best way to "extend" cat5e wire?

[purbeast0]

this weekend my buddy is helping me run wires down another level from my 2nd to lowest level. in my utility room i have an unterminated cat5e wire that is coming out of my wall. my plan is to extend it from that room, and run the new wire down to the lowest level through the walls.

what is the suggested way to connect the unterminated cat5e wire in my utility room to the new wire i'm going to be running? this is in-wall wire so it's the solid core wire.

 

[Cabletek]

They make couplers and such you could use, but if you can make it one wire it is the best idea, each additional piece you add is another potential point of failure in nay network design. You could even put in a dual outlet and connect both ports by using a short jumper so if you ever decided to use the outlet in the other room it was ready to go as long as you lost the first or put in a small hub/switch..

 

[purbeast0]

They make couplers and such you could use, but if you can make it one wire its the best idea, each additional piece you add is another potential point of failure in nay network design. You could even puit in a dual outlet and connect both to a part and use a short jumper so if you ever decided to use the outlet in the other room it was ready to go as long as you lost the first or put in a small hub/switch..

you just made me think that i will just attach ends to both of them and use a patch cable to connect the two ends. and this area is going to be exposed in my utility room so it isn't going to be in the wall, in case something did happen.

so thanks!

 

[Red Squirrel]

I've seen couplers that are punch down, basically same idea as a jack except it's punch down on both ends. You could use that, just make sure it's accessible. (ex: don't shove it behind a wall). If the run suddenly stops working at least you can troubleshoot it.

 

[wirednuts]

solder them if you want to maintain Gb speeds over 100 feet. keep the twists as consistent as you can too.

 

[ScottMac]

solder them if you want to maintain Gb speeds over 100 feet. keep the twists as consistent as you can too.

NO! BAD BAD BAD! WRONG WRONG WRONG ... There is nowhere on a structured cabling plant, of any Category rating, that soldering is acceptable. The heat would so mutilate the insulating material on the wire, that even if you could maintain proper twist and lay conditions (you CAN'T) the spec would be off for anything digital.

And please spare me the "I did it and it worked" BS ... it's not right and there's no way you'd get anywhere near optimal performance.

Insulation Displacement Connectors (IDC) is the only acceptable way to terminate a Category-rated cable; mod plugs or punch-own, that's it.

No soldering, no scotch tape, no twisted together and coated with nail polish, no electrical tape, no screw-down terminals, no twisting the wire around a nail (not even a real high-quality steel nail of any gauge), no twisting around a paperclip and taped to the wall etc., up to and including... no "Scotch-Loks" unless you can meet the other Cat specs for termination.

 

[wirednuts]

i beg to differ. just because soldering is not a common method of terminations, it doesnt mean its not good. there is no way you can tell me a punchdown block holds frequencies better then a proper solder splice. i dont believe it, and if i still had access to an otdr i would prove it.

the reason its not accepted as proper practice is because it would encourage hidden splices. inspectors hate those.

 

[ScottMac]

If you can produce a spec that permits soldering on Category-rated cabling for high-speed data, post it up.

You area dead wrong. It is not so much the quality of the contact, as maintaining the relationship of the conductor-to-conductor distances, the pair-to-pair distances, and the lay of the pair within the sheath. Outside the sheath, you are permitted exactly two occurrences of "naked pair" per span - at the two end termination points (1/2" for up to Cat5e, 3/8" for Cat6 and above).

IDC is "perfect" for these connections, because that is their design. In addition to displacing the insulation, the contact carves its edge into the conductor, effectively sealing out oxidation. That is one (of several) reasons that using stranded cabling (usually used for jumpers) fails in a fixed infrastructure function ... the small multiple strands do not punch in well, and do not make a corrosion resistant physical connection.

BTW: Soldering is not to make a connection secure and tight, it is ONLY to reduce the likelihood of oxidation. The contact quality is a function of the mechanical component of the connection, not the "glue" you cover it with. Soldering, in this context would drive the propagation characteristics of the cable well off the charts.

I'm also wondering if you also solder your fiber connections, and if so, what your "OTDR" fingerprint looks like .... (j/k, I know what you meant).

A TDR would show a much greater impedance lump with a soldered pair, because of the disruption of the C-C spacing and p-p spacing changing the characteristic impedance of the span. Return loss is significantly increased, destructive interference is much higher, crosstalk is significant;y higher ... everything important is degraded.

