Skip to content

02 rp2040_min

Reference: Raspberry Pi "Hardware design with RP2040", minimal design example: USB-C with 5.1k CC resistors, 27R USB series resistors, 12 MHz crystal with 1k series and 15 pF loads, W25Q128 flash, 100 nF per supply pin, 1 uF on VREG_VIN and VREG_VOUT. Values from memory of that document, not re-read. The USBLC6 ESD part and AP2112K regulator are my additions.

Stresses: 0.4 mm-pitch QFN-56 on two layers, 12 decouplers, USB, crystal.

What broke - 16 connections unrouted: a 0.25 mm track cannot leave a 0.4 mm-pitch pad at 0.2 mm clearance. - No room for decouplers on top: QSPI, USB and six supply pins share an edge. - With GND pours present before routing, Freerouting assumed GND was connected; the pours then fragmented: 24 unconnected items, starved thermals. - NPTH hole clearance on the USB-C receptacle. - 12 parity errors on no-connect pins. - The first schematic was an A0 sheet with parts scattered and flags off-sheet.

Tool changes - CLASSES net classes; fine-pitch defaults documented. - Bottom-side placement: PLACE[ref] = (x, y, rot, "B"). - Route everything as traces, pour afterwards (POUR_FIRST restores the old order). - No-connect pads get KiCad's unconnected-(REF-PIN-PadN) net name. - Schematic packer: functional groups, horizontal passives, paper sized to fit.

Second pass, after the review existed

Adding REVIEW limits to this already-passing board failed it on five counts: the flash decoupler was 8.45 mm from the flash VCC pin, the crystal was 6.3 mm from XIN with 13.3 mm of XIN trace, and USB D+ was 7.4 mm longer than D-. None of that shows up in DRC.

  • Flash decoupler moved beside pin 8; crystal, its load caps and the XOUT series resistor pulled in under the bottom pad row.
  • USB D+/D- from the series resistors to pins 47/46 hand-routed as a mirrored pair (TRACKS).
  • First attempt put the series resistors directly over the USB pins. That blocked the escape of the neighbouring supply pins to their bottom-side decouplers: 9 unconnected. Resistors moved 4 mm out, pair kept straight.
  • Freerouting left two tracks 0.1488 mm apart against a 0.15 mm rule. The driver now adds 10 um to every clearance in the DSN before routing.

Left open - USB is length-matched from resistors to chip only; the connector side is autorouted with a 3 mm skew limit. No impedance target on two layers. - Crystal load caps not recalculated for a specific crystal.