how to debug lvs violations

[chanducs24]

hi

i got errors like incorrect nets after i ran lvs..can anyone tell how to debug or trace and which files do i need to compare with?

Thanks,
chandra.

 

[keith1200rs]

Maybe you could post your LVS report? You usually get a spurious elements in the report. Once the matching fails it can look like there is a lot more wrong than there is. Usually I try to find a few obvious errors from the report and then re-run it.

Keith

 

[GI]

Hi chanducs24,

There can be a lot of reasons for an lvs failure. Most probable problems can be:

1 - Shorts
- Check LVS report's nets section.
- # of nets in schematic > # of nets in layout, most probably there is a short in the layout.
- Run LVS with short isolation to detect the shorted nets.

2 - Opens
- Check LVS report's nets section.
- # of nets in layout > # of nets in schematic, most probably there is an open in the layout.
- Locate the open and fix the issue.

3 - missing text layers (calibre.txt file)
- power & ground locations should be given to calibre to allow correct identification of the p/g nets (as a further help for identification).
- Normally the tool can resolve these but biasing it a little helps in terms of runtime & results.

4 - Missing libs in gds or in cdl or missing instance)
- Undefined cells may exist in merged gds or in merged schematic.
- Be sure the gds & schematic are generated without problems.

5 - Missing soft layers (e.g. msub)
- msub layer, which can be used to isolate 2 different grounds may be missing.
- This can produce thousands of errors.

6 - Others
- There can be many more causes, other than above. Above ones are the most common problems encountered.

The best way is to ensure that there is no drc & lvs violation at place and route tool just before gds export (at least fix all shorts & opens). This will provide you a good reference starting point.

I hope these help,
Best regards,
Gokhan
---