Delta calibration bed level 32bits board marlin 2.0

For more then one year i've running my delta machine an anycubic linear plus with a version of marlin 1.1.9 running in a 8bits board without any problems at all, either auto calibration or leveling the bed run flawlessly and i'm able to get a first perfect layer with everything align. These delta machine had a probe that is a switch can bed mounted in place with a magnet when needs to run auto calibration or bed level procedure, when it runs autocalibration its get some delta parameters like delta radius, DELTA_HEIGHT, DELTA_ENDSTOP_ADJ, DELTA_TOWER_ANGLE_TRIM etc.

a few weeks ago i've update the board to a 32bits (SKR 1.3), tmc 2130 SPI, and marlin 2.0.0 bug fix version, since then i wont be able to print anything, cauze the bed its never leveled, I'm running the procedure for auto calibration, everything seems pretty identical to what i'm getting in previous version, the bed level routine give me a mesh that seems to pretty flat with minimal variances between the probes (compared to my hypercube mesh seems more flat), I've set the Z offset a ton of times on bed center. But everytime It start to print first layer seems okay in same parts and other parts are to close or to far from the bed, the print never sticks to the glass.

I'm trying for two weeks fix that problem and start to print again, i've active define RESTORE_LEVELING_AFTER_G28 also activate bed level debug, everting looks okay to me. I'm already order a new probe sensor High Precision Z-Probe to try getting more accuracy results.

Is there something I miss? Is there any problem with marlin? One of the thing that marlin 2.0.0 claims to have improved is delta bed level procedure. Should I forget auto bed level and doing always manual level? For about 2 years I'm able to run bed level without any problems.

When did you download this Marlin version? About two months ago there was a bugfix version in which the Delta settings were broken. If you downloaded it about 1,5 to 2 months ago I would first try to update Marlin to the latest bugfix.

I have since upgrading to the newest Marlin just simply ditched relying purely on G33 leveling and use UBL (unified bed leveling) with one time manual refinement of the measurements. So I do one cold probing set and then fine tune and save it to slot 1. Now when I run a print I have added 'G29 L1' into the startup gcode after G28 so it uses the mesh. Works flawlessly every time. My bed isn't exactly flat and UBL has saved me. It prints flat on 100% of the printable surface.

Somehow G26 isn't working for me, must be another issue with my Marlin version for Delta. They have been messing it up a little bit in the past couple of months with the Delta builds but fixes are done lightning fast, props to the Marlin dev team.

---

http://www.marinusdebeer.nl/

Quote
Ohmarinus
When did you download this Marlin version? About two months ago there was a bugfix version in which the Delta settings were broken. If you downloaded it about 1,5 to 2 months ago I would first try to update Marlin to the latest bugfix.

I have since upgrading to the newest Marlin just simply ditched relying purely on G33 leveling and use UBL (unified bed leveling) with one time manual refinement of the measurements. So I do one cold probing set and then fine tune and save it to slot 1. Now when I run a print I have added 'G29 L1' into the startup gcode after G28 so it uses the mesh. Works flawlessly every time. My bed isn't exactly flat and UBL has saved me. It prints flat on 100% of the printable surface.

Somehow G26 isn't working for me, must be another issue with my Marlin version for Delta. They have been messing it up a little bit in the past couple of months with the Delta builds but fixes are done lightning fast, props to the Marlin dev team.

thanks for your reply, my version of marlin have almost one month yes, i also notice that they have lauch a new realase of marlin 2.0.1 few days ago [github.com], i'll update my version of marlin to that "stable" version and copy my configurations. I've read about UBL bed level for now i always use bilinear bed level, i'll give a try on that and check if changing some point manually it helps to leveling the bed. Even using UBL bed level it necessary to run auto calibration to get mesures of delta printer right?

I'm also order a precision Z-probe in order to read data in a more accuracy way, let's hope it help level the bed. G26 its for print a mesh validation object in order to manually fix some points in the bed that not good enough right? Its not running well on marlin 2.0.1 for deltas?

