Ok, here goes: I’ll setup a system over wifi that allows me to enable / disable, on the fly, wifi while playback ensues.
You tell us when wifi was on or off. Music will still continue to play without break.
the optical isolation (powered by LPSes) bough significant improvement in sound quality - with effects described in my previous post .
Changing low cost optical isolation to Etherregen based isolation bough only very small improvement here.
I’m not using any SMPS on audio power line and also having all power for audio regenerated. That helps to improve transparency to the level that one can easily hear changes done in digital domain (or anything else what is galvanically connected to audio chain)
Although ethernet is natively isolated via transformers, it is not 100% isolation , even when using additional passive isolation device (Pink Faun eth isolator) - when switching to optical it still sounds more clear , tighter and having more top&bottom clarity
It’s -140dB for vanilla copper though. My other quandary is that you could be using a setup that what you are listening to at the moment was delivered 60-90 seconds prior and there is nothing active on the wire.
If the traditional pivot of it’s a layer 1 issue then we have another question to ask. If you have a streamer with local storage does the mere fact of having it plugged in alter the sound. If that’s the case then we should be able to play from direct attached storage and tell when the copper RJE is in or out of situ even when blinded.
1 Like
To clarify:
- my player is either rpi4 or upboard , both with super minimalistic realtime kernel, RFI shielded chips and supercapacitor LPS and in case of rpi also reduced voltages and frequesncies
- both have option to store whole track into RAM before playback
- both are connected via filtered ethernet bridge (running on dedicated NUC with Roon Core) to allow only necessary LAN traffic to the streamer
There are actually 3 different ways I can listen:
- either thru RoonBridge - which makes continuous NW traffic
- via mpd with - which makes semi-continuous NW traffic
- via mpd with playback from RAM with only NW traffic being status info about playback position etc.
Playing from RAM, with ethernet unplugged is only a smidge better than Playing from RAM and having ethernet connected (EtherRegen + Optical isolation). (i would not be able to distinguish in A/B test)
Playing from RAM with ethernet plugged directly into main ethernet switch (circumventing optical isolation) immediately and very noticeably degrades SQ. Surely even no ABX needed.
Now, for the non-RAM based playback (RoonBridge) - whatever i tried it is always noticeably different (worse) than RAM playback - even with very cautious CPU-pinning setup , to minimize jitter on CPU managing audio USB interface. I would dare to say I would be able to ABX them without big issue.
So there are at least 2 different things - one is related to electrical (and EMI/RFI) properties of whole audio chain, including also ethernet and the other one is ability of streamer to suppress network activity impact on SQ.
My LAN is having 3 switches and CAT7 cabling in-wall, about 20 devices connected at time.
P.S. The worst part of this is that all of the described can be heard even with another USB-to-I2S isolator + reclocker (Matrix Audio SPDIF2) between streamer and DAC - which again in theory should not happen at all. 
I decided to come back and reply to comments that I previously decided not to. So here goes…
Bill, for many years I agreed with you… bits are bits and so many of the audiophile favorite tweaks were a waste of money. I did not believe in linear power supplies; switching PSs were just fine… until I tried a linear PS on my SonicTransporter. Immediate improvement in sound. Odd thing was that the main improvement was a widening of sound field. When listening to headphones the instruments that were panned full left and right sounded like they were coming from behind my ears. Interesting and this effect became even more obvious as I made other mods to my system.
Within in a couple of days of installing the fiber mod to the Core, my wife commented that vocals sounded better than ever before… and this was on Sonos Ones. She has no interest in any technical reasons, she just said she heard what she heard, as did I.
Now there are limits to what I believe. That includes “audiophile” power cords. Anyone who believes that several feet of uber pricey power cords plugged into builder grade Romex makes an audible improvement is imagining things. There is nothing audiophile about house wiring, the electrical grid and certainly not power generating stations. I spent 40+ years designing electrical systems for power generating plants and that is most certainly not an audiophile environment!
So, I should invest in setting up a true A/B switching setup, which can be pricey, to judge the difference a $100 mod makes? I think not.
1 Like
Is that the setup you are using to base your opinion that my conclusions are imaginary? If so, perhaps a setup that is capable of higher resolution would change your mind. Just saying…
This reminded me of a rather dramatic failure of a three phase, 900MVA, 230/13kV transformer at a power plant owned by my employer at the time. Happened late afternoon on Halloween, 2005. LV bushing failed, causing the tank to rupture, spilling a lot of oil and igniting a very large fire. Water suppression system took care of the fire after several hours. Took the plant off line for quite a while.
How can that be? Seems that there will be a drop in playback as the Wifi re-establishes the connection.
I should have added that the source for the vocals was the SonicTransporter playing to the Sonos Ones with the fiber optic link to the SonicTransporter installed.
Let me know if you are game to setup your 3rd option with iPerf and SSH. I would like to see what happens when I kick off some traffic during playback to see what it sounds like.
sure i can, I even have some dumps from the past including detailed per-cpu stats from netdata like this one i made years back
you can also tell me which EtherRegen port to use 1gpbs/100mbps
we can have also look into tshark RTT analysis - but this is not telling much, more interesting is the USB interface
Hey, come on, for bitwise transport there’s nothing wrong with either of them! And the CCA even provides your preferred fiber isolation! 
Sure. Again, it’s all the same bits going to the DAC, so I don’t see how the resolution could be “higher”. Yes, the CCA is sample-rate limited, but not the Pi.
All this is to be expected. What I’m wondering about is either instrumented, or by ears, if we generate data traffic with iPerf will it show with properly calibrated instrumentation.
I’m a WireShark ninja and use it all the time in anaylsis. But that’s not going to tell us what is happening at the DAC output.
1 Like
He’s not talking about sample rate and bit depth. He’s calling whatever equipment you have schiit.
1 Like
The legendary highly resolving system, which can reveal veils hitherto unperceived! Another myth, I’m afraid.
You know, it’s some explanation of the supposed mechanism that I really miss. The same bits come out the fiber connection from the CCA that come out of any other streamer; how could the sound be changing?
1 Like
(10 characters)
Maybe the transition “electron - photon - electron” does something dastardly 
2 Likes
Not if you buy my new QUD (Quantum Untangling Device), say goodbye to all that uncertainty and hear the sound as the universe intended, before indecision set in…
2 Likes