ethernet "stub"

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

Threaded View
I have an application that requires two RJ45 jacks on the same ethernet
tr/rx pairs coming from a hub. At any one time only one RJ45 will be
connected to a set of magnetics (and phy and mac etc), while the other will
be unused with no connections. When I use the outermost connector (the one
most further from my hub), I don't expect any problems. If I use the other
connector, the one in between the hub and the second RJ45, will I be
violating any ethernet spec?

 Will the "stub" - the portion of the rx/tx pairs that is left hanging -
cause any problems like reflections as it is not terminated?

 TIA.

bta3








Re: ethernet "stub"
Quoted text here. Click to load it

I wouldn't do it.  I don't have the specs, but
I have a strong feeling that the stub would not
be acceptable.


Re: ethernet "stub"
Quoted text here. Click to load it
will

Definately. The open end produces 100% reflection of the signal, which will
distort the signal in the 'main' line.

Meindert



Re: ethernet "stub"
Quoted text here. Click to load it
one
other
will

But it may still work.
Is this going to be shipped to a customer, or just for in-house 'playing'?



Re: ethernet "stub"

Quoted text here. Click to load it
  I should have explained this better. The application consists of a board
that does a few things (motor control) and plug into a custom backplane. The
backplane connector  has an ethernet (100BT) connection and is connected to
a hub that is not accessable from outside. In certain configurations, the
board is not present but a "simulator" runs on a PC which needs to be
connected to the same ethernet pairs. The "stub" in question is
approximately 1 inch of copper traces.

              |<----  1" ----->|
_____________________
                                                |
     |  x  x  x   ..  x |                     |
     |  x  x  x       x |      ____      |
     |  x  x  x   ..  x | ----|          |   |
     |  x  x  x   ..  x |      |____|    |
     |  x  x  x   ..  x |                    |
                                               |
     connector            rj45      |
                                               |
____________________|  ---  Backplane PCB

    This is an idea retrofit a present design.

Thanks for all the responses.

bta3



Re: ethernet "stub"

Quoted text here. Click to load it

I don't think it makes any sense to do that job by means of an
additional RJ45 jack.  A dummy board to go into that empty slot and
connect it to another RJ45 jack would be a more common and less
worrysome solution.

[Note: your drawing was completely garbled because you forgot to use a
fixed-width font to design it... ;-(]

--
Hans-Bernhard Broeker ( snipped-for-privacy@physik.rwth-aachen.de)
Even if all the snow were burnt, ashes would remain.

Re: ethernet "stub"
Quoted text here. Click to load it

A proper new messages is 7 bit ascii, no font info. So the font setting of
the poster is not important. The font setting of _your_ newsreader should be
set to fixed width if you want to see the drawing right.

Meindert



Re: ethernet "stub"
Quoted text here. Click to load it
Mine is set to fixed width. Still the drawing does not right to me.

Regards.

Elder.

Re: ethernet "stub"
Quoted text here. Click to load it

I meant "does not look right".

Re: ethernet "stub"
Quoted text here. Click to load it
of
should be

O shoot, now I see it. I started type before thinking......:-(

Meindert



Re: ethernet "stub"
You might check the ham forum on usenet. Considering the wave length of
100 mhz is about 10 ft., a 1 inch stub would appear of minimal concern.

Hul

:
:>
:> But it may still work.
:> Is this going to be shipped to a customer, or just for in-house 'playing'?
:>
:>
:  I should have explained this better. The application consists of a board
: that does a few things (motor control) and plug into a custom backplane. The
: backplane connector  has an ethernet (100BT) connection and is connected to
: a hub that is not accessable from outside. In certain configurations, the
: board is not present but a "simulator" runs on a PC which needs to be
: connected to the same ethernet pairs. The "stub" in question is
: approximately 1 inch of copper traces.
:
:              |<----  1" ----->|
: _____________________
:                                                |
:     |  x  x  x   ..  x |                     |
:     |  x  x  x       x |      ____      |
:     |  x  x  x   ..  x | ----|          |   |
:     |  x  x  x   ..  x |      |____|    |
:     |  x  x  x   ..  x |                    |
:                                               |
:     connector            rj45      |
:                                               |
: ____________________|  ---  Backplane PCB
:
:    This is an idea retrofit a present design.
:
: Thanks for all the responses.
:
: bta3
:
:

--
- for email, put the word "keep" in subject line -

Re: ethernet "stub"
Quoted text here. Click to load it

That requirement is really quite a sick one --- how did it come to
that?  This reeks strongly of some PHB having been involved.

Quoted text here. Click to load it

What makes you so sure?  You'll still have some kind of dangling stub
cable, then --- be it just the pins of the RJ45 plug itself.

Quoted text here. Click to load it

You've violated those the moment you mounted the second RJ45 plug
anywhere else but at the end of the cable.  You don't really believe
they make all those prescriptions (like about how long the twisted
pairs may be un-twisted when mounting the plug) just for the fun of
it, do you?

Quoted text here. Click to load it

It will certainly cause reflections, but whether that causes problems
or not is a question of the parameters involved --- particularly the
length of the dangling stub.  As soon as its length comes anywhere
near a bit-time's worth of cable length, you're in serious trouble.
For 100baseT, that's roughly a meter.

In other words: it's high time to very seriously re-evaluate that
requirement.

--
Hans-Bernhard Broeker ( snipped-for-privacy@physik.rwth-aachen.de)
Even if all the snow were burnt, ashes would remain.

Site Timeline