Well, if it's an analogue problem, it's not the L2 device. If it's not broken the packets are intact and there's nothing to see here...
Not quite - there are indeed varying voltage levels involved, but as far as the information represantation of digital signals go they are absolute. In computers it's never actually the simple "on or off" that many assume. Rather, its "high or low". The actual voltages vary significantly between systems. Plus there's no such thing in reality as a perfect square wave for differentiating between high and low pulses. Fortunatley, engineers are (*) a clever bunch, so many ways of dealing with the vagaries of electricity in the real world have been developed.
For example, Ethernet 100BASE-TX uses +/-2.5V for determining 1s and 0s. To send a "1" the transmitter will send +2.5V down the TX+ wire and -2.5V down the TX- wire. To send a "0" It would be -2.5V down TX+ and 2.5V down TX-. The wires send the inverse of one another but always offset by a constant value.
I reality the actual voltages will be affected by noise / interference, but this means nothing to the receiver - it is simply looking for which wire sent the higher voltage, regardless of externally-induced voltage shift. That's it. The receiver could very well see voltages all over the place, but it's trivial to determine if the pulse was a 1 or a 0.
So the voltages vary with environment, distance attenuation, etc, but once the receiver has done its work all the 1s and 0s ARE identical for all purposes of data handling.
In the very rare event that extreme noise makes the determination of a 1 or a 0 not 100% certain, or an error (detected by frame check sequence CRC) is encountered, the receiver basically says "Nope, send that data packet again" and wil keep doing that until complete, intact data is received.
Thus Ethernet delivers either a 100% accurate data packet or none at all. There is simply no way whatsoever for this to affect the audio fidelity of the transmitted music. In the most absolutely extreme case the receiver would keep saying "Nope" and the playback device would experience data stream dropouts, but with buffering this is unlikely.
Can you imagine the internet if this were any different? It would simply not exist as we know it.
* Except when they work for snake oil companies. But I suppose in that case they would be a different sort of clever, having discovered another way to fleece the credulous and uninformed.
Last edited:
The issue is with dacs, clock phase noise, Vref amplitude noise, and or noise modulation of the audio signal in the output stage. Nobody is claiming there is a problem with integrity of digital data. If removing or replacing an ethernet switch fixes the problem, then it fixes the problem. If the digital data was fine, then probably some kind of analog noise problem. Isn't this basic conducted EMI/RFI mechanism stuff?
Last edited:
I am aware of those, but still have no plausible mechanism for that causing a problem. The DAC buffers many seconds of data (such that you can actually unplug and replug the ethernet connector with no dropouts). More that likely a microprocessor in the way as well. It's really clutching at straws from you now.
Nordost is spot on, higher speeds ruin sound quality.
I just converted my entire house to 10BASE2 networking using coax cables. Tore all the CAT7 out of the walls. I'm so done with that low quality garbage.
Now that everything on my LAN has been relaxed down to 10 megabit/s I can assure you that pace, rhythm and timing is nothing short of amazing. My system must have been super stressed out by all those gigabits before. Everything is just more mellow now.
I started my journey to enlightenment by reducing the port speeds to just 10MBit/s on my switch. That did wonders for fidelity already. Then I had an epiphany and tried 10Base2. Believe me: the holy grail is in the coax cable and especially the BNC connectors.
Michael Fremer just left my house. He said he's done with analog. 10Base2 is just that good. He's probably dropping off his record collection at goodwill as we speak.
Proof?
Just trust me. I know what I'm doing here.
I just converted my entire house to 10BASE2 networking using coax cables. Tore all the CAT7 out of the walls. I'm so done with that low quality garbage.
Now that everything on my LAN has been relaxed down to 10 megabit/s I can assure you that pace, rhythm and timing is nothing short of amazing. My system must have been super stressed out by all those gigabits before. Everything is just more mellow now.
I started my journey to enlightenment by reducing the port speeds to just 10MBit/s on my switch. That did wonders for fidelity already. Then I had an epiphany and tried 10Base2. Believe me: the holy grail is in the coax cable and especially the BNC connectors.
Michael Fremer just left my house. He said he's done with analog. 10Base2 is just that good. He's probably dropping off his record collection at goodwill as we speak.
Proof?
Just trust me. I know what I'm doing here.
In the case where there is a streamer its probably doing the buffering. Most dacs don't have even a 500ms FIFO (although a USB interface may do some buffering).
Regarding plausibility, seems to me that if removing and or replacing a switch fixes a problem, then that's sufficient to implicate it in some mechanism that includes problematic sensitivity of the dac. As noted before, most people are not willing to pay extra for a more noise immune dac. For people who want to use wired ethernet and then find there is a problem with sound there may not be a lot of choices. They can go to wireless, but often don't seem to want it. Otherwise there are isolated switches and there are USB isolators (that can go between a streamer and USB dac). If one of those things fixes the problem and the problem cannot be attributed to digital data integrity issues, then seems likely there was an analog noise problem. Even if nobody measures the exact coupling mechanism, it still seems most likely. And of course as also noted previously sometimes people imagine things. However, it seems doubtful that everyone who reports problems with wired eithernet affecting dacs is hallucinating.
Last edited:
we are talking ethernet in not USB in here.
I just re-read the review and noted (ROTFL) that the reviewer also has an etheregen in line before his megabuck dCS DAC ethernet input. And he bought a $2800 power supply upgrade for a $600 uptone box.
So the output of a very expensive L2 switch is going through another less expensive L2 switch into a $40k DAC which ought to be expensive enough to be immune anyway...
My mind hurts from the nonsense I can read here.
I understand that you give 10000 times more than it's worth because that's what you want, but to claim that there are differences for absolute reasons, like for example because it has a case of color x is.....
I remember back in the day, when folks would coat the innards of their Phillips CD player's case and transport with modeling clay. Explanation was that it reduced resonances, allowing the tracking and focus servos an easier time doing their job, which translated to less current demand modulation stress on the power supply, which was affecting the sound in a perceptible way...
At that same era, another fellow told me of how he deliberately injected noise onto an op-amp buffer's power supply rails and couldnt hear any effect at all until they were of a "ridiculous" amplitude. I take that to mean something that would never occur in the normal operation of the power supply with its designed to accommodate loading.
He wrote me personally, as he wanted to express his findings to someone, and he said he was too afraid to post such an idea in the old Audio forum.
Well? Which is it? I've been waiting at least 30 years for the answer.
- Status
- Not open for further replies.