You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Both use sense_resistor 0.1. COSMOS also specifies X/Y home_current 0.8; hold_current and home_current serve different purposes.
Our first-hand stock CC2 notes from firmware 01.03.02.36 independently confirm live X/Y run_current 1.2 and hold_current 0.8. The published Elegoo file has an older 01.01.13.10 header.
The strongest candidates for testing are X/Y run_current 1.2 and extruder run_current 0.8. #387 reports improved extrusion at 0.8 through CANVAS, Elegoo's multicolor feeder, but that test was on CC1. Z run_current 1.0 also has stock provenance, without evidence here of a COSMOS Z torque problem.
Adding stock hold_current values deserves separate validation. Klipper's driver guidance generally recommends omitting hold_current because current transitions can cause unintended motion. Stock provenance alone does not establish that adding it improves COSMOS.
I would leave chopper registers, CoolStep automatic current scaling, homing current and Z mode unchanged pending measurements. COSMOS uses full-time Z StealthChop and driver_SGTHRS 40 (stall-detection sensitivity); stock leaves StealthChop disabled and uses 120. Both configure a physical Z endstop, so this sensitivity difference does not establish a Z homing defect.
Our stock-firmware X/Y tests produced loud growl at 20000 mm/s^2 with full-time StealthChop. A 50 mm/s resonance occurred in both modes; hybrid mode was tested without establishing a general improvement. Homing succeeded with StealthChop configured, which does not validate COSMOS homing or its current transitions.
No proposed change has been hardware-tested under COSMOS. Higher currents increase motor/driver heat. StallGuard sensitivity depends on current, temperature and load; retain home_current 0.8 initially, but verify its application and restoration before assuming homing is unaffected.
These values could support an explicitly untested stock-baseline PR with maintainer hardware validation. They are insufficient to claim #359 is fixed.
Sounds good! The CC1 and CC2 have the same steppers and drivers so the tuning should be identical between them. Please let us know results of your testing 🙂
Thanks for #458. For the record, after it the CC2 run currents are still the CC1 values (X/Y 1.0, Z 0.8, E 0.7 vs stock 1.2 / 1.0 / 0.8), and X/Y still carry the CC1 chopper and CoolStep block. The extruder side looks like the same fault as #387 (0.7 A under-extrudes, 0.8 fixes it). I plan to test the stock currents on my own COSMOS CC2 (DUMP_TMC before/after, repeated homing, motor temps, noise) and will open a separate PR with the results if they hold up, rather than reopen this issue.
First results from my CC2 on COSMOS 26.10.1, as promised. Early data, one printer.
Override tested (printer.cfg, above SAVE_CONFIG): run_current X/Y 1.2, Z 1.0, E 0.8 (Elegoo stock values), and on X/Y the stock chopper values (TBL 2, TOFF 3, HSTRT 5, HEND 0, IHOLDDELAY 8) with CoolStep off (SEMIN 0). No hold_current, no stealthchop_threshold on X/Y/E. Z keeps the 26.10.1 (#458) settings.
DUMP_TMC before (26.10.1 as shipped): X/Y run 0.994 A, CHOPCONF 340281a1 (toff 1, hstrt 2, hend 3, tbl 1), COOLCONF 0000c462 (semin 2). Z 0.795 A, E 0.696 A.
DUMP_TMC after: X/Y run 1.179 A, CHOPCONF 34010053 (toff 3, hstrt 5, hend 0, tbl 2), COOLCONF 0. Z 0.994 A, E 0.795 A. All en_spreadcycle=1, TPWMTHRS fffff.
Homing repeatability: G28 x3 cold, GET_POSITION mcu steps at X128 Y128: spread X 2, Y 15, Z 5 microsteps, so all within one full step (Y only just).
Prints with the override: a 10-layer 60x60 wall tower at accel 20000 (no layer shift, G28 agreed afterwards), then a 37 min PLA part. Both finished fine. Motors felt warm but not hot after the tower.
Not done yet: measured motor/driver temperatures at steady state, noise recordings before/after, a long print. I will add those here, and a PR with the currents only (marked untested on other printers) if they hold up.
Activity
jamesturton commentedon Sep 27, 2026
Also check
run_currentBorjaLL commentedon Oct 31, 2026
Compared COSMOS CC2 machine.cfg with Elegoo's CC2 printer_dsp.cfg.
Configured currents in amperes, run/hold:
Both use sense_resistor 0.1. COSMOS also specifies X/Y home_current 0.8; hold_current and home_current serve different purposes.
