Improving on the stream of a combined streamer dac


I recently (foolishly) compared the streamed sound of my Electrocompanient

ECM 1 Mk II to internally stored digital downloads on the same ( ie not stored in my Small Green Computer server).  I was surprised how much better the downloaded files sounded.  Perhaps, I was naive to think otherwise.  So, I started looking at old audiogon threads to see if I could figure a quick and easy fix to up the sound of the streamer but found little that made sense or was affordable.  My question is whether I should continue to search, or if I should take it for what its worth and just listen to the music.  The rest of my digital front end:  Small Green computer server,  a Trendnet unmanaged switch and the ECM.  I use a Luna Orange ethernet cable from switch to ECM and a Purist Audio cable to the SGC.  Thanks for your advice and happy holidays to all.

rivinyl

Hi @rivinyl,

I’ve read through this thread and I want to validate what you are hearing. You are not "foolish" or "naive" for noticing that local files sound better than your stream. You have stumbled upon the "dirty little secret" of high-end streaming that standard advice about cables and switches often fails to solve.

The problem isn't your Electrocompaniet, and it isn't that "Roon is garbage" as some have suggested. The problem is that our home networks are too fast for digital audio.

The "Why": Bursts vs. Flow

Because a 1 Gbps LAN is so fast, standard protocols (like Roon’s RAAT or DLNA) send audio data in erratic bursts and pulses rather than a smooth, steady stream. Even with a great streamer, your endpoint's CPU has to constantly spike its activity to process brief but intense waves of incoming packets. These rapid fluctuations in processing load cause fluctuations in current draw, which generates low-frequency electrical noise (RFI/EMI) that the DAC is unable to completely filter out.

When you play a local file, the network overhead is lower, the processing is steadier, and the noise floor drops. That is why your downloads sound better.

The "Band-Aids" vs. The Cure

Most of the advice here—buying expensive silver ethernet cables, $1,000 switches, or filters like the ENO2—amounts to various "band-aid" fixes. They try to clean up the noise before it enters your streamer. But they do nothing to stop the streamer itself from generating noise as it processes that bursty network traffic.

The Solution: A 3-Tier Architecture with Diretta

If you want to beat the sound of local files, you need to address the root cause of noise at the source. The most effective way to do this today is by moving from a 2-Tier setup (Server -> Streamer) to a 3-Tier Architecture using the Diretta protocol.

Here is how the architecture works, which many of us are building using simple Raspberry Pi hardware running AudioLinux:

  1. Tier 1: Roon Core. Your heavy server. Keep this far away from your audio rack.

  2. Tier 2: Diretta Host (The Bridge). This machine connects to your LAN. It buffers erratic Roon data bursts and converts them to the Diretta protocol.

  3. Tier 3: Diretta Target (The Endpoint). This small device connects only to the Host via a direct, galvanically isolated link. It connects to your DAC or DDC via USB.

Why This Fixes Your Problem

The Diretta protocol uses a "Host-Target" model where the Host sends data in a continuous, precisely timed stream of transmissions. This "averages" the processing load on the Target, stabilizing the current draw and preventing the generation of electrical noise near the DAC. It also physically and logically isolates the DAC from your noisy network.

I have implemented this using Raspberry Pi 5s. With this setup, CPU usage on the endpoint drops significantly, and the power draw becomes a flat line. The result is an unusual "calm" from streaming. There's a "blacker" background and a natural ease to the music that makes standard streaming sound mechanical by comparison.

Before you spend another dime attempting to mitigate symptoms, I encourage you to try an architecture that sidesteps common sources of digital noise.

Happy listening.

Thanks for your perspective on streaming.  I do have a couple of questions regarding Diretta.  Is the Diretta Bridge (and Host) a device?  If so where does one find such a device.

And secondly, would a filter ( such as the Stack Audio Smooth Lan FIlter) which operates up to 100 mbps serve as a means to limit the speed that the streamer sees?  Or would downgrading one's internet to a lower speed bypass the issues poised by the super high speed internet we purchase (but don't really need )?

Many thanks for an interesting perspective.

You don’t need to downgrade your internet speed. 
All these filters such as smooth lan or Eno have an effect on sound quality. It isn’t always positive. 
Worth watching and reading reviews carefully. 
Hans Beekhuyzen channel on YouTube is a good resource. 
https://youtu.be/PSsKdAcwUL4

I am using a Netgear Orbi mesh system with 2 satellites.  My router is on one side of my house and my audio system is on the other.  I placed a mesh satellite in my Music room and run and ethernet cable (my Linn does not support wifi)  from the satellite to my Linn ADSM/3 streamer/DAC/preamp (1 ft cable). I stream Qobuz using the Linn app.  I also listen to music that I have on a Mac Mini that is used as a music server.  I am in my late 70's so I know m hearing is not as good as it once was, but I cannot tell any difference in sound quality between the 2 sources.

One question came to mind as I wait for the delivery of the Snake River Boomslang ethernet cable.  We didn't discuss the servers role in the overall sound of a digital front end.  I use the Small Green Computer sonicTransporter i5 server.  Would an upgrade at that position have any significant impact?  I have read that some servers have more built in noise reduction etc, obviating the need for better power supply, optical isolation etc.  Just thought I would ask the group while I await the cable.