 

[Red Squirrel]

I'm sure soldering is fine. Heck, I've wrapped a 50 footer cable around a transformer once just to see what happens. Ethernet is more tolerant than most realize. Sure, these things are out of spec but does not mean it's suddenly going to stop working if you bend the rules once. Just don't make it a habit. In this case it's a "fix" so to me it's acceptable, but I would not make it a standard practice.

 

[skyking]

I'm sure soldering is fine. Heck, I've wrapped a 50 footer cable around a transformer once just to see what happens. Ethernet is more tolerant than most realize. Sure, these things are out of spec but does not mean it's suddenly going to stop working if you bend the rules once. Just don't make it a habit. In this case it's a "fix" so to me it's acceptable, but I would not make it a standard practice.

Try reading what Scott has written, understand all the terms and performance metrics of network certification, and then post intelligently.
I don't know 5% of what Scott knows but I do know he won't bullshit you about it.
About 5 posts back the OP figured out what he was going to do, it met the acceptable practices, and he left. He got it right.

 

[Red Squirrel]

Not saying he's wrong, and yes it's against spec/code etc... but what I'm trying to say is just because you are going against certain requirements does not mean it's not going to work. It just means that it is not guaranteed to work as well as the conditions/methods it was certified for.

Though personally in this situation I'd use the coupler I mentioned before using solder, just because it makes the job easier. But if that coupler is not available locally then you can either way 2 weeks for a part to come in, or do the job now and finish it.

 

[skyking]

A pair if keystones and a short patch cable. That's it. No waiting, no bad advice given.

 

[imagoon]

I would use keystones and a patch or a small patch panel depending on needs myself. As for the soldering... it might work but to me that is the same as the people that do things like like swap the power and neutral or the neutral and ground in outlets. It will work... It is not right and can be quite dangerous.

 

[wirednuts]

If you can produce a spec that permits soldering on Category-rated cabling for high-speed data, post it up.

You area dead wrong. It is not so much the quality of the contact, as maintaining the relationship of the conductor-to-conductor distances, the pair-to-pair distances, and the lay of the pair within the sheath. Outside the sheath, you are permitted exactly two occurrences of "naked pair" per span - at the two end termination points (1/2" for up to Cat5e, 3/8" for Cat6 and above).

IDC is "perfect" for these connections, because that is their design. In addition to displacing the insulation, the contact carves its edge into the conductor, effectively sealing out oxidation. That is one (of several) reasons that using stranded cabling (usually used for jumpers) fails in a fixed infrastructure function ... the small multiple strands do not punch in well, and do not make a corrosion resistant physical connection.

BTW: Soldering is not to make a connection secure and tight, it is ONLY to reduce the likelihood of oxidation. The contact quality is a function of the mechanical component of the connection, not the "glue" you cover it with. Soldering, in this context would drive the propagation characteristics of the cable well off the charts.

I'm also wondering if you also solder your fiber connections, and if so, what your "OTDR" fingerprint looks like .... (j/k, I know what you meant).

A TDR would show a much greater impedance lump with a soldered pair, because of the disruption of the C-C spacing and p-p spacing changing the characteristic impedance of the span. Return loss is significantly increased, destructive interference is much higher, crosstalk is significant;y higher ... everything important is degraded.

soldering does not make a connection secure and tight? and any type of splice creates a significant loss. a proper solder job isnt going to make it worse then a punch down splice. i will believe you when i see it. someday ill try it myself if nobody comes up with a link..

 

[wirednuts]

ehow says you can! i win! internet is always right!

http://www.ehow.com/how_8672703_splice-network-cables.html

 

[skyking]

Thanks for putting the credibility of Ehow in crystal clear perspective.

 

[FrankSchwab]

Having once been in the business of building certification devices for installed UTP cable (Microtest, may it rest in the stomach of Fluke Networks forever), I can say with complete confidence that, well, both ScottMac and WiredNuts are wrong, and both are right.

WiredNuts is absolutely correct that Ethernet is far more forgiving of wire quality than the specs are. ScottMac is absolutely correct that it is completely unprofessional to splice a cable, and is unlikely to pass any kind of certification test but he's wrong about Soldering not making a connection "secure and tight". Solder is extremely electrically conductive, and makes very intimate contact with both wires so it creates a great electrical connection even without a mechanical connection (if it didn't, soldering components onto PCB's would never work). As well, solder provides a great mechanical connection - put two 26 gauge wires side by side, solder them generously, then put tension on the wire; it's likely that the wire will fail before the solder joint will.

