[Tfug] Cabling

[Bexley Hall]

On 11/30/2013 1:21 PM, Bender wrote:
>
>> Most of the network nodes in my office are arranged on
>> or *under* a series of "tables"/benches.
>>
>> To keep wires off the floor, all my outlet strips are
>> mounted to the underside of the tables. Makes it a bit
>> easier to get at individual power cords as well as
>> plugging in additional ones. (if I was more "anal"
>> about it, I would label the cords but...)
>>
>> Similarly, I fasten my network cables to the underside of
>> the tables -- along the back edge. Keeps them off the floor
>> *and* out of the way. When a cable comes to the device
>> for which it is intended, it simply peels out of the
>> bundle and falls/rises to connect to the device.
>>
>> OK. Prefab cables come in standard lengths. Using them
>> means I end up with lots of extra slop somewhere. E.g., if
>> I need 8 ft of cable to go from switch to node X, a 7 ft
>> cable is too short and a 10 ft cable brings 2 ft of slop
>> with it.
>
> If ridding the process of slop is worth it for data cables, do you also
> make custom power cables? I'm serious.
> If not, what do you do about the slop on power cabling?

For the most part, there is no "slop" in the power cords.
I have 1', 3' and 6' cords.  A box sitting under the table
typically uses a 3' cord to run power *up* to the underside
of the table top above it.  The same is true of most of the
things on the top of the table (power strips are located
"down the middle" -- front to back -- of the tables).

Bricks that plug into the UPS's sit on the floor beside the
UPS and can use short 1' cords.  The only things that have
long cords (6') are things like my scanners that sit "off in
the corner" of the table (too far for a 3' cord to reach
the power strips.

I.e., none of the devices *on* the table have their service
loops extending anywhere near the floor -- they come down under
the table and then over to the outlets.  Likewise, devices on
the floor have cords rising straight up to power (I had to
move my printers to the top of another table because the
power cords coming up *over* the back of the printers would
interfere with the paper path for the duplexer, etc.)

I am *keenly* aware of all the wires and potential for "slop"!
Remember, in addition to power and network, you also have
all the cables for other peripherals -- SCSI, USB, FW, serial
ports, audio, wall warts/bricks, etc.

>> [BFD! OTOH, if you have a dozen or more nodes in about that
>> many linear feet, all that slop quickly becomes a tangle!]
>>
>> Fine. I can build cables "to length" for each specific node.
>> (lousy waste of time but the clutter that gets eliminated is
>> well worth it!).
>>
>> Aside from the obvious (correct connectors, cable, hoods, etc.)
>> are there any other concerns I should have about this approach?
>
> Don,
> I'll bet you could teach me something.

Yeah, collect stamps -- it requires far less space!  :-/

> What may be obvious to you may not be obvious to others and vice-versa.
> Not all cables are created equal.
>
> What kind of cable and connectors?

I'm using CAT5e riser cable.  The connectors come in a variety of
options:  flat vs. round cable, solid vs. stranded wire.

> Are data sheets available for the connectors?

Catalog pages in the manufacturers' catalogs.  They are largely
"commodity parts".

> You want to know the outer jacket diameter is to spec for the connectors.

The connectors are deformed by the crimping tool to grab the
cable.  Note that there are some variations in cable O.D.
but the conductors *within* are pretty much standardized.  The
jacket variation is often due to "slop" inside the jacket.
Only the thickness of the jacket material affects the crimp
(any "slop" merely causes the excess jacket to bulge out
the *sides* of the cable entry into the connector shell).

> (Otherwise the jacket crimp could get loose.
> Then the cable stays attached with wires holding it.)

You can't tug on the cables, period.  The jacket doesn't form
a good enough strain relief (the jacket has slop in it and
some amount of stretch so tugging can be transferred directly
to the conductors).

And, it's just "bad manners"  :>

> Use Stranded not solid wire presumably?

Yes.  Solid would be less "flexible".

> What material is the outer jacket made of?

PVC

> e.g.
> Would you use connectors on plenum rated cable?

I don't have any applications that need to be rated for plenum
cable.  Even using "riser" cable is overkill but I've got some
left over from the house wiring.

> What kind of crimper do you use?

