[ibis-macro] Open Issues on PAM-4 BIRD

  • From: Walter Katz <wkatz@xxxxxxxxxx>
  • To: "IBIS-ATM" <ibis-macro@xxxxxxxxxxxxx>
  • Date: Tue, 14 Apr 2015 10:05:01 -0400 (EDT)

All,



Based on various comments I received this week, I will rewrite the
PAM4_Mapping section to allow both Format Vale and List, and to allow both
Usage Info and In.



I have not done anything specific about the editorial changes need for
other parts of IBIS to reflect AMI models supporting both NRX and PAM4. I
would like to share what Bob Miller (Avago) wrote:



“Here in Avago-land, we tend to refer to your "mean eye zero crossing" as
the "mean edge transition time", Not sure it's "better", but you may find
it so. This is the start of the symbol time -- "edge" being "edge of the
symbol". The "nominal sample time" is ½ UI away from the mean edge
transition time, nominally centered in the symbol time.”



Expanding on what Bob wrote:



A SerDes waveform is periodically sampled to determine the value of the
waveform between transitions. The time interval between these samples is
the Unit Interval (UI), also referred to as bit_time (the value passed
into the AMI_Init function), and symbol_time. Symbol_time is a more
generic name since a single UI can either represent a bit in NRZ or two
bits in PAM4. clock_times returned by AMI_GetWave are edge transition
times, and are ½ UI before the nominal sample times. For NRZ, the mean
edge transition time is close to the mean zero crossing time. For PAM4,
the zero crossing time is only meaningful for transitions between symbols
0 and 3 and between symbols 1 and 2.



In summary, UI, bit_time and symbol_time are the same and correspond to
the time between the waveform sampled at the receiver latch. Clock_times,
zero crossing time and edge transition time are the same and are defined
as ½ UI before the times that the Rx latch is sampled.



I suspect that this will need non trivial editorial support to modify the
paragraphs in IBIS that I identified in rev 15 of the PAM4 BIRD (also
listed below), however it would be useful for this group to review what I
wrote above first, and then we can proceed with the required editorial
changes to IBIS to support the PAM4 BIRD.


Additional editorial changes required.

Page 177

bit_time

bit_time is the bit time or unit interval (UI) of the current data, e.g.,
100 ps, 200 ps etc. The executable model file may use this information
along with the impulse_matrix to initialize the filter coefficients. The
unit for bit_time is the second. For PAM4 models bit_time shall be the
symbol_time. (red is new)



Page 180

Last two paragraphs assume NRZ. Need to add a paragraph or two explaining
the contents of wave for PAM4.



Page 211

"Rx_Clock_Recovery_Mean" is an AMI parameter of Type either Float or UI,
Format either Value, List, Range, Corner, Increment, or Steps, and Usage
Info which defines a static offset, in seconds or UI, between the
recovered clock and the point half way between the PDF medians of
consecutive eye zero crossings. (red should be removed?)



Page 212

Definition: A static offset between the recovered clock and the
point half way between the PDF medians of consecutive eye zero crossings.
Entries are assumed to be in units of seconds when declared as Type Float.
(red should be removed?)



Should we have a better definition of “eye crossings”



Walter





Walter Katz

<mailto:wkatz@xxxxxxxxxx> wkatz@xxxxxxxxxx

Phone 303.449-2308

Mobile 303.335-6156



Other related posts:

  • » [ibis-macro] Open Issues on PAM-4 BIRD - Walter Katz