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

Mariana Huerta

05/08/2023, 8:37 PM
What does this error mean? I got it when I imported a SPICE netlist to magic
t

Tim Edwards

05/08/2023, 8:40 PM
Offhand, it looks like something wasn't translated correctly either when importing the netlist or in the netlist itself. Can you please post the netlist?
m

Mariana Huerta

05/08/2023, 9:33 PM
opamp_ab.spice
t

Tim Edwards

05/08/2023, 10:00 PM
What version of magic? I was able to import that; it produced a few error messages (one of which I probably need to look into), but it did not generate a Tcl syntax error, and it created a layout with all the devices in it.
Although the result is exceedingly weird. The Tcl statement in question appears to be caused by selecting an instance that was generated by the parameterized cell generation routine, and running the magic commands
array -list count
and
array -list pitch
and somehow coming up with
_@_
for the first one, and an empty string for the second. I'm not sure how either of those happens.
m

Mariana Huerta

05/08/2023, 10:51 PM
I think the p- poly resistors from my netlist are missing from the generated layout for some reason
t

Tim Edwards

05/09/2023, 2:36 AM
res_xhigh_po
isn't an "approved" device, although it exists as a model. SkyWater approves only five specific widths of device and has five models that match the approved widths. But W=20, L=5 is a poor choice for a resistor. Typically, resistor value matching will be very poor unless L >> W (preferably at least 5 times longer than wide). You are choosing an extremely wide and short device, which is poorly matched and far outside of the modeled device's bounds. You would be better off to select a lower-resistance device at something that gives you L >> W. The mrp1 device (
sky130_fd_pr__res_generic_po
) has a resistance of 48.2 ohms/square, so to get the same 500 ohms as your existing resistor, you would need 10.4 squares, so W=20.75, L=2 is a good choice and is smaller than your existing resistor.