1234
1234
MET4 to gate area ratio should be 400. At the end of my finishing section after inserting redundant
vias and re-running route_opt, I load these antenna rules and run check_routes. It detects no
antenna violations.
Straight after I run signoff_check_drc with the runset provided by the fab. It detects a couple of
antenna violations on MET4. It thinks the ratios are ~530. All errors are in the same area of the
design, 4 are all right next to each other, 2 a bit further away but still close. all with MET4 (out of 6
metal layers, MET1-4, METTP, METTPL).
I'm a bit lost here. I'm pretty sure if my pnr script detected the violations it would be able to fix
them, but since it doesn't detect them I'm not sure how to fix them. I could try to manually cut the
routes, but I'm sure there's a better way.
I've compared the antenna rules in my pnr script (and run report_antenna_rules, to check they
applied correctly), with the runset, but I don't really know how to make sense of the runset, the
values given (400.0) are the same, but maybe there's some subtlety I'm missing, for example I'm not
sure how to check the -mode and -diode_mode parameters from my pnr script against the runset.
Both these rules came from the fab so they <should> be correct.
The affected nets are all internal nets named stuff like n2223, so I don't think it's anything to do with
my IO pins, although all 6 of the violations are pretty close to those pins.
In my PNR script if I reduce the max antenna ratio for MET4 to 1 (from 400) and then re-run
check_routes, it still gives no error, this sort of suggests it's not really running the checks at all.
However the output to check_routes does suggest it's using the new value.
Antenna/diode mode:
..
..
..
..
MergeGate == true
...
In ICV the results told me the problem was with nets n2223, ... however when looking at ICC2's error
browser with the results of the signoff_check_drc, it didn't specify nets, but close to where the Xs
were I noticed a pretty long path on MET4. Looking at that it was from an input port.
Warning: Skipping antenna analysis for net blah. The pin blah on cell my_design does not have
enough gate area information. (ZRT-311).
I'd been ignoring that (temporarily) as it didn't seem related to the nets specified by ICV as being the
problem. However two of those warnings mapped to the nets with long paths on MET4.
The solution was to tell the tools to ignore gate sizes on terminals: