Z Bowl Error
Z-Bowl Error (ZBE) happens when the delta model dimensions do not match actual machine dimensions. When delta model/machine dimensions differ, a planned linear movement in a Z-plane will result in a curved Z movement on the actual machine. ZBE vanishes at the X0Y0 origin and increases with distance away from the X0Y0 origin, resulting in "bowl-shaped error". ZBE can be upwards or downwards facing according to the type of error.
We can eliminate a great part of ZBE by adjusting the home angle. However, since all delta dimensions contribute to ZBE, it is not possible to eliminate ZBE with a single measurement. Here is a table that compares the ZBE at 50 mm from the X0Y0 origin for different types of dimensional error.
| Dimension | 50mm ZBE for dimension@-1% | 50mm ZBE for dimension@+1% | Notes |
|---|---|---|---|
| home angle | 0.208 | -0.208 | 1% of 360 degrees |
| gear ratio | 0.062 | -0.071 | |
| f | 0.048 | -0.059 | base triangle side |
| re | -0.049 | 0.038 | effector arm length |
| e | -0.042 | 0.032 | effector triangle side |
| rf | 0.021 | -0.033 | pulley arm length |
We use the Z-probe to measure ZBE by gathering data from a "hex-probe" data set of 8 data points:
- X0Y0 center z-probe
- z-probe at 50 mm, 0-degrees from x-axis
- z-probe at 50 mm, 60-degrees from x-axis
- z-probe at 50 mm, 120-degrees from x-axis
- z-probe at 50 mm, 180-degrees from x-axis
- z-probe at 50 mm, 240-degrees from x-axis
- z-probe at 50 mm, 300-degrees from x-axis
- X0Y0 center z-probe
Typical Z-probes can easily measure 0.050 mm of ZBE.
PREREQUISITES:
- Stepper Calibration
- [Latch Backoff Calibration](Latch Backoff)
- Z-probe operational and configured on pin specified by "syspb"
- Z-probe calibration
Home Angle Auto-Calibration (HAAC) takes about 5 minutes and will calibrate:
- home angle
- bx
- by
- bz
To execute HAAC, just enter the following commands one-by-one. Each will take about 30 seconds depending on your FPD's configured speed and acceleration time. Each step performs a hex probe and the results are aggregated with exponential averaging:
{"pgmx":"cal-fpd-home-medium"}
{"pgmx":"cal-fpd-home-medium"}
{"pgmx":"cal-fpd-home-medium"}
{"pgmx":"cal-fpd-home-fine"}
{"pgmx":"cal-fpd-home-fine"}
{"pgmx":"cal-fpd-home-fine"}Since the results are adaptive, you can change the number of commands you send and use fewer or more as you wish. Empirically, the six commands listed above will converge to a precise home angle +/- 0.3 degrees. Although precise, the home angle may not be accurate given that other dimensions may need calibration.
With SPE correction, HAAC is fairly accurate and provides a good basis for additional, camera-assisted calibration.
Once you're satisified with the HAAC values, save them to EEPROM:
{"sysas":true}
{"sysas":false}The ZBE home calibration sequence consists of six steps that collectively rely on exponential averaging to converge to stable value. The sequence is fairly robust and will handle large as well as small errors in a stable and repeatable manner. For example, the LooseCanon home angle is over 5 degrees different than the Tin Whiskers beta kit. ZBE home calibration starts out with the assumption that it is calibrating a Tin Whiskers beta kit and deduces the LooseCanon home angle from the six calibration steps. In the chart below, Home#1 refers to the initial calibration sequence that deduces the LooseCanon homing angle. Dataset Home#2 refers to a second pass that simply repeats the calibration sequence. It should converge to the same value as Home#1. The chart shows this convergence, and by the sixth calibration step, the two passes are within 2 microsteps of each other.
Auto-calibration may not work for you. We have seen cases where the Z-probe data varies by +/- 0.1mm, which confuses the algorithm, since it's looking for Z-bowl error variances at or below that magnitude.
If all else fails, calibrate home angle manually