A Home Network Experiment: Setup, Equipment Used, and Observations

[claudia5]

I’d like to share a recent home experiment we ran, mainly to document the exact setup, equipment used, and listening process.
This is not a claim about altered digital data, and not an attempt to challenge digital audio theory.
It is simply a transparent report of what we tested and what we observed.

---

Audio System Used​

Source / Streaming

• Bluesound Node ICON (used as a network streamer only)

DAC / Amplification

• Vincent SV-237MK integrated amplifier
• Internal DAC of the amplifier used (coaxial S/PDIF input)

Speakers

• Neat Acoustics SX3i
• No subwoofers connected during the test

---

Network Setup​

Baseline network chain

• Fiber internet into the house
• ISP ONT → router
• Router → standard gigabit Ethernet switch
• Switch → streamer (Node ICON)

Test configuration

• Router → existing gigabit Ethernet switch
• Existing switch → DELA S50 network switch
• DELA S50 → streamer (Node ICON)

Important clarification:

• The DELA S50 was not connected directly to the router
• It was placed after an existing standard gigabit switch

---

Cables Used​

• Standard Cat6 Ethernet cable (baseline)
• Higher-quality Ethernet cable between DELA S50 and streamer
• Later, two very high-end Ethernet cables were tested sequentially

Notes:

• Switching from the standard cable to a better one produced a small additional change
• Replacing that with a very expensive cable resulted in a slightly exaggerated bass balance, which was not preferred

---

Test Conditions​

• Same music tracks (well-recorded reference material)
• Same volume level
• Same playback software
• No DSP changes
• Bit-perfect playback confirmed
• Two listeners present
• Configurations were checked by reverting back and forth

---

Observations (Descriptive Only)​

Compared to the baseline setup, inserting the DELA S50 resulted in:

• More clearly separated layers in complex recordings
• Slightly improved spatial stability of vocals and some instruments
• Low-level details (decays, ambience) felt more continuous
• No increase in treble sharpness or fatigue
• Bass character remained broadly similar

These differences were subtle but repeatable.

---

What This Is Not Claiming​

• Digital data did not change
• This is not a “bits are different” argument
• This is not a universal claim
• This is not presented as proof of measurable improvement

---

Why I’m Posting This​

The purpose is documentation, not persuasion.
This experiment suggests that:

• Some system-dependent effects, if they exist, may not appear as classic additive noise
• Or the observation may be psychoacoustic or interaction-based

No final conclusion is claimed.

---

Closing​

If others have:

• Relevant measurements (J-test, phase noise, time-domain)
• Similar experiments with different results
• Or a clear technical reason why such observations cannot exist even in principle

I’d genuinely like to hear it.

 

[Chrispy]

Suspect could just be volume differences....you don't specify how you level matched

 

[OldTimer]

How about compared to using WiFi? No cable attached.

 

[voodooless]

a clear technical reason why such observations cannot exist even in principle

There are so many layers of abstract between your network hardware and what comes out of your speakers that it cannot possibly make a difference.

Worse: if you played the song multiple times, it is more than likely that the network was never used at all: the song was cached on disc or memory an played from there.

… it’s all in your head!

 

[claudia5]

Suspect could just be volume differences....you don't specify how you level matched

same volume.

 

[claudia5]

There are so many layers of abstract between your network hardware and what comes out of your speakers that it cannot possibly make a difference.

Worse: if you played the song multiple times, it is more than likely that the network was never used at all: the song was cached on disc or memory an played from there.

… it’s all in your head!

Just to clarify: my observation was not about data integrity or “bits changing”.

I fully agree that buffering, caching, and protocol abstraction decouple data timing and ensure bit-perfect playback.

However, software abstraction does not imply physical isolation.
Network PHYs, clocks, power rails, and ground references remain electrically active even when audio data is buffered or cached.

Any potential effect (if present) would be through analog coupling or reference noise — not through altered data.
That’s a hardware interaction question, not a networking one.

I’m not claiming this must be audible, universal, or measurable in every system.
Only that “data abstraction” alone does not logically exclude all physical interaction paths.

 

[Beave]

• Or the observation may be psychoacoustic or interaction-based

Uh, yep.

 

[amirm]

However, software abstraction does not imply physical isolation.
Network PHYs, clocks, power rails, and ground references remain electrically active even when audio data is buffered or cached.

Most players fetch way ahead of the playback. Unplug the cable while music is playing and see if you detect a difference. I have done this a number of times and the result is as predicted: no difference. I have also measured such effects and non exists. See this review: https://www.audiosciencereview.com/...el-bonn-n8-audio-grade-ethernet-switch.12360/

 

[RexrothPigeon]

This experiment suggests that:

• Some system-dependent effects, if they exist, may not appear as classic additive noise
• Or the observation may be psychoacoustic or interaction-based

Unless I'm missing something, I don't see the word "blind" anywhere, and because our innate biases and ocularcentrism are so all-powerful, we just can't really ward them off unless we contrive some tricks (like blind listening) to circumvent them. Therefore I think the only reasonable interpretation is that the differences you think you heard are psychosomatic.

