DIY Ethernet Cable

[plissken]

As a dual CCNP and ACEP (Cisco and HPE Aruba) I know a ton about networking and structured cabling.

 

Unfortunately when I tell you that the people making these recommendations have no idea what they are talking about it starts up a pages long thread were engineering outfits like Texas Instruments and Siemons, with heavily instrumented papers, showing what is going on with Data cabling are somehow ignored.

 

My advice is get UTP CAT5e from Panduit or Tripplite and to ignore the people giving you this advice.

[plissken]

4 minutes ago, lpost said:

I believe one need a tool though which unless you need a bunch of cables I would just buy measured patch from Blue Jeans and call it a day.

 

I know Tripplite guarantees their patch to pass ISO spec.

[plissken]

14 minutes ago, lpost said:

I intended to mention to the OP, all he's done is completely violate the Ethernet spec and yet to its robustness it still works! At least over the short distances in a house or patch cable.

 

Need to mention that if you force your interface to 100Mbit you are only on two pair... No need for modifying the cable.

[plissken]

2 minutes ago, lpost said:

Good to know and have another source.  When I started in networking in the late 80s I had to deal with vampire taps (remember those?), 10base2 'cheapernet' and AppleTalk over 2 wire or was it 4 wire telephone line cord all in the same buildings, all 3 types! I still work for the same company all these years later. We've not gone beyond dual 10Gbit uplinks as we simply don't require it but it's nice to know 40G and 100G is now readily available and 400G is possible. With the exception of upsampling 192k-24bit PCM to even higher rates we still don't need more than 100Mb/s for audio and even most video.

 

Oh how things have improved and in many ways become must simpler.

 

It seems about every 10 years we replace our network core because they approach EOL. End of the month a pair of Nexus 7k are being replaced with a pair of 9606R. The Nexus replaced a pair of Cat6513 with 13 years service and before that a 10M Cisco router running in bridge mode! Silliness.

 

Be safe out there and don't hack Ethernet cables for no good reason.

 

Yep, thinnet over RG59. I also did Appletalk and hand crimped many 110 ohm resistor into an RJ11 for appletalk termination.

 

Novell, Banyan Vines, DECNet, LanTastic.

[plissken]

Enjoy then....

 

[plissken]

1 hour ago, lpost said:

but I manage a large corporate Cisco network with HP Aruba wireless.

 

Are you getting hammered by the chip shortage yet? We are having issues with anything with an SFP cage. We are forecasted out till June now on some SKU's.

[plissken]

I just want to make sure that we are talking about something like this:

 

 

[plissken]

8 minutes ago, TomJ said:

Isnt music and sound a good reason?

 

The potential issue that I have is that this is going to increase the error rate that the switch PHY operates at.

 

I think what is happening is someone started out doing that at 1GBe and found out that it didn't work. Forced the connection down to 100mbps and found out that it did since the tolerance isn't nearly as strict and then also realized you don't need 4 pair, only 2.

[plissken]

Just now, TomJ said:

Exactly! But with solid core AWG23 wire..... 

 

So someone can hear the difference between solid core (horizontal run structured cabling) and stranded patch?

[plissken]

7 minutes ago, TomJ said:

I was in the same situation like you one year ago, then I made some tests only to proof my standpoint.

After this test and recognized i was wrong, I went into the basement and started screaming loudly. 

 

I run wireless so I'm not in a position to worry about wired. I messed around with 10GB Fiber (it was incredibly affordable at $210 all in and that was switch, NIC's, transceivers, OM3 cabling) but when I re-did two rooms I pulled it all out and ran POE for running 802.11ac access points. Implemented K/V/R for roaming, beam steering, and client stick and couldn't be happier with the results with AP's that cost $56 a pop (X3) and the free controller software.

[plissken]

1 hour ago, charlesphoto said:

 

No idea what all that gobbleydook means. In my testing BJC Cat 6a sounded slightly and subtle better than the BJC Cat 6. Lots of users over on the Naim forum that also preferred the 6a to the 6. I've tried generic Cat 7 (Teradak?) and it doesn't sound as good. Keep in mind the shield on BJC 6a is untied at one end, so is floating, compared to the metal shrouds on the connectors of Cat 7 that fully connect the shield, which is what we don't want in domestic hifi. 

 

In a nutshell he's stating that 6A is overkill and the reasons why. CAT6A is CAT6 with FTP around the individual pairs and an overall screen around the 4 pair.

 

We only use CAT6A for our POE multi-gig copper ports.  So our AC and AX AP's.

 

Honestly you only need CAT5e. It can do 10GBe up to 145 feet. At 2.5Gbe it's still 100 meters.

 

2.5GBe is ~280 concurrent 24/192 PCM streams.

