@Stefan Schippers: Something I discovered a few days ago (apologies if this has been seen before and I just missed it or forgot about it): The SkyWater continuous models changed the diodes from D-type components to X-type subcircuits; I knew about that part. What I didn't notice, though, is that the subcircuits redefine the perimeter parameter name from

pj

to

perim

, and the subcircuit calls the model with

pj={perim}

. I patched this up in magic's device extraction, and I patched it up in the original PDK by adding subcircuit definitions for each of the diodes to make them forwardly-compatible with the continuous models, but the xschem diode symbols need the parameter changed from

pj

to

perim

.

Ok, will do that. In the mean time: @Tim Edwards Xyce dies on the following line (

line 90, libs.tech/combined/continuous/models_global.spice

):

.param mismatch_factor*MC_MM_SWITCH*AGAUSS(0,1.0,1) = 0 mismatch_factor*MC_MM_SWITCH*AGAUSS(0,1.0,1) = 0

I also don't understand exactly what this means and surprised why ngspice does not complain.

The continuous models have the same weird units as before, for which the 1μm x 1μm diode has

area=1e12 perim=1e6

.

area=1e12 perim=4e6

ok, thanks

Right. : )