[SI-LIST] : Catching the Corners

About this list Date view Thread view Subject view Author view

From: Abe Riazi ([email protected])
Date: Sat Mar 18 2000 - 12:28:07 PST


Dear Scholars:

  There are numerous PCB and modeling parameters frequently utilized as
inputs to a SI simulator program. Associated with such parameters are
tolerances representative of fabrication processing variations. The
manner by which these PCB and modeling elements are varied and combined
constitutes an important topic.

  To verify a high speed design under all manufacturing situations, it
is considered a necessity to perform as a minimum simulation of the
following two worst case parametric corners:

Fast corner: a combination of "fast" PCB with "fast" drivers/receivers.
Slow corner: composed of "slow" PCB with "slow" drivers/receivers.

  Furthermore, simulation of the typical corner may be carried out for
purpose of correlation with physical measurements.
  The fast, typical, and slow corners are also referred to as minimum,
nominal and slow, or strong, typical and weak respectively.

 The PCB parameters which are varied for various simulation corners
include:
  Substrate dielectric constant and thickness, trace widths, thickness
of plane and signal layers.

  The modeling parameters which are changed consist of:
  Pin and package parasitics, I-V and V-T curves.

 For fast corner timing, it is recommended to set the substrate
dielectric constant to the lower value, trace widths to smaller
tolerance, the insulator thickness to their upper tolerances. GND plane
thickness and signal thickness to the upper tolerance.

 For slowest possible configuration, the parameter setting should be
reverse of the fast corner settings. That is the dielectric constant and
trace widths should be increased upper limits while the insulator
thickness, GND plane thickness, and signal layers thickness should be
decreased to lower tolerances.

  Let us consider few examples:

  Example 1. Substrate (FR-4) dielectric constant ( Er ):
                  Fast corner: Er = 4.2
                  Typ corner: Er = 4.5
                  Slow corner: Er = 4.6

   Example 2. PCB Impedance ( Zo ):
               Fast Corner: Zo = 67 Ohms
              Typ Corner: Zo = 58 Ohms
              Slow Corner: Zo = 49.0

   Example 3. Signal propagation Delay ( Tpd):
             Fast corner: Tpd = 1.8 ns/ft
             Typ Corner: Tpd = 1.95 ns/ft
              Slow Corner: Tpd = 2.0 ns/ft

   Example 4. Passive components:
            Resistor:
            Fast Corner: 68.9 Ohms
            Typ Corner: 68.0 Ohms
            Slow Corner: 67.3 Ohms

           Capacitor:
            Fast Corner: 90 pF
            Slow Corner: 110 pF
          (i.e. Fast corner involves highest Resistors, largest
Inductors, and smallest Capacitors)

 Although, the SI professionals are in good agreement that normally it
is insufficient to simulate only the Typ corner and it is crucial to
examine the design under the Min and Max requirements; there is no
unanimity of opinion as to how the variable elements should be permuted
in order to catch the worst case corners. Consequently, there may be
disagreements regarding some statements or examples presented in this
communication.

  It should be also added that in some cases the Monte Carlo,
Sensitivity sweep or What-if analysis may be also required towards
exploration of a solution space.

  In closing, simulation of the Fast, Nominal and Slow corners may be
carried out to examine tolerance effects. Variations of numerous PCB
parameters such as substrate dielectric constant and thickness, trace
widths, the plane and signal layer thick nesses have to be taken into
account. The variable driver and receiver parameters encompass the
operating temperature range, supply range, and manufacturing process
variations. A good behavioral model should contain Min, Typ and Max
data for the parasitic sections, I-V curves and V-T tables to allow the
most accurate evaluation of each simulation corner.

   Thanking you in advance for your valuable comments and with sincere
regards,

  Abe
 [email protected]
 [email protected]

**** To unsubscribe from si-list or si-list-digest: send e-mail to [email protected]. In the BODY of message put: UNSUBSCRIBE si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Thu Apr 20 2000 - 11:35:45 PDT