Repository navigation
feat(rp2040): 5 kHz stepping, hardware info, SSD1306 display, CI coverage + TMC hold-current fix - #309
Merged
Conversation
Board ID defines (2101/2201), pins-file selection arms, local-config arms, and validation gates for ARDUINO_ARCH_RP2040. Adds the [common_rp2xxx] platformio section (Earle Philhower core via the maxgerhardt platform wrapper) with rp2040_jackw01 and rp2040_skr_pico environments, and unignores Configuration_local_examples/. Co-Authored-By: Claude Code <noreply@anthropic.com>
The RA/DEC END_SWITCH_ACTIVE_STATE and ENDSWITCH_BACKSLEW_DEG constants were only defined inside the '#else' of '#ifndef USE_*_END_SWITCH', i.e. only when that axis's end switches were enabled. EndSwitches.cpp references both axes' backslew constants whenever either feature is on, so a DEC-only (or RA-only) end switch configuration failed to compile on every platform. Co-Authored-By: Claude Code <noreply@anthropic.com>
- EPROMStore: flash-backed EEPROM section for the Philhower arduino-pico core (API-identical to the ESP32 path) - InterruptCallback: repeating hardware-alarm timer via the Pico SDK add_repeating_timer_us() API for the 2 kHz stepper control interrupt - b_setup: USB CDC flow-control bypass and connect wait, GPS/UART1 pin remapping, fan PWM, JackW01 shared-EN handling, and 115200 baud TMC serial on RP2040 (57600 kept on AVR/ESP32); AZ/ALT DIAG pinMode calls now guarded like RA/DEC (pins optional per board) - c_buttons/f_serial: explicit serial polling - the Philhower core does not reliably run serialEvent() for USB CDC - Mount: TMC2209 toff=4 on RP2040 only (toff>=3 required for StealthChop slow-decay phase); toff(1) unchanged on AVR/ESP32 - DayTime/Declination/Latitude/Longitude: defaulted copy assignments to silence -Wdeprecated-copy on the RP2040 toolchain - SSD1306_128x64_Display: wrap Arduino.h/SSD1306Wire.h includes in PUSH_NO_WARNINGS for the arduino-pico core headers Co-Authored-By: Claude Code <noreply@anthropic.com>
Pin definitions for the JackW01 Pi Pico printer carrier board and the BigTreeTech SKR Pico v1.0 (Klipper-derived mappings, TMC2209 UART address strapping documented per slot), plus a tracked example local configuration for the SKR Pico. Ported from PR #298 (EricSchubert); the orphaned bare-Pico pins file from that PR is intentionally not carried over. Co-Authored-By: Claude Code <noreply@anthropic.com>
Adds rp2040_jackw01 (24 solutions) and rp2040_skr_pico (96 solutions) to the CI build matrix with conservative BOARD_SUPPORT restrictions (no LCD menu display, no gyro, no focuser; TMC2209 UART-only AZ/ALT). The SSD1306 info display matrix is extended to rp2040_skr_pico. Co-Authored-By: Claude Code <noreply@anthropic.com>
The platform branch of getMountHardwareInfo() only knew ESP32 and AVR, so RP2040 builds reported "Unknown" in the hardware info string. Co-Authored-By: Claude Code <noreply@anthropic.com>
logv() issued seven separate print() calls per log line. On the arduino-pico USB CDC each write re-enters the USB mutex and the TinyUSB FIFO path, which multiplies exposure to the core's write-loop edge cases. Compose the line first and send it with one write. Co-Authored-By: Claude Code <noreply@anthropic.com>
- Wait for the console to connect before logging in debug builds: the CDC TX FIFO holds only ~250 bytes and with no host attached the remaining boot output is dropped, so a debug build boots invisibly. Release builds are unaffected (logging compiled out, boot headless). - Make Serial.ignoreFlowControl() opt-in via RP2040_IGNORE_CDC_FLOW_CONTROL (default off): forcing TX while no host drains the port can wedge the TinyUSB write loop while it holds the USB mutex. Needed only for hosts that open the port with DTR=false (e.g. OATControl). - Drop the RP2040 serial poll from serialLoop(): serialEvent() handles CDC RX on the Philhower core (verified on hardware with a display-less build answering :GVN#). Known issue (under investigation, likely arduino-pico/TinyUSB): builds with sparse debug levels (e.g. STEPPERS|MOUNT|INFO) can wedge within milliseconds of boot; DEBUG_NONE and DEBUG_ANY builds are stable on the same hardware. Co-Authored-By: Claude Code <noreply@anthropic.com>
Add a STEPPER_CONTROL_INTERVAL_MS configuration knob (overridable in local config) with a 5 kHz default on RP2040; AVR stays at 2 kHz where higher rates interfere with serial communications. Also fix the RP2040 repeating-timer interval semantics: a positive delay_us measures from callback END to next start, silently adding the callback runtime to every period. Use a negative delay for a fixed start-to-start rate, matching the AVR TimerInterrupt behavior the stepper controller was designed for. Co-Authored-By: Claude Code <noreply@anthropic.com>
