Best practice for splicing cat5e

Thread Starter

Imnosardine

Joined Feb 10, 2020
7
I need to add about 50ft to a cat5e run that's about 300 feet by the time it will terminate. It's a pretty low volume of data that it needs to transmit so I guess I'm not super worried. I know its not a good idea but I wasnt involved in the planning, just need to make it work as well as I can. My thought is to use a punch down splice box enclosed in a weatherproof box since its outside. I could also put the rj45s on the ends of the cables and use a coupling like that but I really dont know what would be more reliable. I appreciate any input. It will also be following all code requirements for protection by using rigid pipe to protect the cable as it comes out of the ground to the box then emt to the inside of the building and at least a nema 4x box maybe higher rated since it's in a bit of a vulerable spot for water and dust.
 
What speed is this running?
You're at the end of the range spec at 300 feet, and not all equipment can make it that far in the first place, and a splice can really make a mess.
Ethernet can be having lots of problems with data errors and retries but you won't know aside from speed seeming really slow and sometimes data just vanishes or a device crashes.

I find punchdowns don't work reliably, data gets lost due to the breakout from cable to RJ45 and jumper, then back to RJ45 and the other cable. You don't want to upset the balance and signals and cause reflections. If this was low speed telephone or RS-485 there would not be a problem. RJ45's aren't the greatest connector, they can oxidize esp. in outdoor JB and cause troubles.

If you can simply solder the extension cable to the end of the existing cable, while keeping the twisted pairs grouped together (not flying loose leads) and the conductors symettrical and in-line is most important. So soldering the ends flat on top of each other, not a Marrette kind of thing. I'd use heatshrink tubing.
If the cable was stretched during pull, it can ruin the twisted pairs and make the cable useless despite an ohmmeter saying continuity is good.
 

Thread Starter

Imnosardine

Joined Feb 10, 2020
7
That makes sense. The location that it's in I definitely need to stick it in a box regardless though. So maybe just getting a regular weatherproof box and doing strain relief with a solder sealed up with heat shrink inside the box? I'm guessing that I need to strip the bit of copper that will be soldered, do I need to do anything else other than try to untwist as little as possible?
 

Thread Starter

Imnosardine

Joined Feb 10, 2020
7
I was thinking about a punch down box like that mounted in a weatherproof box. I kind of have to agree with the other guy though that they can be unreliable. Now I'm not sure how that translates to signal loss once it's done right though. I'm a bit unsure of how soldering wouldn't have the solder points touch in heat shrink and it wouldn't be nearly as big a deal to redo a punch connection to get it right. I'm not sure of the exact speed but its just s piece of equipment sending a 5 digit number to a printer less than once a minute. Dunno how that translates to speed.
 

SamR

Joined Mar 19, 2019
2,006
When I used to splice it, it was done in a box with a ground. Either terminated or butt-spliced and the cable shield grounds also spliced and connected to the box ground. If the max distance was to be exceeded an amplifier was included in the box. When an amplifier needed the cables were simply terminated and plugged into the amp. Amps were in boxes to prevent problems with "Inquisitive fingers" causing data interruptions.
 

eetech00

Joined Jun 8, 2013
1,865
I need to add about 50ft to a cat5e run that's about 300 feet by the time it will terminate. It's a pretty low volume of data that it needs to transmit so I guess I'm not super worried. I know its not a good idea but I wasnt involved in the planning, just need to make it work as well as I can. My thought is to use a punch down splice box enclosed in a weatherproof box since its outside. I could also put the rj45s on the ends of the cables and use a coupling like that but I really dont know what would be more reliable. I appreciate any input. It will also be following all code requirements for protection by using rigid pipe to protect the cable as it comes out of the ground to the box then emt to the inside of the building and at least a nema 4x box maybe higher rated since it's in a bit of a vulerable spot for water and dust.
The proper way to splice CAT5e cable is with a Cat5e or better certified punch down block. Keeping the terminations as short as possible to avoid NEXT issues. When your done splicing, you’ll need to test with a CAT5e cable tester. Be sure to terminate all 8 conductors.
The max distance for a
Horizontal cable run is 100 meters (328 ft) so 300 feet is Ok. If the cable is longer than 100 meters, then the reliable detection of Ethernet collisions is lost. The amount of data transmitted is irrelevant. If the cable isn’t pulled and run properly, errors will cause problems with applications using the data.

