Jim,
Thank you for the comments. The reason I stressed the full circumferential
shield termination within the connector (usually within a back shell or
termination clamp/ring system) is because that is what is required to meet
MIL-STD-461 (and some DO-160 civilian aircraft) radiated emission requirements
for many aerospace systems. With so many hams having problems with LAN and
digi equipment interference issues due to having their antennas close to (or
even inside) the house/shack, I am quite certain that there are plenty of
instances that actually need RFI suppression to the same degree as the military
systems. This would be a lesser problem if people could place their antennas
tens to hundreds of feet away from their computer and network systems. I can
assure you that failure to maintain a full circumferential, low-Z shield
termination on cables has been the cause of many system test failures. There
is also a real need to make certain that pc card ground planes have low-Z
current paths to chassis; that's part of a good design.
YES, the path for shield grounding MUST be to the chassis of the associated
equipment. In this case, "grounding" can mean "common or chassis return", as
opposed to "earth ground". A lot of ITE equipment operates with a 2-wire line
cord, so there is no "earth" ground to be found. However, when one considers
the loop area of the RFI-conducting path, the primary concept that matters is
total length (and net Z) between the device chassis (which contains the noise
makers) and the cable shield. Having a connected peripheral device in which
the LAN signal does not tie to chassis via a short, direct path is nothing more
than having a probe antenna sticking out at the end of the coax.
The test frequency range for radiated emission MIL-STD-461 (and DO-160) testing
is typically 30 MHz to 1 GHz. In most cases, the most important portion of the
spectrum needing help is 30 MHz to about 300 MHz, and the stated 100 to 200 MHz
certainly falls within that as a subset. If 2m is the only VHF band that the
original poster of this thread operates, then that limited frequency range is
all he needs to consider. However, for those active on 6m, 2m, 1 1/4m, and
70cm, the need can be much greater - depending, of course, upon the placement
of spurs on the LAN. Mix 43 still works well for most of these frequencies,
but the amount of placed 43 mix parts may have to be varied when multiple
frequency bands are involved.
As a side note, I am active on all of the bands listed, but except for my HT,
all antennas are at least 50 to 100 feet from the LAN cables and devices. I
have yet to hear a problem on any of those bands.
One added note: My previous post talked about the special RJ-45 with full
shroud shielding available from a vendor. A big reason for that device needing
to be part of the original design for a system is that it is not small. You
probably won't see these connectors on laptops, pods, or similar devices. I am
also guessing that they are very costly.
73, Dale
WA9ENA
-----Original Message-----
>From: Jim Brown <jim@audiosystemsgroup.com>
>Sent: Mar 1, 2013 6:27 PM
>To: rfi@contesting.com
>Subject: Re: [RFI] RFI suppression cores for 147 MHz
>
>On 3/1/2013 2:09 PM, Dale wrote:
>> 1) It MUST have a full circumferential grounding contact to the chassis of
>> the equipment into which the cable gets connected. Except for direct hard
>> wiring, this does not happen with the conventional RJ-45 jack, "shielded" or
>> not; 2) The net path to chassis at each end MUST be low enough in impedance
>> to effectively route the currents induced into the shield to the chassis.
>> The "rule of thumb" for effectively low Z is 1/8th to 1/20th wavelength at
>> the highest frequency of interest, with shorter path length preferred.
>
>Hi Dale,
>
>A couple of thoughts. First, a circumferential connection is less
>critical if the length of the interruption of the shield and the
>interruption of the twist is VERY short, As I see it, it's a question
>of degree -- the shorter the opening as a fraction of a wavelength, the
>greater the shielding effectiveness. Second, if the shield wanders
>around the circuit board before it gets to the shielding enclosure,
>there's a Pin One Problem that will couple RF both ways to/from the
>shield to the equipment. Third, if there's an effective shield contact
>at one end but not the other, noise from the equipment with the good
>connection may be suppressed pretty well, whereas noise from the other
>end will not be. Fourth, I suspect that some (much?) of the VHF trash
>that escapes from UTP (ordinary CAT5/6/7) may be differential mode. One
>way to figure this out is to use a LOT of #43 cores on both ends and
>observe the resulting noise reduction. Those cores are killing common
>mode, not differential mode. I suspect that both are present.
>
>SO -- IMO, having the shield go to the shielding enclosure by a very
>short path is generally far more critical than whether the connection is
>circumferential.
>
>And this particular discussion is specifically addressing 100-200 MHz,
>which is where #43 is most effective. Above that, another material like
>Fair-Rite #61 and other materials from other companies are the weapon of
>choice.
>
>For those new to the Pin One Problem, there are several tutorials on my
>website that address it in considerable detail. And in that context,
>it's important to realize that there are Pin One - Like Problems
>associated with ANY cable shield, and with the Green wire to the Mains
>Power outlet.
>
>73, Jim K9YC
>_______________________________________________
>RFI mailing list
>RFI@contesting.com
>http://lists.contesting.com/mailman/listinfo/rfi
_______________________________________________
RFI mailing list
RFI@contesting.com
http://lists.contesting.com/mailman/listinfo/rfi
|