A mount powered from an external supply boots with no USB host attached, and Serial (tud_cdc_connected) never becomes true until a host asserts DTR. The previous infinite while (!Serial) spin left the firmware parked before any initialization indefinitely - no drivers, no timer, no commands - whenever it was powered without an already-open console. Wait at most 3 seconds for a console, then boot headless. Debug builds that boot headless lose the boot log (CDC drops writes with no host), but reset-with-console-open still shows the full boot. Co-Authored-By: Claude Code <noreply@anthropic.com>
The OAE firmware merge in V1.13.18 (#274) added ihold(31) after rms_current(mA, 1.0f) in the hardware-UART RA and DEC driver setup. IHOLD=31 is the top of the 5-bit current scale (~980 mA RMS per coil with 0.11R sense and vsense active), so at standstill the driver holds the motor at nearly three times the configured operating current instead of matching it - hot motors and a supply draw about 4x the expected value. The rms_current(..., 1.0f) call on the line above already sets IHOLD = IRUN, which is the documented intent ("hold current = run current"); the override only clobbers it. Note the readback log "Actual rms_current" cannot catch this: it reads cs2rms(IRUN) and never reflects IHOLD. The software-serial driver variants never got the override, so this also restores consistency between the two setup paths. Affects all TMC2209_UART boards using hardware serial since V1.13.18. Co-Authored-By: Claude Code <noreply@anthropic.com>
…lies Make Serial.ignoreFlowControl(true) unconditional on RP2040 instead of opt-in via RP2040_IGNORE_CDC_FLOW_CONTROL. Hosts that open the port without asserting DTR (OATControl and other C# SerialPort apps) were getting all firmware output - including Meade replies - silently dropped, since tud_cdc_connected() is the DTR bit. The opt-in hedge guarded against a write-path stall this core version cannot produce: SerialUSB::write() breaks immediately once the CDC TX FIFO is full and no DTR host is connected, and drops after 1 second without progress otherwise (framework-arduinopico SerialUSB.cpp). So unconditional bypass is safe for headless boots while making the mount usable from every host application without changes. As a bonus the first ~257 bytes of the boot log survive a headless boot and reach the first host that connects. Co-Authored-By: Claude Code <noreply@anthropic.com>
Develop now carries the RP2040 port via the rebuilt #298. Take our side where this branch holds the verified deltas (5 kHz stepping and timer semantics, TMC hold-current fix, RP2040 hardware info, SSD1306 info display, CI matrix entries, single-write logging), and develop's side for the ConfigurationValidation DIAG advisories and the .gitignore ordering. Co-Authored-By: Claude Code <noreply@anthropic.com>
…branch The PassthroughCmd arm of the action chain used a single = instead of ==, so reaching that arm assigned the String and evaluated it as truth: the branch always fired, and any unrecognized action silently entered passthrough-command mode instead of being ignored. On the RP2040 toolchain this also fails to compile (-Werror=parentheses via the Philhower core String header), which is how the CI matrix caught it. Co-Authored-By: Claude Code <noreply@anthropic.com>
The generated matrix solutions previously included A4988 and standalone driver types for the RA/DEC axes of both RP2040 boards. Those driver types need MS pins, which the boards do not route (on-board TMC2209s wired for UART only), so the affected solutions failed to build in CI with "Missing pin assignments for MS pins". Restrict RA_DRIVER_TYPE and DEC_DRIVER_TYPE to DRIVER_TYPE_TMC2209_UART for both boards, matching the AZ/ALT restrictions that were already in place. Co-Authored-By: Claude Code <noreply@anthropic.com>
Documents the default pin assignments for the SKR Pico and JackW01 boards: power (including the USB-jumper/VIN contention gotcha), steppers and TMC2209 UART addresses, USB serial behavior, SSD1306 info display (with the SDA=5/SCL=4 default colliding on the SKR Pico), GPS on the SKR Pico left header, Hall homing and DEC endstop inputs, the FAN1 output, and the features that are not available on RP2040. Co-Authored-By: Claude Code <noreply@anthropic.com>
