AIF2APD Revision History

The following notes describe changes, fixes and enhancements to AIF2APD. Where version numbers are skipped it is normally due to the fact that the version never left the development area.

2.29 [12/17/2004]

AIF Import - Fixed Die Outline

The AIF Importer was not using the die outline offset (CENTER) information so die outlines were always automatically centered about the die pad extents -- even when the input data had the die outline not centered. [reported by SanDisk]


2.27 [10/26/2004]

Flipped Substrate

Added support for flipped substrate. Prior version ignored the mirrored flag and produced incorrect AIF output.


2.25 [04/26/2004]

Soldermask Support

Version 2.25 of AIF2APD enables the extraction of soldermask data. The user can specify any subclass to be a solder mask and this data will be output to the AIF file. Since AIF does not support the concept of solder masks and we don't wish to add more complexity to the AIF format, solder masks are exported in the same fashion as ring data. The latest version of Bondgen supports reading such rings. [requested by SanDisk in order to document solder mask defined pads.


2.24 [04/26/2004]

BGA Outline

Added support for a polygon to define the BGA outline. (Requested by SanDisk for LGA documentation)

Fiducial Extraction

Can extract metal areas as fiducials if they have been marked with a custom property (AIF_FIDUCIAL) which is a boolean and set=1. Requested by SanDisk for substrate documentation.

Fixed Bug in Shape Pads

Corrected a bug when extracting padstacks defined by a SHAPE.


2.22 [04/08/2004]

Updated the AIF Import Module

Updated the AIF import module for 15.0/15.1. These versions of the program created a slightly different stackup.

Load more than one die package

Supports importing more than one die. You can now import several die (one at a time ) via AIF.

New Windows Install

A new one-click InstallShield installion has been supplied for updates on Windows. (Unix users will still have to install manually - but hey, they like it that way ...)


2.21 [04/01/2004]

Selecting Components to Export to AIF

Modified the APD2AIF interface to display component refdes rather than symbol names, so that multiple insertions of the same symbol may be exported selectively. MCM example by SanDisk with multiple insertions of capacitors.


2.20 [03/31/2004]

Extracting Package Outlines

Modifed the package outline code to read layers ASSEMBLY_TOP/BOTTOM and use symbol bounding box center rather than component insertion point. This was to correct outline placement shift. Example from SanDisk.

2.19 [03/21/2004]

Extracting Package Outlines

Modified the code for calculating package outlines based on packages with multiple paths.


2.18 [02/24/2004]

Bug Fix in Staggered Die Pads

Fixed a bug when exporting AIF from a staggered die where the die padstack was defined only between the two wire layers. Dropped wires on connected to the lower levels of the die pin pad stack are now detected.


2.17 [02/12/2004]

Bug Fix in Multi-Chip Die Support

Fixed a bug when exporting AIF in a multichip design. Duplicate entries were made in the AIF output when a wire went from one die pad to a different die pad. This has been fixed.


2.16 [01/07/2004]

Bug Fix in Multi-Chip Die Support

Fixed a bug when exporting AIF in a multichip design. If the bond pad name in the MCM database was null (i.e. non-existent) then the output AIF file failed to include a placeholder ( - ) for the missing bondfinger number in the netlist section. (Reported by Fluent Technologies.)


2.15 [12/22/2003]

Multi-Chip Die Support

Added support enabling the export of mutliple die in a single package. The output is an extension to the AIF II spec and will require some work on AIF II readers to properly display both die. It also allows just the die information to be extracted and not the package info.


2.14 [06/23/2003]

Duplicate Pads in Flipchip

fixes a bug when processing a flip chip design where some of the bumps were duplicated in the AIF file output. ( Problem noted by IDT)

Die with Polygonal Pads

fixed a bug in exporting AIF where the die used polygonal pads instead of square or rectangular ones. (Problem reported by Fluent)


2.13 [02/28/2003]

Output Data Resolution

Previous versions of the AIFOUT module might not use all of the available resolution for the coordinate data. This version now outputs as many places of data as is found in the APD database. (reported by Amkor)


2.12 [02/24/2003]

Revised for Updated APD

The last ISR for APD 14.2 and also version 15.0 modified the behavior of one of the SKILL functions used by the AIF Import routine causing the import routine to stop working. This version fixes that problem and remains backward compatible by testing the version/ISR of APD and applying the proper call depending on the version.



2.11 [02/06/2003]

Import Bug Fix

Fixed an import bug reported by TI.


2.09 [11/27/2002]

Oblong Pad

Fixed a bug in the oblong pad definition that caused it to appear rotated by 90 degrees.


2.08 [08/13/2002]

Net Name Support Enhanced
  1. Modified the net name checking algorithm to allow any characters except !'; and [] at the beginning of the net name.
  2. Modified the netlist creation code to quote net names, allowing previously illegal characters to be correctly read into APD.
  3. Modified the AIF export code so that it no longer replaces now legal characters in net names.

requested by PMC Sierra in order to keep compatibility with netnames supplied by IC design tools.

2.07 [03/21/2002]

Update for Better Flip Chip Support

Updated the APD2AIF routine to support "traces" between die pads and fingers. This is to support extraction of net and connectivity when using APD for die pad redistribution work. The solder bumps are treated as "fingers" and the traces from periperhal die pad to solder bump are treated as wires. Previous versions assumed that a wire could only have one segment. This has been changed to support multiple segments but only the first and last vertex point are transferable to AIF.


2.06 [03/19/2002]

No-connect bond finger number fix

Fixed a bug where no connect fingers did not have their finger number written to the AIF file, when executing APD2AIF. Reported by Amkor Technology.


2.05 [02/15/2002]

Bug Fix

When AIF2APD finishes, the design name is changed to that of the AIF file.

Net Name Formatting Fix

A bug associated with the formatting of net names was fixed.


2.04 [02/13/2002]

Change to Extract Rules

In accordance with Cadence recommended practice, all vias with the BOND_PAD property are considered to be bond fingers. This is a necessary and sufficient condition. This eliminates the need for bond finger identification by the user. Hence, we have removed the dialog where the user was requested to identify the bond finger padstacks. APD2AIF can now automatically identify bond fingers by the BOND_PAD property. Conversely, if you do not attach the bond_pad property to your bond fingers they won't be extracted by apd2aif.


2.03 [02/08/2002]

Modified how ring nets being exported from APD, so that they meet AIF naming standards.

Added a new function to create and run scripts on the fly. This allows us to run APD commands in silent mode when no Skill interface is available. This removes much of the remaining screen flashing during AIF2APD.

Ring creation is now implemented using Skill to directly write to the database, which makes it faster and cleaner than using APD commands.


2.02 [12/14/2001]

-----------------

Improved look and speed of AcsAifToApd by use of the following functions: axlLayerCreateCrossSection (x-section), axlDBCreateSymbol (place parts), axlDBCreateLine/axlDBCreateVia (wirebonds and fingers) and axlDBPutProp (finger properties). Prior to this we had to use a "script" to achieve the same functions.

Removed viacount.il as it was no longer being used.


2.01 [12/10/2001]

Rebuilt Code for Integration into APD

This is a rewrite to the apd2aif and aif2apd Skill code in order to enable Cadence to cleanly and efficiently incorporate the code into the 14.2 and later releases. The code has been rewritten as a "context."


Version 1.14 November 3, 2001

Fixed Missing Ring Problem
AIFOUT dropped sections of rings - problem was due to an unexpected connectivity relationship between the ring metal. The converter has been enhanced to deal with this possibility. [Reported by Lightspeed/Unicap]