Mesh bed levelling will work always provided:

• your printer is repeatable (does not have any unwanted play)
• the z-probe works well
• your version of the firmware is not buggy

The disadvantage of mesh bed levelling is that it does not fix dimensional accuracy. It only makes your first layer stick.

Quote
hercek
Mesh bed levelling will work always provided:

• your printer is repeatable (does not have any unwanted play)
• the z-probe works well
• your version of the firmware is not buggy

The disadvantage of mesh bed levelling is that it does not fix dimensional accuracy. It only makes your first layer stick.

What do you mean by that? The process of calibrate a delta its also ajusting the endstop, tilt the bed etc. you're saying its not recomendable calibrate the delta via firmware?

There are two different kinds of calibrations:

• simple mesh based levelling: this does not try to fix endstops, tower positions, tower tilt, or diagonal rod length; instead it just measures z-error in a regular mesh and adds a linearly interpolated correction to the z-position based on the measurement so that one layer is planar; this method does not do anything to fix YX error
• calibration (often based on root-mean-square error minimization) which measures bed in a given number of points (it can be (and often is) a different shape than a regular mesh) and tries to compute the correct endstops, tower positions, tower tilt, or diagonal rod length so that the one layer is planar; this method typically fixes also XY error

Quote
hercek
There are two different kinds of calibrations:

• simple mesh based levelling: this does not try to fix endstops, tower positions, tower tilt, or diagonal rod length; instead it just measures z-error in a regular mesh and adds a linearly interpolated correction to the z-position based on the measurement so that one layer is planar; this method does not do anything to fix YX error
• calibration (often based on root-mean-square error minimization) which measures bed in a given number of points (it can be (and often is) a different shape than a regular mesh) and tries to compute the correct endstops, tower positions, tower tilt, or diagonal rod length so that the one layer is planar; this method typically fixes also XY error

Thanks for your explanation. My question is how those two method are implements in firmware like marlin? What is the steps to achieve that? My routine to level and calibation the bed is first, run auto calibration (second method you mention) these I tink correct endstops height, tower position etc.. and then bed level routine which I think it's the first, it only create a mesh of Z-position to achive interpolate the position for creating the first layer. It's that?

AFAIK, only RepRap firmware on a 32-bit Duet board implements the 2nd method in firmware. You need to do it semi-manually with other firmwares using e.g. this.

The 1st method is implemented in Marlin. I do not know it's state now. I used Marlin the last time about 4 years ago. Mesh based compensation was barely introduced at that time.

Does RRF run on a SKR v1.3?

---

http://www.marinusdebeer.nl/

Quote
Ohmarinus
Does RRF run on a SKR v1.3?

Yes, [reprap.org].

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
hercek
AFAIK, only RepRap firmware on a 32-bit Duet board implements the 2nd method in firmware. You need to do it semi-manually with other firmwares using e.g. this.

The 1st method is implemented in Marlin. I do not know it's state now. I used Marlin the last time about 4 years ago. Mesh based compensation was barely introduced at that time.

Unfortunately my board its only compatibility with marlin! its a bigtreetech skr v1.3 but i'm glad to know about these methods. I'm updated now for marlin 2.0.1 and using a FSR sensor to level bed I hope its work well.

Quote
masterviana

Quote
hercek
AFAIK, only RepRap firmware on a 32-bit Duet board implements the 2nd method in firmware. You need to do it semi-manually with other firmwares using e.g. this.

The 1st method is implemented in Marlin. I do not know it's state now. I used Marlin the last time about 4 years ago. Mesh based compensation was barely introduced at that time.

Unfortunately my board its only compatibility with marlin! its a bigtreetech skr v1.3 but i'm glad to know about these methods. I'm updated now for marlin 2.0.1 and using a FSR sensor to level bed I hope its work well.

Apparently it runs on an SKR, see my question. Also, I have had poor results with an FSR.. Sadly.

---

http://www.marinusdebeer.nl/