I bought mine many years ago, made by Amp? IIRC.  Over $100 back
then.  I used to have one for 6P6C connectors, too.  But, that
seems to have grown legs... :<

You can buy inexpensive crimpers.  But, it seems like as
cost goes down, hand strength must go *up*!  (seems the
inexpensive models lack much mechanical leverage and you
*really* want to get the crimp made in ONE shot)

> Do you test the cables after assembly? What kind of test?

I have a cable tester that verifies continuity, checks for
swapped conductors, pings hosts, etc.  It catches the
mistakes I am likely to make (i.e., two adjacent wires
shifting positions as I feed the cable end into the connector
shell).

In theory, I could also use it to tell me which cable is
which -- but, as they are all plugging into the same switch,
who cares!?

> Do you care about what colors go where?

Of course!  :>  Not only do you have to make sure the "ends
line up", but you have to make sure each specific signal is
twisted with its particular mate.  (different pairs actually
have different twist ratios!)

In my case, Wh/Or, Or, Wh/Gr, Bl, Wh/Bl, Gr, W/Br, Br
(I think this is the 568B scheme?)  The PITA is the green
pair "split".

> If so,
> To what standard are the cables assembled to?

They are wired as above.  As to "standards", IiWfD (If it Works
for Don).

> The only thing you couldn't practically do at home is assemble cables
> with injection molded boots / strain relief / hoods.

You can purchase hoods -- you just have to remember to slide them on
the cable before you crimp the second connector on!  <:-(

But, they don't really stay put.  You probably need to dab some
adhesive to hold the hood on the cable jacket lest it slip back
away from the connector shell (exposing the damn little plastic
locking tab that wants to snap off always!)

> However, it might be possible to DIY pump some thermoplastic, (say hot
> glue gun glue) into the void in the connector. Using a fixture made of
> silicone you could also fashion a strain relief perhaps.

In my case, the cables don't see much "action".  It's a PITA for me
to move a piece of equipment (my body is becoming less flexible
with age so crawling around *under* the tables is much less appealing).
So, boxes tend to stay put.  And, no real need to unplug stuff from
switch (unless switch dies).

I have one "spare" port on the switch that I've run out to a
dangling cable so I can plug <something> into the network for
testing.  I had originally planned on putting an "outlet"
on the underside of the table for this purpose but that seemed
redundant (just plug a cable into the switch!).

> Wonder if urethane is also a possibility, but the cure time might be a
> pain.
>
> People in the IT/Networking business probably don't have the ability or
> desire to build good custom cables, IMO.

Making cables is a royal PITA.   *Any* kind of cable!  And, after
a few, you get bored out of your mind and start making stupid
mistakes.  Plug in tester... CRAP!  Snip off connector.  Start over.

Thankfully, network cables have "one" pinout.  Working with
things like EIA232 serial ports (*real* ones, not the silly
IBM PC version) you end up with a bunch of different pinouts
all of which are mutually incompatible.  So, you spend a
fair bit of time sorting out what you *need*... then *making*
it... then figuring out how to *label* the thing so you
know what it is, later!

[I've worked around this by standardizing on 25 pin M-F cables.
Plug any number of them together to lengthen the overall cable.
Then, *adapt* one of the ends (based on the gender you need)
to the OVERALL pinout you need.  The "adapters" are just small
"widgets" -- basically, two connectors in a "shell" that have
been wired as a "mini cable".  This allows the functionality
to be labeled on the shell.  And, hanging a "25 pin extension
cord" off the end -- EITHER end -- just makes that mini cable
into a *long* cable!  :>   I can't tell you how much work that
has saved me over the years!  And, I've effectively got dozens
of (mini!) cables that fit in a shoebox!]

> Besides, they're busy configuring, running or setting up & tearing down
> the damn boxes all day. Why would they want to have concern about
> possibility for problems with home made cable?

Exactly.  And, it doesn't *cost* them (personally) anything to
buy a cable (cheaper than paying them to make one!).

But, more importantly, they have the *room* for "slop" -- in the
equipment racks, raised floor, etc.  And, actually *need* big
service loops so they can move equipment (e.g., on drawer slides)
while connected.

And, probably dick with their cabling more than I would (because
its more accessible, generally).

[I can't move anything without uncabling.  Even if I left a
service loop for power and/or network, there are so many
other peripherals that tie to the machines that don't or
can't have "service loops"]