong, but because the timing and noise context degrades its ability to convert cleanly and accurately.

This is why many audiophiles notice dramatic differences when switching between different DAS, even if each DAS promises bit-perfect output. The difference lies not in the data, but in the environment created for the DAC conversion.

A High-End DAS Matters

With the DAS influencing the operating conditions for the DAC, then the quality of the DAS matters just as much as the DAC itself. A high-end DAS is about serving local files and re-serving internet streams by buffering and generating them in a low-noise environment:

• minimising electrical and radiated noise
• managing clocking relationships
• controlling operating system activity
• applying thoughtful layout, grounding, and shielding strategies
• providing a clean, stable and low-noise digital audio conduit to the DAC.

This contrasts with a typical ‘streamer’, which relies on a music server that is not designed for emotional fidelity, let alone sound quality, that is hundreds or even thousands of kilometers away, streaming over a chaotic and noisy internet.

Once the conversion to analogue has happened, nothing in your stereo system can get the emotional engagement back. This is why DAS quality is so crucial to audiophiles that listen to digital audio, in the same way that the source quality has always been crucial to the playback of recorded music.

The Oladra Perspective

Oladra treats the DAS as a single complete system so that our design can take a total system approach. Using partial solutions such as a network switch or an isolator/regenerator can help, but will fail to deal with all of the noise sources and noise pathways that can pollute the DAC environment.

We design and manufacture our own computers, thereby eliminating the interfaces on off-the-shelf motherboards that we don’t need, and allowing us to select optimal protocols, use high quality signal drivers, tightly control impedances, minimise interactions between traces and optimise the platform for real-time digital audio delivery. From motherboard design to firmware control to software stack, our goal is to ensure the DAS stays out of the way of the music. This is how we achieve consistently engaging sound, improving the sound of any competently-designed DAC.

Conclusion

The DAC is only as good as the environment it operates in. And that environment is hugely influenced by the DAS. By paying attention to the total path from data source to DAC input, and all other potential pathways for noise, we can address all of the upstream choices that can affect downstream conversion. See our other articles for more detail on how design choices inside the DAS, such as power, firmware, layout, and clocking, determine its impact.

3. DAS Total Design System

The DAS has a major impact on the electrical and timing environment the DAC operates in. Its influence on audio quality can be profound. This paper provides an overview of the core engineering challenges in designing a high-performance DAS and introduces four areas of deeper technical insight explored in subsequent articles.

The DAS Environment

The DAS is the most electronically active device in the playback chain. It houses the CPU, RAM, storage, network and interface circuits, which all produce various forms of electrical noise. The DAS generates and transmits the digital audio signal to the DAC. The quality of this signal and the noise that hitches a ride along with it directly influences the timing integrity of the DAC conversion process.

The Role Of The CPU

The CPU runs the audio playback software and manages the operating system, network stack, and file access. Its activity determines most of the electrical noise generated inside the DAS and the timing of audio data output. High-performance CPUs can improve responsiveness and reduce software-induced jitter, but they also generate more power noise. Selecting the right CPU and motherboard architecture, including chipset design, memory access, and USB or PCIe layout, is critical to striking the right balance.

Power Supply Architecture

Power supplies in a DAS must be both quiet and dynamic. Many people assume that linear power supplies are better, but in a digital system with varying current demands, a well-designed switch-mode power supply (SMPS) can perform even better. It can supply clean, fast current to demanding subsystems, reducing voltage sag and noise-induced modulation of timing. Designers must also ensure effective power distribution within the DAS to avoid crosstalk and ground noise.

Audio Output Design

The output interface determines how audio data reaches the DAC. USB, I2S, AES3, S/PDIF and Ethernet each have strengths and weaknesses. Their implementation can matter more than the protocol itself, with physical layout, power isolation, transmission integrity, and clock management all affecting sound quality.

Clocking And Timing

Clean timing is as important as clean data. The DAS typically contains the master clock that initiates audio transmission, and any phase noise (jitter) in this clock or its distribution network can degrade musical flow and spatial coherence. Designers must carefully manage clock quality, power supply, and layout, and decide whether the DAC will assume master clock control (where the DAS clock is slaved to the DAC’s clock).

