Homemade cat5 cable using existing phone line fails.

Well, what I was trying to say is that if you can't amortize the cost of a crimping tool over many cables, the one home-made one will cost more than a manufactured cable. If you factor in the cost of tools, materials, the yield rate and your time then it's never worth it to crimp your own ethernet cables, as you point out.

This plug has 8 terminals that matches the exact dimensions to a normal cat5 plug from what I'd measured using a precision caliper. If there is chemistry involve that makes this unsuitable for network use then what will it be?

>

The drawings in your PDF show too much untwisted wire to meet CAT5 spec. CAT 3 maybe, and it might work at 10MB.

Here's a pic. There's MUCH to MUCH untwisted wire.

Buy a real patch cable.

Look REAL close at the gold plated tines. If the two or three are offset, then its for solid conductor. If the two are in line, its for stranded cable.

>

Thank Eli Whitney for that.

A 4 pin handset plug will fit in the jack, so why don't you solder that up and use it?

The particular plug you refer to was designed a long time before Cat-5e was a standard.

It's O.K. In fact, rather than use any kind of standard, take you 8 conductors and put them down in any color sequence you like. Just match them on the other end and check them with a "tester" for continuity. If it works, you did a good job. If it doesn't you can post to 4 usenet groups and then argue with the standards.

Oh, you already did.

Carl Navarro

In comp.dcom.cabling Sam Nickaby wrote in part:

>

The network "chemistry" is balanced signalling. It is the magic that allowed clever network engineers to run 100 MHz over 100m of wild country ten+ years ago when motherboard designers could hardly get 50 MHz over 20cm on a PCB with unbalanced signals.

In this case, your plug has projections that may block other ports or might not fit in the jack properly.

Less obviously, it has long parallel unbalanced conductors. AKA "untwist". A crimped plug has about 1 cm of untwist. This low-tool plug has at least 3 cm.

But this isn't as fatal as trying to run silver satin at 100 MHz.

-- Robert

Forgive me - I kind of lied. I dug out what's left of the 100-footer, and it might actually be cat 3 - it's not the flat "silver satin" stuff, it's round, and beige, with solid wire with about 1 or 2 twists per foot. The connectors, I scrounged from a previous installation that they were trashing. They were crimped, on 8-conductor solid (real cat 5) wire; I cut it off about 4" from the end, then cut the 4 unused wires all the way back to the connector body, and just twisted them together and wrapped black tape around them.

Worked fine, until somebody ran over it with the fork lift. )-; And my laptop died, so there's not much point in wiring the trailer any more. )-; )-;

Thanks! Rich

Sam Nickaby wrote on 3/11/2006 5:41 AM:

Just get a couple of these and dont worry about where you put the DSL modem.

Jim

No chemistry, physics. There are *lots* of DC and AC electrical properties that are every bit as important as the physical properties for correct operation. Your plug is probably OK as far as the DC properties, but there's no way it's going to come anywhere close to meeting the AC properties required for a reliable 100BaseT connection.

-Larry Jones

Hmph. -- Calvin

A cable like that costs $3. Why are you doing this?

"

"

Having a correct pin out doesn't a Cat5 cable make. It has to do with signal handling. And this one will not pass muster.

Each wire pair in a cat 5 cable has a different number of twist, making each pair have a different capacitance per foot. The regular phone wires have no twist and no pair wires. The phone wires are meant to pass dc signals and the cat 5 wire pass modulated frequencies, so without the wire pairs in the cable with the twist - the 2 modulated frequencies crosstalk and interfear with each other.

Ya know, if you keep posting to this thread, you can make it to 4 months. The sad part is that Sam (the OP) is probably dead serious in how he did this. Two 4C silver satin base cords shoved into a retrofit 8C shell like we used for solid wire to an AT&T 8 pin plug and when the tinsel wire wouldn't fit he used a soldering iron.