For histories sake, 10Base-T (1990) and Token Ring originally worked on Cat-3 cables (Actually, 10Base-T has been demonstrated over Barbed wire ) with the advent of the Cat-5 spec in 1991 which support them and looked forward to a 100 MHz future, most of the building wiring done during the rise of the internet was done with Cat-5. When 100Base-TX was finalized in 1995, Cat-5 was already in place to support it. 1000Base-T came about in 1999, with a design goal of working with Cat-5 cable because that's what the world was wired with - but it was only barely able to do that. The Cat-5e spec was published that year to help support 1000Base-T. Cat-6 came about in 2002 and drastically tightened cable construction and installation specs. Please don't make me remember terms like "Equal level far end cross-talk" - the nervous twitch will come back.

The cable and installation specs are designed to make sure that 100% of installed cables will work perfectly with the appropriate networking technology - with margin to spare in case the wire drapes over a magnetic ballast for a fluorescent bulb, crosses in front of a 2.4 GHz WiFi access point, or has a sharp bend. Also note that most networking protocols will do their damnedest to get a packet through, and you may not even notice if your network is retrying 10-20% of packets due to data corruption.

If you do something redneck like soldering a connection and wrapping it with electrical tape, your connection will most likely be rejected by any certification test because doing so will grossly impact the characteristic impedance at that point. However, 100Base-TX will likely work just fine - as long as that solder junction isn't what's draped over the ballast or stuck in front of the AP, and as long as you're fairly careful about what you do (e.g. not untwisting the cable more than an inch or so, and not melting through the jacket), and as long as you don't try to run networking protocols that are barely supported by the installed wiring (1000Base-T over Cat-5, for example, would probably be very unhappy about this repair).

There is no "acceptable" method to splice two cable ends together in the field, but if you were going to try, I'd start with something like http://www.firefold.com/RJ45-Junction-Box-CAT5E . It'll probably work beautifully.

 

[ScottMac]

ehow says you can! i win! internet is always right!

http://www.ehow.com/how_8672703_splice-network-cables.html

To the best of my knowledge, Ehow is not yet accepted as any kind of standards maintenance body.

There's some websites that'll give you good, solid information on the impending alien invasion too ...

 

[ScottMac]

Having once been in the business of building certification devices for installed UTP cable (Microtest, may it rest in the stomach of Fluke Networks forever), I can say with complete confidence that, well, both ScottMac and WiredNuts are wrong, and both are right.

WiredNuts is absolutely correct that Ethernet is far more forgiving of wire quality than the specs are. ScottMac is absolutely correct that it is completely unprofessional to splice a cable, and is unlikely to pass any kind of certification test but he's wrong about Soldering not making a connection "secure and tight". Solder is extremely electrically conductive, and makes very intimate contact with both wires so it creates a great electrical connection even without a mechanical connection (if it didn't, soldering components onto PCB's would never work). As well, solder provides a great mechanical connection - put two 26 gauge wires side by side, solder them generously, then put tension on the wire; it's likely that the wire will fail before the solder joint will.

For histories sake, 10Base-T (1990) and Token Ring originally worked on Cat-3 cables (Actually, 10Base-T has been demonstrated over Barbed wire ) with the advent of the Cat-5 spec in 1991 which support them and looked forward to a 100 MHz future, most of the building wiring done during the rise of the internet was done with Cat-5. When 100Base-TX was finalized in 1995, Cat-5 was already in place to support it. 1000Base-T came about in 1999, with a design goal of working with Cat-5 cable because that's what the world was wired with - but it was only barely able to do that. The Cat-5e spec was published that year to help support 1000Base-T. Cat-6 came about in 2002 and drastically tightened cable construction and installation specs. Please don't make me remember terms like "Equal level far end cross-talk" - the nervous twitch will come back.

The cable and installation specs are designed to make sure that 100% of installed cables will work perfectly with the appropriate networking technology - with margin to spare in case the wire drapes over a magnetic ballast for a fluorescent bulb, crosses in front of a 2.4 GHz WiFi access point, or has a sharp bend. Also note that most networking protocols will do their damnedest to get a packet through, and you may not even notice if your network is retrying 10-20% of packets due to data corruption.

