Archived from groups: comp.dcom.lans.ethernet (
More info?)
Walter Roberson wrote:
> In article <cbvetg0uvp@news2.newsguy.com>,
> J. Clarke <jclarke@nospam.invalid> wrote:
> :Let me make sure I understand what you're saying here. You say you have
> :a
> :"second punch block that has RJ connectors out the back end". Are you
> :saying that you have cables punched down on the block with male RJ-45
> :connectors on them, or are you saying that you have a punchblock with
> :RJ-45 female connectors on an attached panel, in other words a patch
>
anel?
>
> If I have picked up the terminology correctly, then we currently
> have this chain in our structured wiring system:
>
>
> A) device NIC with RJ-45 female receptacle
> B) cat5 cable with male RJ-45 on each end
> C) labeled walljack with female RJ-45 receptacle
> D) high-quality professionally-installed (AT&T certified) drop cables in
> proper trays etc.
> E) drop cables end at fixed labeled positions on a 110 punch block
> F) patch cable with unknown type name. Each end has 4 flat vertical tines
> that push between the plastic bumps of a 110 punch block. Similar such
> endings but wider (more tines) endings are also available -- e.g. our
> [few] ISDN lines use the wider version.
Most likely a standard 110 to 110 patch cable. See if the cable is
imprinted "CAT5".
> G) Above unknown patch cable is connected to any available labeled
> position on a second 110 punch block.
> H) the same kind of high quality cable as is used for the drop cables
> comes out the back of the second 110 punch block, and runs over a metre
> or two, each cable ending in a male RJ45.
Is that cable punched down on the 110 block? If so then it's trouble
waiting to happen--if it's stranded it won't make reliable contact on the
110, if it's solid then it's eventually going to break--may take a long
time if it's seldom disconnected but eventually it will go. Every time you
bend it even a little it work-hardens a little bit--eventually it
work-hardens to the point that it breaks instead of bending.
> I) These above are neatly bundled with little slack, each just long
> enough to plug into sequential female RJ45 ports on our switches.
> The expectation of the wirers was clearly that it would always be
> the same male RJ45 plugged into any given switch port, with all
> variability handled by which position on the drop 110 block that we
> connect to which position on the switch 110 block.
>
>
> Thus, if we wish to connect datajack D15 to switchport B22, we do so by
> putting one of the those unknown flat-tine connectors between the
> position that D15 is connected to on the drop 110 block, over to the
> position that B22 is connected to on the switch 110 block.
>
>
> We have installed additional drops (that terminate on 110 blocks),
> and we have installed additional switches that we do NOT have a 110 block
> for as yet.
>
> My co-worker's proposal is that we do not install that switch 110 block
> at all: that instead, we buy some new patch cables that have the
> flat tines on one end, and have male RJ45 on the other end, and that then
> when we wish to connect up a datajack, we do so by patching from the
> drop 110 block position ("E", above) directly over to an available switch
> port's female RJ45. That would be a different patch cable than "F",
> and would skip "G" and "H" above. We would also lose the neatness
> property mentioned in "I". At the moment, the cable tangles are
> restricted to those unknown patch cables of point "G" above
> (between the two 110 blocks); if we were to make the change, then
> the tangles would exist between the positions on the drop 110 block ("E")
> and the switchports.
This would from a signal propagation viewpoint be a better solution, also
likely to be a bit more reliable due to using the correct type of stranded
cable with the special connector taking care of making secure contact with
the 110 block.
> :From a signalling point of view the fewer discontinuities (not the right
> :word--for some reason I've gone blank on the right one--hate it when that
> :happens) in the cable you have the better--each one is a potential
> :failure
>
oint and each one degrades the signal a little. If you're going from
> :device to jack to punchblock to punchblock to switch that's 8
> :discontinuities. If you go from device to jack to punchblock to switch
> :then that's 7, which is slightly better.
>
> Good point.
>
> I think perhaps the two points that I have been most concerned about are:
>
> 1) additional mechanical cable strain at the switch, considering that
> we wouldn't be able to neatly bundle the new-type cables; and
Wouldn't worry about it. Switches (at least rack-mountable ones--don't know
about the desktop cheapies) are designed to be used with patch cables
plugged into patch panels, which are typically rat's nests no matter how
neat you try to make them.
> 2) whether the signal quality would be affected. The current cables
> that plug into the switch are 8-wire Cat5. I don't know, though,
> whether the patch cables that currently go between the 110 blocks
> are technically Cat5 -- they are not very thick, and do not give me
> the -impression- of being 8-wire cables [but they might be internally.]
> If they are not really Cat5, then it would seem to me that they should
> be kept as short as practical.
A cross-connect cable doesn't have to be 4 pair. At least not in the spec
version that I have.
> But then, I realize now that I don't know how they function at all; the
> connectors that go onto the punch blocks are 4 tine, not 8 tine, which
> would imply that there cannot be end-to-end signalling involving all 8
> wires of Cat5. 4 grounds, I guess? But if so, then would a cable with
> the 4-tine connector on one end be able to provide proper Cat5 grounding?
> Could we count on the switchport RJ45 end to provide any necessary
> grounding?
100TX only needs 2 pairs. Each pair has a signal conductor and a ground.
If you go to gigabit then you'll need all 4 pairs. As long as you've got
the same pairing from one end to the other you've got no problems there.
> As you can probably tell, I am inexperienced about physical-layer
> issues. Not quite so inexperienced that I've never heard of NEXT,
> but I haven't wired anything more than RS232.
Bottom line on this is that aside from the issue of orderly appearance, if
the cable run from E to the switch is less than 5 meters, then to my way of
thinking your coworker's proposal would be the way to go. If it's more
than 5 meters then the "right" way to do it would be to go from E to a
patch panel and then use ordinary CAT5 patch cables from there to the
switch. Alternatively, you could go to a 110 block and use the type of
cables your coworker suggests.
In an ideal world, E would be a patch panel at the switch location instead
of a 110 block.
--
--John
Reply to jclarke at ae tee tee global dot net
(was jclarke at eye bee em dot net)