🚀 go-pugleaf

John has given (IMO) some good advice here - take it..... but I must make a
couple of comments.

Jon Plews (jonplews@dial.pipex.com) wrote:

: Geoff Blake wrote in message ...
: >Robin Faichney (robin@nospam.demon.co.uk) wrote:
: >: I've only networked PC's in the same room, using readymade patch cables,
: >: but now I have upstairs and downstairs flats to contend with.  I've heard
: >: that making connections is a big deal.  Can someone reassure me that if
: >: I buy some wall sockets and some cat 5 cable, with ordinary DIY skills,
: >: I should be able to put a reliable network together?  Is there a special
: >: tool required?
: >
: >Provided you get the connections right :-) I cabled up this house and work,
: >so it must be easy - however I am colour-blind so I had my daughter make the
: >connections :-)
: >
: >Seriously, the important thing is to use the right pairs in the right places
: >and to keep the twist going to about 0.5 inch of the IDC (insulation
: >displacement connectors) used in the wall plates.
: >

: Half an inch is alot; you should try to keep the pairs twisted
: right up to the termination.

I was referring to the about 0.75" spacing between the rows of IDC's -
however, these are probably the "telcom pairs".

: The odds are that you won't notice any difference, even with a
: fair length untwisted, but it doesn't hurt to be as anal as
: possible about it.

From what I've found, the (lack of) twist is significant @ 100MHz. For the
price difference, I'd stick to 10MHz hardware but cable to 100MHz, i.e. as
John said continue the twist as close to the IDC as possible.

: To illustrate the point I had a monitor power cord resting on
: an RJ45 plug.  The cord was therefore very close to some untwisting
: of the pairs.  Whenever I turned the monitor on, the hub would show
: a collision occuring.  If I moved the power cord a few centimetres
: away (but still in contact with the patch cable) there was no glitch
: when the monitor was powered up.

Could this not be due to the induced voltage caused by the de-gaussing field
saturating the hub or nic receivers (probably common mode voltage)?

: The sockets were some of the first ones I'd ever done, so they were
: probably not the best made connections.

: >You will need an insertion tool, a proper one costs about 25 GBP, but a
: >cheapie provided/available with DIY telephone wiring kits will serve.
: >

: I disagree about the cheap plastic tools.  Unless you're doing a
: single socket (literally 8 connections only) I'd say forget using the
: plastic tool that you get in kits -- I found them to be absolutely
: crap.

YMMV, but I have found them satisfactory in an emergency.

: I use a punchdown tool from Maplins (BP42V @ �15) and it does both
: type Krone and 110 blocks reliably.

: I've used a "proper" Krone bladed punchdown tool and I really didn't
: get on with it, YMMV.

:-)  I use a couple of RS tools which I believe are badge-engineered Krone
tools - same as the network wiring guys use! 

: One tip I have is to strip the sheathing well back.  I made things
: alot harder on my first attempts because I kept the cables too tight.

: I strip the sheath, tie the cable down and then punch the wires
: down.  The longer the unsheathed pairs are the easier it is to
: punch them down with minimal untwisting.  Remember that it doesn't
: matter in the least that one pair is a few inches longer than
: another.

Spot on!

: Do use the right type of cable as well.  Although you can get
: punchdown blocks for stranded cable and RJ45 plugs for solid
: core cable, they are not common (RJ45 plugs for solid core
: cables seem to be more available than sockets for stranded
: cable though).

: I've got away with using solid core cables and normal RJ45
: plugs for stranded cables in the past, but punching down
: stranded cable into solid core IDCs is completely unreliable.

For the effort in (summoning the resources, namely one 18 year old daughter,
for) putting on RG45 plugs, I buy the drop cables, they are hardly cost more
than a pair of plugs.... 

: Another good idea is to test *all* of the pairs, not just those
: used for ethernet.  You never know when you might need those
: extra pairs for another ethernet node or a telephone line.

Now there's a thought :-)


Geoff
-- 
    Geoff  Blake        geoff (at) palaemon . co . uk         linux 2.0.36
    Chelmsford           g8gnz @ g8gnz . ampr . org           sparc - i586
    Please, only use the  .ampr.org address if you know what you are doing
      Intel create faster processors - Microsoft create slower processes