SRAM dummy pmos extracted parameter has changed

[d-m-bailey]

In previous version 8.3.376 W/L=0.07/0.15 matches newer sources sky130_sram_1rw1r*.spice

.subckt EL_G3_sky130_fd_bd_sram__openram_dp_cell br1 br0 bl1 bl0 vdd wl1 wl0 gnd a_38_n79#
+ a_400_n79#
...
X1 a_38_133# wl0 a_38_133# vdd sky130_fd_pr__special_pfet_pass ad=0.0175 pd=0.39 as=0 ps=0 w=0.07 l=0.15

In newest version 8.3.389 W/L=0.14/0.08 matches older source sram_1rw1r_32_256_8_sky130.spice

.subckt EL_G3_sky130_fd_bd_sram__openram_dp_cell br1 br0 bl1 bl0 vdd wl1 wl0 gnd a_38_n79#
+ a_400_n79# 
...
X1 a_38_133# wl0 a_38_133# vdd sky130_fd_pr__special_pfet_pass ad=0.0175 pd=0.39 as=0 ps=0 w=0.14 l=0.08

spice snippets are from extracted versions.

It may be easier/more appropriate to change the source spice. @mguthaus just so you're aware.

[RTimothyEdwards]

@d-m-bailey : I think the newer sources have had the device size altered so that it matches what magic produced and needs to be changed back to what it was originally. The correct device size is W=0.14, L=0.08, although the size of a device that is not a properly constructed FET with source and drain is somewhat undefined. The lack of a terminal was what was causing magic to generate the questionable value of W/L. I had to change the code to specially handle the case of the calculation of W and L for a device with one terminal missing.

[d-m-bailey]

Great! simple solution. Is that something that @mguthaus's group should handle or a change to open_pdks?

[mguthaus]

We just saw this issue the other day as well so I flagged my students and we can fix it.

We might have another issue as well but we will see when we adjust this.