Software Matters

Software architecture affects both timing and noise. A lightweight operating system with a clean audio stack and an optimised buffering strategy reduces CPU wake-ups and timing instability. Playback software must be stable, efficient, and avoid unnecessary system calls. Careful tuning here makes a measurable and audible difference.

Voicing And Final Tuning

All digital components, such as CPUs, clocks, and interfaces, produce electrical noise. Even after extensive efforts to reduce that noise, some remains. The key to achieving truly exceptional sound lies in how that remaining noise is managed, and so one final step remains.

Voicing – a process of listening and adjusting.

What we’ve discovered is this:

• The frequency of the noise matters. Noise that overlaps with the system’s data or clock frequencies, especially around the sample rate, does the most harm. By adjusting the design to shift noise to higher, less critical frequencies, we get significantly better sound. The timing becomes cleaner, and the DAC performs more accurately.
• The interaction between noise sources matters. Different parts of a system generate different frequencies of noise, and these can interact in ways that create interference at new frequencies. We tune the system so that these interactions are minimised. This isn’t just noise reduction, it’s about shaping the noise to make it less disruptive.
• The type of connection matters. Higher-speed interfaces like Gigabit Ethernet and PCIe generate much more electrical activity than slower ones like USB or 10Mbps Ethernet. That activity increases system noise and makes it harder to maintain clean timing. In our designs, slower and simpler interfaces are often used because they allow the system to deliver music with greater clarity and coherence, but this must be balanced with practical considerations of availability of parts and compatibility with other devices.

Together, these approaches are what we mean by voicing the system. It's the final and most important stage of the design process, and it can only be employed in a Total System DAS, not in a system of cobbled-together parts. Shifting noise spectra can significantly improve the naturalness of the sound and profoundly impact the emotional impact of the music on the listener. This can require multiple adjustments to the DAS design to get things right. Oladra has been perfecting its techniques in this area since 2009 and this is the main reason for Oladra’s reputation for communicating the emotional content of the music better than any other DAS brand.

More Design Detail

3.1: Core Design – How CPU and chipset design decisions affect DAS performance.

3.2: Audio Output Design – Making the most of the digital audio connection.

3.3: Clock Design & Slaving – The complexities of clock implementation.

3.4: Power Supply Design – The fundamental influence of the DAS power supply.

3.1 Core Design

The design of a Digital Audio Source (DAS) starts with the selection of CPU and motherboard architecture. These choices underpin every electrical, thermal, and timing behaviour of the system. Performance, noise, and system layout are all directly shaped by this foundation.

CPU Choice

The CPU determines the system’s responsiveness and controls playback, file access, and network communication. A more powerful CPU delivers smoother performance and better multitasking, but usually at the cost of higher electrical noise. Noise from high-frequency power draw and switching activity spreads across the system and can affect sensitive audio circuitry.

High-end DAS designs must find a sweet spot: using enough CPU power to ensure stable and timely operation, while limiting peak load and background activity. Some designs also throttle CPU cores or use software controls to manage activity cycles.

Architecture differences have so far favoured Intel, but ARM is promising:

Intel (x86) Key Characteristics For Audio

• Clock & PLL Quality: Intel CPUs generally feature lower jitter in internal clocks and more stable PLLs, especially important for USB, PCIe, and Ethernet performance.
• Mature Power Management: Their SpeedStep/C-states are predictable and can be manually tuned or disabled to reduce noise in real-time audio contexts.
• Superior USB Host Implementations: USB is deeply integrated into Intel chipsets, meaning:
  • Lower jitter at host level
  • Better isochronous handling
  • Fewer packet retransmissions
• Thermal Stability: CPUs on 10nm and 7nm+ process nodes (e.g., Alder Lake and Raptor Lake) maintain excellent thermal behaviour, helping clock oscillators and RAM timing stability.

Chipset and I/O Topology

The CPU doesn’t work alone. Motherboard chipsets manage communication between CPU, memory, and peripheral interfaces. Their internal layout affects how signals and noise are routed through the board.

Key considerations include:

• Avoiding shared data buses for high-speed interfaces (e.g., USB and PCIe)
• Minimising latency and electrical interference on USB audio paths
• Preventing internal switching noise from Ethernet or SATA controllers bleeding into audio subsystems
• Selecting chipsets that allow disabling of unnecessary features in BIOS or firmware

Faster Interfaces Generate More Noise

Interface choices can be critical. If you build a basic computer you want to use the latest, greatest and fastest interfaces. But this is precisely the opposite of what you should use for digital audio.

