Re: deploying patch cords

On 7/2/2025 12:01 PM, Liz Tuddenham wrote:

Don Y <blockedofcourse@foo.invalid> wrote:

The 'sandwich' arrangement will be easiest for a blind person because
they only have to remember one linear order of the ports:
>
ABCDEFG...
abcdefg...

>
I think so.  Cables never cross so you don't have to trace a cable
under/behind another (like in the interleaved approach).
>
The downside is the top row of the top device and bottom row of the bottom
device are missing adjacent mates.
>
A possible solution is to build one of the devices as a "single row"
(instead of the dual row common in switches).  This would require placing
one above/below each row of an opposite device.  And, thus ensures the
top and bottom devices have nearby mates.
>
[This seems to be a good take-away!]
>

There is a risk with the 'sandwich' system that someone could plug an
output to another output in the same row.

>
I think they can be relied upon to "follow" the adjacent patch cords
to know when they have advanced or fallen to another row.
>

Another possibility would be to use a matrix with shorting plugs and no
connecting leads.  If it is imperative that only one-to-one connections
are permitted, the sockets could be break jacks to interrupt the
connection to all the subsequent jacks in that row or column.  (The
disadvantage of break jacks is that there are a lot of series contacts
to go faulty.)

>
I want to allow for the possibility -- in an unusual situation -- for
1 to be broken from 1 and mated to 82, instead.  E.g., if the mated 1
malfunctions.
>
But, this would be a "singleton" connection that could easily stand out
among the more orderly other connections.
>
The sandwich illustrated relies on lots of very short patch cords.
This keeps things dressed nice and tidy.  But, may require care in
their manufacture -- e.g., if you plug in one end and have only
a few inches of service loop to ensure the other end is *oriented*
in the correct direction.
>
It also means you'd need a few "exception handlers" -- longer cords
to deal with the 1-to-82 case alluded to above.

 It sounds as though your problem is very similar to the problems faced
by the designers of manual telephone switchboards a century ago.  I
would look around and see if it was possible to pick up some NOS
switchboards - or at least read up on some of the old journals: Bell
Labs, POEEJ.

I think the biggest takeaway is to design things that are "one row"
of 8P8C's.  This gives you the most flexibility in interconnecting
them without "tangle" or "overlap".
Also, ensure there are no markings that dictate a particular
orientation.  So, a column is just as feasible as a row.
I've already made provisions to facilitate *locating* specific
cables as well as *indicating* ones of interest.
Next, a reexamination as to whether 8P8Cs are the "right" connectors
(just because data centers use them doesn't make them suitable for
use in other applications!).   The telephone "plugboard" would be
nice (no orientation problems) but may be an issue electrically.