[plissken]

34 minutes ago, charlesphoto said:

And according to my wife, one only needs a Sonos or Dot that one can ask Alexa...

 

I know what one needs for transmission of internet data obviously, but I was just passing on my experience with sound, not data integrity. Please don't always turn cabling threads into I.T. pedantic-ism. 

I was responding to the question about what Garis was saying.  You do you.

[plissken]

2 minutes ago, TomJ said:

I'm glad not to be alone with this phenomenon - in most forums, I'm being labeled as crazy

 

You are getting push back because you aren't willing to admit that expectation bias may be at work. I did a blind test with someone that had a cryo-treated, Telegartner terminated, cable, vs $12 cable from Amazon using their Tidal account. What the cable owner didn't realize is that Tidal will cache entire tracks. He had 300Mbit internet service. He put on his favorite piano track that was ~11 minutes long. I waited 30 seconds and pulled the cable out and showed him that the entire song was stored locally and that you could even pause/play/ff/rr.  Many have no idea how data works.

 

Once they they couldn't see what was going on, they went from 100% 'correct' sighted listening, to only getting their $500 custom cable correct 40% of the time.

 

See this video as an example of how it might be done:

 

 

[plissken]

5 minutes ago, recepky said:

I changed the ethernet cable while music playing. of course there is buffer and computer stıoes some of data. when I play a few songs from a minute ago I started to recognize positive differences. just a vocal and violin will be enough to understand whats happening. 

 

Can you reword this if possible? I don't want to assume that you are saying that the buffered data is somehow different depending on the cable that was used to fill the buffer.

[plissken]

7 minutes ago, TomJ said:

As I said already, its not about the data integrity and jitter of ethernet signal, but about noise.

But this discussion is senseless. 

 

No the discussion isn't senseless. Noise is an immutable, objective, verifiable property.

 

Extrapolating this further: Noise can only be present when the cable it inserted correct?  Ergo if I have a properly designed system, using a service like Tidal, then at the very least, someone should be able to tell when the cable is connected/disconnected while playback occurs regardless of the amount of buffering.

 

Taking it even one step further: This should also affect playback of files on local storage? Since it is noise creeping into the system regardless of music sourced over the wire or even direct attached storage?

 

 

[plissken]

1 minute ago, recepky said:

I really can not explain whats happening by engineering aspects. If somebody has curiosity He must be try it by himself to hear. In the world of audiophile anything can be true or false depending on your system speakers ears etc. 

 

I have tried a few permutations:

 

EtherRegen vs my Aruba 2530, a $330 WW Starlight 12' Ethernet Cable vs ~300' generic Hypertek Cat5E.  I even posted a view tracks captured to my ADC where I plugged and unplugged the network cable during playback, posted links to the files at AS and no one could pick when any of the changes were made. Not the # of changes or when the change was made.

[plissken]

17 minutes ago, TomJ said:

So why are you using a 200$ switch, when a 20$ Netgear will also do the job?

 

I'm a dual CCNP (Route/Switch and Collaboration) and ACEP. I have a rack full of lab gear: Aruba 2530's, Aruba 3810's, Aruba Mobility Controller 7210, Aruba Mobility Master 1K, Cisco Nexus 4K's, Cisco 3850X, 2960's, 2630's, ASA's, Cisco Wireless Controller, Fortigate FW's. So all in my rack I have about $100,000 retail worth of lab gear.  I'm able to purchase not for resale at significant discounts due to partner status with various vendors. I run Eve-NG also but sometimes you just need real gear.

 

Servers are HPE Proliant Gen7/Gen8 360DL's and 380 with 120GB RAM and 32 core XEON with SSD RAID5. Crapload of broadcom NIC's.

 

For my home network (production) it's a $60 ebay Netgear GSM7352S v2. It's a fully managed L3 switch and it's the one I did a 3 part tutorial series on subnet design for home use that I posted at AS. It also has SFP and SFP+.

 

Server is a Dell R620 with 64GB RAM and 4 port nic with two ports running a VM of PFSense for firewall.

 

Outside of all the networking I'm a DIY speaker builder and been doing that just as long as I have been doing the tech stuff. My first speaker build was some Karlson 1/4 wave folded horns with a Crown stack to drive them.

 

 

[plissken]

Not that I want to contribute to taking a perfectly good cable and ensuring that it goes out of spec but that's not Belden Bonded. Belden make both types just like car manufacturers make many models for different needs.

[plissken]

16 minutes ago, jabbr said:

Huh! Is there an actual use case for unbonded cable?

 

Cost. Ethernet only cares how the end of the cable spec is met:

 

Gamechanger Copper CAT6/POE 200M and hits spec