[RFI] Ethernet RFI
jim at audiosystemsgroup.com
Sun Jan 23 16:15:19 EST 2005
On Sun, 23 Jan 2005 13:18:08 -0600, Ed -K0iL wrote:
>But only one of those would place the "twisted pairs" into the
>correct EMC signal configuration.
Not only is correct wiring critical for EMC, it is also quite
critical for performance as an Ethernet cable.
Understand this basic concept. A CAT5 cable (or CATx, for that
matter) consists of FOUR TWISTED PAIRS. EACH TWISTED PAIR IS A
100 OHM TRANSMISSION LINE. (CAPS FOR EMPHASIS). Further, each of
the pairs is assigned a different twist ratio (twists per ft,
called the "lay") to further reduce crosstalk between the pairs.
The standards that define Ethernet systems define which colors are
assigned to which ratios, and to which pins on the connector.
When you connect Ethernet equipment with this cable, you are
connecting four 100 ohm transmission lines between that gear. Not
all of those pairs are used by most equipment, and by all
Now, if you connect the wires to the wrong pins (let's say the
blue/white where the orange/white should go), several things are
wrong. First, it's no longer a twisted pair, it's two wires, so
the noise rejection (and noise radiation) go up. Second, the
impedance is no longer 100 ohms, so the data transmission gets
unreliable, causing errors, retransmissions, and slow speeds.
Third, the crosstalk between pairs goes sky-high, causing more
errors and data problems.
Bottom line -- you cannot test data cables with an ohmeter, you
MUST test them with a specialized data-transmission test set that
measures end to end loss, crosstalk, and related parameters. These
test sets cost big bucks.
Jim Brown K9YC
More information about the RFI