High-speed interfaces like Gigabit Ethernet and PCIe:

• Switch at higher frequencies (e.g. GHz range).
• Require more aggressive edge rates (fast voltage transitions).
• Engage more lanes and differential pairs, each generating electromagnetic noise, ground bounce, and power rail modulation.

This creates:

• More broad-spectrum noise, including harmonics and intermodulation products.
• Greater coupling into adjacent circuits, including audio clocks and power domains.

Slower interfaces like 10Mbps Ethernet, SATA3 or USB2.0:

• Switch less often and more gently.
• Generate less switching noise, and what they do produce is often at lower, more benign frequencies.
• Are less likely to excite resonances or couple into DAC-sensitive areas.

Faster Interfaces Increase Activity

When you use a faster interface, the chipset must work harder:

• More buffers and machine state toggling.
• DMA (direct memory access) activity increases.
• CPUs and controllers may wake more frequently or stay in higher power states longer.

That additional activity creates more low-level internal noise, especially:

• Ground plane turbulence.
• Power supply ripple.
• Clock domain crosstalk.

All of this affects the electronic environment of the DAC, especially in systems where the DAC is USB-attached or shares a ground or power source with the host.

Faster Interfaces Are Less Deterministic

Faster links like PCIe and GbE don’t lead to cleaner timing. Gigabit Ethernet is inherently packetized and bursty, and a 10Mbps stream can deliver data more smoothly, creating less burst noise.

USB has better isolation than PCIe and can be run through simpler and more predictable firmware layers (especially on Linux ALSA), whereas PCIe is deeply integrated into the system fabric, with higher interrupt load, memory access, and timing sensitivity.

Lower bandwidth I/O often behaves more like a steady stream, while higher bandwidth I/O behaves more like a storm of bursty electrical events.

In general, more speed leads to more noise, more activity, more complexity, and less predictability. The added noise finds its way into the clock domain, reference voltages, or ground plane of the DAC.

That is why some of the best digital audio designs intentionally use lower-speed Ethernet, USB instead of PCIe, and slower CPU cores and bus speeds.

Memory Layout and RAM Behaviour

RAM access can be a major source of noise and timing disruption. The physical layout, memory speed, and controller behaviour all influence performance.

DAS designs typically favour:

• Using lower-frequency RAM to reduce power noise
• Single-rank configurations to limit switching activity
• Minimising simultaneous memory and I/O demands to reduce interference

USB and PCIe Implementation

Many DACs rely on USB input, making the DAS USB implementation critical. Poor layout, shared buses, or power contamination can result in degraded timing.

Designers should:

• Use dedicated USB controllers where possible
• Isolate USB power paths
• Route USB signals with controlled impedance and short traces
• Physically separate USB and high-current power paths
• Carefully plan PCIe lanes to avoid crosstalk and ensure low-noise expansion.

Thermal and Mechanical Layout

Heat influences clock stability and power behaviour. High temperatures lead to increased phase noise and component stress.

Key strategies include:

• Passive cooling with careful heatpipe or heatsink design
• Using chassis ground as part of EMI control
• Ensuring even airflow to avoid hotspots on the motherboard

BIOS/UEFI Configuration

The BIOS can be configured to disable unused ports and features, reducing background activity and improving timing predictability.

Recommended settings include:

• Disabling onboard audio, Wi-Fi, Bluetooth, and unused SATA/USB ports
• Locking CPU frequency (turning off turbo modes)
• Enabling power-saving states that reduce unnecessary switching

Conclusion

Optimising CPU and motherboard architecture in a DAS is not about chasing benchmarks. It’s about creating a quiet, stable, and efficient computing platform that supports the timing and noise performance of the downstream DAC. The most revealing systems show clear differences between designs, and getting this foundation right is essential.

Addendum: CPU and Motherboard Architecture Design Insights

CPU Selection and Power Profile

• Choose a CPU with low thermal design power (TDP). High-TDP processors can introduce excess switching noise and complicate power delivery.
• Consider mobile-grade CPUs where practical. These offer adequate performance with reduced electrical noise.
• Avoid high-TDP CPUs that require aggressive power management or cause unnecessary background switching activity.

