ccruss@deneb.UUCP (03/13/87)
There were a couple of questions about voltages on the Ethernet cable and
how collisions were detected. Hopefully this may help.
Ethernet or IEEE 802.3 uses an access method know as Carrier Sense,
Multiple Access with Collision Detection (CSMA/CD). This simply means a
device can talk on the cable as long as no other device is currently
talking on it and if two devices talk at the same time, it is detected,
and they both stop and try again later.
In 802.3 the carrier is a data edge, a transition of 0 to 1 or 1 to 0,
on the cable. This edge is guaranteed to occur within 75 and 125 percent
of a bit period, measured from the center of the last bit. A method know
as Manchester encoding is used on 802.3. In this encoding scheme, the
first half of the data bit is the inverse of the bit value, and the
second half is the true bit value. Manchester encoding is used because
it allows easy recovery of the clock, and creates a datastream that is
equal number of high and low states on the cable.
A logic high is represented on the cable by 2.05 volts. A logic low is
represented by 0 volts. Since the datastream consists of equal numbers
of high and low, the average voltage on the cable when a signal is
present stays right around 1 volt. When no device is using the cable,
the average voltage is 0 volts. When two devices are using the cable at
the same time, there are no longer equal numbers of high and low states
and the average voltage rises above 1 volt. The transceiver will send an
alarm to the device when the average voltage exceeds a certain threshold
(typically 1.5 volts). The alarm indicates a collision to the device and
the device stops sending data. Retransmission delays are in multiples of
512 bit times and range up to 1023 times this unit of delay.
Russell Hobby
Data Communications Manager
U. C. Davis
Computing Services BITNET: RDHOBBY@UCDAVIS
Davis Ca 95616 UUCP: ...!ucbvax!ucdavis!rdhobby
(916) 752-0236 INTERNET: ucdavis!rdhobby@ucbvax.Berkeley.EDU rdt@houxv.UUCP (03/15/87)
Question: How long after the 50% 1's density is violated does it take for the collsion detector to go off?
wam@cdx39.UUCP (Bill MacLeod x7520) (03/19/87)
In article <551@houxv.UUCP>, rdt@houxv.UUCP (R.TRAUBEN) writes: > Question: How long after the 50% 1's density is violated does > it take for the collsion detector to go off? in this article, I answer: beween 9 and 29 bit times. That's up to 2.9 microseconds, depending on how far away and feeble the interfering station is... IEEE 802.3 working group on System Topology could help you here; I understand they even have a spreadsheet for delay budgets... -- Bill MacLeod telephone: 617/364-2000x7520 Email: ...{cthulhu,inmet,harvax,mit-eddie,mot[bos],rclex}!cdx39!wam Smail: Codex Corporation; Mailstop C1-65; 20 Cabot Blvd; Mansfield MA 02048 send two boxtops and one thin dime for standard non-disclosure agreement...