19

I got this cable with a homeplug AV 200 device - its labelled cat 5e

enter image description here

But looking at the plug end has me rather dubious about it.

enter image description here
(There's only 4 strands there, not the standard 8 - and missing pins, so I suspect these cables are designed to do so. The wire jacket is also surprisingly loose).

Another angle. This isn't a standard connector, no, someone has went and made a non standard connector specifically for this

enter link description here

I figured the cable was crap so I cut into it - some of you were wondering if they had merely left out the additional pairs - and the answer is no - its basically half a cat 5 cable.

enter image description here

I'm currently using it for some legacy equipment that runs on fast ethernet and the plug it came with and it seems to work. However, I'm curious - what're the downsides of such a cable? What do I need to look out for? In what situations would this cat-5 cable be unsuitable for use in anything other than a ethernet cat o'nine tails?

enter image description here

Improve this question
11
  • 3
    +1 nice question. GigE and beyond is common enough now that the fact only 2 pairs are required for Fast Ethernet is no longer "common" knowledge.
    – Phil
    Commented Jan 14, 2014 at 6:53
  • 1
    @DanielRHicks: Most modern equipment is 1000megabit, and nearly all the cables I've seen are 8 wire. I'm currently running some tests, and am seeing rather dramatically bad results with this specific cable.
    – Journeyman Geek ♦
    Commented Jan 14, 2014 at 11:55
  • 1
    Like I said, few folks have any need for that kind of speed. You're apparently one of the few.
    – Daniel R Hicks
    Commented Jan 14, 2014 at 12:06
  • 1
    The cable seems to be running at 10 mbps from what I can tell. The speed test I used on the newer cables, and took minutes has been running for hours on this.
    – Journeyman Geek ♦
    Commented Jan 14, 2014 at 12:43
  • 2
    Cat5e cable must have 8 strands of copper. The person who terminated this cable was obviously on drugs.
    – pauska
    Commented Jan 14, 2014 at 14:05

5 Answers 5

Reset to default
18

Standard 100 base ethernet only uses two pairs so this works... what is a little worrisome is it appears someone did it on the cheap and saved a couple of cents by omitting the other pairs. That said, 1000 base ethernet may use all four pairs and this cable can't handle that.

Improve this answer
3
  • 1
    To add to this - apparently fast ethernet needs 1 2 3 and 6 connected - which is what we see here. I'm going to have to gank one of these cables for testing on my gig-e gear. I'm curious to whether it'll be detected as a inferior cable and the network will throttle itself down.
    – Journeyman Geek ♦
    Commented Jan 14, 2014 at 6:38
  • I'm working on my own answer on this - you're theoretically correct, but running some speed tests find some interesting findings I'd like to share.
    – Journeyman Geek ♦
    Commented Jan 14, 2014 at 11:53
  • 1
    The extra nasty bit is that 1000BaseT uses only the 2 main pairs for negotiation, so devices can still negotiate gigabit, but it won't pass frames properly. Worst of all, some poor implementations (I'm looking at you CableCo custom firmware) will not drop to 100base when no frames make it across.
    – Laikulo
    Commented Jan 10, 2016 at 20:00
15

I ended up doing some tests myself, and the results were... frankly appalling.

TLDR: Throw out the cable.

Now on to some science.

To address some concerns that were stated in the comments - the initial testing was done on a undamaged cable - I only cut it open after the speed tests. I've repeated the tests with an identical cable from the same kit, and have found the same results. I've also repeated the tests with the 'proper' cat 5 cables with 4 other cables.

I tested this by connecting two systems to the same switch, and swapping out the cable on one of them. I also used the free version of LAN Speed Test to do some 'quick' performance testing. I'm running the test from my laptop to a share on a 3tb 7200 rpm drive on my desktop. I'm testing the cables at this point, not overall bandwidth, under conditions reasonably similar to real-world situations for me.

I was thinking of doing a direct system to system speed test as well, without the bridge, but looking at my results, that would be a clear waste of time.

Let's start off with the most obvious things.

enter image description here

The 4 stranded cable isn't capable of gigabit ethernet (no surprise there). In fact, it can't even do fast ethernet.

The speed test took a while to complete... in fact, I had dinner in the time it was running and the total time was about 2 hours to write and about as long to read.

enter image description here

It's pretty close to plain vanilla 10baseTX speeds by the looks of things.

Less obviously? No IPV6. I have no idea what's with that. I've had enough oddness with legacy gear that I'm not even going to touch on this.