Chipset and I/O Integration

• Use chipsets that offer the minimum required functionality. Unused peripherals and high-speed interfaces (like PCIe lanes or GPU cores) may still be active and generating noise.
• Select integrated I/O that simplifies layout and avoids additional clock domains or switching regulators.

BIOS Configuration and OS Interaction

• Disable unused peripherals (e.g. audio codecs, wireless, serial ports) in BIOS to reduce background processes.
• Lock processor clock speeds where possible to eliminate frequency scaling jitter.
• Tune OS scheduling and process priorities to ensure deterministic system behaviour.

Memory and Storage Design

• Choose low-power RAM and avoid memory configurations that require high-speed interfaces or aggressive prefetching.
• Prefer SSDs with low controller activity and minimal background write behaviour.
• Locate storage away from sensitive analog areas or high-speed data lines.

System Layout and Grounding

• Use short, consistent ground paths between CPU, memory, and output sections.
• Ensure the grounding strategy supports separation of digital noise domains from audio output references.
• Avoid split planes unless absolutely necessary. Use a unified ground strategy where timing and audio coherence are priorities.

System Determinism and Scheduling

• Pin critical processes to fixed CPU cores (core isolation) to avoid thread migration jitter.
• Eliminate background services and unpredictable interrupts.
• Consider using real-time Linux kernels or stripped-down OS builds tailored for audio.

Heat Management

• Design passive or low-turbulence cooling that avoids fan-induced EMI or vibration.
• Ensure thermal design maintains the processor in its optimal steady-state frequency and voltage without throttling.

In summary, a high-end DAS must be treated not just as a digital transport, but as an integrated signal conditioning system. CPU selection and motherboard design are central to this task, and must be addressed as carefully as the more obvious output stages. Every signal path, clock domain, and power rail must be understood in context of how it contributes to, or interferes with, the music signal's timing and integrity.

3.2 Audio Output Design

The connection between a DAS and a DAC is critical. It's where the digital audio stream crosses into the DAC's clock domain, and the way this handoff is conducted has a significant effect on sound quality.

A common claim is that “The DAC reclocks everything, so jitter or noise upstream doesn’t matter.” This would be true if:

• The DAC receives a perfect data buffer via a medium with no timing sensitivity,
• The DAC is 100% isolated from the source (electrically and logically),
• The DAC has perfect clock recovery, with zero sensitivity to input variations.

But none of these conditions are fully met in any real-world system. This is what makes the design of the DAS critical for high-end audio.

This article outlines the differences between output interfaces, including Ethernet, USB, I²S, S/PDIF, and AES3, and explains why implementation matters as much as interface choice.

The Role of the Output Interface

The DAS doesn’t just deliver digital data, it also delivers electrical noise, timing irregularities (jitter), and radiated interference. A good output design aims to address all of these, so that the DAC can operate in a stable, low-noise environment.

Each interface handles timing and noise differently, and these characteristics influence how much of the DAS performance reaches or disrupts the DAC.

Ethernet

Ethernet is asynchronous and packet-based. The DAC typically buffers and re-clocks the incoming stream, which in theory decouples it from upstream timing errors.

However, the big downside of Ethernet is that it introduces its own issues in the worst possible place – inside the DAC:

• High-speed PHYs generate switching noise and EMI
• Electrical noise can propagate through ground paths and shielding
• The physical layer is often powered by noisy internal supplies
• Streaming traffic is bursty and processor-intensive

While Ethernet offers convenience and broad compatibility, it is inherently the noisiest of the digital audio interfaces. Great care is needed in grounding, isolation, and buffering to achieve good sonic results.

S/PDIF and AES3

These serial interfaces embed the clock within the audio signal, meaning the DAC must extract timing from the stream. This makes them more vulnerable to jitter and phase noise being introduced upstream.

However:

• They are simple, transformer-isolated, and easy to implement cleanly
• With careful output design, timing jitter can be kept low
• In some systems, the DAC can slave the DAS clock to avoid clock recovery altogether

AES3 typically performs better than S/PDIF due to balanced transmission, galvanic isolation and superior connector/cable hardware. Both can sound excellent when expertly implemented.

