Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: XAUI IO specs


The past can come alive.  For the recent discussion of
	 Vdif vs Vdif peak to peak,
Howard Johnson had a wonderfull and crisp definition of these terms for 802.3z when the same subject came up.

His e-mail is reproduced below, as I can not improve upon it.  

Tom Mathey
A reflector lurker

Date: Tue, 4 Mar 97 14:10:30 
To: stds-802-3-hssg@xxxxxxxx
From: Howard Johnson <howiej@xxxxxxxxxx>
Subject: Peak-to-peak Differential Voltage

Regarding the recent flurry of messages about the differential amplitude specification for clause 39 (short copper), I would like to point out that the existing specification may be met with a differential output (that's two signals) each of which has a 550-mV logic swing.   That's right.  Only 550 mV is required.

The picture below depicts a typical circumstance, involving two complementary, wimpy, PECL-like drivers.    (a fixed-width font, like Courier, is required to display this picture)

-(A)----        ----------------------------   (3.75 volts, logic HI)
        \     /               |
         \   /                |
          \ /                 |
           /              (+550 mV peak differential amplitude)
          / \                 |
         /   \                |
        /     \               |
-(B)----       -----------------------------   (3.20 volts, logic LO)

(1) Each signal has a peak-to-peak range of 550 mV  (3.20 to 3.75)
(2) The differential signal (A-B) ranges from +550 mV to -550 mV
(3) The PEAK differential signal is 550 mV
(4) The PEAK-TO-PEAK differential signal is 1100 mV  (this meets the minimum requirement in clause 39).

The term PEAK-TO-PEAK always means the difference between the most positive and most negative readings of a particular signal.  In this case we have:  (+550 mV  - (-550 mV)) = 1100 mV.  To say that you cannot meet the clause 39 specification is to say that your individual outputs cannot each source 550 mV peak-to-peak.  It would be *very* surprising to me to hear that any CMOS vendor could not meet this specification.  In conversations with many individuals this morning I have determined that, as Bob Rumer suspected, many individuals have mis-interpreted the phrase "peak-to-peak differential amplitude".  In preparing for a possible re-wording, I have dug out the following references from other 802.3 clauses.

NOTE: in 802.3 10BASE-T,, the standard defines the PEAK differential amplitude (not PEAK-TO-PEAK).

NOTE: in 802.3 100BASE-T4,, the standard defines the PEAK differential amplitude (not PEAK-TO-PEAK).

NOTE: in 802.3 100BASE-TX, we reference an ANSI standard, ANSI X3.223 (FDDI TP-PMD clause, which defines  the PEAK differential amplitude (not PEAK-TO-PEAK).

NOTE: in 802.3 100BASE-T2,, the standard defines the PEAK differential amplitude (not PEAK-TO-PEAK).

Perhaps, since we are writing an 802.3 standard (not a Fibre Channel document), we should convert table 39-1 to use units of PEAK differential amplitude, instead of PEAK-TO-PEAK.  Relavant entries for table 39-1 might then read:
         Peak differential amplitude (MAX) (NOTE a)  1000 mV
         Peak differential amplitude (MIN) (NOTE b)   550 mV

       NOTE a: The magnitude of the differential signal may at no time exceed this value.

       NOTE b: The magnitude of the differential signal between normalized bit positions 0.29 and 0.71 (see figure 39.3) may at no time fall below this value.