open-source-silicon.dev

set the channel description: Channel for discussing and sharing information about high-frequency design using the SKY130 process

<@U02A4K5TG5B> Following up on the conversation we had on the June 10 Meetup, I came across this link which summarizes a couple of open-source tools for EM simulations:

<https://www.matecdev.com/posts/differences-fdtd-fem-mom.html>

I am curious to see how these tools benchmark against ADS-MoM or HFSS.

The main issue is integrating the layout + meshing into them in an human-friendly way that doesn't consume a whole day just to get a single design iteration... that's in fact what's still to be seen

I bet. You may need to run some funky scripts to do so.

I will check with some people to see if they have experience on this and let you guys know.

There seems to be some past discussion on EM solvers in the <#C01774RT7GV|serdes> channel, maybe worth checking out

Reading the whole thread there, folks haven't still focused on a specific solver, though.

After a long while (mainly being involved with chipathon) I'm back at the task of setting up a minimally operational OS EM toolchain. The main problem that needs solving for this to happen is solving the integration between different tools. ElmerFEM solves the issue of electrostactic and magnetostactic simulation (and thus more accurate parasitic extraction and custom device modelling). But there is no quick answer for high-frequency full-wave solver. I don't know if ElmerFEM can do that or not. OpenEMS has the limitations of the FDTD solver, and I don't know how would it perform in simulating more exotic structures.

Magic works well in PVS and its RC extraction is enough for the active cells (and coupling that to ElmerFEM magnetostactic sims would provide data on the inductive parasitics).

Another thing that might be interesting, or a point of departure, is the SiEPIC tools
<https://github.com/SiEPIC/SiEPIC-Tools>
Apparently that's a FDTD solver that uses klayout as layout editor. I think it is feasible to edit a simulation PDK around the sky130 process and use that to run layout-driven simulations]

Hey all!
Anyone know how (and if) Meep can run simulations on non-optical devices, like microstrips?