Impact of Routing Layer Direction on Completion Rate
Two-layer board test comparison: top layer horizontal + bottom layer vertical vs top layer vertical + bottom layer horizontal
Impact of Routing Layer Direction on Completion Rate
Many PCB designers focus on obvious rule parameters like trace width, spacing, and via size when using auto-routing tools. However, the routing layer direction can significantly impact routing results more than you might expect.
This time we conduct a real test comparing two different layer direction settings on the same two-layer board DSN project, to see how swapping layer directions changes completion rate, via count, and overall routing topology.
Why Does Layer Direction Affect Routing Results?
In two-layer board design, routing layer direction conventions are typically:
- Scheme A: Top layer horizontal (0°), bottom layer vertical (90°)
- Scheme B: Top layer vertical (90°), bottom layer horizontal (0°)
It looks like just a simple swap, but from the auto-routing algorithm's perspective, this change directly affects:
- Search space: Different direction constraints limit the algorithm's path choices
- Via necessity: One direction might require more vias for turning
- Critical area congestion: Signal-dense areas may behave differently under different directions
- Overall routing topology: May completely change main net routing strategies
This is why different designers prefer different direction settings — they may have seen different optimal results on different projects.
Test Setup
Test Project
We use a real two-layer board project for comparison:
This project has the following characteristics:
- Uneven signal distribution with severe cross-direction interference
- Mixed digital and analog circuits
Direction Setting Comparison
Scheme A: Top layer horizontal + Bottom layer vertical
- Top layer (Layer 1): Horizontal (0°)
- Bottom layer (Layer 2): Vertical (90°)
This is the most classic two-layer board setting, widely used in industrial and consumer products.
Scheme B: Top layer vertical + Bottom layer horizontal
- Top layer (Layer 1): Vertical (90°)
- Bottom layer (Layer 2): Horizontal (0°)
Swap direction setting to see if different routing results emerge.
Other Conditions Remain Consistent
- Design rules (trace width, spacing, via size) completely identical
- Component positions and net connectivity completely identical
- Algorithm parameters and search strategy identical
- Runtime environment identical
Scheme A: Top Layer Horizontal + Bottom Layer Vertical
Start routing using the classic direction setting.
Routing Results:
| Metric | Result |
|---|---|
| Completion Rate | 99% |
| Total Routing Time | 80s |
| Unconnected Nets | 2 |
Observations: - Vertical traces in certain areas forced to avoid congestion - Certain critical nets show obvious detours
Routing Result:
Scheme B: Top Layer Vertical + Bottom Layer Horizontal
Swap layer directions and re-route.
Routing Results:
| Metric | Result |
|---|---|
| Completion Rate | 99% |
| Total Routing Time | 30s |
| Unconnected Nets | 2 |
Observations:
- Top layer vertical traces form longitudinal arrangement
- Bottom layer horizontal traces show different distribution
- Overall allocation becomes better
Routing Result:
Analysis and Insights
Why These Differences Appear?
In this two-layer board project, changing layer directions produces different routing results. Main reasons:
- Algorithm search strategy: Different direction constraints change how the algorithm evaluates path costs
- Critical net routing preference: Certain main nets more easily form shortest paths in certain directions
- Area congestion distribution: Different directions may shift congestion pressure to different regions
FAQ
Q: How should I set this parameter in actual use?
A: PcbAutoRouter has strong compatibility. As long as it's not a board with severe horizontal-vertical interference, use default parameters. Only when encountering severe horizontal-vertical interference like this test case should you consider adjusting this parameter.
Specifically:
- Most projects (80%+): Directly use the default top horizontal + bottom vertical and get great results
- Uneven signal distribution projects: If one direction is particularly dense, try swapping to see the effect
- Difficult routing projects: Before trying direction swap, first check if design rules are too strict — that's usually the main cause
- Experienced routing designers: Can proactively choose more suitable directions based on project characteristics
Q: Do I need to re-export when swapping layer directions?
A: No. Adjust directly in the PcbAutoRouter routing menu.
Conclusion
This two-layer board comparison test shows:
- Layer direction does affect auto-routing results, potentially causing obvious completion rate differences in certain projects
- But for most normal projects, this parameter isn't the primary issue — design rule reasonableness usually matters more
- When encountering particularly difficult routing projects, adjusting layer direction can be a simple and effective solution
- PcbAutoRouter's tolerance is large; design reasonableness is more critical than parameter optimization
Most importantly: Don't over-optimize parameters. Ensure basic design rules are correct, and most projects can route smoothly.