USB

USB audio is asynchronous. The DAC sets the pace, and the DAS sends data on request. This allows very precise timing, in theory, but practical results depend on effective implementation to address certain issues.

• USB is a complex protocol with high packet rates and continuous activity
• It requires significant processing overhead and is sensitive to ground noise
• Power lines and data lines can couple noise if not well isolated

I2S

I²S is not a transmission protocol, it is a signal-level interface designed for internal circuit-board-level audio connections, and so to be used in a DAS/DAC scenario it needs to adopt a suitable transmission protocol to deliver data over a distance. It benefits from the fact that it carries data, clock, and word select lines separately.

• I²S allows the DAS to synchronise to the DAC’s master clock
• It avoids the need for clock extraction or packet decoding
• When transmitted over cable (e.g. via HDMI sockets), great care must be taken with impedance matching, shielding, and line length

Hierarchy of Performance

From extensive testing across several DACs, when each interface is properly implemented, we found we preferred the outputs in the following order, with I2S offering the most direct and accurate signal handoff. This translates to more natural sound and emotional connection without any sonic compromise to detail or dynamics:

I2S > USB > AES3 > S/PDIF > Ethernet

This ranking assumes a high standard of design at both ends of the interface. In reality, poor implementation can easily flip the order.

That’s why interface design matters at least as much as interface choice.

Designing for Performance

No matter which interface is used, we apply consistent principles:

• Ultra-low-noise, dedicated regulators for each interface
• Shielding and physical separation to reduce coupling
• Reclocking where appropriate
• Controlled impedance routing
• Clock slaving support when available

The goal is always to protect DAC performance by delivering the cleanest, most stable signal possible.

Why This Matters Sonically

The interface defines what reaches the DAC: not just the data, but also the clock edges and the noise spectrum surrounding them.

A well-designed output preserves timing, reduces intermodulation, and lets the DAC operate as intended, producing more natural tone, better dynamics, and deeper emotional engagement.

Addendum: Audio Output and Connection Design Insights

Interface Selection Strategy

• Understand that not all interfaces are equal. I2S, USB, AES3, S/PDIF, and Ethernet vary in timing precision, noise susceptibility, and how they relate to the DAC's internal clock.
• When each is implemented to a high standard, there is generally a hierarchy: I2S is best, followed by USB, then AES3, S/PDIF, and lastly Ethernet.
• The above holds true only when implementations are clean. Flawed implementations change the hierarchy.

Clock Relationship and Interface Behaviour

• Determine whether the interface is clock master or slave. Interfaces that require the DAC to extract the clock (e.g. S/PDIF, AES3) impose extra timing uncertainty.
• Consider whether the DAC can provide clock data back to the DAS (clock slaving), improving timing performance for I2S and sometimes USB.
• Avoid clock domain crossings where possible or handle them with high-integrity reclocking solutions.

Signal Integrity and Transmission

• Pay close attention to trace impedance, cable type, and connector grounding. Digital signals are affected by transmission quality.
• Match line driver strength and rise/fall times to the load and cable length.
• Avoid radiated emissions by careful shielding, connector choice, and layout geometry.

Electrical Noise and Leakage Paths

• Identify how power, ground, or digital reference signals may be shared across DAS and DAC.
• Use galvanic isolation where needed, but don’t assume it solves all noise problems. Leakage through other cables or via shielding paths is common.
• Physically separate the output modules or ports to limit mutual coupling.

Modularity and Output Switching

• Allow users or designers to select output types via software or hardware jumpers.
• Disable unused outputs to reduce emissions and avoid coupling into active paths.

Compatibility and DAC Tuning

• Understand that DACs have different receiver characteristics. Test outputs across common DACs and refine transmission behaviour to suit real-world scenarios.
• Provide clear guidance to users on optimal cabling, settings, and interface selection based on their DAC, in particular to only make one digital audio signal cable connection at a time.

The audio output of a DAS is not a commodity interface. It must be optimised with the same care as clocking or power design. The choice and execution of the output connection is not the end of the signal path, it is the beginning of the DAC's job, and the quality of that beginning defines the system's musical integrity.

