[802.3ae_Serial] Ideas for fixing stressed receiver sensitivity test
Ideas for fixing stressed receiver sensitivity test, from a discussion of
the serial ad hoc, Wednesday 14 November, led by Tom Lindsay. Notes by Bob
Zona. With thanks to all participants for such a constructive approach.
1) Measure crossing height instead of DCD (correlation of crossing height
and DCD may depend on rise/fall time)
2) Is attenuation ~= ISI?
3) Total DJ = DCD + SJ
4) sigma is less harmful than W
5) Separate jittered eye (with fast edges) and vertically closed eye tests
6) Use bandwidth test with lower and higher limits for receiver and
eliminate ISI component of stressed sensitivity test
7) Vertical BERT scan
8) Use swept SJ to test receiver jitter tolerance. Can we develop supporting
evidence that this is sufficient? How do we correlate this with our
transmitter specs?
9) Keep looking for ideas on calibrating equipment for jitter bathtub and
current stressed eye tests
10) Use scope to allocate jitter in stressed test setup during calibration
11) Build a quantifiable stressed source using only measurable or
calculatable components from Piers's matrix.
And a couple of notes from Piers: the green stressed eye shown in the draft
standard illustrates DCD without significant amounts of other DJ. It was
suggested that a scope shot of a more realistic stressed eye could be used
in the standard. Tom and Jonathan had slightly different definitions of
DCD. Maybe we need one of those dictionary lists of what our words mean.
Below the line, I have given the matrix in tab-delimited form which (After
removing the line breaks caused by the reflector) you can paste into a
spreadsheet. There are six columns. Also I have attached it as an Excel 4
file. Let's see if it passes the reflector!
Piers
-------------------------------------------------------
Eye impairment Current value Useful? Do-able? Measurable
(quantitatively)? Quantified source?
SJ LF <~10 MHz Yes? Yes Yes Yes
SJ HF >~10 GHz 0.05 UI for W Try relative (noise stressed jitter
tolerance) measure like ITU-T? Yes Yes?
sigma 0.015 UI Yes White noise Yes? Find a (upper?) bound
ISI (VECP) e.g. 2.2 (-L) Redundant with OMA? If no DCD? Yes
Accuracy and pattern? Not yet
DCD 6 ps How much? PP thinks 6 ps is too much Yes? Look at
crossing height instead?
OMA Yes Yes Yes Yes
Other DJ 0.25 UI Yes No!
"W" effective DJ Yes By SJ (not all) and other DJ No
-------------------------------------------------------
Stressed_eye.xls