IBIS BIRD63.2 Comments/Questions

From: Bob Ross <bob_ross@mentorg.com>
Date: Wed Feb 09 2000 - 16:46:22 PST

To Stephen Peters and IBIS Committee:

I am finally starting to look at BIRD62.2 and BIRD63.2 and have these
comments. My first comments are on BIRD63.2.

Bob Ross
Mentor Graphics

1. In general, I have problems with any slew rate definition. We
have already mis-used it for the [Ramp] keyword where the only
meaningful portion is dt, and we interpret it as the 20% to 80% value.

The BIRD63.2 definition is

  (Tester_Vhigh - Tester_Vlow) / (deltaT for the above tester voltages)

Presumably, these values are unreduced 0% to 100%.

According to SSTL-2, for 2.5 V, EIA/JESD8-9, Sept. 1998, pg. 4,

  Slew = (VIHmin(ac) - VILmax(ac)) / (deltaT for the above (ac) voltages)

where the delta ac voltage could fall inside the delta tester voltage
according to the diagram. This would produce a different unreduced values

                             _______ Tester_Vhigh
                            /
                           /<- VIHmin(ac)
                          /
                         /<- VILmax(ac)
       Tester_Vlow ____/

Does JEDEC have an official definition of slew rate?

Where do you get the Tester_Vhigh and Tester_Vlow values if they
  are not the (ac) values? Are they documented in data sheets?

If Tester_Vhigh and Tester_Vlow are given, then the numerator is
redundant - a source of error.

The 0% to 100% interpretation is inconsistent with the [Ramp]
intepretation and also with a potential connector model interpretation.

2. According to the JEDEC document, the slew rate is a minimum value
(i.e., faster signals are allowed). Should this be noted?

3. Should the Tester_Vhigh and Tester_Vlow values be deltas from
Tester_Vref to be consistent with the BIRD62.2 subparameters?

Tester_Vlow is listed as -1.0 V in the first example.

4. I do not like mixing differential and single-ended definitions
by value. A new subparameter should to be defined if the Vth is to
be interpreted for differential operation.

I presume it is possible for a particular buffer in some technology
to be used both in differential and single-ended mode. Sometimes
simulators can control whether single-ended or differential is used.

The differential example has the Tester_Vlow and Tester_Vhigh
voltages relative to the differential voltage - an inconsistency.
The slew rate voltage does not conform to the slew rate definition
in the differential example. Is this intential? - is it always
a single ended definition?

5. Cosmetic: I would prefer the proposed Sub-Parameter names to be
the following:

  Vth_tester
  Vhigh_tester
  Vlow_tester
  Slew_setup_tester
  Slew_hold_tester

etc.

for the following consistency reason. So far, all of the subparameters
of IBIS which are of the form:

  Xxxx_qqq = numerical value

or

  Yyyy_zzz word

follow the format of uppercase first letter and lower case remaining
letters (e.g., Si_location, R_pkg, Model_type, R_fixture, V_fixture_min,
S_overshoot_high, etc.)

Moving V as the first letter gives the major definition of the
subparameter and preserves this convention.

We still remain case sensitive, so having a consistent convention
is important. This was violated in BIRD63.2 where Tester_Slew_Setup
was defined, but Tester_Slew_setup was used in the examples. So you
see the problem.

6. Under Usage rules, Tester_Skew_* is used instead of "slew".
Received on Wed Feb 9 16:48:02 2000

This archive was generated by hypermail 2.1.8 : Fri Jun 03 2011 - 09:52:30 PDT