›

How to Identify Layer Types in a Complex GDS File for a cross- section?

YessnrYessnr
in Ruby Scripting

I’m new to layout design and was assigned a task to generate a cross-section for a very complex layout. I have a GDS file with multiple cells and a layer list, but the layers aren’t labeled. I wrote a script to retrieve the layer names, but none were found. Is there a way to determine the layer types (e.g., metal, via, poly) programmatically? Can a script help identify these automatically?

Comments

  • StefanThiedeStefanThiede

    You should know for which technology this gds was built, no ?
    If you have stdcell routing with via, you can deduce metal/via layers by “looking” at the layout.

  • YessnrYessnr

    Hi, thanks for getting back to me!
    Honestly, that idea didn’t even cross my mind—lol. But aside from that, I managed to put together a working script. It’s not exactly how I envisioned it, but after a lot of trial and error, I was able to extract some of the layers.
    I’m still waiting to hear back from my boss to find out which software was originally used to create the layout.

  • MatthiasMatthias
    edited August 3

    Hi @Yessnr,

    A GDS does not have layer names or layer descriptions. It's just numbers.

    But every Design Manual (the most basic document of a PDK for a specific technology) has a table called "Drawing layers" or something like that, with Design system layer names and GDS layer/datatype numbers.

    For example here it's right on page 5: https://github.com/IHP-GmbH/IHP-Open-PDK/blob/main/ihp-sg13g2/libs.doc/doc/SG13G2_os_layout_rules.pdf

    Matthias

  • dick_freebirddick_freebird

    A script would have to be "taught" by someone
    or something that "knows" the meaning of each
    layerNum.

    A skilled layout or design engineer would have
    some amount of the required "gut". A PDK layer
    table of course would be most certain.

    I think I've seen that there is a cross section,
    layer order table that can be populated to
    make a cutaway (how large, how correct,
    how process-effects-aware, dunno - like CMP
    would look way different than non-planarized
    flows' ILDs).

Sign In or Register to comment.
Powered by Vanilla