advertisement


Audiophile Network Switches for Streaming ... really ?

Sue Pertwee-Tyr said:
My suspicion - no measurements so don’t ask - is that the system ground and/or mains Earth is a route for noise. Given the MHz frequencies we’re talking about, it doesn’t seem impossible that noise on the line, either data or power lines, could find its way onto the ground, within the device, does it?
Click to expand...
I think we got to here before and I agree it must be some noise leaking on to the gnd or power, which is getting into the audio band. However, the supposition had been IMPs but no one has come forward to say, or show where in the audio band they are.

It does seems that it is the susceptibility of the DAC to noise at play here, which points to a poorly designed DAC, noise rejection wise.
 
Sue Pertwee-Tyr said:
My suspicion - no measurements so don’t ask - is that the system ground and/or mains Earth is a route for noise. Given the MHz frequencies we’re talking about, it doesn’t seem impossible that noise on the line, either data or power lines, could find its way onto the ground, within the device, does it?
Click to expand...
That's why you first look for noise and intermodulation products on the output of a device (the DAC being the pivotal point here). Also, it's why studios use differential signals that are off the ground (and why I now insist on balanced kit).

I could point to test results showing it's widely possible these days to operate a modern DAC with a digital input supplying complex digital audio signals plus inevitable digital noise, on mains power, and not see any intermodulation products or noise on the DAC output above a level a reasonable person would consider audible. DAC makers have learned to practice the techniques in a good technical paper @tuga posted earlier.

Just because it's possible does not mean it's always achieved, of course. And all sorts of quibbles can be thrown at any evidence if you have a position to advocate, so this thread is not one for reasoned argument, I think.
 
I remember a certain iconoclastic amp maker whose name shall not be mentioned hereabouts, used to argue that in the case of power supply regulation, the cure could be worse than the disease. Excessive regulation sat on dynamics. Doesn’t stop some manufacturers still regulating the **** out of their products.

In the case of digital devices I could envisage there having been listening tests which indicated that measures to reject the nth degree of noise compromised the SQ; that’s not ‘broken’ it’s a design choice. From which it could follow that further reducing the noise reaching the unit could be beneficial.
 
John Phillips said:
That's why you first look for noise and intermodulation products on the output of a device (the DAC being the pivotal point here). Also, it's why studios use differential signals that are off the ground (and why I now insist on balanced kit).

I could point to test results showing it's widely possible these days to operate a modern DAC with a digital input supplying complex digital audio signals plus inevitable digital noise, on mains power, and not see any intermodulation products or noise on the DAC output above a level a reasonable person would consider audible. DAC makers have learned to practice the techniques in a good technical paper @tuga posted earlier.

Just because it's possible does not mean it's always achieved, of course. And all sorts of quibbles can be thrown at any evidence if you have a position to advocate, so this thread is not one for reasoned argument, I think.
Click to expand...

That's for offering some sanity.

Jussi Laako measures DAC response into the low MHz because he claims that noise will fold back into the audio band. Low noise / IMD is one of the things he strives to achieve with his upsampling algorithms/modulators/noise-shapers. He's is an expert in the field and I trust him. I am also able to evaluate the pratical effects of his software by upconverting DSD into my RME in "NOS" mode.

In my view, this thread has evolved away from its title and is no longer about Audiophile Switches but about network digital audio.
I think that my approach can be classed as sceptical and rational. Ask anyone here, many call me a measurement-obsessed objectivist.
 
This is the RME when fed 44.1 (the ADI-2 DAC doesn't upsample):

jKpf2tQ.png






And DSD256 (DSD direct):

v0OEYGR.png


https://audiophilestyle.com/forums/topic/55601-any-users-of-the-rme-adi-2-dac/#comment-926151
 
Sue Pertwee-Tyr said:
My suspicion - no measurements so don’t ask - is that the system ground and/or mains Earth is a route for noise. Given the MHz frequencies we’re talking about, it doesn’t seem impossible that noise on the line, either data or power lines, could find its way onto the ground, within the device, does it?
Click to expand...
Given that Ethernet signals are balanced and transformer coupled it does seem impossible.

What none of the advocates have suggested is an explanation as to how the DAC part of a streamer can tell the difference between tiny amounts of putative noise and the large amount of UHF signal of the Ethernet.
 
Author @mansr will confirm but I think the following quote applies:

"Non-linear effects create harmonics of the input frequencies. Any harmonics above the Nyquist frequency will fold back as aliases. Operating at a higher sample rate reduces the amount of such aliasing since less of generated harmonic content then falls above Nyquist."
 
tuga said:
Wrong guess. ASR mentality.
Click to expand...
"These are measured from analog outputs of the entire DAC, not just the chip, the chip output would have notably more"

Looks like I was right.

And please be careful making any judgement of my mentality or thoughts.
 
Last edited:
Great, so an rme dac shows noise at 500khz and multiples thereof. But nothing of note at 250khz, i assume, unless it would have been shown.

So now all we need is to show a preamp folding this 500khz back or a power amp folding this back. Because there's **** all in the audible band of the rme output, because its imd is measured to be down at -120db 20-20khz.

If the IMP in the audible band remain below audibility then what happens above the audible band by definition can't be audible.

If all these crappy pre amps and power amps with bandwidth from dc - light exist let's see their out of band noise measurements. Otherwise it's just hand wringing whataboutism.

Who knows maybe Spectral's amps are the antithesis of naims clipped input frequency circuits.

I'm all for seeing all audio components measurements taken from 10hz - 200khz, that seems like a reasonable bandwidth for assessing imd.
 
Warszawa said:
There is nothing special about the Zyxel switch. It’s only flavour of the month because somebody opened a couple of “audiophile switches” only to find the same Zyxel board with a crystal oscillator stuck on for… reasons.

You can guarantee these audiophile brands chose the Zyxel because it’s compact and cheap, not because of any special audiophile qualities the engineers at Zyxel stumbled upon.
Click to expand...
And why would they choose any other board? It's a network board.
 
davidsrsb said:
Sadly to make a living as a HiFi designer you either have to be a good self publicist or move around a lot. Very few engineers succeed in running their own brand these days. There are plenty of marketing experts who pretend to be engineers
Click to expand...
Or know how to find one.
 
Zombie said:
Why do you even argue when you don't have any measurements to offer?
How can noise travel WITH the data? I haven't seen any link to a reliable source for this claim, just 69 pages with mostly blabla.
Click to expand...
Why do you argue when you don't have any measurements to offer yourself?
Noise travels along the same cable as the data. It's the whole reason someone once started experimenting with using a network switch for audiophile purposes in the first place. It can also enter the switch. It can also enter other devices in the chain but a switch won't/can't have any effect on this.

Are you denying that noise (RFI/EMI) is a thing which exists and impacts sound quality?
 
sq225917 said:
Great, so an rme dac shows noise at 500khz and multiples thereof. But nothing of note at 250khz, i assume, unless it would have been shown.

So now all we need is to show a preamp folding this 500khz back or a power amp folding this back. Because there's **** all in the audible band of the rme output, because its imd is measured to be down at -120db 20-20khz.

If the IMP in the audible band remain below audibility then what happens above the audible band by definition can't be audible.

If all these crappy pre amps and power amps with bandwidth from dc - light exist let's see their out of band noise measurements. Otherwise it's just hand wringing whataboutism.

Who knows maybe Spectral's amps are the antithesis of naims clipped input frequency circuits.

I'm all for seeing all audio components measurements taken from 10hz - 200khz, that seems like a reasonable bandwidth for assessing imd.
Click to expand...
Much better said than I attempted at :)
 
You must log in or register to reply here.


advertisement


Back
Top