PCB Layout Initial DRC / Route-Feasibility Notes
Current PCB setup
- Board outline: 160 mm × 100 mm rectangle, 1.5 mm corner radius.
- Stackup: JLC7628-style 4-layer stackup.
- Intended layer usage: all components currently on top side for initial placement.
Initial placement summary
- J1 USB-C is placed on the right edge.
- J2–J9 audio input jacks are distributed along the left edge.
- AK5558 U1 is placed in the central-left analog region.
- XMOS XU208 U2 is placed in the central-right digital/USB region.
- QSPI flash, reset switch, xTAG header, and XMOS oscillator are near U2.
- AK5558 audio MCLK oscillator X1 is near U1.
- AFE op amps and per-channel passive input networks are placed between left-edge audio jacks and U1.
- Regulators are in the digital/power region.
Verified XU208 footprint requirements
Research against XMOS/LCSC/DigiKey confirms the correct package is:
- Part: XU208-128-TQ64-C10
- Package: 64-pin TQFP exposed pad
- Body: 10 mm × 10 mm
- Pitch: 0.50 mm
- Lead span: 12 mm nominal
- Exposed pad / heat slug: approximately 5.3 mm, tied to GND
- Matching footprint name:
TQFP-64_L10.0-W10.0-P0.50-LS12.0-BL-EP5.3
Custom footprint / custom part attempts
A custom KiCad footprint asset was created and added to the project:
- Asset:
XU208-128-TQ64-C10_Custom_TQFP64_EP5p3.kicad_mod
- Structure: 64 lead pads plus one exposed pad, pad 65, with no extra internal unnetted pads.
- The asset was applied to the U2 footprint node.
- The U2 footprint node shows the custom asset in layout inspection.
A custom KiCad symbol was also generated:
- File:
XU208-128-TQ64-C10_CUSTOM.kicad_sym
- Contains all 65 pins with names matching the existing XU208 pin map.
- Imported together with the custom footprint as
XU208-128-TQ64-C10_CUSTOM.
However, the imported custom-part project appears empty when inspected and did not become a usable replacement UID for U2 in the current project.
XUF208 alternate test
A temporary XUF208-256-TQ64-C10A part was added as U99 and placed away from the design to test whether its TQFP64 footprint avoided the XU208 footprint issue.
Result:
- Route-feasibility still reported generated internal pads on U99.
- Therefore XUF208 TQFP64 appears to share the same Flux/LCSC generated footprint behavior.
- U99 was removed after the test.
ERC / DRC / route-feasibility results
ERC
- ERC was clean in the previous checked set.
DRC
- DRC reports airwires, which is expected because routing has not started.
- No general U2-specific DRC issue is reported by the standard DRC query.
Route-feasibility blocker
- Route-feasibility still reports the original generated internal pads overlapping U2 exposed pad.
- This suggests the routing checker still uses generated sub-pad data from the original part/footprint asset, not the overridden instance footprint nor the attempted custom imported part.
Fix/replacement attempts made
- Marked generated internal pads inactive.
- Shrunk generated internal pads.
- Moved generated internal pads away from the exposed pad.
- Disabled the generated
Mounts container.
- Attempted deletion of generated pad nodes.
- Re-applied Package / Package or Case Code metadata to the verified footprint name.
- Imported the Flux/LCSC files page as new component projects; imported projects did not expose a usable replacement part UID.
- Tested an alternate LCSC XMOS variant using the same package footprint; it showed similar generated internal pad behavior and was removed.
- Created and applied a custom KiCad footprint asset; the layout node shows the new asset but route-feasibility still reports the old generated pads.
- Created and imported a custom KiCad symbol plus custom footprint as a full part; the imported part project was not usable for replacing U2.
- Tested
XUF208-256-TQ64-C10A; it showed the same generated-pad behavior and was removed.
Current routing readiness
- Do not start autorouting yet.
- The placement is ready as a floorplan, and ERC is clean, but routing is blocked by route-feasibility's U2 footprint anomaly.
Recommended next action
- Do not migrate to XUF208 TQFP64 unless the library footprint is fixed first.
- If avoiding the XMOS TQFP footprint issue is mandatory now, evaluate a non-XMOS alternative such as STM32H743VIT6 LQFP100, accepting that this requires a larger firmware and USB HS PHY migration.
- Otherwise wait for a corrected Flux part/footprint asset for XU208/XUF208 TQFP64.