open-source-silicon.dev

Hi <@U016EM8L91B> As a follow-up of <https://open-source-silicon.slack.com/archives/C016UL7AQ73/p1658056242392849|this> thread, do you think i should try to update my open_pdks install and try again device mismatch simulations?

It would be helpful to know if there are just some devices whose parameters are not being handled properly, or if it is a more widespread issue with the installer.

<@U016EM8L91B> I am doing a clean new open_pdks install, will do a detailed test on the various devices and follow up on this.

test_mismatch.png

<@U016EM8L91B> I have finally done a clean install of `open_pdks`, removed the `share/pdk` altogether to avoid old ghosts, recloned from scratch the repository, and done configuration and install. Tested mismatch simulation (20 runs) for `nfet_01v8, nfet_01v8_lvt, nfet_g5v0d10v5, nfet_03v3_nvt, nfet_05v0_nvt` and `pfet_01v8_lvt, pfet_01v8, pfet_g5v0d10v5, pfet_01v8_hvt`. The simple test simulation measures the `Vgs` at `1uA` for all transistors in 20 consecutive runs. All transistors show variations. So I don't know why my previous install was broken.

<@U01819B63HP>:  That's good to know.  I did have a concern about the possibility that resistors might not show variation, from what I was seeing in the models.

<@U016EM8L91B> will do a test on resistors. Thank you.

<@U016EM8L91B> A quick test on `res_iso_pw` and `res_xhigh_po_0p69` shows indeed that the resistor value is identical in all mismatch runs. I have inserted the resistors, biased at 1uA like the transistors and measured the voltage across them. May be some other switches needs to be set. Not looked at the model files yet.

What I saw in the model files was that the mismatch parameters didn't get handled by the script, so that there were no GAUSS() or AGAUSS() functions in the parameter equations like there should have been.