Hello, I had a small issue with PDN generation. Fr...
# openlane
p
Hello, I had a small issue with PDN generation. Fragments of vertical VDD grid were generated, but only slightly touching horizontal traces, with no chance to put via. VDD net is highlighted on the picture. VSS rails are correctly not generated here. This causes LVS issues at precheck script (and FP_PDN check fail, but I had it disabled by default script 😞 ). I solved this issue by moving the bottom macro a little bit up to block this space. Is this something worth reporting to openlane or a standard problem? And additional question - is there some simple way to define additional PDN blockages in free space (outside of macros) in wrapper?
m
@Piotro I think for generic blockages, you can use
GRT_OBS
. For PDN blockage around macros, there’s also
FP_PDN_HORIZONTAL_HALO
and
FP_PDN_VERTICAL_HALO
which defaults to 10.
👍 1
m
Why does this fail LVS? Presumably the halo or the macro itself prevents the vertical strap from extending far enough to put a via but that doesn't seem like a LVS issue.
p
It was failing because these vertical straps are not connected to anything else and formed separate nets (vdd_uq*) during extraction, causing net mismatch
(on LVS check manually run on local precheck tool)
m
It sounds like you need to coordinate your PDN halo then to eliminate the strap
👍 1