should magic be able to extract mosfet subcircuit with exact source and drain area/perimeter when the entire cell is just the mosfet, without the ext2spice -d option? I'm getting all zeros without the -d option and the -d option result doen't seem to make sense at all. The nodes and connections after extraction works great tho, just wondering how it works and how magic detects diffusion areas so I can double check whether I'm defining anything wrong

I deprecated the "-d" option recently and had magic find the gate and source areas and perimeters individually for each device, dividing among the number of devices connected to the same area when needed. You didn't indicate what result you were getting. The "`*psilicide`" for the drain and source indicates

psilicide

plus its contact type(s), so the contact section needs to be properly defined or the contact area might not get counted. Your extract section is missing a

lambda

statement and does not use `units microns`; if the width and length of the device are coming out right in the netlist, then the

lambda

is probably okay. I implemented the subcircuit parameters relatively recently and may have never tested it with a tech file that doesn't use

units microns

, which might also impact how the values are computed. Finally, what version of magic are you using?

image.png

@Anderson Hsieh: The current magic version is 8.3.460 and I did some significant work on separating out individual source/drain areas and perimeters that very likely post-date your version, so you probably want to update first.

ok I will do a update right now

Good to know!

@Tim Edwards I shrunk the diffusion area down to 2*200 grids each side (source/drain), this is how the numbers turn out. Each grid is 0.5um * 0.5um = 0.025um^2 but the numbers seems to still be in lambda, not microns. this is with the

units microns

statement. Any chance you may know what went wrong?