Sam, buddy, you goofed. In order to solder wires like that, you need a telco approved soldering iron. Preferrably the 250 watt American Beauty that we used on the frame. Your second mistake was that you really only needed one base cord, since it only uses 4 wires. The reason your setup didn't work is because, even though it passed resistance tests, the wires are too small. Go down to the hardware or big box store and get some doorbell wire. Two runs of that ought to work just fine and get some tape so you can tape the 2 one pair cables together, or look for 4 wire doorbell cable...if they make it.

Now, since you haven't troll^h^h^h^h posted in a few months, I'm guessing that you figured out your error and everything is working fine.

Carl "this is message 34 in the thread" Navarro

"James Thompson" hath wroth:

The cazapitance per foot is about the same for each pair (I just measured it on a 500ft roll) at about 30pf per ft. For CAT6, each pair is solvent welded together at a constant distance making it constant capacitance. The relatively loose twist rate does not contribute any change in cazapitance.

The purpose of the different twist rates is to reduce coupling between pairs. The basic specification is about 11dB NEXT or Near End Cross Talk. The twisting drastically reduces the external coupling from each pair. If the pairs were all wound at the same rate, they would have a much larger number of points of contact between adjacent pairs. It's these points of contact that cause the most coupling between pairs due to simple capacitive coupling at the point of contact. Reduce the number of points of contact and the coupling goes down.

By using non-twisted pair telco wire instead of CAT5, the worst case cable is created. It has the maximum points of contact (the entire length), the worst case NEXT as it makes a great distributed transformer, and the worst loss because the non-twisted pairs will radiate somewhat more.

I've actually seen commercial cables with such wiring. It's legitimately CAT3 cable (which usually has a silver colored jacket) and has 4 unpaired wires in the jacket. Some vendors (Asante) used to bundle those with their MacIntosh ethernet (AAUI) adapters. I found an entire skool computer lab connected with the junk. It works so-so with 10baseT-HDX, but fails miserably with FDX (full duplex) or

100baseTX. I tossed something like 50 cables in the trash and replaced them with home made CAT5 cables. Unfortunately, someone fished them out of the trash and they found their way back all over the skool. Sigh.

I've also used 25 pair telco wire for ethernet. That works just fine for 10baseT-HDX, but also fails on the better or faster protocols. Because it's twisted pairs, it doesn't leak too badly. Like the real CAT5, the various pairs are twisted at different rates to reduce crosstalk. I've never bothered too measure the impedance or NEXT as I'm sure it varies and isn't even close to the CAT5 100 ohms. However, it works well enough for 50ft or less runs.

A single piece of four conductor cat3 phone wire can be used to make a 10baseT ethernet connection. Pins 1, 2, 3, and 6 on the RJ45 connector are used. I use this setup to connect to an old computer in another part of the house. Below are some links showing a four wire setup.

Did you ever look at it? The different TPI is pretty visable on any cat5 I've installed/used over the years.

Lets just cut some open I have here and count?

Lets see, I cut off a 10" long chunk and in my sample I have..

Blue Pair = 1.0 TPI Orange Pair = 1.3 TPI Green Pair = 1.6 TPI Brown Pair = 0.7 TPI

7 twists in 10" vs. 16 twists in 10" is pretty visable to me.

Jamie hath wroth:

Summary of EIA-568 at (because the EIA wants $750 for the printed or CDROM version of the standard):

Pair Assembly The pair twists of any pair shall not be exactly the same as any other pair. The pair twist lengths shall be selected by the manufacturer to assure compliance with the crosstalk requirements of this standard.

Every CAT5 cable that I've worked with has had different twist rates on each of the 4 pairs. ANSI/TIA/EIA-568 does NOT specify the twist rate, but simply leaves it to the manufactory to optimize. If you can meet it with identical twist rates, I'll be suitably impressed.

Hmm, that's strange, our factory made Cat5 for years and still does on a lower volume. last time i checked, all pairs are twisted the same. and its the differential TX and Rx that helps with the reduction of noise etc..

Ok, I always figured it was a capacitive coupling problem, and is why they twist them at different rates. Kinda like having tuned l/c tanks so the signals would not crosstalk so bad. But what you said about contact points, make a lot of sense too. Always open to new info. JTT