LVS expert! Issue with LVS-ing Standard Cells

When I CDL in a standard cell, Cadence messed up the drain and source
connections of the transistor as the following:

..SUBCKT NAND2 A B Y
MM6 VDD! B Y VDD! P W=3u L=180.0n M=1
MM7 VDD! A Y VDD! P W=3u L=180.0n M=1
.....
.....
..ENDS

Instead of "Y B VDD! VDD!" for the P-channel devices, it swapped the
the drain and source connection to "VDD! B Y VDD!".

For simulation, it seems okay because it's a switch to VDD. However,
does it pose any problem for LVS??

Thanks!!

 

What you describe is much like what happens during device extraction
from layout. There is no way to orient the source and drain along the
current flow, so the connections are assigned arbitrarily. LVS
understands the S/D permutability of a MOS device and handles it without
any trouble.

The only case where I have seen problems is when a user wanted to
consider the parasitic characteristics of the S/D regions as
differentiators. It is not possible to know if LVS has done a S/D swap,
so you cannot check the parasitic values are correct. I've only seen
this in memory cores, which are so rigid in their design there is little
point to be checking them.

 

most likely your pmoses have a completely simetric layour for source and
drain, so that the distinction is electrical, but not physical.
Only in some special devices, with drain enhanced FET (for reduced hot carrier
effects) or gradual doped channels (for maximum speed) do you have a physical
difference between S and D.

 

Thank you all! Good information.