Here I will trust your knowledge about ethernet. Still the theorie could work. When we see what is happening in the network as a side effect. I was grabbing in my memory. Where we tested a R2R / NOS dac and some I2S dac's. These dac work on an external clock which of course also can suffer and be influenced by noise as well.

When the only thing that can cause for example dark sound is the DAC itself. And let me first tell there must be several kinds of noise of which at least a few aint the problem, because they are canceled. Everything is always very pure, the only thing that troubles is the dark sound. This must mean the data is due to jitter only half transferred to analog. For example half way the clock cycle.

Then the question is why will this only happen after several hours and never immediately. Then it must be a clock that is slowly drifting and cutting data in parts and only convert half bits to analog.

I think a clock will drift due to another clock that's driving. This seem to be a clock in the ethernet system

Yes, i agree. Your delayed dark sound makes no sense in my mind.

The digital noise is rather associated with the opposite of "dark", sort of :-).

Note that for the DAC it is only ones and zeros in different combinations that it works with. There is no way that one certain aspect of the sound can be impacted, if its not part of the recording/data itself, since all the DAC sees are ones and zeros and try to recreate to analogue as good as it can.

The digital part can only affect how precise the DAC can sound (depending on the noise) but not impact it actively to give a certain flavor/tonality of sound, based on some type of malfunction. The digital data is always perfect.

So, if the tonality changes and becomes dark, then that must be on the analogue side of the DAC conversion (in my mind) since there is nothing in the digital data that tells the DAC to sound dark suddenly.

It also doesn´t make sense in my mind that you suddenly get more noise into the DAC. Mostly because that usually won´t give a dark sound, from my experience at least. It would be another type of impact that we usually associate with noise (degraded DAC performance).

Thanks! I will think about this deeply for the coming days.

I just have done some test's and processed some thoughts regarding noise aspects. My first thought is that when I only use switch mode power supplies. That there's no trouble with dark sound. The sound is really not that good with switch mode power supplies, but there's no trouble.

Theoretically it could be possible that a high grade powersupply could magnify some sort of noise which  already came with the stream from elsewhere. Also the ethernet signal itself is some kind of noise for an audiosystem. When it becomes a really stable blockwave who says the DAC can handle that. In the end the audio clock runs at another speed when it would drift up towards the 25Mhz would be a problem.

Well there are already much more toughts that I could write down, but I will give it some time first. Because many experiences still doesn't fit. 🙂

One more thing. My current dac also plays for about 5 seconds after I disconnected the ethernet. The sound never changes at that moment. The DAC always keeps the same quality as which the buffer is filled.

This is a really difficult thing to connect with the noise aspect