[RFI] HomePlug Experience?

Wed Oct 13 10:13:12 PDT 2010

I'd like to add a few delayed comments from the Iowa cornfields.  I do have
wired Cat 5 in my house and no problem with RFI except to an HT set on
145.39 MHz (the local repeater, for which I am trustee) when I get within a
few feet of the LAN cabling (unshielded), computer, or Ethernet terminal.  

The original problem that was cited (and I have lost that posting) had to
do with terminal equipment being installed on the same tower as a stack of
10m beams.  First, if you think about it, with the basic clocks used in 100
Base-T, that was the worst possible band for which to have any antennas
near an Ethernet-equipped terminal.  Next, the close proximity of the
equipment to antennas means that emissions must be severly suppressed, and
while it souinded as if the equipment installers tried, they missed the
boat with handling the shielded Cat5 cable.

I have been involved in aerospace EMI/EMC for more than 10 years, and I can
tell you that RJ-45 connectors do not cut it for meeting either MIL-STD-461
or RTCA DO-160 requirements.  They do meet FCC (obviously), but the FCC
Class B levels are at least 20 to 30 dB higher than the aerospace specs. 
For aerospace, no RJ-45s are used - they are replaced with either mil-type
circular connectors or fully shielded subminature-D connectors.  Both types
MUST be fitted with EMI backshells and in many cases, the shielded twisted
pairs of cable are covered with a second shield of overbraid that is tied
to chassis via the backshells.

Running the cables through conduits on the tower will help IF, and only IF,
the following is done:

1.  Cables to be fully shielded pairs per Cat5 or better

2.  The conduits to be all metal and joined with compression connectors,
NOT set-screw types (even better if the conduit sections are welded
together without connectors).  The conduits must be well-grounded to the
tower at top, bottom, and several points in between.

3.  The equipment on the tower must be in fully enclosed metal housings
that are grounded to the tower.  If they are not so encased now, have the
company that owns them get the gear into weather-proof metal enclosures. 
Make certain that only the RF connector is accessible outside the case and
make certain that NO RJ-45s are mounted on the case.  Do not allow any
cables to enter the case via a grommet.  Entry for signal and power must be
via a conductive-body connector with backshell to terminate the cable
shields.  The mating connector on the equipment must be conductive and
bonded to chassis ground.  

4.  As someone else stated, a shield connected at one end is NOT an
electromagnetic shield.  Be certain that the shields are all grounded at
both ends of the cable run using backshell termiantions.  Do not use
pigtails for shield grounds and do not allow a single RJ-45 to be exposed
on or anywhere near the tower.  The "shielded" RJ-45s actually radiate very
well because they can not provide a low enough terminating impedance for
the shields.  That is a major reason that I did not bother to run shielded
Cat5 in my house.  My antennas are far enough away that noise pickup is not
a problem, and the shielded cables would only have added cost to the LAN. 

These may sound like severe measures.  Consider this: the new generation of
commercial aircraft (Airbus A-380 and A-350, plus the Boeing 787) are all
flying LANs.  Without treating Ethernet cables in the manner described,
those planes could not operate.  The situation with Ethernet equipment on a
tower with 10m band antennas is no less severe.  In my opinion, the
equipment owner needs to either make these types of changes or move to
another tower with VHF or UHF antennas (and hope that none of the clock
spurs and harmonics fall on critical ferquencies).  The emissions level
should drop by 20 dB or more if everything is done correctly.  That may not
take the in-band noise down to S-1, but there should be enough improvement
to render the band usable again.  If not, either the equipment emissions
are way too high, the installation of the cables is still not correct
(shields not properly grounded with LOW impedance), or the cables and
equipment are just too close to the 10m antennas and should be relocated.  

73, Dale
WA9ENA

> [Original Message]
> From: Roger (K8RI) <k8ri at rogerhalstead.com>
> To: <rfi at contesting.com>
> Date: 10/7/2010 11:53:51
> Subject: Re: [RFI] HomePlug Experience?
>
>
>
> On 10/7/2010 1:39 PM, Jim Brown wrote:
> >    On 10/7/2010 3:42 AM, Hare, Ed W1RFI wrote:
> >>> I do know the CAT5e and CAT6 do radiate making the use of 2-meter HTs
> >>>>   problematic within about 75 feet of the cable runs.
> >> That almost sounds like something could be wrong. At ARRL HQ, we have
an extensive Ethernet and radio operation continues pretty much noise free.
> > I don't use wired Ethernet here in CA, but I did in Chicago.
> With the photography and avi work I do the rolling backups have almost 
> made the wired gigabit network a necessity.
> >   2M HTs
> > heard a LOT of noise in close proxmity to the CAT5 wiring in the room
> > that was my combined shack/office,
> This is where our problems differ.  I hear no interference in either 
> shack.  I can even set the HT on the computers.  IF I move it around 
> this one, I can just find a very weak signal with the HT antenna right 
> up against the steel tower case. Out by the tower it was causing 
> problems between two 5W HTs that were 75' to 150' apart.
>
> The CAT6 in in PVC conduit that is buried about 15' to the East of the 
> tower and runs between the garage and shop. The actual route for that 
> CAT6 is about 8' from the switch, down through the floor, across the 
> basement ceiling in PVC, up, through the plate into the garage, up the 
> South wall to the ceiling, across the ceiling to the North side of the 
> garage, down the wall, through the plate, down about 2' with 1' of that 
> below ground, then 37' North underground to the South wall of the shop, 
> up to the plate, through the plate and into 1" grounded EMT, and then up 
> about 3' to the metal junction box. I should add that there is one hard 
> wired phone line in the bundle as well.
> > and I did NOT notice an improvement
> > when replacing low quality stuff with Belden  Mediatwist (a very high
> > quality CAT6/7 cable).
> Likewise mine are custom made CAT6 cables from the switch in the house 
> to the shop.
> OTOH I have made my own with the same results. (I had forgotten about
that)
> There are 3 runs with two terminated at computers and a spare that is 
> not terminated.
> That spare could be a problem, they are color coded so I can try the 
> spare connected and disconnected from the switch.
>
> 73
>
> Roger (K8RI)
>
> >    My noise level there was far too high to hear it
> > in a roof-mounted vertical 50 ft or so from that wiring.  This is
> > broadband noise I'm talking about, not the modulated carriers that
> > clustrer around 14030, 21052, etc., which can be effectively reduced by
> > multi-turn ferrite chokes.
> >
> > 73, Jim K9YC
> > _______________________________________________
> > RFI mailing list
> > RFI at contesting.com
> > http://lists.contesting.com/mailman/listinfo/rfi
> >
> _______________________________________________
> RFI mailing list
> RFI at contesting.com
> http://lists.contesting.com/mailman/listinfo/rfi