Bench testing on the SKR Pico v1.0 revealed that GP20/GP21 cannot be used for the I2C info display: both only drive on-board MOSFETs (HB/FAN3 and the opto-isolated Laser output) and are never exposed as raw GPIO on any connector. Wiring a display to those power pins hangs the I2C bus and with it the firmware loop (USB stays enumerated, serial goes deaf). - Default the info display to GP0/GP1 on the SKR Pico (the only header-accessible I2C0 pair, on the left-side header) and to GP20/GP21 on the JackW01; ConfigurationValidation now rejects display pins on GP20/GP21 for the SKR Pico, and the display+GPS combination (shared pair) - matrix_build: stop generating skr_pico display+GPS solutions (would now fail) - Example config, wiring doc and wiki page updated accordingly Bench-verified: SSD1306 on GP0/GP1 renders, serial stays responsive. Co-Authored-By: Claude Code <noreply@anthropic.com>
…upported The TMC2209 DIAG outputs are irrelevant for the OAT firmware on these boards (StallGuard-based autohome is not supported - Hall sensor homing uses the endstop inputs instead), so the wiring tables now name the column plainly 'Endstop input' and the not-supported lists spell out the limitation. Example config comments no longer mention RA_DIAG_PIN. Co-Authored-By: Claude Code <noreply@anthropic.com>
The optional AZ/ALT DIAG pin checks printed a pragma message on every RP2040 build (both boards leave AZ/ALT DIAG unwired). StallGuard-based homing is not supported by the OAT firmware on these boards - Hall sensor homing via the endstop inputs is the supported path - so the messages are pure noise. Remove the checks; the pins stay optional. Co-Authored-By: Claude Code <noreply@anthropic.com>
…ersion Boot console updates (addConsoleText/updateConsoleText render with a null content callback) never drew the header, so the logo only appeared via ISR-driven renders in the ~0.5s window between stepper-timer start and console-mode end - invisible on a fast RP2040 boot and fragile everywhere. Extract drawHeader() from drawScreen() and call it from the base render() on every console-mode frame, so the logo and product name show from the first boot line through BOOT COMPLETE on all platforms. Also uppercase the FW version line: Bitmap3x5 covers only chars 32-90 and release-please now stamps VERSION with a lowercase v prefix, which rendered as an out-of-range glyph. Console lines already uppercase. Bench-verified on the SKR Pico (SSD1306 on GP0/GP1). Co-Authored-By: Claude Code <noreply@anthropic.com>
The 19200 ASCOM default exists because the 8-bit AVR platforms cannot keep up; the RP2040 USB CDC port has no such constraint. Add SERIAL_BAUDRATE_FAST and select it automatically on ARDUINO_ARCH_RP2040, and set monitor_speed accordingly for the RP2040 environments. AVR and ESP32 are unchanged. Co-Authored-By: Claude Code <noreply@anthropic.com>
The old comment said the alarm API works "without relying on FreeRTOS or a second core", which reads as if FreeRTOS were an option on this platform. It was meant as a contrast to the ESP32 port, which steps from a FreeRTOS task. State that explicitly and note the Philhower core has no RTOS. Co-Authored-By: Claude Code <noreply@anthropic.com>
andre-stefanov
requested changes
Oct 17, 2026
- SSD1306 display header: provide no-op PUSH/POP_NO_WARNINGS fallbacks (mirroring inc/Globals.hpp) so Arduino IDE builds, which lack the PlatformIO -include of macros/gcc/Macros.hpp, can still compile it. - updateInfoDisplay(): declare isSlewingAny inside the RP2040 arm; it was unused (and -Werror=unused-variable under -Wall) on display-enabled AVR/ESP32 configs. - Utility logv(): restore flush() on all non-RP2040 platforms so AVR/ESP32 logging behavior is unchanged; only the arduino-pico USB CDC path skips it (CDC flush can stall without a reader). - Example config: fix AZ_CIRCUMFERENCE pi typo; align the stepper interval with the 5 kHz PR default (10 kHz noted as possible but unverified). - Docs: drop the nonexistent USE_AUTOHOME flag reference. - .gitignore: remove duplicated negation lines from the develop merge. Co-Authored-By: Claude Code <noreply@anthropic.com>
andre-stefanov
approved these changes
Oct 17, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Follow-up to #298 (merged in 2f77d61): everything the maintainer-side verification branch carries on top of the merged RP2040 port, bench-verified on a BTT SKR Pico v1.0 over multiple hardware sessions.
