LAN cable terdinations

[Adam2]

 

I had a company come round to sort the patch panel and some wall connections etc. The panel is as per the images on the back of the 48 port panel (supplier says intended to be wired as T568B)

 

 

 

I took one of my wall plates off as I wanted to see if they had wired up as T568A or B as I needed to do another wall plate myself. When I took an existing one off I noticed the orange cables were swapped - weird, almost T568B. 

 

 

 

So, I plugged a pre-made long patch lead into the wall outlet and used my patch cable test device to connect this into the patch panel - it shows 3 pairs as not good and 1 pair as good. so weird - thought it would be 1 pair bad.

 

So - swapped the oranges around and now I use my tester and I have 2 pairs good and 2 not showing any connection WTF???

 

Looked at another faceplate - wired the same with oranges seemingly crossed. On one of these I have a laser printer connected and the switch (Unifi) reports this as being a fully operational 1Gb connection - which I guess means 4 pairs need to be working correctly.

 

Clearly I'm very confused - any pointers appreciated to calm my confusion 🙂 Else will call installer in morning to see what he thinks.

 

 

 

[SteamyTea]

terdinations

 

Sounds painful.

I once found out why turds are tapered, they stop the bottom going bang.

[TonyT]

Make sure you krone push down on the conductor at least twice to ensure it’s pushed into the termination.

 

cheap cat cable tester available for under a tenner

 

 

[Adam2]

Thanks yes have a cable tester which I used and had results above. Just cannot understand how the wall panel can be wired incorrectly as per the colour on the module yet do 1gb connection and cable tester reports 2 pairs with no signal maybe I'm using it wrong or something but it is simple just 2 cable connectors and some lights to show each pair...

[joth]

1 hour ago, Adam2 said:

it shows 3 pairs as not good and 1 pair as good. so weird - thought it would be 1 pair bad.

 

So - swapped the oranges around and now I use my tester and I have 2 pairs good and 2 not showing any connection WTF???

 

Can you say which pairs were bad, and what sort of badness they had?

Remember with T568B the pairs are not in order of the pins in the RJ45, which is what the tester will be reporting. So for example if green pair are reversed, you'll see the tester sing out an order of "1 2 6 4 5 3 7 8" because green are on pins 3 and 6. It would be easy to see this as 2 bad pairs, but it's only one.

 

All that said....  doesn't explain your original issue with orange swapped.

 

To confirm, the photo from the back of the patch panel is actually from the port with the issue? It's possibly not all ports were done the same if an error slipped in.

 

Edited October 18, 2022 by joth

[PeterW]

1 hour ago, Adam2 said:

On one of these I have a laser printer connected and the switch (Unifi) reports this as being a fully operational 1Gb connection - which I guess means 4 pairs need to be working correctly.

No - for full Tx/Rx you only need 2 pairs connected, usually 1&2 and 3&6. The remainder are available to PoE etc 

 

Useful guide here  

 

https://www.intel.com/content/www/us/en/support/articles/000007187/ethernet-products.htm

 

1 hour ago, Adam2 said:

When I took an existing one off I noticed the orange cables were swapped - weird, almost T568B. 

 

Not odd - you need to read the numbers carefully as they don’t go 1-8 on the back of a socket, the PIN numbers are on the sticker so you need to match them to the panel. 
 

[Adam2]

OK thanks - so more info 🙂

 

The tester I have is this one  basic - just shows a LED for each pair 1/2, 3/6, 4/5, 7/8 - writing this makes me notice a typo in the original message which I'll update if poss. When I connect the following:

 

1 - Patch panel port with terminations as per images (supplier says is "B")

2 - Cable from patch panel into tester

3 - Other end of tester - cable running to wall plate - wall plate wired as per "B" image on the keystone module

4 - Cable at back of wall plate terminated into patch panel (1) 

 

Tester shows 2 green LEDs pairs and 2 LEDs not lit at all - I did an experiment to see what condition causes no LED vs red LED -seems red LED is a pair that is crossed and no LED means one of the wires in a pair is not connected. 

 

 

When patch cables are used on their own - so a single cable plugged into both ends of the tester - I get 4 green LEDs indicating all pairs working as expected

 

From looking at patch panel looks like all wired the same.

[HughF]

I'd suggest a slightly more premium tester if you have a lot of this to sort out (or in the future) - we use a lot of the cheap ones at work, that just count through the LEDs on each unit But they do show all 8 connections.

[MJNewton]

1 hour ago, PeterW said:

No - for full Tx/Rx you only need 2 pairs connected, usually 1&2 and 3&6. The remainder are available to PoE etc 

 

For Gigabit (1000BASE-T) you need all four pairs for data transmission.

 

PoE in such a network is carried using 'phantom power' common mode DC voltage which doesn't interfere with the transformer-coupled AC data transmission on the same wires.

Edited October 18, 2022 by MJNewton

[TonyT]

I like peak testers for my electronics, and they do this kit

 

https://www.rapidonline.com/Peak-UTP05E-Ultimate-Atlas-IT-Set-UTP05-IDT8A-IDT8B-IDT8C-RJA11-and-RJC08-85-2079?IncVat=1&utm_source=google&utm_medium=surfaces&utm_campaign=shopping feed&utm_content=surfaces across google&gclid=EAIaIQobChMIu_nK1dfq-gIVDe3tCh1icwvQEAQYCCABEgI3ZfD_BwE

 

it’s good has a selection of numbered plugs, so when you have lost cable identification, you plug them into the sockets, go to the patch panel, and it identifies each one as you plug it in, saves a bit of time.

