SPACE

HOWTO

                    CADENCE AND GDSII HOWTO
                    =======================


Introduction
============

This file describes how to create a CACD process description from a Cadence
process description, as well as how to obtain layout data from a Cadence
repository.

NOTE: This file is under development, and at the moment only covers the
conversion of visual mask styles and the conversion of layout information
using GDSII files. We are also looking into the possibility of automating
(part of) the conversion.


Visual styles
=============

In the Cadence process directory, there exists a "draw file" named
"display.drf", which describes how elements (masks etc.) are visualized in
the various Cadence tools. We will roughly describe here how this file is
structured, so that you can convert the visual information to our CACD format.

The file "display.drf" contains a number of sections. Each section starts
with a name, and the contents of each section have been placed between
parentheses; semicolons are used to introduce a comment.
Use your text editor/ viewer to examine this file.

The section "drDesignPacket" relates four style parameters to a symbolic
name (a "packet" name). The four style parameters are (1) the stipple,
(2) the line-style, (3) the fill, and (4) the outline of the element being
visualized.  These style parameters are further defined in different
sections of the "display.drf" file.

An example of the "drDefinePacket" section is given below:

   drDefinePacket(
    ;( DisplayName  PacketName  Stipple  LineStyle  Fill   Outline )
     ( display      dots_white  dots     solid      white  white )
     ( display      packet2     solid    solid      black  black )
     ( display      packet3     blank    solid      slate  slate )
     ( display      packet4     blank    solid      white  white )
     ...
   )

Thus when you refer to a packet name (e.g. "dots_white") you refer to the
specific style defined for that name (e.g. white dotted fill, with white
outline).

The valid values for the "fill" and "outline" parts are described in a
"drDefineColor" section:

   drDefineColor(
    ;( DisplayName  ColorName  Red   Green   Blue   Blink )
     ( display      white      255    255    255 )
     ( display      yellow     255    255    0   )
     ( display      slate      140    140    166 )
     ( display      black      0      0      0   )
     ( display      blinkRed   255    000    000    t )
     ...
   )

The format of the color-descriptions should be self-explanatory.
Note that a "t" in the last column can be used to make the color blink.

The possible values for the stipple pattern are described in a
"drDefineStipple" section. We will not describe the format here, since it
is quite self-explanatory. Just have a look at that section using your text
editor/viewer. The possible values for the line-style are described in the
"drDefineLineStyle" section, which is also not described in this document.

Now, have a look at the Cadence technology file. This is a file in the
process directory, with extension ".tf". The basic format is similar to
that of the draw file (i.e., sections start with a name, and their contents
are between parentheses).

The "layerDefinitions" section describes the layer names, and relates these
names to the "packets" which we descibed above. The section contains some
more information (necessary for Cadence tools), but we will focus here
merely on the information we need in our CACD process description.

The "layerDefinitions" section contains a subsection "techLayers",
describing the layer names, and their internal numbers used by Cadence.
An example is given below:

   layerDefinitions(
    techLayers(
    ;( LayerName   Layer#   Abbreviation )
    ;( ---------   ------   ------------ )
     ( nw          10       nw  )
     ( co          11       co  )
     ( cos         21       cos )
     ( ins         22       ic  )
     ...
    )
   )

Note that the layer numbers given here are NOT the numbers used in the
GDSII stream (those numbers are in the "streamLayers" subsection of the
"layerRules" section, as described below).

The "techDisplays" subsection relates layer names to a visual style,
using the packet name.

   layerDefinitions(
    techDisplays(
    ;( LayerName  Purpose  Packet      Vis Sel Con2ChgLy DrgEnbl Valid )
    ;( ---------  -------  ------      --- --- --------- ------- ----- )
     ( nw         drawing  dots_white  t t t t t )
     ( co         drawing  packet2     t t t t t )
     ( cos        drawing  packet3     t t t t t )
     ( ins        drawing  packet2     t t t t t )
     ...
    )
   )

The lines containing the keyword "drawing" in the second column define the
normal drawing style for each layer. We will ignore here columns 4 through 8
(containing the characters "t" in this example).

