home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.electronics
- Path: sparky!uunet!infonode!ingr!b30!catbyte!medin
- From: medin@catbyte.NoSubdomain.NoDomain (Dave Medin)
- Subject: Re: Coax Question
- Message-ID: <1992Jul27.224416.20628@b30.ingr.com>
- Sender: medin@catbyte (Dave Medin)
- Reply-To: medin@catbyte.b30.ingr.com
- Organization: Intergraph Corporation, Huntsville AL
- References: <1992Jul20.140143.19341@yang.earlham.edu> <1992Jul22.052519.901@athena.cs.uga.edu> <1992Jul25.120158.600@cmkrnl.com>
- Date: Mon, 27 Jul 1992 22:44:16 GMT
- Lines: 52
-
- In article <1992Jul25.120158.600@cmkrnl.com>, jeh@cmkrnl.com writes:
- |> In article <1992Jul22.052519.901@athena.cs.uga.edu>,
- |> mcovingt@athena.cs.uga.edu (Michael A. Covington) writes:
- |> > [very nice explanation of "reflections" due to impedance mismatches]
- |> > That's why we terminate Ethernet cables with resistors. It prevents
- |> > reflections.
- |>
- |> well, yes, and this is important when the cable is "long". But there's another
- |> reason. The resistors (or equivalents) must also be there with arbitrarily
- |> short Ethernet cables, since the "carrier" on Ethernet is merely voltage
- |> applied through a resistor. The 25 ohms (net) resistance to ground seen by the
- |> transceiver (50 ohms at each end) forms the other half of a voltage divider. If
- |> two transceivers try to transmit at the same time the voltage on the cable is
- |> higher than would be expected with just one transmitter -- and that's how
- |> collisions are detected.
- |>
- |> Hence, very short (a few meters) Ethernets will work just fine with a 25-ohm
- |> resistor at one end!
-
- Just a few corrections...
-
- Ethernet taps are TRUE current sources (ideally behaved like infinitely
- high-value resistors), not voltage applied through a resistor. Otherwise,
- they would screw up the impedance of the line and cause possible
- reflections. The voltage produced on the line is that developed by the
- current source flowing through the termination resistor alone (or
- the immediate cable impedance--properly terminated coax behaves
- like a resistor of its termination value). You're correct that
- collisions are detected by an abnormal DC voltage on the line (a
- negative voltage with respect to the shield due to how the current
- sources are polarized).
-
- The line is terminated at each end by 50 ohms to preserve the
- characteristic impedance of the line and prevent reflections. The
- net impedance to a signal entering the line is 25 ohms, as the signal
- can flow either way, encountering a 50 ohm load in each direction.
-
- The application notes of AMD's Ethernet transceiver chips are a great
- tutorial for those wishing to learn more about the physical Ethernet
- interface.
-
- --
- --------------------------------------------------------------------
- Dave Medin Phone: (205) 730-3169 (w)
- Intergraph Corp. (205) 837-1174 (h)
- M/S GD3004
- Huntsville, AL 35894 Internet: medin@catbyte.b30.ingr.com
- UUCP: ...uunet!ingr!b30!catbyte!medin
-
- ******* Everywhere You Look (at least around my office) *******
-
- * The opinions expressed here are mine (or those of my machine)
-