 

 

[joth]

1 hour ago, PeterW said:

No - for full Tx/Rx you only need 2 pairs connected, usually 1&2 and 3&6. The remainder are available to PoE etc 

 

Useful guide here  

 

https://www.intel.com/content/www/us/en/support/articles/000007187/ethernet-products.htm

 

No, gigabit ethernet definitely requires all 4 pairs to be correctly wired through. See the T4 section of the page you linked, or various other pages discuss it too. E.g

https://www.practicalnetworking.net/stand-alone/ethernet-wiring/#:~:text=The first major difference is,when building a Crossover cable.

https://en.m.wikipedia.org/wiki/Ethernet_over_twisted_pair

 

Or from 20 years ago:

https://arstechnica.com/civis/viewtopic.php?t=690470

[MJNewton]

4 hours ago, Adam2 said:

Looked at another faceplate - wired the same with oranges seemingly crossed. On one of these I have a laser printer connected and the switch (Unifi) reports this as being a fully operational 1Gb connection - which I guess means 4 pairs need to be working correctly.

 

Fairly common when it comes to Gigabit adaptors and doesn't necessarily mean the wiring is 'right'. The spec requires a certain amount of auto-negotiation to cater for the different permutations of device connectivity and whilst the bare minimum is to cover the use of straighthrough vs crossover cables most implementations take this liberal approach to the extreme such that all pins can be internally re-routed to enable successful connectivity to take place. It can't deal with missing connections through, or at least not without a degradation of performance by falling back to a low spec level (which don't require all four pairs to function, such as with 10/100BASE-T).

Edited October 18, 2022 by MJNewton

[Adam2]

14 minutes ago, joth said:

No - for full Tx/Rx you only need 2 pairs connected, usually 1&2 and 3&6. The remainder are available to PoE etc 

 

Ok which is what I originally expected. So that bega the question as to why my switch (unifi 24 port POE) reports a Gb connection when the wall plate looks to have bad wiring. Feels like some key info is missing. In morning will wire put a device on the end of a port with what looks like switched over orange cables and then on one that I've terminated as per the B layout image and see what the switch thinks about that. I wonder if it does some magic to determine incorrect wiring or something clever.

[Adam2]

Thanks @MJNewton will look at switch spec and do my test above to learn a bit more 😀

[MJNewton]

1 minute ago, Adam2 said:

Thanks @MJNewton will look at switch spec and do my test above to learn a bit more 😀

 

It's likely not something that'll be adequately explained/advertised in the specs - it's more a low-level function that corresponds to a equally low level coverage in the IEEE specs too. A switch might advertise something along the lines of 'auto-crossover detection' or similar, but that doesn't necessarily mean it'll go further and cope with any randomly-wired cabling you connect it to.

[MJNewton]

To add: my advice would be to 1) check the T568B colours on your patch panel and faceplates do actually connect to the correct pin in the socket as per the specs (it is not unheard of for cheap products (no offence intended - they're what I use!) to have such issues/errors, and 2) double-check that the T568B wiring spec has been followed throughout the installation (all to easy for mistakes to creep in, such as the reverse polarity orange pair you found, or for the adjacent T568A colour code to have been inadvertently followed somewhere). 

[joth]

9 hours ago, MJNewton said:

It's likely not something that'll be adequately explained/advertised in the specs - it's more a low-level function that corresponds to a equally low level coverage in the IEEE specs

 

Auto MDI-X (auto crossover) is well specified, and mandatory part of 1000base T.

It specifically deals with the TX and RX pairs being swapped, but doesn't explain the cores being inverted within a pair.

 

Gigabit ethernet uses AC signalling so there's the possibility the negotiation completes even with a polarity inversion. It means one pair is phase inverted so highly likely to mess with max bandwidth performance and maybe the PoE function too.

Tldr the gigabit light lighting up is not proof of a perfectly wired cable. 

 

@Adam2

You definitely need a new testing protocol as the existing one that works according to pairs of pins on the connector is not giving a sensitive or specific diagnosis.

 

Tip: is buying a new tester go with a 9 LED one with screening (ground) continuity test too, just avoid ever needing to buy another should you ever need that, LOL.

 

Edited October 19, 2022 by joth

[MJNewton]

54 minutes ago, joth said:

 

Auto MDI-X (auto crossover) is well specified, and mandatory part of 1000base T.

 

I was talking about the specs of the switch being able to detect/correct polarity inversion, as that's what Adam2 sounded like they were going to check (but as I said it's unlikely that that sort of detail that would be included as it's a very low-level function).

 

Quote

It specifically deals with the TX and RX pairs being swapped, but doesn't explain the cores being inverted within a pair.

 

Auto MDI-X doesn't, no, but the IEEE 802.3 spec absolutely does and clearly states it as an intended goal:

 

Quote

40.1.4 Signaling

1000BASE-T signaling is performed by the PCS generating continuous code-group sequences that the PMA
transmits over each wire pair. The signaling scheme achieves a number of objectives including:

 

[...]

m) Ability to automatically detect and correct for incorrect polarity in the connections.

[...]

 

'PMA' if the Physical Medium Attachment sublayer and essentially acts as a abstraction layer that decouples the interface backend and the physical port and is where all the 'magic' happens to deal a whole raft of medium-related issues that might arise (such as reverse polarity pairs as stated).

[Adam2]

Ahh the fond memories of the ISO 7 layer model 🙂

 

Thanks for all the valuable info, always good to learn. I think the cable installer just had a brain fart and systematically crossed the oranges in the faceplates - he was being assisted by his old man so possible that there was some miscommunication. Even though seems to work I'll swap them to align with the labeling for piece of mind and at the same time check the  + punch down again to make sure all is good - not too many anyway.

 

May also go buy a more comprehensive tester - not long until Christmas!