open-source-silicon.dev

We are getting errors from GF that indicate MT.2a indicating a top metal spacing of 0.46 however the gf180mcuC.tech has 0.36 in it - has anyone encountered this?

This looks like an error in the open_pdks pre-processor script.  The block in the ifdef that is supposed to be active for 0.9um and 1.1um thick metal 5 is being passed over.  I had not noticed that.  Thank you for pointing it out.

Seems like my preprocessor only likes `defined(X) || defined(Y)` and not `defined(X || Y)`.  Should be easy to fix.

I just pushed a fix for this on open_pdks (<http://opencircuitdesign.com|opencircuitdesign.com>) version 1.0.397.

By the way, if you're not aware:  GF gave us some push-back on our choice of the 0.9um thick top metal for the 5 metal stack used in GF-MPW-1, asserting that it causes delamination issues on the pads when doing wire bonding, and suggesting that we use the 1.1um thick top metal instead.  Although the DRC rules are the same for 0.9um and 1.1um top metal, the parasitics are a bit different, so I elected to designate a new PDK variant called `gf180mcuD` using the 1.1um thick top metal 5, to differentiate it from the one we used for the GF-MPW-1 tapeout.

Do you have a github link to pull by chance?

<@U01AS2SDR7S>:  Done.  Update is available on github now.

<https://github.com/efabless/caravel-gf180mcu>

This is <https://github.com/RTimothyEdwards/open_pdks>

<@U016EM8L91B> is this going to fix the other errors that GF indicated?

we don't seem to have any errors on our portion

<@U016EM8L91B> Sorry if this is a odd question but where would your `preproc.py` file fit in the caravel flow?

<@U01AS2SDR7S>:  Ideally, all files in open_pdks can be written with preprocessor directives that allow the files to be filtered through the definitions declared in the Makefile.  Thus, the file `gf180mcu.tech` is not a magic tech file but is a _template_ for all magic techfiles related to `gf180mcu`.  It gets filtered through the preprocessor using definition options like `-DTHICKMET1P1` and produces a tech file that is specific to one set of process options.

Please also check other occurances of the preprocessor branches, I fear some other rules like e.g. antenna rules might be affected of bad preprocessor rules as well.

<@U016HSAH6AE>:  I looked;  only the DRC rules for top metal and a couple of other very minor things (already forgot what) were affected.