Oh, the mechanism is well described and studied; placebo.
Why don't we just get it, it's about a subjective feeling of difference, damned be evidence and logic. It's mystic.
Removing unused pairs clearly turns the cable from a simple conductor to a digital signal processor capable of performing high speed computations and altering the transmitted data in a positive way. This, as opposed to introducing random and destructive errors.
It is, of course, virtually impossible for the casual listener to perform a real ABX test on network cables. Thus all comparisons are sighted and biased. But "highly trained bias" enables ultra sensitivity to changes too small to measure.
And now, we return you to planet Earth. Where actual science still works in the face of ever-persistent hogwash.
It is, of course, virtually impossible for the casual listener to perform a real ABX test on network cables. Thus all comparisons are sighted and biased. But "highly trained bias" enables ultra sensitivity to changes too small to measure.
And now, we return you to planet Earth. Where actual science still works in the face of ever-persistent hogwash.
In 1970s, many considered the cable does not make a difference; in 1980s, many thought digital source does not make a difference since a bit is a bit. Today, the effects of cables and jitter are widely acknowledged.
Now, more and more hifi NAS/switch have been introduced to the market. Not all consumers are stupid.
Now, more and more hifi NAS/switch have been introduced to the market. Not all consumers are stupid.
No, but the hifi NAS and switches are targeted at the stupid ones. Just like the $50 gold plated HDMI cables they try and sell you with a new TV.
Dave has explained the reason in the original post. The two pairs are not unused. The idea to cut them is to enforce the transmission working under the rate of 10M/100M, instead of 1000M.
Scientists can't be as stupid as anyone else?
Scientists are open to new idea and do experiments
They do, and show results of said experiments.................if they want to be taken seriouslyScientists are open to new idea and do experiments
If there ever was a post demonstrating complete ignorance about data transmission, this must be the mother of them all!
Jan
He didn't really explain anything and I can't understand the point or logic of not connecting half the pairs and forcing the connection speed to 10/100.
Please go speak to your university computer science professor so he/she can explain to you the mechanisms of data exchange over ethernet.
10/100baseT uses only two pairs. If that's the interface in use (a 10/100 NIC or switch at either end), there's no need (or difference) in cutting the unused wires as they are already unused, already effectively "cut". Unless there's Power over Ethernet (PoE) in use, and then the unused pairs are used.
1000baseT requires all 4 pairs. If you really want to transmit high speed data, you can't cut anything. If you do, you can toss out that Gigabit switch you paid up for, because it will now drop back to a standard "fast" 100mb switch. As will every Gigabit NIC wired that way. You can also throw out the Cat6 cable, as it's no longer improving anything.
Uncorrupted data is uncorrupted data. You cannot change the "sound" of a data path with any degenerative passive device, you can only corrupt the data, which always...without exception...manifests as noise, or complete data loss.
Jitter in Ethernet data exchange is simply a non-factor. Data is transmitted in packets, bad packets are retransmitted. The "stream" is nothing of the kind, it's in chopped up bursts to begin with. The whole mess is assembled and re-clocked out at the receiving device. You can jitter the heck out if it, and if the data is recoverable, the jitter is clocked out. If not, the packets are resent until they're good, or the data stream stops. That's why there's a buffer. And the buffers are huge for streaming applications because the data rate is all over the map, way beyond any amount of jitter. You can see it graphically by doing a simple Internet speed test. speedof.me gives the, real single-threaded picture. There's nothing stable about data speed. If the cable gets the packets there, and they contain no errors (checked also, btw, be definition of the system), the job is done.
1000baseT requires all 4 pairs. If you really want to transmit high speed data, you can't cut anything. If you do, you can toss out that Gigabit switch you paid up for, because it will now drop back to a standard "fast" 100mb switch. As will every Gigabit NIC wired that way. You can also throw out the Cat6 cable, as it's no longer improving anything.
Uncorrupted data is uncorrupted data. You cannot change the "sound" of a data path with any degenerative passive device, you can only corrupt the data, which always...without exception...manifests as noise, or complete data loss.
Jitter in Ethernet data exchange is simply a non-factor. Data is transmitted in packets, bad packets are retransmitted. The "stream" is nothing of the kind, it's in chopped up bursts to begin with. The whole mess is assembled and re-clocked out at the receiving device. You can jitter the heck out if it, and if the data is recoverable, the jitter is clocked out. If not, the packets are resent until they're good, or the data stream stops. That's why there's a buffer. And the buffers are huge for streaming applications because the data rate is all over the map, way beyond any amount of jitter. You can see it graphically by doing a simple Internet speed test. speedof.me gives the, real single-threaded picture. There's nothing stable about data speed. If the cable gets the packets there, and they contain no errors (checked also, btw, be definition of the system), the job is done.
Last edited:
Your patience and willingness to continue to explain this to people who are so willfully ignorant is commendable, but I am concerned about your own sanity. Please take time for some deep relaxation while listening to your bit-perfect streams through your standard, non-audiophile Ethernet switch.