In order to convert the visual style from a Cadence process description
into the CACD format, you should consider the following information:

  1. In the "*.tf" file, look for the "techDisplays" subsection of the
     "layerDefinitions" section.
  2. Look at the layer name (in the first column), and look at the packet
     name (in the third column). Ignore lines containing something other
     than "drawing" in the second column.
  3. In the "display.drf" file, look for the "drDefinePacket" section, and
     look for the packet names found in step 2.

If you use "spock" to create your CACD process description, then you can
enter the visual style in the graphical user-interface. Note that the CACD
format does not have the descriptive power of the Cadence format, and hence
you should make an approximation (e.g., only one shade of blue is allowed
in the "dali" editor, thus you should map different shades of blue to either
different colors, or to different fill patterns; some creativity is required).

If you do not wish to use "spock", then you will have to create a "maskdata"
file containing the visual information. How this is done is described
elsewhere (see the file "process.pdf", in the share/doc directory, or
click the link "Introducing a New Process" at:

   http://www.space.tudelft.nl?page=documentation

Also, you can type "icdman maskdata" for information about the maskdata format.


Importing GDSII files from Cadence
==================================

When you wish to import a GDSII file into our CACD environment, you need
to use the tool "cgi". Type "icdman cgi" for detailed information.

Unfortunately, a GDSII file does not contain layer names, but it uses
numbers instead. Thus, in order to read a GDSII file, we will need to map
these numbers to layer names. This mapping is specified in the file "*.tf"
in the Cadence process directory. Look for the subsection "streamLayers"
of the section "layerDefinitions". An example is given below:

   layerDefinitions(
    streamLayers(
    ;( layer   streamNumber   dataType   translate )
    ;( -----   ------------   --------   --------- )
     ( nw          10             0          t     )
     ( co          11             0          t     )
     ( cos         21             0          t     )
     ( ins         22             0          t     )
     ...
    )
   )

We are mostly interested in the first two columns of this description.
The first column contains the layer name, and the second column contains
the corresponding number used in the GDSII file. The third column contains
the "datatype", which is usually 0, but you may also encounter different
values.

Now, you can again use "spock", or edit the relevant information manually.
In the CACD process directory, a file named "bmlist.gds" defines the
mapping for the CACD tools. This file contains a line per layer.
Each line starts with the layer name, and is followed by the stream number.
An example is given below. This example corresponds to the exact same
description in the Cadence format above.

   nw       10
   co       11
   cos      21
   ins      22

(You may use a "#" mark to start a comment.)
Thus, you can see that basically the "bmlist.gds" file comprises the first
and second data column from the "streamLayers" section above.

Beware that if you are using both "spock" and manual editing, your manual
editing work may be overwritten when you re-generate your process files
using "spock". If you use only manual editing, or only use "spock", then
this is irrelevant.

Note that each layer name referenced in this file must also appear in the
"maskdata" file. Failure to do so will result in errors by the "cgi" tool.

For records in the GDSII stream containing a datatype other than 0, you may
specify the datatype after the stream number:

   ext:label 24 9

Usually, when the datatype is nonzero, a suffix ":label" is appended to the
mask name, to signify that these GDSII records should be mapped to a label
on the specified layer. A suffix ":term" can also be used, and indicates
that these GDSII records are mapped to a terminal on the specified layer.
Type "icdman bmlist" for details.

Now, with a proper "maskdata" and "bmlist.gds" file in place, we can import
a stream from a file named "file.gds" using the "cgi" tool. Simply change
to your project directory, and type:

   % cgi file.gds

Furthermore, note that when you merely have a GDSII file, you can use the
"cga" tool to get an ascii translation of the records contained in the file
(type "icdman cga" for information). If you filter out the lines containing
the word "LAYER", you get an overview of the gds layer numbers contained in
the file.  Example:

   % cga file.gds | grep LAYER | sort -u

However, in this way you cannot obtain any information on the mapping of
the layer numbers.