Cat5 cable - can it stand kinks?

[Murmansk]

Two flat in a large Victorian house - I'm planing to take a Cat5 cable from the router on floor 2 down to the flat on the ground floor via the underside of the banisters.

In the lower flat it'll terminate in an RJ45 type socket and there'll be an 8 port switch. From there it'll go round to RJ45 sockets in the flat and I'll have a Sipgate VOIP account with a Cisco ATA wired to the phone sockets.

I'm wondering if this all sounds feasible and particularly if the fact that the Cat5 under the banisters will do quite few tight bends, will this slow the broadband at all? I know it's not a hosepipe but still!!!?

[Graham J]

Google suggests: http://www.lanshack.com/cat5e-tutorial.aspx

Specifically:

Do
Make only gradual bends in the cable where necessary to maintain the
minimum bend radius of 4 times the cable diameter or approximately 1"
radius (about the roundness of a half-dollar).

Do Not
Allow the cable to be sharply bent, twisted, or kinked at any time. This
can cause permanent damage to the geometry of the cable and cause
transmission failures.

Do
Always test every installed segment with a network cable tester. (i.e.
something that looks for reflections and crosstalk, as well as just the
correct connectivity.)


Do Not
Use staples on UTP cable that crimp the cable tightly.


So long as you make smooth bends rather than sharp corners you should be
OK.

--
Graham J

[Graham.]

I once got castigated (ooh er missus) on uk.diy for saying this, but
it didn't change my opinion.
Whatever you may hear to the contrary, kinks and twists in Cat5 will
not *of themselves* make any measurable difference to your data
throughput.

Now please don't misunderstand me, I am not saying it's desirable to
have kinks, ignore the minimum bend radius or squash the cable with
staple-gun staples, because all of those can cause physical damage to
the conductors or their insulation which can and will kill the signal,
but it's not the kinks that cause that per se.

Think of a typical RJ45 socket and plug interface (or indeed a patch
panel)
The cable is connected to the IDC and the pairs are turned through a
right angle the way to the plug. all the way through the socket plug
interface, the cores are parallel instead of being twisted.

Is there a limit to the number of such interfaces in a segment of Cat5
cable? Probably in the specs, but I've never seen anyone worry about
that as much as they worry about the odd kink.

10Base2 coax cable that was popular before UTP was likely less
forgiving about being kinked.


--

Graham.

%Profound_observation%

[Graham J]

Graham. wrote:
> On Mon, 16 Jan 2017 13:02:15 -0800 (PST), Murmansk
> <stai...@gmail.com> wrote:
>
>> Two flat in a large Victorian house - I'm planing to take a Cat5 cable from the router on floor 2 down to the flat on the ground floor via the underside of the banisters.
>>
>> In the lower flat it'll terminate in an RJ45 type socket and there'll be an 8 port switch. From there it'll go round to RJ45 sockets in the flat and I'll have a Sipgate VOIP account with a Cisco ATA wired to the phone sockets.
>>
>> I'm wondering if this all sounds feasible and particularly if the fact that the Cat5 under the banisters will do quite few tight bends, will this slow the broadband at all? I know it's not a hosepipe but still!!!?
>
> I once got castigated (ooh er missus) on uk.diy for saying this, but
> it didn't change my opinion.
> Whatever you may hear to the contrary, kinks and twists in Cat5 will
> not *of themselves* make any measurable difference to your data
> throughput.

What they will do is make a measurable difference to the performance of
the cable, in terms of crosstalk and reflections. A proper network
cable tester will demonstrate this. Usually it will summarise the
performance as a "headroom" figure - typically a few dB. Provided that
there are only a few kinks and twists there will be some headroom.
Clearly a cable nearing the maximum 100 metre length with kinks and
twists all the way along its length may be sufficiently degraded that
the headroom drops to zero.

> Now please don't misunderstand me, I am not saying it's desirable to
> have kinks, ignore the minimum bend radius or squash the cable with
> staple-gun staples, because all of those can cause physical damage to
> the conductors or their insulation which can and will kill the signal,
> but it's not the kinks that cause that per se.
>
> Think of a typical RJ45 socket and plug interface (or indeed a patch
> panel)
> The cable is connected to the IDC and the pairs are turned through a
> right angle the way to the plug. all the way through the socket plug
> interface, the cores are parallel instead of being twisted.

The reason for twisting is to ensure that both wires in the pair follows
as nearly as practical the same path through the environment and that
the distance between the wires remains constant, ensuring that the
characteristic impedance of the pair remains constant along its length
thereby minimisiing reflections. Any noise current induced into one
wire will be induced equally into the other. The data transmitted along
the pair is differential, so the noise currents being common mode will
cancel out. Twisting is simply an easy way to ensure this physical
stability. Where the cores are parallel, as in the plug and socket, the
mechanical construction ensures stability.

Some impedance mismatch and noise susceptibility is acceptable at each
junction, and the standards assume two flexible patch leads and a fixed
cable for a complete link - thus:

Device - patch lead - fixed wire - patch lead - device
........ 5 metres ... 90 metres ... 5 metres .........

Thus there are a maximim of 5 plug/socket disturbances in the 100 metre
total length from one device to the other.

> Is there a limit to the number of such interfaces in a segment of Cat5
> cable? Probably in the specs, but I've never seen anyone worry about
> that as much as they worry about the odd kink.
>
> 10Base2 coax cable that was popular before UTP was likely less
> forgiving about being kinked.
>

--

Graham J

[Graham J]

Graham J wrote:
[snip]

>
> Device - patch lead - fixed wire - patch lead - device
> ........ 5 metres ... 90 metres ... 5 metres .........
>

Correction ... not 5 but ...

> Thus there are a maximim of 4 plug/socket disturbances in the 100 metre

[Murmansk]

Thanks for all the interesting info, gives me plenty to go on

[Graham.]

On Tue, 17 Jan 2017 15:13:50 +0000, Graham J <gra...@invalid.com>
wrote:

That's interesting, thanks.

I think we can agree though, that in a typical domestic installation,
liberties can be taken without any significant consequences.


--

Graham.

%Profound_observation%

[Vir Campestris]

On 17/01/2017 10:05, Graham J wrote:
> Where the cores are parallel, as in the plug and socket, the mechanical
> construction ensures stability.

AIUI a field with a gradient will introduce a different signal in one
side of the socket compared to the other. This doesn't matter when it's
only a few inches, but would on the whole 100M. Which is why we don't
use ribbon cable.

We used to have SCSI ribbon cables - 50 way - and they were twisted
pairs too. Every couple of feet they untwisted so you could put an IDC
connector in - then carried on again twisted.

Andy

[Woody]

Go dig on fleabay and you will find super thin CAT5 which is about 6mm
wide and about 0.5mm thick for not very much money.

I have used it to get my network from one room to another by pushing
the cable down the edge of the carpet and across the (brass finish)
door strip under the top on top of the carpet. Works a treat.



--
Woody

harrogate3 at ntlworld dot com

[Nick]

I wired my house with solid core cat 5e.

After wiring it worked ok at 100Mb/s but not at gigabit. I found some
kinks in the wire and after I straightened them out it worked at
gigabit. So yes, it is sensitive to bends if you want gigabit.