Ethernet Cable Strangeness

[Lee DeRaud]

Sometimes tech moves in mysterious ways...

My TV is old and not-so-smart, but I recently got a new BluRay player that also handles Netflix, Amazon video, etc. Unfortunately it's about as far from my WIFI router as it can get and still be in the same house, so the connection is kind of twitchy.

I finally decided to run an Ethernet cable, following the same path as the cable coax, total run maybe 80 feet. Snaked the cable, put RJ45 plugs on both ends, hooked up the tester: no joy. Replaced both ends multiple times, kept getting 'no signal' through conductor 4 (blue).

I figured something had happened to the cable itself while fishing it under the kitchen cabinets or through the wall between the garage and the study where the hub is located. So, before running another chunk of cable, I cut it at the halfway point and threw on a couple of plugs. Tested TV to cut point: good, problem must be in the other piece. Tested hub to cut point: also good. Huh?!? I put an Ethernet coupler between the two ends, cable now tests good hub-to-TV.

Ok, problem solved, but I'm starting to think they should have named it "Eithernet", because it either works or it doesn't, with no particular logic. Can anybody suggest a plausible explanation for what happened?

[Steve Demuth]

Lee,

Were you actually wiring it for ethernet? The blue pair isn't actually used in plain 10Base-T or 100Base-TX ethernet. Unless you were running power over ethernet (POE), or for some reason are trying to run 1000Base-T, its just a wire there along for the ride. If it's POE, the "signal" is just 48V DC.

Which brings another question: which standard was your tester testing for?

[Jim Becker]

If it didn't work, it's most likely that how you terminated the cable is the issue. Putting the RJ45 connectors on the cable is an "art" and requires keeping careful track of and maintaining proper twist of the individual pairs right up to the connector...all with just two hands and ten fingers. Not sure what you used for the connectors but the simple clip on types often found at the home centers are not all that reliable to install. In most cases, a consumer is best served by buying a pre-terminated cable, quite honestly. It took me quite a few iterations to learn how to correctly terminate cables years ago (I was in the IT and telecom space for most of my career) and even when I do it now, I have to pause and collect myself...and I have a pro-level crimp tool.

And it's correct the blue (center) pair isn't normally used for Ethernet...only the orange and green pair. Brown (outside pair) also usually remains unused. But all must be terminated correctly.

[Lee DeRaud]

Lee,

Were you actually wiring it for ethernet? The blue pair isn't actually used in plain 10Base-T or 100Base-TX ethernet. Unless you were running power over ethernet (POE), or for some reason are trying to run 1000Base-T, its just a wire there along for the ride. If it's POE, the "signal" is just 48V DC.

Which brings another question: which standard was your tester testing for?

The wired Ethernet out of the Spectrum router/hub is 1000Base-T, although that's certainly overkill for the BD player. (I can't tell offhand what speed the BD player is actually using from its menus. I can probably check on the router side if it actually matters.)

As far as the tester, it's just a cheap DataShark widget I got at Home Depot, but it seems to work on all the "factory" CAT5e/CAT6 cables I have on hand. It just sequentially scans the 8 wires...easy to tell if a pair is reversed as the LEDs will light up out of sequence. The "error" I was seeing was that it simply skipped over #4 during the scan.

[Lee DeRaud]

If it didn't work, it's most likely that how you terminated the cable is the issue. Putting the RJ45 connectors on the cable is an "art" and requires keeping careful track of and maintaining proper twist of the individual pairs right up to the connector...all with just two hands and ten fingers. Not sure what you used for the connectors but the simple clip on types often found at the home centers are not all that reliable to install. In most cases, a consumer is best served by buying a pre-terminated cable, quite honestly. It took me quite a few iterations to learn how to correctly terminate cables years ago (I was in the IT and telecom space for most of my career) and even when I do it now, I have to pause and collect myself...and I have a pro-level crimp tool.

Not my first rodeo: the crimp tool, cable, and connectors are the good stuff. :-) (I went with the feed-through connectors so my old eyes can check the order before I hit it with the crimper.) The tester is pretty good about spotting flipped pairs etc.

I've buggered up enough connectors snaking pre-terminated cables that I figured, if I'm going to end up replacing one end, I might as well do both. Not to mention that the pre-made stuff only comes in two sizes: too short and too long.

