Many chips have very precise layout requirements. You can draw the nets but they will not be the correct shape according to the datasheet so you create a polygon filled zone to overlay on the traces to expand them to the correct shape. You then assign these filled zones to a particular net that is connected to the chip.
If you then do multichannel layout using these polygons, the polygons do not get copied to the target rule areas.
My post was not quite correct. On further inspection this is the problem:
Rule zones are apparently copied correctly - the shape and layer is correct.
BUT, the rule zone net is NOT CORRECT
The rule zone for the targets is the same net as the reference. e.g. U2-pin1, even if there is U5 in the target rule zone (and U2 is only in the reference).
Even better: when you now do a copper pour, SOMETIMES the U2-pin1 rule zone is merged with the U5-pin1 chip net in the target where U5 exists. Sometimes it isn’t.
This is definitely a bug. I will try and get you a test circuit tomorrow.
if it’s of any use, I just tried a simple test with a zone for an internal node and there were no issues. It was replicated correctly. I tried earlier with global nodes (power rails, typical “zone” use case) and didn’t run into any issues either.
I’m using 9.0.0.
PS: just mentioning one possible “gotcha” that I keep running into:
Zones are not drawn as solid fill beyond the board outline (edge cuts). Is it possible that you tried replicating beyond the borders and the zones are created but not completely drawn?
In this case, you’d still see them as hatched lines, zooming in.
Uh-oh… this may be a different issue but it seems there - are- some gremlins after all.
Each target area contains two replicas of the sheet’s internal node - one for the original sheet, one for the target sheet.
Should probably make this a separate topic but anyway, attached my observation:
That’s exactly the problem and explains why I am sometimes seeing the copper pour merging and sometimes not; it is because you have U2-pin1 and U5-pin1 filled zones with equal priority fighting one another.
Edit: the workaround for this is not great because it means deleting every replicated source zone in every target zone.
Probably yes. It’s difficult to keep up with all the small problems of the new multichannel feature. Small things will be forgotten soon unless documented properly in the issue database. Even if something isn’t a bug in the end, the issue can be closed easily.
I tried to log in to Gitlab but it wouldn’t let me and kept saying “Verify you are a human” in an infinite loop even when I clicked the box it wanted me to.
So please could someone else file the bug report using my sample project?
Description:
In multichannel design, target areas receive identical copies of rule zones from the reference area in addition to correct rule zones derived (not identical) from the reference ones
Steps to reproduce:
Place a component in a hierarchical sheet.
Go to layout and layout the single component of the first hierarchical sheet with an additional hand-drawn polygon rule zone named after a net of one of the component pins (you will need to untick “hide automatically generated net names” for this).
Create a rule area around the rule zone and set the placement to the first hierarchical sheet.
Copy just the rule area 4 more times nearby, non-overlapping.
Draw an edge cuts rectangle around all the rule areas.
Do Multichannel → repeat layout.
Check the rule zones in the target areas. See that there are unwanted duplicate rule zones of the reference area, as well as desired renamed rule zones.