User LayoutEditorFull Tuesday 30th June 2026
I am having an issue when applying radii to parts with them coming out with jagged edges on the arc rather than smooth edges. I thought I had found some settings that were not too bad a compromise (around 2 deg and 8 Bezier iterations) but seem to be randomly having more issues lately on small parts (1-10um radius). In extremes even on larger radius arcs (100um) this can result in a 1nm notch at the end of the curve. I have also accepted that I cannot simply take a curved shape and resize it like in other tools as undetectable tiny see saw edges become amplified to roughness so I always now start from a manhattan shape and apply a radius to that.
I am currently running build 20260102 but will update to 20260610 shortly.
Any suggestions on a general setting for curves that can avoid such roughness?
Jürgen LayoutEditorFull Tuesday 30th June 2026
The behavior you are describing -- especially the fact that it worsens on small scales (1 - 10 µm) and causes a 1 nm notch on larger arcs - is a classic symptom of grid snapping caused by the GDSII Database Unit (DBU) and its resolution limits.
This is a fundamental characteristic of the GDSII file format and is completely independent of the LayoutEditor version you are running. The DBU is a setting embedded within each individual GDSII file/library upon creation.
Because GDSII stores all coordinates as integers, LayoutEditor must snap every calculated vertex of a smooth curve (whether using Bezier iterations or circular arcs) to the nearest grid point. If your database unit is set to the standard 0.001 µm (1 nm), a rounding error on a tiny 1 µm radius is relatively huge, creating exactly those "jagged" or "see-saw" edges.
Here are a few insights and suggestions to address this:
1. The Database Resolution Limit
The most direct way to get smoother curves mathematically would be to increase the resolution when creating a new GDSII file (e.g., setting the DBU to 0.0001 µm / 100 pm).
Important Manufacturing Constraint: Please note that resolutions finer than 1 nm are not supported by many photomask manufacturers (mask shops / foundries). If their fabrication equipment operates on a 1 nm grid, any sub-nanometer data will just be rounded again during mask data preparation, reintroducing the roughness. You should check with your vendor before changing this setting.
2. Optimize Curve and Bezier Settings for Small Scales
If you must stick to the 1 nm GDS grid, high-density curve settings can actually work against you on micro-scales:
For small radii ( < 1 µm): Having too many points (like 8 Bezier iterations) on a tiny arc forces the vertices to sit incredibly close to each other. When the point-to-point distance approaches the 1 nm DBU grid, rounding errors dominate and cause the jaggedness. Try reducing the number of iterations slightly for very small shapes to let the grid snapping blend in better.
User LayoutEditorFull Wednesday 1st July 2026
Useful insights, thank you.