If you do something redneck like soldering a connection and wrapping it with electrical tape, your connection will most likely be rejected by any certification test because doing so will grossly impact the characteristic impedance at that point. However, 100Base-TX will likely work just fine - as long as that solder junction isn't what's draped over the ballast or stuck in front of the AP, and as long as you're fairly careful about what you do (e.g. not untwisting the cable more than an inch or so, and not melting through the jacket), and as long as you don't try to run networking protocols that are barely supported by the installed wiring (1000Base-T over Cat-5, for example, would probably be very unhappy about this repair).

There is no "acceptable" method to splice two cable ends together in the field, but if you were going to try, I'd start with something like http://www.firefold.com/RJ45-Junction-Box-CAT5E . It'll probably work beautifully.

Having once been in the business as an engineer in Anixter's Interoperability Lab, which also included the Cable / Component qualification Lab (we also tested the MT products -and others- before accepting them into the product mix), I didn't say it wouldn't work, I said with was BAD and WRONG (I intentionally left out "stupid", though it is / was accurate); any metal conductor will likely get something through. The assumption is that having gone through all the expense and fun of stringing the cable, the implementer / user would expect maximum performance .... get what they're paying for. That ain't gonna happen by soldering the conductors. It's not. Neither is barbed wire, coat hangers, a pair of electrician's scissors or a variety of stuff we tested and displayed at a BICSI show. We also put 155m copper ATM through a knotted up ball of Belden MediaTwist (that was also run over by a couple cars and a truck) ... worked OK, but not up to spec.

Squirrel: Wrapping 50' around a transformer introduced common mode noise ... twisted pair handles constant-level common-mode noise with no problems. It's the spiky impulse noise that flaky fluorescent baluns and old bulbs produce and / or unbalanced pair that cause the most problem.

(quote)WiredNuts is absolutely correct that Ethernet is far more forgiving of wire quality than the specs are. ScottMac is absolutely correct that it is completely unprofessional to splice a cable, and is unlikely to pass any kind of certification test but he's wrong about Soldering not making a connection "secure and tight". Solder is extremely electrically conductive, and makes very intimate contact with both wires so it creates a great electrical connection even without a mechanical connection (if it didn't, soldering components onto PCB's would never work). As well, solder provides a great mechanical connection - put two 26 gauge wires side by side, solder them generously, then put tension on the wire; it's likely that the wire will fail before the solder joint will. (end quote)

Soldering , in the electronic / electrical context is NOT meant for the physical connection, other than to reduce oxidation or corrosion. On a PCB the component leads are in contact with the traces, and the solder ensures the connection remains stable.

Laying two conductors side-by-side and slathering them with solder is poor technique, at best. In most cases, the conductors are twisted together in some fashion; that's the electrical connection, not the solder (though solder is conductive, of course).

There are several acceptable methods for joining / splicing a span; some of the easier ones have been covered above (two connectors & a coupler, two plates and a jumper ...) Soldering, in any form or fashion, is not one of them. It never was, and is likely never will be. Not for high-speed data.

And, as I've said many times before: what you do on your personal home network is your choice ... nobody cares what you do to yourself / your stuff. Y'all can solder until you catch a buzz on the fumes ... it doesn't matter to me, it doesn't affect me or what I do even a little bit.

Where I have an issue is suggesting it to the public that doesn't understand enough to know how bad it is. Adherence to standards means that some poor bastard down the line won't have to suffer the stupid and lazy techniques that some choose to implement (even when it is explained to them WHY it a bad idea). If anyone tries this junk tech on a business LAN and / or "for money" should be keel-hauled, drawn & quartered.

 

[ScottMac]

(Quote) For histories sake, 10Base-T (1990) and Token Ring originally worked on Cat-3 cables (Actually, 10Base-T has been demonstrated over Barbed wire )(End Quote)

Frank: The original LatticeNet (Synoptics) ran over plain phone wire ... not Category-anything ... just plain old phone wire (DIW).

Token Ring ran over "Type 3" not Cat 3. Type 3 was an IBM spec for really nice plain old
phone wire (and they had some propagation specs to apply).

I worked with all of it at the time and got an IBM certification for their Structured Cabling System.

