web analytics

Why a good switch does matter for streaming audio

Results

Contents

We did two measurements, as mentioned: on the power supply side and on the Ethernet side. Both are interesting. The Ethernet side you connect to the streamer. The power supply you might connect next to the rest of the system. Both can potentially affect the reproduction. It depends on several factors.

We have labeled the pictures with the measurement results. One is called ‘voedingskant’ (power supply) and one is called ‘switchkant’ (ethernet). That way you know which one is which.  You will see that there are still differences visible. And that especially the power supply is an essential part of the performance. The lower the noise on the power supply side, the better the initial situation.

Cisco standard 8 ports

Swich Mode PSU

PSU side max PSU side min Switch side max Switch side min
-16 dBm -67 dBm -15 dBm -67 dBm

Linear Power Supply

PSU side max PSU side min Switch side max Switch side min
-9 dBm -67 dBm -37 dBm -67 dBm

English Electric 8 Switch

PSU side max PSU side min Switch side max Switch side min
-10 dBm -62 dBm (Waves) -27 dBm -62 dBm (Waves)

Silent Angel Forester power supply

PSU side max PSU side min Switch side max Switch side min
-17 dBm -67 dBm

NuPrime SW8

PSU side max PSU side min Switch side max Switch side min
-9 dBm -66 dBm (spikes) -29 dBm -66 dBm (spikes)

Paul Pang Single

PSU side max PSU side min Switch side max Switch side min
-10 dBm -68 dBm (Waves) -33 dBm -67 dBm (Waves)

Paul Pang Quad

PSU side max PSU side min Switch side max Switch side min
-25 dBm -67 dBm -25 dBm -67 dBm

SoTM sNH-10G

PSU side max PSU side min Switch side max Switch side min
-9 dBm -67 (Waves) -30 dBm -62 dBm (waves)

What does this all say?

Noise is definately not something we want to introduce into a good hi-fi system. Noise creates all sorts of issues: a veil, unrest, loss of definition. Every manufacturer tries to eliminate noise from his or her design. If we then introduce noise again with a switch behind the streamer, we’re just being foolish.

When audiograde switches made their appearance, we did frown, to be honest. But after some trials with fiber/glass and good power supplies, we also heard a difference. And this is reproducible. Are they shocking differences? We do not think so, but the fact is: sometimes small steps are also important.

Who looks at these measurements, will see that there are just differences in performance. And that both the Cisco with linear power supply and the Paul Pang single switch perform well. In terms of common mode better than the English Electric and NuPrime which are just 10 dBm above it. In terms of price, these four are not that different.

By the way, what is funny is that the English Electric and NuPrime are the same (just like the Silent Angel… all made by Thunder Data) except for the power supply…. The NuPrime measures differently (slightly better) anyway, proving that power supply has a direct impact. The Cisco with switch mode and a linear power supply is the same story.

What also stands out is that the Paul Pang Quad measures the same on the power supply side as it does on the switch side. Very good isolation, in other words. A negative outlier is – in our opinion – the SOtM. The price is not in proportion to the performance. We attribute this purely to the cheap switch mode power supply that they provide.

Quick win

What we recommend for a “quick win” within a standard network, is to place a decent switch (dLink / Cisco / Netgear) close to the streamer. A decently built one already filters better than a cheap brand. Also connect a good power supply to the switch. Think of a decent Ifi power supply, sBooster or if you want to go really crazy, a Hypsos. You can hear these effects in our Live Switch Test.

Leave a Reply

}