https://open-source-silicon.dev
Join Slack
Menu
Home
Archive
I am getting just 6 DRC errors, but the odd thing is that it is in something that is repeated hundre...
# magic
j
I am getting just 6 DRC errors, but the odd thing is that it is in something that is repeated hundreds of thousands of times (filler) yet only one seems to have a problem:
Copy code
user_project_wrapper
----------------------------------------
P-diff distance to N-tap must be < 15.0um (LU.3)
----------------------------------------
 2912.325 2693.035 2912.585 2693.905
 2912.855 2693.035 2913.115 2693.905
 2913.705 2693.035 2913.965 2693.905
 2911.865 2691.695 2912.125 2692.565
 2912.855 2691.695 2913.115 2692.565
 2913.705 2691.695 2913.965 2692.565
----------------------------------------
[INFO]: COUNT: 6
[INFO]: Should be divided by 3 or 4
It did get "Unknown layer/datatype in boundary, layer=236 type=0" errors when reading the sky130 cells but I doubt that this was a problem
what is around these coordinates
closer up
I think I found it by zooming out a bit and only showing layers 65/44 and 94/20 - it is indeed slightly different from all the rest
image.png
this is where my macros stop so the alignment of the tap cells got out of sync, and at the end of this single row it missed one
@Tim Edwards moving my macros a bit horizontally might make these DRC violations go away. But it is taking a few hours per run now (due to all those filler cells). I think that in this case the slightly longer tap distance won't hurt anything. Do you agree?
I ended up moving my macros 4µm to the left. I think that will solved the top without hurting the bottom
t
Okay, although it is also true that the tap distance wouldn't be an issue to worry about.
j
4µm to the left made it worse - I then got metal4 violations since it routed one thin vertical piece too close the the power straps. I moved it 7µm more to the left and nearly aligned the tapcell pattern as if my macros weren't there (not exactly, but close enough). It is still doing DRC but I see one thing piece of metal4 that is still too close
t
Personally, I would go with what the placement tool produced, and not worry about the errors. Another solution is to tweak the openlane setup to reduce the tap cell placement distance until it stops creating those errors, but frankly, that's overkill.
Open in Slack
PreviousNext
Powered by

What is Linen?

Linen is a search-engine friendly community platform. We offer integrations with existing Slack/Discord communities and make those conversations Google-searchable.

Powered by