While there is nothing in the specification that says you can’t splice, I wouldn’t. I would pull a new cable of the correct length.

Regarding best practice, best practice is to avoid splices. :)

eT
 

djsfantasi

Joined Apr 11, 2010
6,334
I second rerunning a cable of the proper length with no splicing, neither a soldered splice or using a punch-down block. You may be within cat5e limits, but the splice will reduce that maximum by an unknown amount. It may work at first, but the splice is an unknown, uncontrollable variable.

Run a new cable of the proper length.
 

SamR

Joined Mar 19, 2019
2,006
the splice will reduce that maximum by an unknown amount. It may work at first, but the splice is an unknown, uncontrollable variable.
Hence a CAT5 line amp. I designed and had built the fiber optic backbone for a moderate-sized plant covering hundreds of acres with at least a dozen operating areas plus shops and warehouses. It connected the central office computer servers and communications internet gateway to the entire operating area of the plant which outside of the main offices had ~75-100 computers and network printers connected to it. This was a common problem running from the fiber hubs to remote spots in an operation area. Arguably the best method is not to splice the cable but there are times the max length must be exceeded in a plant operations environment. Also note that you cannot run CAT5 in an electrically classified hazardous area without it being in sealed conduit, hence fiber is preferred and much cheaper than running conduited cables.
 

SamR

Joined Mar 19, 2019
2,006
Last time I had a contractor make an overlong cable run it was immediately obvious that the data was corrupt and he had exceed the length limit. There is no margin for error when it comes to MAX length per run and I would avoid coming even close to it. CAT5, 6, 7 cable is relatively cheap but I don't know how it compares cost/ft to fiber now. Cost is the same to run it and back when I did it you needed trained technicians to polish the end, pot it into the termination and test it. That was the big cost compared to cable. All in all, fiber is much preferred for long runs. CAT5 is fine for moderate office runs. Also, any cable amps/repeaters need power protection from EMI. They were designed for office/campus environments and not noisy industrial ones. They typically come with a wall wart power supply that in a noisy industrial environment I would have the contractor put into a grounded box supplied by a dedicated Isolator/Surge Suppressor cabled to the box in grounded flex.
 

kubeek

Joined Sep 20, 2005
5,718
100Mbit ethernet 100baseT is actually designed for 120m - a 100m hroizontal run plus 20m for for "vertical" run and cabling in the cabinets, if I remember correctly. So a good cable with solid core and a nice splice should work just fine. We used to test our switches with a 150m spool of non-inuctively wound cat5e (no splice though). And even the gigabit worked.
 

eetech00

Joined Jun 8, 2013
1,865
100Mbit ethernet 100baseT is actually designed for 120m - a 100m hroizontal run plus 20m for for "vertical" run and cabling in the cabinets, if I remember correctly. So a good cable with solid core and a nice splice should work just fine. We used to test our switches with a 150m spool of non-inuctively wound cat5e (no splice though). And even the gigabit worked.
[/QUOTE

But when designing a network cabling system, best practice is to design per specification. The specification for CAT5/6 twisted pair ethernet max cable length is 100 meters. If it’s longer , it won’t certify, so If you install longer cable lengths, you are inviting problems.

For example, if a $2000 switch is suspected of malfunctioning, one thing routinely asked is “is your cable system certified”. If you’ve run cable lengths longer then specification, it will fail certification. The manufacture won’t support your claim, and rightfully so..

eT
 

djsfantasi

Joined Apr 11, 2010
6,334
My experience is in an office, not industrial environment. Our space was one big rectangle. Straight runs to offices on the other side would have exceeded/ been close to the max run specification of Gbit Ethernet. Early test runs failed their qualification. And we ran dual fiber from one side to the other. Used fiber cards from the manufacturer, HP,
 

kubeek

Joined Sep 20, 2005
5,718
actually my memory serves me wrong, and the actual standard length is 90+10m as I now checked. Then it means we tested it with 120m cable, which is still more than OPs 107m, so I still believe that for a one off where you can test it sprawled on the ground before putting it into the conduit it is still ok.
Also I believe that if you have gigiabit devices on the ends then they should be much better at coping with subpar wiring when set to 100mbps then a normal 10/100mbit device would.
 

djsfantasi

Joined Apr 11, 2010
6,334
actually my memory serves me wrong, and the actual standard length is 90+10m as I now checked. Then it means we tested it with 120m cable, which is still more than OPs 107m, so I still believe that for a one off where you can test it sprawled on the ground before putting it into the conduit it is still ok.
Also I believe that if you have gigiabit devices on the ends then they should be much better at coping with subpar wiring when set to 100mbps then a normal 10/100mbit device would.
I agree... but you are paying for gigabit devices in order to get 100mbit operation. Sounds financially foolish to me.
 