 

[voodooless]

Any potential effect (if present) would be through analog coupling or reference noise

Do you hear noise? Do you hear it change if you swap the device? If not then that is not happening.

 

[claudia5]

I think we may be mixing two different questions here.


One is data integrity — and here we all agree: buffering, caching, and bit-perfect transmission ensure the audio data itself is identical. Unplugging the Ethernet cable during playback proves that point.


The second question is electrical noise coupling, which is orthogonal to data transport. Even with galvanic isolation, Ethernet interfaces still have parasitic capacitance, shared PSU paths, and ground/reference interactions. This allows conducted or radiated noise to exist in the system without altering a single bit.


Whether such noise is measurable or audible depends entirely on system-specific factors (DAC architecture, PSU design, grounding, clock sensitivity, analog stage immunity). In many systems it will be completely negligible — in others it may not.


So the correct statement is not “noise cannot exist because the data is buffered”, but rather:
noise can exist, even when data is bit-perfect — though it is often too small to matter.


That distinction is important, and it’s independent of blind testing or subjective claims.

 

[RexrothPigeon]

What? You claimed audible differences:

Observations (Descriptive Only)​

Compared to the baseline setup, inserting the DELA S50 resulted in:

• More clearly separated layers in complex recordings
• Slightly improved spatial stability of vocals and some instruments
• Low-level details (decays, ambience) felt more continuous
• No increase in treble sharpness or fatigue
• Bass character remained broadly similar

These differences were subtle but repeatable.

 

[RexrothPigeon]

Also, you need to read this:

Official policy on use of AI

We have had recent issues come about regarding use of AI in member posts. I personally think this technology can be useful to us. And at any rate, hate having to police members left and right on what they post. That said, here are my thoughts: 1. If you are going to use AI, make it...

www.audiosciencereview.com

 

[Keith_W]

Well the answer is pretty simple - regardless of what happens upstream, what matters is what comes out of the DAC. The network cable could pass through Godzilla's arse and out his mouth - if no difference is measured at DAC output, then there is no difference, period. You don't even need fancy test equipment - you can null the DAC output and if it nulls, there is no difference.

 

[claudia5]

Also, you need to read this:

Official policy on use of AI

We have had recent issues come about regarding use of AI in member posts. I personally think this technology can be useful to us. And at any rate, hate having to police members left and right on what they post. That said, here are my thoughts: 1. If you are going to use AI, make it...

www.audiosciencereview.com

i use AI to analyze my rew results ,to read graffs,to write better in english,to undestend things i ear from my setup ,and more

 

[RexrothPigeon]

i use AI to analyze my rew results ,to read graffs,to write better in english,to undestend things i ear from my setup ,and more

Just follow the rules. But more importantly, you can't say a single thing about "what you hear from your setup" since your listening "experiment" is sighted, which means you are using your eyes, your brain, your preconceptions, etc. and not your ears. It's simple.

 

[claudia5]

Well the answer is pretty simple - regardless of what happens upstream, what matters is what comes out of the DAC. The network cable could pass through Godzilla's arse and out his mouth - if no difference is measured at DAC output, then there is no difference, period. You don't even need fancy test equipment - you can null the DAC output and if it nulls, there is no difference.

I agree with you within the limits of linear, time-invariant analysis.
A null test is definitive for detecting additive noise and static amplitude differences.


Where it becomes incomplete is with time-domain and phase-related phenomena (e.g. jitter-induced modulation, reference noise coupling), which may not show up in a simple null or averaged FFT, yet can affect low-level detail through masking.


This is not about bit errors or mystical upstream effects — it’s about whether the chosen measurement captures all relevant mechanisms in a real DAC implementation.

 

[RexrothPigeon]

I agree with you within the limits of linear, time-invariant analysis.
A null test is definitive for detecting additive noise and static amplitude differences.


Where it becomes incomplete is with time-domain and phase-related phenomena (e.g. jitter-induced modulation, reference noise coupling), which may not show up in a simple null or averaged FFT, yet can affect low-level detail through masking.


This is not about bit errors or mystical upstream effects — it’s about whether the chosen measurement captures all relevant mechanisms in a real DAC implementation.

lol this is not using AI for your english. This is using AI to straight up compose responses. You didn't write or even think of this.

pure ai slop

 

[voodooless]

Also, you need to read this:

Official policy on use of AI

We have had recent issues come about regarding use of AI in member posts. I personally think this technology can be useful to us. And at any rate, hate having to police members left and right on what they post. That said, here are my thoughts: 1. If you are going to use AI, make it...

www.audiosciencereview.com

While we’re at it, I liked the conclusion it had:

Real experiments don’t read like legal disclaimers.

 

[claudia5]

What? You claimed audible differences:

I didn’t claim a universal or proven audible difference.


I reported a subjective listening observation, acknowledged potential bias, and discussed whether known physical mechanisms could in principle explain such observations — which is a different discussion entirely.


Conflating “I heard something under specific conditions” with “this is audibly different in all cases” is a category error.