3.3 Clock Design & Slaving

Clock Design

Attention to clocking in the DAS is important, despite the common claim that ‘the DAC reclocks everything, so upstream clocks do not matter’.

Even if clocks are logically separate, they are often physically coupled.

• Jitter doesn’t only come through timing lines, it comes through:
  • Voltage rails
  • Ground return paths
  • EMI/EMC interference
• Upstream systems with noisy VRMs, poor clock layout, or aggressive power states can inject noise that rides the same ground or power domains as the DAC's oscillator or PLL.

This causes phase modulation, timing skew, and internal jitter at the DAC level, despite subsequent reclocking.

• Voltage and ground noise can reduce the stability of the oscillator that provides the timing reference to the clock circuit.
• Radiated noise will interfere with the clock circuit traces and cause phase noise (random variation in timing).
• Radiated noise will interfere with the analogue circuits.

Consideration should be given also to intermodulation and non-linear effects inside the DAC.

• DACs are not perfectly linear. They are analogue circuits with sensitive current mirrors, switches, and timing comparators.
• Noise from upstream sources, even if "inaudible," can cause:
  • Intermodulation with the clock signal
  • Modulation of the reference voltage
  • Disturbances in comparator thresholds

These effects don’t show up as “lost bits”, but as:

• Shifts in transient response
• Smearing of microdynamics
• Flattening of timbral depth

Clock Saving

Clock slaving means making the DAS follows the DAC’s master clock when generating its digital output. This gives the DAC full control of timing and can reduce jitter at the conversion point.

Without slaving, the DAS runs its own clock, and the DAC must adapt, often by extracting the clock from the signal or using buffers and resamplers. These methods work, but are less precise.

Why It Matters

Jitter is a form of timing error. Even tiny deviations in the arrival time of digital audio samples can introduce audible distortion, particularly in high-end systems.

When the DAC is clock master, and the DAS is slaved to it, the handoff becomes cleaner:

• No clock extraction is needed
• The DAC's timing isn’t disrupted by upstream jitter
• Fewer layers of buffering or resampling are required

This allows the DAC to perform its job with fewer compromises, producing more coherent and emotionally engaging playback.

Interfaces And Slaving

Not all interfaces support clock slaving. Here’s how it works for each:

- I2S

This interface allows full clock slaving. The DAC can send its master clock back to the DAS, and the DAS uses it to time the I²S output.

This is the cleanest method. It avoids clock extraction and keeps all signals in phase. If the physical implementation is clean (short lines, correct impedance), I²S with clock slaving will give the best performance, but separation of the DAS from the DAC requires longer lines. Transmission protocols introduced to connect I2S over a longer distance need to be appropriate and expertly implemented to preserve the clock precision.

- USB

USB Audio Class 2 is asynchronous, meaning the DAC sets the timing by requesting data from the DAS. While this resembles slaving in practice, it is mediated through buffers and drivers, not a direct clock signal. This indirectness adds complexity, and the results depend heavily on the USB stack and implementation quality.

- S/PDIF And AES3

These embed the clock in the data stream, meaning the DAC must extract the timing. This is inherently less precise, and is subject to phase noise introduced upstream.

Some DACs allow slaving the DAS through a return clock path, but unfortunately few DACs implement this. If available, this removes the need for clock extraction and significantly improves timing.

- Ethernet

Ethernet is packet-based and decoupled from timing. The DAC rebuilds the clock using buffers and typically synchronises to a local master clock.

There is no direct way to slave a DAS to a DAC via Ethernet. In systems using protocols like AES67 or Ravenna, synchronisation via network timing protocols (e.g., PTP) is possible. But this is only relevant in a recording scenario, and not useful in a DAS/DAC scenario because they only reduce long-term drift, and do not offer precise audio sample timing.

Slaving Is Not Always Better

Clock slaving can improve performance if:

• The DAC has a high-quality master clock
• The clock signal is cleanly returned to the DAS
• The DAS output can lock reliably to that clock

Poor implementation, noisy return paths, unstable PLLs, or poor layout can make things worse.

In practice, clock slaving is only beneficial when it’s done very well. Otherwise, a well-implemented asynchronous system (e.g., USB with clean power and reclocking) may outperform it.