MisterBill2

Joined Jan 23, 2018
5,738
If the splice, whatever you do, does not work at the screaming speed needed for fast HD video, SLOW DOWN. At a lower data rate those impedance bumps are much less obvious, and things can work adequately. It is always a trade-off between speed and accuracy, or reliability, or something else. But slower data is able to ignore a lot of noise.
 
I need to add about 50ft to a cat5e run that's about 300 feet by the time it will terminate. It's a pretty low volume of data that it needs to transmit so I guess I'm not super worried. I know its not a good idea but I wasnt involved in the planning, just need to make it work as well as I can. My thought is to use a punch down splice box enclosed in a weatherproof box since its outside. I could also put the rj45s on the ends of the cables and use a coupling like that but I really dont know what would be more reliable. I appreciate any input. It will also be following all code requirements for protection by using rigid pipe to protect the cable as it comes out of the ground to the box then emt to the inside of the building and at least a nema 4x box maybe higher rated since it's in a bit of a vulerable spot for water and dust.

I was thinking about a punch down box like that mounted in a weatherproof box. I kind of have to agree with the other guy though that they can be unreliable. Now I'm not sure how that translates to signal loss once it's done right though. I'm a bit unsure of how soldering wouldn't have the solder points touch in heat shrink and it wouldn't be nearly as big a deal to redo a punch connection to get it right. I'm not sure of the exact speed but its just s piece of equipment sending a 5 digit number to a printer less than once a minute. Dunno how that translates to speed.
before you get yourself in a lot of trouble, figure out what you need to do.

Some idiot, may have decided that CAT5 is for a doorbell. it's suspicious that it's one number.

Who is taking the responsibility for the splice?

This almost sounds like R232. If it is, you better know because 300' might not be good. It might say be RS485. It may not even be 10baseT.

Are the bean counters trying to save money and don;t know what they are doing.

This reminds me of a problem I had. I was told to spec a UPS for a PLC that only needed to run about 2-3 minutes or enough time for the generator to kick in. Management made noise. My selection was too expensive.

Something made the project more expensive that I uncovered. Not all UPS's are suited for generator operation. You generally need an always-on UPS because the frequency of the generator is unstable. I asked the right questions. besides that, it was for a safety critical piece of equipment - a PLC designed to shut-down a reactor in case of a power failure. The gases were Hydrides an very toxic.

If your application was R232, I would not even consider "CAT5". There are potential "ground loop" issues and with that distance, storms can raise the ground potential. A storm took out a computing center because grounds from one end of a 200' building were different during a storm.

RS485 might be better.

What about potential EMI and RFI.

SO, if they are going to tell you how to do it, then THEY take responsibility. Make sure that is clear.

I've done that. I said, if you don't let me clean up the rats nest, I'm not going to work on it. i wanted power run to an enclosure and I was told no. they had a whole bunch of "building centric" things like emergency shutdown switches, valves, hood air velocity detectors, gas cabinet air velocity detectors, fire alarm contact etc. and they could still connect the wires via barrier strips.

I could not even separate the two components to work on them. No drawer slides either. A 4RU and a 5RU high enclosures mounted in a rack. The system I did mounted the building stuff on the wall in an enclosure.

They decided to do a "sequential interlock" where a "glitch" would shut down the system and you would not know why it happened. It didn't latch.

The second thing they did was to not measure the hood velocity. I knew from experience that you had to clean the velocity detector every nice in a while. The one I did, I set the readout to read in %. As the sensor got dirty, you would see the % change and you would know when to clean it. The low velocity would also latch. A gust of wind would shut down our reactor and we would immediately know why.

I said I would compromise: I want 120 VAC power to this box to power the velocity detector and you can have you barrier strip as log as I connect it to a 37 pin connector and all the "building stuff" goes into this enclosure that's mostly empty, then we can move the rack enclosures when we need to so they can be worked on.

There my have been 4 wires that connected the enclosures together and 2-3 pairs that went to the reactor. A local gas detector and the output of the pressure sensor.
 
Top