[Doug Garson]

Maybe when you cut the cable in the middle you cut thru the one defect in the cable, quick go buy a lottery ticket.

[Lee DeRaud]

Maybe when you cut the cable in the middle you cut thru the one defect in the cable, quick go buy a lottery ticket.

Yeah, what are the odds...

[Frank Pratt]

I firmly believe in sticking to EIA/TIA 568 standards, cause if you don't, eventually you get bit. Firstly, putting male plugs on the ends of solid copper cable is against spec, and for good reason. I'm in the business & even with professional installers doing the terminations, the failure rate is shockingly high. You're supposed to terminate both ends with RJ45 jacks & then use patch cables to connect the jacks to the equipment. Sometimes a customer just wants them done with a plug on each end, but we've had too much trouble with that.

Another no-no is putting a splice mid-run. A link is defined as the run of solid copper cable terminated in an RJ45 jack at each end and a patch cable at each end. Couplers or splices will put the link out of spec. Will it work? Probably, but that's just one more failure point, and it will probably cause some degradation. A kink in the cable can certainly cause the link to fail.

We test every drop we install for wire mapping & bandwidth. Except for the occasional screw up with wire mapping when one of the guys gets a couple of wires mixed up, a failure is extremely rare. Once it tests good, it stays good. Not the same when using male plugs.

I do believe that a copper link is superior to relying on WIFI. The only things that we use WIFI for in our house is for mobile devices.

[Jim Becker]

Sorry, Lee, I didn't know you were experienced at it. Could just be a bad connector or a bad crimp...redoing the ends absolutely makes sense!

[Lee DeRaud]

[UPDATE] I had the BD player hooked up to the hub with the cable laying out on the floor, everything working fine. Finished lunch and went back to clean things up. Disconnected the BD end, ran the cable through from the back of the TV cabinet, and (paranoid as I am at this point) put the tester on it again. And the LEDs on it started doing random stuff like I'd hit it with a hammer.

And then a light dawned. Opened up the tester and the battery was the kind of no-name crap that usually ships with widgets like these, left over from when I got it three years ago. I popped in a fresh 9V and everything started working again. Best guess is, the cable was fine the first time I terminated it.

So the lesson is, if you get the same test results after multiple attempts to fix a problem, make sure the test gear is working. Now I'm getting flashbacks of the day decades ago when I learned that a bad 'scope lead can make a perfectly good square wave look like random noise.

[Jim Becker]

I think that a battery thing like that is a direct corollary of the "is it plugged in?" troubleshooting step...

[Lee DeRaud]

I think that a battery thing like that is a direct corollary of the "is it plugged in?" troubleshooting step...

Yeah, but not as obvious: if the LEDs are blinking, "dead battery" usually isn't the first thing that comes to mind.

I'll admit, they were a LOT brighter with the new battery though.

[Aaron Rosenthal]

I smiled all the way through your “aha” moment Lee.
Many years ago we lived in South Africa. There was a certain mentality about the lowest common denominator type of worker.
The fable went thus: A worker was given an orientation as to the sound difference between hitting a railroad car wheel which was sound, and the dull thud of one which was defective.
The worker was given a new hammer, and was sent out to the yard to check the carriage wheels.
He condemned over 1000 carriages before someone realized he had a defective hammer.

[Steve Demuth]

[UPDATE]
So the lesson is, if you get the same test results after multiple attempts to fix a problem, make sure the test gear is working. Now I'm getting flashbacks of the day decades ago when I learned that a bad 'scope lead can make a perfectly good square wave look like random noise.

This is a neat little illustration of a lesson even the most experienced teams seem to need to learn over and over. We manage over 300,000 devices and 2000 applications in hundreds of buildings across many campuses in multiple states. When the Operations Center reports an issue, verifying the telemetry should always be part of the protocol. When debugging an actual outage, verifying that the telemetry is reporting accurately on what's happening should always be part of the protocol. But the truth is, if the telemetry is faulty it will take at minimum twice as long to discover the root cause of any problem, because nobody ever suspects it until they have exhausted every other possibility.

[Myk Rian]

Yep, you have to make sure the test equipment is ok.
Part of my job was the repair, calibration, and certification of test and measuring tools. I needed a 100', certified tape measure, and bought a Starrett steel tape, with Cert paper and trace numbers.
I was getting funny readings while testing several tape measures. They were all too long. Turns out the "Certified" Starrett started at 3 3/8".