Hi there, has been a while since used the ngspice models for sky130 and I may be missing something obvious. Why would below work with lines 21,22 (when swapping the comments) but not 23,24?

21 *X0 vss vout vin vss sky130_fd_pr__nfet_01v8 w=650000u l=150000u
 22 *X1 vdd vout vin vdd sky130_fd_pr__pfet_01v8 w=1000000u l=150000u
 23 X0 vout vin vss vss sky130_fd_pr__nfet_01v8 w=0.65 l=0.15
 24 X1 vout vin vdd vdd sky130_fd_pr__pfet_01v8 w=1.0 l=0.15

Additionally, I remember the models use the

option scale=1e-6;

but the MOSFET bins seem to be in meters, not sure how this gets resolved in the end, any clues?

>>> grep lmin cells/nfet_01v8/sky130_fd_pr__nfet_01v8__tt.pm3.spice |tail
+ lmin = 1.5e-07 lmax = 1.8e-07 wmin = 3.9e-07 wmax = 4.2e-7
+ lmin = 2.0e-05 lmax = 0.0001 wmin = 3.6e-07 wmax = 3.9e-7
+ lmin = 8.0e-06 lmax = 2.0e-05 wmin = 3.6e-07 wmax = 3.9e-7

This is commit f62031a1be9aefe902d6d54cddd6f59b57627436, from https://github.com/google/skywater-pdk-libs-sky130_fd_pr.git And I'm using ngspice-36

As for the first question: The order of pin connections got changed. MOSFETs should always have pins in order (D G S B), and while drain and source can be swapped without issue, in your example, the drain, gate, and source have all been reordered.

Thanks a lot, I got so fixated on the scaling aspect that a missed a simple typo. Also, I was checking the manual, it seems that the semantic for the scale option is to only scale the device parameters so it should have no consequence on the ranges.