š OpenLane Flow (RTL ā GDS)
This chapter explains how the RTL developed in previous chapters is transformed into a manufacturable ASIC layout (GDS) using OpenLane and SkyWater SKY130.
The focus is on understanding the flow, not just running commands.
šÆ Goal of This Chapter
By the end of this chapter, you will understand:
- What happens at each stage of the OpenLane flow
- Why each step is required
- How to interpret basic PPA (Performance / Power / Area) results
- How digital control logic maps onto physical silicon
š§° Toolchain Overview
The OpenLane flow integrates several open-source tools:
- Yosys ā RTL synthesis
- OpenROAD ā Floorplanning, placement, CTS, routing
- Magic / KLayout ā Layout viewing and DRC
- Netgen ā LVS (Layout vs. Schematic)
All steps target the SkyWater SKY130 open PDK.
š OpenLane Directory Structure
A typical OpenLane design directory looks like this:
openlane/
āā vi_control_core/
āā config.tcl
āā pin_order.cfg
āā runs/
The config.tcl file defines clock parameters,
utilization targets, and design constraints.
āļø Key Configuration Parameters
Important parameters in config.tcl include:
DESIGN_NAMEVERILOG_FILESCLOCK_PORTCLOCK_PERIODFP_CORE_UTIL
Example (conceptual):
set ::env(CLOCK_PORT) "clk"
set ::env(CLOCK_PERIOD) "20.0"
set ::env(FP_CORE_UTIL) 50
These values directly affect timing, area, and routability.
š§ Step 1: Synthesis
During synthesis:
- RTL is translated into a gate-level netlist
- Flip-flops, adders, and multipliers are mapped to standard cells
Key outputs include:
- Gate count
- Estimated timing
- Area estimate
This is the first point where silicon cost becomes visible.
š§± Step 2: Floorplanning
Floorplanning defines:
- Core area size
- Aspect ratio
- IO pin placement
- Power grid strategy
For control ASICs:
- Moderate utilization (40ā60%) is recommended
- Simple rectangular layouts are usually sufficient
š§ Step 3: Placement
During placement:
- Standard cells are placed inside the core area
- Timing-driven optimization is applied
At this stage, you can observe:
- Cell density
- Early timing slack
- Congested regions
ā± Step 4: Clock Tree Synthesis (CTS)
CTS inserts clock buffers to ensure:
- Low skew
- Balanced clock distribution
For this design:
- Single clock domain
- No gated clocks
This simplicity greatly improves robustness and timing closure.
š£ Step 5: Routing
Routing connects all placed cells:
- Global routing
- Detailed routing
Key checks include:
- No open or shorted nets
- Acceptable congestion
- Reasonable wire lengths
š Step 6: DRC and LVS
Final sign-off checks:
DRC (Design Rule Check)
Ensures the layout follows all manufacturing rules.
LVS (Layout vs. Schematic)
Ensures the layout matches the synthesized netlist.
Passing both checks is mandatory for tapeout readiness.
š PPA Analysis
After completion, evaluate:
- Performance: Maximum achievable clock frequency
- Power: Estimated dynamic and leakage power
- Area: Core and die size
For educational control ASICs:
- Performance margins are usually generous
- Area is dominated by arithmetic units
- Power is modest due to low operating frequency
š§ Educational Insight
Seeing RTL transformed into geometry is a critical learning milestone.
At this point, you should be able to:
- Point to where the PID logic resides on silicon
- Relate control complexity to chip area
- Understand timing as a physical, not abstract, property
š Completion
You have now completed the full journey:
Control Theory
ā Fixed-Point Arithmetic
ā RTL Design
ā FSM & PWM
ā OpenLane
ā GDS
This is the essence of practical digital ASIC design.
š¼ Layout Inspection (OpenLane ā Magic)

The figure above shows an intermediate placed-and-routed layout generated by OpenLane and visualized using Magic.
This view is used to inspect:
- Standard-cell placement
- Power rails
- Routing density
It is intended for flow understanding and inspection, not as the final tapeout deliverable.
š Next Steps (Optional)
Possible extensions include:
- Adding SPI register interfaces
- Integrating test and debug features
- Exploring alternative clock periods
- Tapeout via MPW shuttle services
š Congratulations
You have reached the end of the core documentation.
If you understand every chapter in this project, you understand how to design, verify, and implement a real digital control ASIC.
ā”ļø Next
Proceed to post-layout verification:
ā”ļø Gate-level Simulation (Functional)
The next chapter verifies logical equivalence between RTL and the post-place-and-route gate netlist.
ā¬ ļø Navigation
- š Documentation Index
- š Project Top