What We Do

Our DAS output options include support for clock slaving for I2S, as well as AES3 and S/PDIF. DAC compatibility is critical and so we endeavour to work with relevant DAC manufacturers to develop and test our interfaces. Implementation focuses on:

• Minimal jitter in the slaved output
• Shielded, impedance-controlled clock return paths
• Phase alignment between clock and data lines

Where slaving is not included with the installed output option, we focus our design on ultra-low jitter and clean isolation.

Why This Matters Sonically

With slaving, the DAC’s internal timing circuit is in full control. Music flows with greater ease and coherence. Transients are sharper and rhythmic tension feels more alive.

It doesn’t change the data, it simply allows the DAC to breathe, without compensating for upstream instability.

Addendum: Clock Design Insights

Clock Type and Specification

• Select oscillators not just for ppm accuracy, but for extremely low phase noise in the audio-relevant frequency ranges and for high levels of electrical and temperature stability.
• Understand that absolute frequency accuracy (e.g. atomic clocks) is far less relevant.

Clock Location and Distribution

• Place the master clock as close as possible to the component that initiates audio data output.
• Avoid long PCB traces from clock source to destination. Reflections and delay distort clock integrity.
• Use controlled impedance and differential routing if possible.

Clock Power and Isolation

• Power clocks with extremely quiet and well-isolated supplies. Clock circuits are sensitive to both power noise and ground instability.
• Separate clock grounds from noisy digital domains, while ensuring they do not float dangerously.

Clock Multiplication and Division

• Avoid noisy PLLs and frequency multipliers unless absolutely necessary. Clean base clocks are preferable to multiplied ones.
• If multiple frequencies are needed, use ultra-low-noise clock distribution chips designed for precision audio or instrumentation.

Clock Domain Management

• Avoid unnecessary crossings between clock domains. Each crossing introduces the need for reclocking or buffering, increasing jitter.
• Where crossings are necessary, use FIFO buffers or asynchronous interfaces with very high clock precision.

DAC Clock Slaving Strategy

• Consider whether the DAC will act as clock master. If so, enable return clock paths from the DAC to the DAS, where interfaces allow.
• For I2S and USB, slaving to the DAC’s master clock can dramatically reduce output jitter.
• For S/PDIF and AES3, the DAC extracts the clock, so upstream jitter directly affects DAC timing.

Environmental and Mechanical Considerations

• Reduce temperature swings and vibration around the oscillator; both affect stability.
• Physically isolate clock modules if needed, using damping materials or separate subchassis.

Clock design is an invisible art. The ear is more sensitive than the oscilloscope, and clock errors don’t show up in frequency response graphs. A good DAS design prioritises clean, well-positioned, and well-powered clocks as a fundamental contributor to sonic realism. You may not even be able to hear the difference, but if you enjoy music at all, you will feel it.

3.4 Power Supply Design

Power quality shapes everything in a DAS. It defines how quietly and accurately each subsystem behaves and ultimately affects how well the DAC performs.

This article explains the key principles behind DAS power supply design, with a focus on keeping electrical and magnetic noise low, timing clean, and performance consistent.

The Origin Of All Noise

Every DAS function, from CPU instructions to memory refresh to Ethernet streaming, starts with power. If that power is noisy, every subsystem becomes a noise generator.

Ripple, transient instability, ground noise, and magnetic flux leakage all radiate through the DAS, via conductive, capacitive, and inductive paths. Each of these mechanisms contributes to degrading the DAC's ability to preserve timing and detail.

Linear vs Switched vs Battery

Linear power supplies are simple but limited. They produce smooth output under static loads, but under dynamic conditions, like a CPU ramping up, they become unstable and they cannot respond quickly enough to fast digital loads.

Battery supplies typically also lack fast transient response, and their grounding reference can drift or create unbalanced return paths, compromising timing consistency. Batteries are also neither noise-free nor stable under load, and they introduce charging interference and long-term inconsistency.

The best designs employ switched-mode power supplies (SMPS), designed specifically for computer audio, employing synchronised switching, careful control of magnetic flux paths, and extensive filtering and shielding. While generic SMPS designs are noisy, especially in the ultrasonic band where DACs are sensitive, expertly designed switched-mode power supplies offer superior regulation and lower impedance. But the most important benefits are stability combined with fast transient response, which crucially reduces noise and defines data transitions more precisely in time.

