On Thu, 2008-02-21 at 08:40 -0600, Ed Gray W0SD wrote:
> First of all thanks for the help! I have some new ideas on things to
> try. I is very much appreciated.
>
> I am using cat5e cable, not shielded/screened. I have the pairs correct
> using 568A and use EZ-RJ45 connectors and the expensive Platinum tool
> for crimping. The Cat5e cable runs from the router to the computer are
> direct, no wall outlets and punch downs. Every run is an end to end run.
Carefully verify this as any single run out out of 'pair' will be a
serious noise source
http://www.incentre.net/incentre/frame/ethernet.html
>
> I need to do some investigating here as the problem seems to come from
> the computer network card area but I am not sure what you mean on
> grounding as it is a plastic plug but of course there is the ground wire
> in the twisted pairs.
There is no ground wire in cat 5, it consists of 4 balanced lines
(pairs). the presence of a ground wire in the twisted pairs would create
unbalance.
Cat5 is actually a pretty decent RF transmission line and is
characterised out to 100Mhz. There is a bit more than meets the eye
about it's spec.
http://en.wikipedia.org/wiki/Category_5_cable
The socket at the PC or router end is sometimes recessed and does not
make contact at the socket body with the chassis ground, this leads to
common mode problems on the cat5 Run.
this is usually easy to fix by soldering the shell of the socket to the
chassis slot with a very short wire.
> >
> > 3/ are the Ethernet cards in the PC's grounding the RJ 45 socket body
> > correctly at the backplane of the PC, some do and others just have the
> > socket floating which is bad from an EMC perspective, no matter how well
> > the case is screened this creates a 'pin 1 problem '
>
> I don't think it is on this end based on the sniffer approach. The
> router is in a plastic case. I forgot to mention with the DSL modem
> turned off but the router on the birdies are the same as with the DSL
> modem on.
This test is irrelevant, the Birdies are a fundamental part of the
correct operation of cat5, your issue is not that the birdies exist but
that they get 'out' of the system and cause radio issues
The device that they are connected to does not have to be powered up to
provide the unbalance required (or additional 'antenna') to make the
system leaky.
You could try eliminating the router and connecting them one by one
directly to the DSL modem to see if you have any particular culprit on
the leaky stakes
> > 4/ the router end is more likely to be the issue, try an ethenet hub, a
> > decent(ish) one in a metal housing. does the nose go away / reduce
> > significantly?
> > Power up your DSL router, any changes?
> > now connect your ethernet hub to your DSL router with a short patch
> > lead, does the problem come back ? If so choke all wires leaving the DSL
> > router, including the phone and power cords. don't put it's plastic case
> > on top of or close to anything that can act ans an antenna for it
> >
> The noise seems to be coming from the Computer network card area in the
> computer but it only occurs when the CAT5E cable is connected to it with
> the router operating.
Sure, no router connected then no networking, no networking then no
birdie but nearly all the emc issues I have seen with Ethernet were at
the router end.
is the Shell of the RJ45 socket on each computer directly connected to
the chassis of the computer by metal finger stock or similar right at
the socket?
here are some pictures of Ethernet cards (scroll down)
http://tinyurl.com/2ggfll
Note where the network socket meets the back-plate, there needs to be a
good connection here, some cards do not have a good connection here,
it's ok to use (carefully!) a big hot soldering iron and some copper
wire to take care of RF grounding at this point
> > 5/ still got issues? use a nearfield loop (single 2" closed loop on the
> > end of some coax) with something like an FT817 to see what the noise is
> > using as an antenna
>
> I am between a rock and hard spot as I just built and have moved into a
> fantastic new radio room and ran all the cat5E cable from the router
> through the basement, up the inside of the wall, across the ceiling and
> down the wall and under a false floor to the operating bench. Again
> this is an end to end run with an EZ-RJ45 connector on each end. That
> is where you can put all the twisted pairs through the connector and the
> wires all stick out the end. The expensive crimper cuts them off and
> rachet crimps them. Supposedly the advantage is you have each wire
> almost exactly the same length and the twisted part of each pair comes
> as close as possible to where it is crimped. I don't think you can do
> any better than what we did.
But you have to undo the the wist a little to make it 568A standard, you
did do this, didn't you? If you have all colour pairs parallel then the
signal balance is destroyed on 2 of the data lines and it's going to
radiate, a lot.
http://www.practicallynetworked.com/img/568a.jpg
I ma not familar with the EZ-RJ45 system but I have made literally
thousands of rRJ45's up with the standard crimp tools / plugs
the picture shown here with the EZ-RJ45 plug is in the wrong colour
order and would be a serious EMC problem
http://www.iautomate.com/ezrj45.html
568B is the dominant standard here in europe in places I have worked but
they are interchangeable standards providing the same standard is used
end to end on each run.
http://www.incentre.net/incentre/frame/ethernet.html
> From a practical stand point these runs can not be replaced with out
> doing an extreme amount of destruction to the new construction.
>
> As I write this I have come up with an idea. I could move the DSL modem
> into the new radio room along with the router. I could then have very
> short runs of CAT5E cable and if still a problem I could even tried
> shielded CAT5E cable.
I have more cat5 than you can shake a stick at around here, providing
the equipment on either end is maintaining the balance and reasonably
good from an EMC perspective it will not radiate very much at all.
there are less obvious paths, such as the birdies piggybacking on the
USB cables (bad socket grounding again)
or on the Router power cable (crappy router not using proper xformer
balanced inputs or lousy PCB layout in a plastic housing )
the less obvious paths are where the sniffer loop comes in
> >
> > 6/ consider rearranging your network topology to have shorter runs of
> > Cat5 to a central hub (or switch) and replace the long run with fibre, I
> > have seen 100Mb Media converters (ethernet to multimode fibre) for as
> > little as 30 euro each on ebay. A single terminated run of multimode
> > fibre will not cost too much these days in a custom length and will be a
> > bargain in a 'generic' length
> >
> >
> > All of the Ethernet problems I have seen to date are resolvable,
> > domestic DSL routers are often the worst offenders because they don't
> > take care to ensure very accurate line balance on the cat5 pairs,
> > grounding is ineffective from an RF perspective, the phone line is often
> > involved and the power supply provides effectively no RF isolation from
> > the mains, all of these issues can be addressed
>
> Mean while I am going to get a wireless card or USB wireless connection
> and based on my wireless laptop should get rid of the problem while I
> work with the above ideas and see if I can resolve it.
Wireless will make the problem go away but I'll wager this is a
resolvable issue.
73
Brendan EI6IZ
--
Don‘t complain. Nobody will understand. Or care. And certainly don‘t try
to fix the situation yourself. It‘s dangerous. Leave it to a highly
untrained, unqualified, expendable professional.
_______________________________________________
Topband mailing list
Topband@contesting.com
http://lists.contesting.com/mailman/listinfo/topband
|