I also used to argue against the LatticeNet concept, instead arguing in favor of coax (Ethernet). "Why would you trust your high speed data (10Mbps) over some flaky phone wire? Coax is shielded and secure ..."

Of course, times changed, structure cabling developed and evolved, thanks largely to Anixter's "Levels" programs, and onward we go ...

 

[imagoon]

I agree with ScottMac on posting that stuff on the public forums. I serious never understand why so many people put so much effort in to doing it wrong. Inside your house doesn't affect me but posting "solutions" like that get read and eventually becomes "accepted." I mean look at the obsessions most people seem to have with crimping 8P8C connectors on solid core cables [for any reason mostly, except for T1 cards etc.] It is not an accepted practice and most people do it because it is "cheaper" when it typically isn't or because there cousin's roommate's sister's boyfriend said to do it.

It really isn't hard to do it correctly. It actually takes more time and effort to do it wrong but people seem to insist on doing it.

 

[drebo]

The only time I'd ever splice a cable together would be to do something like this: http://www.installerparts.com/5057-cat5e-junction-box-110-punch-down-type.html

That said, in the case of the OP, I'd just terminate the cable to a keystone (or patch panel if there were lots of them) in the utility room and then do the same for the new cable down to the basement. Then just connect them with a patch cable. That way, if he ever wanted to add a device in the utility room, it'd be easy to add a switch and not have to worry about reterminating anything.

Soldering wires together or using crimp splices will probably work, but it'd get you fired from any professional cabling gig. It also takes a LOT more time than just punching it down into a keystone.

 

[FrankSchwab]

Frank: The original LatticeNet (Synoptics) ran over plain phone wire ... not Category-anything ... just plain old phone wire (DIW).

Token Ring ran over "Type 3" not Cat 3. Type 3 was an IBM spec for really nice plain old
phone wire (and they had some propagation specs to apply).

You're absolutely correct about Latticenet, and the original 10Base-T. As I recall (perhaps incorrectly), Category-3 was essentially a standard to homologate current practice of telephone twisted pair, and had essentially the same characteristics.

And, as I recall (again perhaps incorrectly), 16 Mbit Token ring was often run over installed Cat-3 or Cat-5 wiring and RJ-45 jacks (no one ever used the Cat-4 wiring defined for 16 Mbps token ring). It may not have been the way that IBM wanted it installed, but it surely worked and was surely used. Our certification products at the time had a specific tests to verify that Token Ring would work on the installed cable.

 

[ScottMac]

You're absolutely correct about Latticenet, and the original 10Base-T. As I recall (perhaps incorrectly), Category-3 was essentially a standard to homologate current practice of telephone twisted pair, and had essentially the same characteristics.

And, as I recall (again perhaps incorrectly), 16 Mbit Token ring was often run over installed Cat-3 or Cat-5 wiring and RJ-45 jacks (no one ever used the Cat-4 wiring defined for 16 Mbps token ring). It may not have been the way that IBM wanted it installed, but it surely worked and was surely used. Our certification products at the time had a specific tests to verify that Token Ring would work on the installed cable.

Before Cat3 was Type3, before Type3 (defined by IBM) it was plain old twisted pair.
Some where in the middle was the Anixter "levels" program that guaranteed, on paper, that cable certified to the stated level would perform to that level. A sample from every batch was tested; if it failed, the entire batch was returned to the manufacturer (who re-sold it to someone else). I believe Anixter still does this.
The Levels program was recognized as a Good Thing by the industry, a committee was formed, and the Category rating system was created.
IIRC, the current Levels program still has thresholds above the comparable Category rating.

Four meg TR "required" Cat4 or better, 16 meg "required " Cat 5 or better.
Cat 4 was pretty short-lived, people were still learning what all that Category stuff was all about, and with Cat 5 following shortly after, smart businesses jumped to the higher standard. Businesses that went Cat4 to "save money" instead of Cat5 (Cat5 was more than 50% more expensive) got to pay again when Fast Ethernet and ATM hit the streets.
Back in the day, people cheaped-out buying Radio Shack coax instead of Certified Ethernet grade (and performance suffered). Today people come up with stupid cabling tricks (like soldering) with UTP, and performance suffers.

 

[skyking]

anything before cat3 the phone guys call "jake". I've been forced to do some really bad things by circumstances beyond my control, to get some 10baseT going. I won't post about them here