I actually tested 2 random 'to spec' cat5 cables from my spares bin. Both have UL marks, and had similar enough performance that I'm only going to post one set of results. These detect as 'proper' gig-e cables. Both were freebies from routers, so it's not like I'm comparing apples and oranges.

enter image description here

Gigabit Ethernet Sweetness!

And have performance that's pretty reasonable - taking into account various overheads like the test writing to a file share.

enter image description here

Yes, that's about 200 seconds total vs about 4 hours. And Ipv6 works perfectly.

Even with the devices it came with - which could easily reach up to 80-100Mbps, the cable might have fallen short. It's clearly not cat 5 e as labelled, and in my specific use, it's clearly unacceptably slow. If I wasn't planning on documenting this, I'd probably have stopped the test for the other cable sooner.

Improve this answer
4
  • It could also be that your gigabit hardware falls back to 10mbps operation even if 100mbps would be fine. Or that when you cut into the cable to check for the wires you damaged it. You'd need a cable tester to be sure if it meets 100-BaseTX specs... but I agree with your conclusion to junk it; surely not worth the expense of cable testing gear.
    – derobert
    Commented Jan 15, 2014 at 21:24
  • I tested it before cutting into it - I have another one of these somewhere, and I figured I could just use this as hookup cable for other projects.
    – Journeyman Geek ♦
    Commented Jan 15, 2014 at 23:54
  • Any reason why IPv6 would fail? The cables themselves have no business dealing with OSI layer 2 and above; they only operate at 1. In fact, IPv6 should be agnostic of which ethernet technology is used (possibly even any switched-bus/bus should work)
    – nanofarad
    Commented Oct 10, 2014 at 23:25
  • Nope, there's no reason for IPv6 to fail. However at that point, I was more interested in documenting performance than chasing down another quirk.
    – Journeyman Geek ♦
    Commented Oct 10, 2014 at 23:47
4

Most of the time, these cables still include the other 2 pairs in the cable itself, but they don't crimp them. So you can make them real Ethernet cables with the proper tools.

I assume they do this so they can sell "Gigabit" Ethernet cables at a higher price.

Improve this answer
6
  • 2
    This was free with a piece of equipment, and no, the cables are completely missing. I'd assume its a way to cut cost, but I'd think that the cost of buying special, crippled ethernet cables in bulk is more than buying normal uncrippled ethernet cables
    – Journeyman Geek ♦
    Commented Jan 14, 2014 at 1:49
  • 2
    @JourneymanGeek If you manufacture on a large enough scale, then you do save a bit - copper is expensive. The more typical cost-cutting method is to use thinner wires - which presents a problem over any real length. Actually, if they used properly specced wires for the pairs that do exist, I think I'd prefer this method. Of course, a fully to-spec cable would be much better...
    – Bob
    Commented Jan 14, 2014 at 3:15
  • 2
    @Bob Not including the full 4 pairs is violating the Cat5e standard. It's not exactly common to see this.
    – pauska
    Commented Jan 14, 2014 at 14:07
  • These could be cheap-o patch cables for "unified cabling", as a good portion of those carry phone signals...
    – NickW
    Commented Jan 14, 2014 at 14:51
  • 1
    This was bundled in with a network device, specifically a homeplug AV kit, so this was never meant to carry phone signals. The cables were intentionally designed this way, however right down to the plugs, and would have never been used with a phone. Amusingly, a slightly higher end device by the same maker came with some very nice cat 5 e with VERY nicely strain relief and what seem to be shielded connectors.
    – Journeyman Geek ♦
    Commented Jan 15, 2014 at 7:16
3

Cat5 is a standard defining the physical layout and type of conductor not the number of conductors. Things distinctive to cat5 are things like the twist, length, and size of the conductors inside the cable.

Two pairs is all that is needed for 100mbit Ethernet. Your poor performance is probably the result of a bad crimp not that the cable is inherently bad. You can also cause bad performance by doing things like cutting the sheathing on the cable and then changing the physical orientation of the conductors. 1g+ ethernet requires all four pairs.

Cat5e is an improvement to the Cat5 standard which specifies additional testing and improved quality control. A Cat5e cable is not necessarily better than a Cat5 cable.

Improve this answer
1
  • 1
    Well, I didn't assume this was out of spec initially, and certainly not until testing. That said, the standards I am looking at (ANSI/TIA/EUA 568B) explicitly refer to "4 pair 100 ohm UTP category 5e". I do think testing cables for transfer speed, and basic functionality is something I need to do with freebie cables like this in future.
    – Journeyman Geek ♦
    Commented Jan 15, 2014 at 9:37
0

It’s for whatever router/modem you purchased. I have two that came with an older Netgear router. The other was to a TP link range extender.

Improve this answer
0

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .