Power supply for ethernet switch?

@mclinnguy

Well, I’ve re-read the article on Alpha Audio entitled “Hard evidence: a network affects playback quality” and am little the wiser.  Alpha Audio is a Dutch audio blog started by Jaap Veenstra, a journalist with an interest in audio.

Deep in the comments on the article there is a statement that the jitter was measured on the output from the clock feeding the input to the Sabre ESS DAC included in the streamer.

In response to the first comment

“You say that these minor differences in switch jitter are “audible”. How did you determine this? What science is behind this claim?”

moderator Martijn replied

“We use two very special fleshy flaps that are attached to the sides of our heads to determine audible differences. The science part is that we do not claim to do science”.

The issue for me is that this last statement is patently true, but conflicts with the headline “Hard Evidence”, and the way that the results have been used in this topic.

In the past, the Alpha Audio blog has used high-street kit speakers to assess sound quality.

Here’s part of what I think they did.  They took a fairly low-end streamer and “placed it in the RF-shielded measurement box”. Also in the box seems to be other active electronic components of the test measuring system, plus an Ethernet connection, probe cables and an extension lead.  Almost the entire cover (Faraday cage?) of the streamer has been removed.

Anyone familiar with microwave ovens knows that a Faraday cage can keep RFI in as well as out!  Also that wires entering the cage can carry noise.

They used the streamer’s external 5-Volt switching power supply.  The streamer includes a CPU running at more than gigahertz speeds, plus a wireless hot spot.  In other words, the streamer itself is potentially electrically extremely noisy.

They picked three common Ethernet switches and a couple of extra power supplies.  They did not say what speed Ethernet they used, but Gigabit would make sense.

If we exclude airborne RFI (big if) the obvious source of jitter is noise carried on the power supply lines between the switches and the streamer.  Considering the number of switched mode power supplies in the chain, this is hardly surprising.

It is possible that some electrical noise is carried by the Ethernet cable, probably of the unshielded twisted pair type.  A good baseline would have included measurements without the Ethernet cable, and without the switches and their power supplies.  I did not see any such baselines in the results.

Now for the claim that there is a difference of 30 picoseconds of jitter.  Run the DAC on two channels at 192-kHz to get a sample rate of 384-kHz.  That’s about 2.6 million picoseconds per sample.  So 30 picoseconds represents a timing error of about 1 part in 100,000.  That is smaller than the resolution of amplitude on a CD.  I would title the article “No evidence: a network affects playback quality”

@mclinnguy ….

@kingbr, I also have an inexpensive TP Link switch in my network attached to my network player/dac.  Interested by testimonies of better audio performance from improved network components, I made a couple of relatively inexpensive changes as a test.

I replaced the stock TP Link switch mode power supply (SMPS) with a linear regulated power supply from Jameco, cost $12-$14 plus shipping.  I compared the two power supplies (PS) on my oscilloscope with the same current draw as the switch.  The stock SMPS PS output an awful racket of sawtooth noise while the Jameco linear regulated PS looked like a smooth line.  To my ears the Jameco sounded better too.  Music sounded smoother and less peaky with no loss of detail.  I heard better definition, sense of space, and placement of instruments.  On a 1-10 scale I’d rate the improvement a 2, not huge, but certainly worth $20 or so.

Now I’m sure an audiophile PS from Small Green Computer, iFi, or other source would be even better and audiophile purpose switch even more so.  However, the Jameco PS was an inexpensive way to see if a PS made a difference in my network and audio system.  It did.  An option to consider.

Incidentally, the other change I made was adding an iFi LAN iSilencer at the input of the network player.  Another small improvement IMO.

I found that an Sbooster BOTW 12V power supply gave me a nice improvement with my Melco S100 switch.