Changes - vvuk/eddy-ng GitHub Wiki

28-Apr-2025

  • Restructured the entire page layout to be more accommodating for GitHub Wiki
  • Added more detailed instructions for Mellow Fly SHT-36v3, also applicable to other CAN toolboards with integrated LDC1612 sensors.

22-Feb-2025

  • Added HOME_Z=1 optional parameter to TAP. If set, tap will now home the Z axis to the computed tap position instead of setting a G-Code offset.
  • Support Z offset modifying and saving. After a tap (with or without HOME_Z=1), if the Z offset is manually modified, calling Z_OFFSET_APPLY_PROBE (via the save icons in Mainsail/Fluidd) will update tap_adjust_z.
  • Added max_errors configuration to allow changing the number of sequential sensor errors that are to be ignored, instead of aborting the operation. The default is 0, meaning that any sensor error will abort tap/homing (this was the previous behaviour).
  • New status values: tap_adjust_z: current tap_adjust_z value. last_tap_z: the last raw Z tap value (before adding tap_adjust_z).
  • Wiki documentation updates will follow.

11-Feb-2025

  • Fixed a bug that would cause tap to be detected later than it should. The value is the same, but the toolhead was moving more than it needed to be.
  • Properly update last_probe_value status value, which is accessible to macros after a EDDY_NG_PROBE/EDDY_NG_PROBE_STATIC command.

8-Feb-2025

  • Introduced EDDY_NG_SETUP, which attempts to set up and calibrate both homing and tap configurations automatically. If it doesn't work, the regular process can be used.
  • EDDY_NG_PROBE_STATIC with HOME_Z=1 will no longer print the probe results to the console. (This is typically used during homing, and the results are not very meaningful to view in the middle of homing.)

1-Feb-2025

  • Changed the calibration Z start to 15.0 from 5.0. This significantly improves accuracy in the 5-10mm range, which is helpful for doing QGL or Z-Tilt adjustments higher up off the build plate in case drastic adjustments are needed. It's recommended to recalibrate, but not necessary. There is no need to recalibrate a separate tap drive current, if you have one.
  • Added support for Mellow Fly boards. Use sensor_type: mellow_fly. If you already already using a Mellow Fly board with the generic ldc1612 sensor type, you must recalibrate if you switch your sensor type! Don't home Z without recalibrating!
  • Improved PROBE_EDDY_NG_PROBE_ACCURACY to do something useful and report useful results. This command this will run a probe every 0.5mm starting from Z (5.0 by default) down to 0.5mm and report results.
  • Introduced write_tap_plot boolean config, to disable writing the plot if desired. Computing the plot takes a few seconds on underpowered CPUs.
  • Fixed Already sampling! error after a failed QGL.
  • Fixed a miscalculation for LDC settle time, which could cause the sensor to occasionally incorrectly report sensor errors.
  • Fixed old-API run_probe return value (for Kalico)

28-Jan-2025

  • Made butter mode the default. scipy is not required to use it as long as the filter parameters are kept to the default, and the data rate is 250 (default) or 500. To use the previous detection mode, set tap_mode: wma, though there should be little reason to do so. If you have a tuned threshold, consider starting from the default (250) and re-tuning.

26-Jan-2025

  • Introduced tap_samples, tap_max_samples, and tap_samples_stddev to specify the number of samples to average together, the maximum number of samples to take, and the maximum standard deviation any group of tap_samples needs to have in order to be used. This can greatly increase the accuracy of tap results by throwing out outliers. Default values are 3 samples, 5 max samples, and 0.025 standard deviation.
  • Introduced tap_mode to specify different approaches for calculating the tap. Each mode has a different threshold value, so if you switch modes, either remove your threshold to use the default, or calculate it beforehand by using the TAP command with an explicit MODE and THRESHOLD. The two modes are:
    • wma: Computes tap position based on moving averages of the frequency and their derivatives. Very sensitive to noise.
    • butter: Use a Butterworth filter to compute tap position. Much more accurate than wma and less sensitive to noise, it can greatly improve tap accuracy. Recommended, and is likely to become the default soon.
  • Removed scipy_* options. These became the butter tap mode, which is now computed on the MCU.