Avoiding Magnetic and Conductive Crosstalk

Good power design considers not just what voltage is delivered, but how magnetic and electric fields propagate through the system.

It is important to contain and redirect magnetic fields from transformers and inductors, ensure loop areas are small to minimise flux leakage, and to shield sensitive areas.

Conducted noise is reduced with local decoupling, multilayer power planes, and star grounding. Flux interactions between supplies and data paths must be carefully controlled through physical separation and routing of traces.

Fast and Predictable Delivery

A DAC’s jitter response can worsen dramatically if the DAS behaves erratically, even for a few microseconds. Oladra switched-mode power supplies are designed for fast recovery and low impedance across a wide frequency range, and minimal overshoot or ringing under load.

The aim is stability, not just average performance, because predictable behaviour across varying demands improves DAC clock stability and increases sound coherence.

Multi-Rail Isolated Domains

A DAS needs multiple supply rails, with each rail being designed for its specific role, ensuring isolation where needed, for example between processor, memory, I/O, and audio interfaces.

Cross-domain coupling is avoided by isolating regulation stages, filtering shared returns, and designing for ground domain control.

Why This Matters Sonically

Poor power allows timing instability, radiated EMI, and intermodulation that blurs the signal arriving at the DAC.

Clean power lets the DAS behave quietly and consistently, enabling the DAC to reproduce music with better precision, dynamics, and emotional clarity.

The result isn’t subtle. With good power design, the music feels more fluid and immersive, as if more life is present in the playback.

Addendum: Power Supply and Distribution Design Insights

The five target outcomes are:

• Fast transient response – to handle dynamic CPU and I/O loads without voltage sag or recovery delay.
• Low output impedance – to prevent fluctuations that modulate clocks and data lines.
• Minimal conducted and radiated EMI – to stop switching nodes polluting other circuits.
• Magnetic containment – to prevent fields from leaking into audio paths.
• Subsystem isolation – to avoid noise cross-coupling between domains.

Power Supply Type Selection

• Do not assume linear power supplies are superior. Modern SMPS, when expertly designed, can outperform linear supplies in noise performance and regulation.
• Evaluate the total system behaviour, not the textbook ideal. Battery supplies, for example, can introduce a range of performance limitations or instability.
• An expertly designed SMPS can be faster than a linear supply, quieter than a battery (after filtering), more compact and more efficient.

Local Regulation and Domain Separation

• Use cascaded regulation stages to locally control voltage domains close to their point of use.
• Avoid sharing regulators across noise-sensitive and noise-generating domains.
• Use ultra-low-noise LDOs only where necessary; they may not outperform good switched-mode regulators if layout or grounding is flawed.

Grounding and Reference Strategy

• Plan a unified grounding structure that avoids parasitic current loops.
• Ensure digital, analog, and interface grounds interact in controlled, predictable ways.
• Avoid star grounding where fast-switching domains are present, as it can introduce floating potentials.

Isolation and Containment

• Use galvanic isolation where functional separation is required (e.g. Ethernet PHYs, USB isolators), but ensure the isolation barrier doesn’t introduce leakage or timing issues.
• Maintain physical separation between high-current and low-level circuits to reduce radiated and conducted coupling.

Power Path Routing and Connector Choice

• Avoid narrow traces or underspecified connectors, as these can create voltage drops and heat.
• Minimise the length of supply and ground returns. Symmetry and proximity improve EMI performance.
• Ensure each return path matches the forward path to avoid imbalance.

Noise Control and Filtering

• Apply filtering at both input and point-of-load stages.
• Use ferrites, common-mode chokes, and multilayer ceramic capacitors (MLCCs) with care. Poor placement can make them ineffective.
• Evaluate conducted and radiated emissions as a system.

Supply Response and Load Behaviour

• Assess power supply behaviour under transient and nonlinear loads, not just steady-state.
• Systems with CPUs, storage, and networking introduce complex, dynamic load profiles.

The power supply is the backbone of any DAS, and it cannot be treated as a passive utility. It is an active participant in system behaviour. Design choices here are rarely obvious, sometimes counterintuitive, and must be verified in the context of overall audio performance, not in isolation.