Min / Max Models

This email topic:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Scott McMorrow raised the issue that modern connectors have tolerances and
that minimum and maximum models are needed along with typical models. He
also presented some ideas on syntax. Arpad suggested multiple connector
models."
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

I think the multiple model approach is probably best suited for the
specification.

The specification was made to accept matrices that can be used interchangeably.
 As such, to add min / max matrices would be easy to do in the specification
today.

However, I do have a couple issues with maybe some of the basic premise of using
min/max matrices. Please keep in mind, I do not understand the scope of where
the min / max parameters are proposed to be used.

* If you were to use the specification to model a connector with different
length pins (or a package where the leadframe pins to the corners are longer
than the others), I would contend that there should be a model that represents
the entire set of major discontinuities (and use the swath operator where it is
applicable). Each discontinuity would be modeled as a coupled representation to
all other near by pins. I do not think in this case that a min/max should be
used.

* Connector tolerances.... hmmmm... I am not sure what tolerances are of
concern... but I would like to make a couple general comments.

Impedance tolerance.... this is effectively handled by cascaded matrix. Each
matrix defines the impedance (LCR) of a portion of the length of the entire
connector.... including the approximate respect with regards to time.
Impedance is probably better handle in this way as opposed to min/ma CL values

Capacitance tolerance and Inductance Tolerance:
Many people model a connector as a single lumped circuit (i.e. a SLM) ... A
single lumped model could very well suffer a severe "overburdening" effect with
this parameter due to the distributed nature of an interconnection. Cascaded
would be less so...

And maybe a question?
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
* What is min/max? 3 sigma... if you mention this, you by definition build in
the tolerance of the measurement process.. Is it worst case based on production
tolerances... Is it worst case based on as manufactured tolerances... What is
the temperature / humidity range? What about contact resistance and how it may
or may not vary under different shock loads.... How do we make sure that it is
consistent between vendors. This could open up a big can of worms.

Also, for what it is worth, I know that there are connectors with mechanical
dimensions on the prints that are "Critical To Function" (or CTF) that ONLY are
there to maintain electrical design goals and tolerances.

_gus: 630-969-4617
Received on Wed Nov 24 05:36:58 1999