I made a patch cable!

[spidey07]

Well against all my better judgement I just HAD to make four temporary 50 foot patch cables today using solid belden datatwist 5e.

Crimped some ends on 'em - man those crimps looked perfect. Hooked 'em up to the omniscanner and POW. 3 out of 4 failed the cat5 tests.

Just for giggles I hooked sniffer upto a switch with the fault cables - got 100/full link and everything, could even transfer some files. Errors all over the place.

Guess I remember why I suggest not making patch cables, huh? Had to learn my own lesson all over again.

 

[miguel]

Uhhhh, I've made probably hundreds of patch cables. Maybe only 1% had problems. Maybe you should practice...

 

[spidey07]

Ever test them?

<edit> I meant certify them

 

[miguel]

I tested them using (darn, can't remember the brand) a hand-held network analyzer that has a function to test cat5 cables. They are also in use in a production LAN.

<EDIT>The store-bought cables are not made by rocket scientists, you know... </EDIT>

 

[ScottMac]

NO, the "store bought" cable are usually made by machine (to higher tolerances) then tested (automatically) to verify compliance.

DataTwist was kind of an intermedite challenge though, I bet, because you have to split the pairs (each pair is bonded together...make for a more consistant impedence though the length of the cable).

I can usually crank out a pretty good patch cable using solid conductor. Stranded, well, I just hate the sh*t. I'd rather eat a plate of Liver than do a stranded cable. I can do 'em, but I'd rather not.

FWIW

Scott

 

[RagManX]

When I was a networking guy, I learned to make cabling. Had 100% success rate with CAT-5 on shared 10Meg networks. When I went to a place with siwtched 100 Meg networks, I had a success rate of about 15%. After getting the cabling guys there to teach me how to use the Fluke to test my cabling, I took the time to learn how to do cabling the right way, and by the time I left, I was better 90% on making cables. For me, it just took awareness that those specs really meant something and then taking the time to learn them that made the difference. I'm pretty sure you already know the specs, so for you it is probably just an issue of practice, which you probably don't want to do, since in the general case, you probably have little need to make cables.

I sure remember how frustrating it was trying to fix cables without the Fluke, when you didn't even know what the real problem was.

RagManX

 

[nightowl]

At an internship I had to make quite a few > 50ft cables and each one worked fine at 100 Full-Duplex. I have always made my own cables, not a single problem yet. I do take my time and always cut off any excess on the cable if I untwisted too much. I even tested them with a PentaScanner and all were verified Cat5 complient.

 

[Nutz]

<< Uhhhh, I've made probably hundreds of patch cables. Maybe only 1% had problems. Maybe you should practice... >>



Ditto. I usually use our Fluke afterward though. Only once in a blue moon do they go "bad" after installation.

By the way, I'm red & green color deficient (blind/off with certain hues).

Come to think of it, it may not be the method but the material or tools being used. I remember when we got this crappy batch of CAT-5 in. Nobody could me them work, and if they did they failed within a week.

 

[spidey07]

cool.

my point was really more about how you can make a cable that looks great, gets a 100 meg link (you'll get that on just about any thing), appears to work - but yet is full or errors and doesn't pass muster.

 

[Mucman]

How do you certify cables? I have made about 40 so far for our network room and I am just testing them with a $100 tester from Home Depot. They are working great right now, but I understand that if I made them poorly they can fail 6,12 months down the road.

I really enjoy making cables I find it very relaxing.

 

[Abzstrak]

People can make good cables, I've actually gotten pretty good at it, but it does take alot of practice... I dont make them often but probably can about 3 outta 4 that'll pass certification with flying colors. I think the main key is taking your time and like ragman said, knowing the limits of the specs is important.

Its truely not worth my time though, its alot cheaper to buy cables than have me make them ;-)

Spidey is making a good point though, you MUST certify them cause they can look perfect and run like crap.

 

[Mucman]

how much do those cable certification testers cost?

 

[CBuxton]

Good thread, I do a lot of cabling and have a VERY simple testor. Thinking that "if it connects it works" is a really bad theory. Can somebody recommend some cable testors? How about on that is entry level, cheap, mid level, and high end? Thanks!

 

[ScottMac]

Most of the cable testors capable of Certifying (I mean, like legal certification..something you could go to a courtroom with) are US$2500.00 up to the tens of thousands.

The cheaper ones will do continuity and position, maybe even a high frequency burst, maybe even some TDR stuff (tells you the length of the span)...but aren't able to fully ceritfy the cabling above cat3 or 5.

What I'm getting at, I guess, is there's "Certification,"...and there's certification...The expensive testers certify the cable...and you can be high-ninety-something-percent confident that the cable meets or exceeds category specs.

The less-expensive testers will scan the cable and tell you it's connected, the pairs are in the right order, and the cable is probably about XX meters long...and more than likely, it'll work.

WaveTek, Fluke, MicroTest are all pretty expensive (more than a couple thousand dollars).

FWIW

Scott

 

[Mucman]

Good thing I saw this thread! I was being ignorant and assumed that if my cables worked that they were fine. Part of my project was to make about a 100 patch cables, but I realised I was using solid cable (which is great for doing the structured cabling in our network room) but is not appropriate for patch cables.