Thanks for your concern but my sanity left me years ago.
I continue to construct large professional Audio over IP studio systems without audiophile switches. In fact, an audiophile switch would not work in these at all, we need enterprise-class managed switches to handle all the uncompressed audio streams flying around without clogging up a dumb switch. So yeah, I'm listening to non-audiophile switches.
If you need to confirm my lack of sanity, this should do it:
Insanity: Doing the same thing over and over while expecting different results.
There's a fine and blurry line between infinite patience and insanity.
It's all zeros and ones, how hard could it be? Seriously though it either works or it doesn't, there is no in-between. Only the final step into the DAC can have jitter. From the Qobuz server, wherever it is, to my Raspberry Pi, I have no control how it gets there, how many hops on how many subnets in however many packets. And I don't care as long as it gets there. And most of the time it does, and when it doesn't it just stops playing if it takes long enough. Or sometimes you get that digital stutter when it keeps trying.
Oh, I use the horror of horrors, wifi. No cables to fuss with and it still delivers a bit perfect file. If it didn't you wouldn't be reading this after I click post!
Oh, I use the horror of horrors, wifi. No cables to fuss with and it still delivers a bit perfect file. If it didn't you wouldn't be reading this after I click post!
Last edited:
Thank you.I tried to cut 4&5,7&8.
It worked very good to my system.
I use Tidal thru roon optimized core kit.
Fanless,liner psu,headless PC.
Of course it did. And it worked very good before the cut too. You simply have no means to make the comparison.
I have to say, whilst I've been sat at home these last few weeks, reading this thread really did make me chuckle.
I'm okay with physical phenomena (what a word!) and I am happy to argue along and debate what might be important in sound quality - I do this in hifi shops and then some kindly person brings over a different pair of speakers that really are more engaging, despite the other, what I thought were boring speakers, still selling a million copies. We're both happy.
But it is interesting to see what has been written here. I think only Jaddie has really said anything that was right and technical with other people supporting similarly.
I've worked in instrumentation for most of my life. At one stage I worked on a unit designed to read data at a remote source and bring it back to the server to be archived for analysis later. This data was 32 bits and it was very fast. The data rate was virtually wire-speed of fast Ethernet - around 88 Mbps. We thought we would have to use TCP to get the data back safely as the world told us that UDP would be dreadful! But TCP would limit our top speed to well below wire-speed so we wanted to avoid it.
During our testing we went through any number of cheapo switches to see where we were going to have problems using UDP. We tried hubs (just for a laugh, not that you can buy them anymore), PoE switches, managed switches, un-managed switches, switches with lots of ports, switches with few ports, switches with built in power supplies, switches with wall wart supplies, switches plugged into different brand ethernet cards.
On a private, even busy network (our fast ethernet would uplink over gigabit ethernet to be collated at the PC), we never lost a UDP packet for any reason other than physical connection. The random data at each end was always perfectly matched.
In the end we tried the same over WiFi with much the same result though at that time WiFi was a bit naff so the data throughput was not something we could cope with.
That was my story.
Here's something else for you to think about - be careful cutting the spare wires in your ethernet cable as although gigabit needs all four pairs, it still negotiates its speed capabilities over the first two pairs without checking if the other pairs exist or not! If the phy isn't auto MDIX it's gonna break!!!
I'm okay with physical phenomena (what a word!) and I am happy to argue along and debate what might be important in sound quality - I do this in hifi shops and then some kindly person brings over a different pair of speakers that really are more engaging, despite the other, what I thought were boring speakers, still selling a million copies. We're both happy.
But it is interesting to see what has been written here. I think only Jaddie has really said anything that was right and technical with other people supporting similarly.
I've worked in instrumentation for most of my life. At one stage I worked on a unit designed to read data at a remote source and bring it back to the server to be archived for analysis later. This data was 32 bits and it was very fast. The data rate was virtually wire-speed of fast Ethernet - around 88 Mbps. We thought we would have to use TCP to get the data back safely as the world told us that UDP would be dreadful! But TCP would limit our top speed to well below wire-speed so we wanted to avoid it.
During our testing we went through any number of cheapo switches to see where we were going to have problems using UDP. We tried hubs (just for a laugh, not that you can buy them anymore), PoE switches, managed switches, un-managed switches, switches with lots of ports, switches with few ports, switches with built in power supplies, switches with wall wart supplies, switches plugged into different brand ethernet cards.
On a private, even busy network (our fast ethernet would uplink over gigabit ethernet to be collated at the PC), we never lost a UDP packet for any reason other than physical connection. The random data at each end was always perfectly matched.
In the end we tried the same over WiFi with much the same result though at that time WiFi was a bit naff so the data throughput was not something we could cope with.
That was my story.
Here's something else for you to think about - be careful cutting the spare wires in your ethernet cable as although gigabit needs all four pairs, it still negotiates its speed capabilities over the first two pairs without checking if the other pairs exist or not! If the phy isn't auto MDIX it's gonna break!!!
- Status
- Not open for further replies.