https://open-source-silicon.dev logo
m

Matt Venn

06/04/2021, 8:46 AM
Can someone explain to me why this works:
l

Luis Henrique Rodovalho

06/04/2021, 12:29 PM
The transistor dimensions are strange. Are there scaling options?
t

Tim Edwards

06/04/2021, 1:31 PM
@Luis Henrique Rodovalho: Ignore the strange dimensions. It's what I managed to get working with magic given the SkyWater scale factor that is internal to the model files. I will clean up that code some time but it's a low priority.
l

Luis Henrique Rodovalho

06/04/2021, 2:03 PM
It was another doubt I had. I've extracted some single transistors from a layout and the scale options were very strange. I've corrected it manually in order to simulate it in ngspice. From now on, I'll let it as it is.
@Matt Venn If you really need wider transistors, it's better to use multipliers. Instead of a 66 single transistor, you should use 2x33, because really large transistors are not characterized by the foundry and the models just extrapolate the parameters.
m

Matt Venn

06/04/2021, 2:33 PM
It's just for understanding at this point, thanks
t

Tim Edwards

06/04/2021, 2:37 PM
@Luis Henrique Rodovalho: It's not 66um, it's 0.66um.
l

Luis Henrique Rodovalho

06/04/2021, 3:00 PM
Wasn't the bin issue already fixed for the models in the latest PDK version? The older version models had strict values for simulation, as the transistor dimensions couldn't be outside the bin values. https://github.com/google/skywater-pdk-libs-sky130_fd_pr/blob/f62031a1be9aefe902d6[…]d6f59b57627436/cells/nfet_01v8/sky130_fd_pr__nfet_01v8.bins.csv The 0.65/0.15 um is in the bin values.