I managed to convince my boss to purchase the patch cables from Anixter (~$550 )), but when I asked the Anixter sales guy about buying a spool of stranded cable, his reply was that they don't sell bulk stranded cable because nobody can make them up to a certified spec! Even though this is my first real network I am building I plan on doing it right!

I just hope my punch-downs are good. I will take pictures when I am done so ScottMac can critique (or slam ) my cabling job.

 

[Tallgeese]

Layer 1 = Your Whole Enchilada Relies On Good Cabling

 

[DocDoo]

<< Well against all my better judgement I just HAD to make four temporary 50 foot patch cables today using solid belden datatwist 5e >>


Since SOLID CAT5 takes different style RJ45 ends (the pins are different), could that made the difference?

Of course I am assuming something without knowing all the details...

PS. Almost forgot, that comment miguel made...made me laugh.

 

[gogeeta13]

I have made about 25 cables, some small and some 50feet+. I never used a tester or anything and they all work fine. I used bulk solid cable and some $8 pack of boots and connectors from microcentor. All of my cables transfer fast as hell, and I have not gotten one bit of corruption. I only use them on a 100bt full duplex network work, but they are fine for me!

 

[spidey07]

don't muck with the physical layer.

to gogetta - how do you know you are not getting any corruption? Seriously - how do you know?

 

[gogeeta13]

<< don't muck with the physical layer.

to gogetta - how do you know you are not getting any corruption? Seriously - how do you know? >>



Well, I consider nearly 24/7 data transfer use, and about 3-4 ISO tranfers a day(and they all work, and pass CRC and MD5 SUM checks) a pretty good sign. Frankly, i could care less if they arent 100%, 99%is fine for me. And who cares if they are certified? I have bought belkin patch cables that transfer at 50% the speed of my home-made ones.

 

[CBuxton]

How could one go about testing your wiring infastructure w/o purchasing a $4000 Fluke testor? I've done cable installation for a couple of small businesses and did the following: MediaTwist Cat6 to Hubbell Cat6 jacks to Hubbell Cat6 patch panel, with purchased MediaTwist 1' patch cables to the switch, and 6' MediaTwist cables to the desktops. Is doing the punchdowns on the jacks or patch panel LESS susceptible to cabling errors vs. crimping down RJ-45 ends? I'd be curious to see how "good" the cabling is. Obviously I'm having NO problems connecting over the network, but as we've seen this doesn't necessarily mean that there aren't problems. I've transfered several large files across the network and it sure SEEMS faster than other 100Mbps networks that I've been on, but I'd like to know if I have any problems. Ideas? Thanks for a good topic!

 

[randomlinh]

for home use, i just do it cause patch cable just doesn't seem worth it... #1, i needed to run wire from upstairs to the basement.. buying the reel of 1000 was cheapest overall. made some patch cables and what not (home and lan use). I test by doing transfers and using netcps (yeah, professional eh? ) and average around 7-8MB/s w/ low pings. Now, might not be acceptable for professional use.. but nice and affordable our wannabe networks

 

[toshiba3020]

Almost every cable I have made myself works great. I have real problems with belkin cables though. They die all the time.

 

[Mucman]

For the people who are stinking up for cable making - I think the point Spidey07 is trying to make is with reference to professional networks. The reason why I decided to purchase cables instead of making them was because of the following :

1 - Anixter does not sell stranded cable. The guy on the phone mentioned the issues that have had with people making their own patch cables and it was not worth it. I was
contemplating making patch cables out of solid cables but re-thought my situation because of reason 2.
2 - Working as a web-host reliability is our number one issue. We have clients that notice a minute of downtime. What do you think they will say when their server was down because a flaky cable got bumped the wrong way? Making over 100 cables, and only testing them with a $100 cable tester is not a good way to ensure our advertised 99.9% guarantee. When linking up thousand and thousands of dollars worth of equipment, why be cheap and make your own cables?
3 - Making a 100 cables would take me a while

If the cables are for your basic home network then making cables is fine. It's convenient and it's nobodies arse on the line when something goes down. When you have networks where you got traffic going over those cables 24/7, it's another story.

btw, I just received my shipment of cables today How do you like a green and yellow colour sceam?

 

[spidey07]

Thanks for the vote of support Mucman.

I'm really trying to drive home the importance of physical layer here. I have been building networks since 1989 and the number one reason bar none above all for poor performance or just flaky connectivity is cables. I've charged many a company many a dollar to come in and do a network analysis (full week of consulting) only to recommend "fix your cable infrastructure." From my experience (100+ network consulting jobs) at least 70 percent of their problems were physical layer related. This fact is undisputed. Just because you get a link light and can communicate does in no way shape or form means you have a good cable.

So, this thread was hopefully helpful to folks who have had intermittant connectivity problems or slow performance. Lesson learned -

IF YOU HAVE ANY KIND OF PERFORMANCE PROBLEMS OR GENERAL WEIRDNESS WITH NETWORK COMMUNICATIONS - CHECK THE PHYSICAL LAYER FIRST. NETWORKS SHOULD BE TROUBLEFREE.