Better Bed Probing - Optical Probe with Black Paint under PEI

Relatively speaking, David Crocker’s differential optical probe is an outstanding depth sensor, with better than 3um repeatability. Still, for optical probes, limiting the optical path is desirable. Irregular air bubbles and red material several millimeters down is not quite ideal.

To finally obtain a bed/probe combination with well-defined accuracy, paint was experimentally applied before PEI sheets. As it turns out, brake caliper paint has a nice combination of adhesion, adhesive compatibility, glossy finish, and moderately high-termperature performance.

Two complete heated beds were painted, as well as one bare borosilicate glass plate for machine calibration.

Performance already seems much better. Topology readings on TazStiff were correctable to within ~0.28mm. Previously, some of the data points would be oddly ~0.5mm out of line with nearby readings. Although the overall average measurement was generally accurate, this seemed to skew the data enough to make alginment across the entire bed to better than ~0.8mm challenging. Notably, this seemed much less problematic with the Duet electronics, which seemed to take better measurements through an analog input.

After testing to 135degC, bubbling under the PEI sheet is typical, comparable to the factory assembled bed I obtained a few months ago. Many tiny water bubbles emerged after exceeding 95C, as well as a few regions of larger bubbles near the top and right sides.

Appearance, although not generally a priority on my list, is awesome. Not only does the black matte surface look rather like chalkboard, but clearance between nozzle and bed is much easier to see.

Topology data from TazStiff (RAMBo electronics) is as follows. Especially note the corrected bed results, for which linear deformations have been compensated out.

Recv: Eqn coefficients: a: -0.00741610 b: 0.00046591 d: 0.76572694
Recv: Mean of sampled points: -0.56699995
Recv: planeNormal x: 0.007416 y: -0.000466 z: 1.000000
Recv: Bed Height Topography:
Recv: +-----------+
Recv: |...Back....|
Recv: |Left..Right|
Recv: |...Front...|
Recv: +-----------+
Recv:  +1.02200 +0.63700 +0.07700 -0.40300 -0.78800
Recv:  +0.95200 +0.47700 +0.03450 -0.45800 -0.86050
Recv:  +0.85700 +0.40200 -0.14300 -0.43800 -0.95050
Recv:  +0.69450 +0.29200 -0.08800 -0.40550 -1.02550
Recv:  +0.70700 +0.65950 -0.00550 -0.50050 -0.74550
Recv: Corrected Bed Height vs. Bed Topology:
Recv:  +0.23525 +0.28779 +0.16533 +0.12287 +0.17541
Recv:  +0.19600 +0.15854 +0.15358 +0.09862 +0.13366
Recv:  +0.13175 +0.11429 +0.00683 +0.14937 +0.07440
Recv:  +0.00000 +0.03504 +0.09258 +0.21261 +0.03015
Recv:  +0.04325 +0.43329 +0.20582 +0.14836 +0.34090
Recv: Bed Level Correction Matrix:
Recv: +0.999973 +0.000000 -0.007416
Recv: +0.000003 +1.000000 +0.000466
Recv: +0.007416 -0.000466 +0.999972

From TazMega (Duet electronics), topology has not been thoroughly re-measured yet, but the repeatability of four-corners measurements against the calibration glass is generally within 3um.

Bed probe heights: -0.100 -0.138 0.135 0.092, mean -0.003, deviation from mean 0.118
Bed probe heights: -0.097 -0.138 0.133 0.092, mean -0.002, deviation from mean 0.117
Bed probe heights: -0.090 -0.140 0.135 0.092, mean -0.001, deviation from mean 0.117