BLTouch - M558, Accuracy test & strange behaviour

3Dreamer

Hi all,
in order to check the accuracy of the BLTouch probe mounted on my machine I've run some bed probing sessions while using two different BLTouch settings (M558).
Specifically:

Bed size: 800 x 400 mm
Number of probe points: 25
Probe area: 2100 cm²
Bed temperature: 45 °C
BLTouch settings 01: M558 P9 F200 H5 T8000 B0
BLTouch settings 02: M558 P9 F150 H5 R0.3 T8000 A5 B1

I've run 10 probing session using "Settings 01" and 10 probing session using "Settings 02", totally collecting 500 measures.

So I've calculated the Standard Deviation for each probed point (for the same Setting) and subsequently the arithmetic average between the standard deviations values, obtaining those results:

For Settings 01: 0.0756
For Settings 02: 0.0821

Aside from the big absolute values (not so good in order to print a 0.1 mm first layer height) I've noticed a couple of "strange" things:

Ideally, slowing down the probing speed ("F") while adopting the parameters "R", "A" and "B1" should have led to a better results, but it does not seem so.
The 3D physical map of the surface of the bed follows the 3D map of the Standard Deviation values for each point, as if the BLTouch instrumental error depends by the travel made by the BLTouch pin while probing!
Here attached you will find a graphic representation about this.

I realize that 10 bed probing for each Setting will not produce good representative statistical samples, anyway your considerations about the results will be will be much appreciated!

Thank you!

Veti

for repeatability checking with sd calculation see
https://forum.duet3d.com/topic/6962/m48-measure-z-probe-repeatability-and-print-to-serial-output/4

is it a genuine bltouch? the clones are know to be less accurate.
are you probing with heater/fan on? the bltouch is prone to interference.

3Dreamer

@Veti
Thank you Veti, I'll do the repeatability test that you pointed out.
BLTouch is genuine. Heaters/Fan OFF. About the bed, as you can see from the two M558 Settings that I've posted, I've run tests using B0 and B1 commands.

Phaedrux

I've found reducing my Z jerk and acceleration and speed for the duration of probing has helped with repeatability. Perhaps give that a test?