Our first-hand stock CC2 notes from firmware 01.03.02.36 independently confirm live X/Y run_current 1.2 and hold_current 0.8. The published Elegoo file has an older 01.01.13.10 header.
The strongest candidates for testing are X/Y run_current 1.2 and extruder run_current 0.8. #387 reports improved extrusion at 0.8 through CANVAS, Elegoo's multicolor feeder, but that test was on CC1. Z run_current 1.0 also has stock provenance, without evidence here of a COSMOS Z torque problem.
Adding stock hold_current values deserves separate validation. Klipper's driver guidance generally recommends omitting hold_current because current transitions can cause unintended motion. Stock provenance alone does not establish that adding it improves COSMOS.
I would leave chopper registers, CoolStep automatic current scaling, homing current and Z mode unchanged pending measurements. COSMOS uses full-time Z StealthChop and driver_SGTHRS 40 (stall-detection sensitivity); stock leaves StealthChop disabled and uses 120. Both configure a physical Z endstop, so this sensitivity difference does not establish a Z homing defect.
Our stock-firmware X/Y tests produced loud growl at 20000 mm/s^2 with full-time StealthChop. A 50 mm/s resonance occurred in both modes; hybrid mode was tested without establishing a general improvement. Homing succeeded with StealthChop configured, which does not validate COSMOS homing or its current transitions.
No proposed change has been hardware-tested under COSMOS. Higher currents increase motor/driver heat. StallGuard sensitivity depends on current, temperature and load; retain home_current 0.8 initially, but verify its application and restoration before assuming homing is unaffected.
These values could support an explicitly untested stock-baseline PR with maintainer hardware validation. They are insufficient to claim #359 is fixed.
jamesturton commentedon Nov 1, 2026
Fixed in #458
jamesturton commentedon Nov 1, 2026
Sounds good! CC1 and CC2 use the same steppers and drivers so they should be tuned identical. Please let us know the results from your testing 🙂
jamesturton commentedon Nov 1, 2026
Sounds good! The CC1 and CC2 have the same steppers and drivers so the tuning should be identical between them. Please let us know results of your testing 🙂
BorjaLL commentedon Nov 1, 2026
Thanks for #458. For the record, after it the CC2 run currents are still the CC1 values (X/Y 1.0, Z 0.8, E 0.7 vs stock 1.2 / 1.0 / 0.8), and X/Y still carry the CC1 chopper and CoolStep block. The extruder side looks like the same fault as #387 (0.7 A under-extrudes, 0.8 fixes it). I plan to test the stock currents on my own COSMOS CC2 (DUMP_TMC before/after, repeated homing, motor temps, noise) and will open a separate PR with the results if they hold up, rather than reopen this issue.
BorjaLL commentedon Nov 1, 2026
First results from my CC2 on COSMOS 26.10.1, as promised. Early data, one printer.
Override tested (printer.cfg, above SAVE_CONFIG): run_current X/Y 1.2, Z 1.0, E 0.8 (Elegoo stock values), and on X/Y the stock chopper values (TBL 2, TOFF 3, HSTRT 5, HEND 0, IHOLDDELAY 8) with CoolStep off (SEMIN 0). No hold_current, no stealthchop_threshold on X/Y/E. Z keeps the 26.10.1 (#458) settings.
DUMP_TMC before (26.10.1 as shipped): X/Y run 0.994 A, CHOPCONF 340281a1 (toff 1, hstrt 2, hend 3, tbl 1), COOLCONF 0000c462 (semin 2). Z 0.795 A, E 0.696 A.
DUMP_TMC after: X/Y run 1.179 A, CHOPCONF 34010053 (toff 3, hstrt 5, hend 0, tbl 2), COOLCONF 0. Z 0.994 A, E 0.795 A. All en_spreadcycle=1, TPWMTHRS fffff.
Homing repeatability: G28 x3 cold, GET_POSITION mcu steps at X128 Y128: spread X 2, Y 15, Z 5 microsteps, so all within one full step (Y only just).
Prints with the override: a 10-layer 60x60 wall tower at accel 20000 (no layer shift, G28 agreed afterwards), then a 37 min PLA part. Both finished fine. Motors felt warm but not hot after the tower.
Not done yet: measured motor/driver temperatures at steady state, noise recordings before/after, a long print. I will add those here, and a PR with the currents only (marked untested on other printers) if they hold up.