Fixes
src/Mount.cpp): removes theihold(31)override introduced with the OAE merge in V1.13.18 (V1.13.18 - Updates #274). IHOLD=31 is the top of the 5-bit current scale (~980 mA RMS per coil with 0.11 R sense and vsense active), so at standstill the driver held motors at nearly 3x the configured operating current instead of matching it. The adjacentrms_current(mA, 1.0f)already sets IHOLD = IRUN, which is the documented intent. The readback log cannot catch this (rms_current()reads back IRUN only). Affects allDRIVER_TYPE_TMC2209_UARTboards using hardware serial since V1.13.18. Bench: 1.4 A at rest dropped to 0.12 A on a 470 mA motor at 75 %.src/InterruptCallback.cpp): the repeating alarm now uses a negative delay. Per the Pico SDK, a positive delay measures callback-end to next-start (adding the ISR's own runtime to every period, with load-dependent jitter); negative gives true fixed start-to-start rate, matching the AVR TimerInterrupt semantics the stepper code was designed around.src/Utility.cpp,src/b_setup.hpp):logv()composes each line into oneprint()call instead of seven writes racing the TinyUSB FIFO, and theignoreFlowControl(true)call documents why unconditional bypass is safe on this core (SerialUSB::writebreaks immediately once the CDC TX FIFO is full with no DTR host connected; 1 s no-progress cap otherwise).Configuration.hpp,ConfigurationValidation.hpp,matrix_build.py, example config, docs): per the BTT schematic, GP20/GP21 on the SKR Pico only drive on-board MOSFETs (the HB/FAN3 power outputs and the opto-isolated Laser interface) - they are never exposed as raw GPIO on any connector, and wiring an I2C display to those power pins hangs the bus and the firmware loop (USB stays enumerated, serial goes deaf). Info-display pins now default per board (SKR Pico: GP0/GP1 on the left-side header, shared with the default GPS wiring; JackW01: GP20/GP21), and compile-time guards reject GP20/GP21 display pins on the SKR Pico as well as the display+GPS combination. The CI matrix no longer generates that combination.InfoDisplayRender.hpp,SSD1306_128x64_Display.*, all platforms): console-mode text updates now draw the header (logo + product name) on every frame instead of relying on a ~0.5 s ISR render window at the end of boot; and the FW line is uppercased - theBitmap3x5font has no lowercase glyphs and release-please's lowercasevversion prefix rendered as an out-of-range glyph.ConfigurationValidation.hpp, docs): the optional AZ/ALT*_DIAG_PINpragma messages printed on every RP2040 build; StallGuard-based homing is not supported by the firmware on these boards (Hall sensor homing via the endstop inputs is the supported path), so the checks are gone and the docs say so explicitly.DEBUG_LEVELguards use!= 0instead of> 0(semantically identical today; robust shouldDEBUG_NONEever change value).Features
Configuration_adv.hpp,b_setup.hpp): newSTEPPER_CONTROL_INTERVAL_MSknob (0.2 ms on RP2040, 0.5 ms / 2 kHz elsewhere - AVR keeps 2 kHz where higher rates disturb serial). Override in a local config to experiment.Constants.hpp,Configuration.hpp,platformio.ini): newSERIAL_BAUDRATE_FAST, selected automatically onARDUINO_ARCH_RP2040with matchingmonitor_speed. The 19200 ASCOM default exists because the 8-bit AVR platforms cannot keep up and is unchanged for them.getMountHardwareInfo()reportsRP2040(src/Mount.cpp), so:XGM#returns a populated hardware field instead of an empty one.platformio.ini,SSD1306_128x64_Display.hpp,matrix_build.py): adds the OLED library to[common_rp2xxx], compiles the display TU again (drops thesrc_filterexclusion), wraps its includes inPUSH_NO_WARNINGS/POP_NO_WARNINGS, and extends the matrix SSD1306 constraint torp2040_skr_pico.matrix_build.pygains AZ/ALT driver variants and GPS for the SKR Pico.docs/RP2040-WIRING.md): pin tables for both boards - steppers, display, GPS, Hall homing, DEC endstops, fan, power cautions.Configuration_adv.hpp):*_END_SWITCH_ACTIVE_STATEand*_ENDSWITCH_BACKSLEW_DEGare now defined unconditionally - DEC-only endstop configurations previously failed to compile on all platforms.Bench verification (SKR Pico v1.0)
:XGM#hardware infoSerialPortapps open ports) - no host-side changes required anywhereSTEPPER_CONTROL_INTERVAL_MSreverts it in one line if needed.Still to bench: Hall sensor homing (RA/DEC) and DEC endstop switches, pending sensor wiring.
🤖 Generated with Claude Code