MBL print area #4183

leptun · 2023-05-01T10:18:41Z

Similar implementation of the print area UBL from the Buddy firmware, but instead of using M555 for defining the print area, G80 uses additional parameters. The G80 print area is defined with X, Y, W and H, just like with M555. If the print area is not specified, the whole bed is probed. Gcode future and backward compatibility is maintained since old firmware will ignore these extra parameters.

With this PR, if a print area is provided, the 7x7 MBL will only probe those points. This results in a quicker, precise MBL where precision is needed. Besides the print area probed points, the rest of the 7x7 mesh uses interpolation of the data sorted in eeprom during Z calibration so that the toolhead can safely exit the print area and not cause a nozzle crash with a warped bed.

Tasks after merge

Update rep-rap Gcode documentation
Update PrusaSlicer profiles
Update translation (kill(PSTR("Mesh bed leveling failed. Please run Z calibration."));)
Add O in G80 [ N | R | L | R | F | B | X | Y | W | H ] list

IMG_1223.MOV

IMG_1227.MOV

In case the 3x3 mesh in eeprom from Z calibration is missing/invalid, the 3x3 points are also probed besides the print area:

IMG_1221.MOV

Another interesting feature enabled by this PR is the ability to load and activate the mesh stored in eeprom without probing any point ~~(besides point 0)~~. This can be achieved with G80 W since it results in the print area being completely outside the bed.

Closes #3224

leptun · 2023-05-03T11:03:11Z

In order to use the feature, change the G80 call in the slicer startup script to this:

G80 X{first_layer_print_min[0]} Y{first_layer_print_min[1]} W{(first_layer_print_max[0]) - (first_layer_print_min[0])} H{(first_layer_print_max[1]) - (first_layer_print_min[1])}; mesh bed leveling

MBL 7x7 must also be enabled on the printer. Otherwise, the entire bed will be probed with 3x3.

Firmware/Marlin_main.cpp

leptun · 2023-05-05T17:52:23Z

I pushed a lot of changes to the PR. I fixed the point counter behavior. It now computes in advance how many points will be probed, so it no longer skips the counter on the status screen.
I also improved a lot the eeprom mesh code and the code that determines the approximate range where the current probed point should be located. I am now using the upsampled eeprom mesh (in the past it was just hardcoded to current_position=0.f for all points not on the 3x3 mesh for some weird reason). So now I finally fixed the issue where certain warped beds have lots and lots and lots of rechecks on certain points. This was especially an issue on MK2.5(S) since the bed there was generally more warped.
I also improved how the Z is lifted to the next point. Previously it was computed relative to the height of the last measured point. Now I compute it based on the upsampled mesh from eeprom. This results in a much more consistent MBL on warped beds, while still being super fast.
I also no longer have to probe the first point. Only the print area needs to be probed now. So G80 W doesn't probe any point now, but this time for real 🙂. I tested it on my machine with a calibrated eeprom mesh and the first layer was still spot on.

Commands I use for testing:

G80 R1 N3                       ; regular 3x3 MBL
G80 R1 N7                       ; regular 7x7 MBL
G80 R3 N7 W                     ; load eeprom MBL. No probing
G80 R3 N7 R1 X20 Y10 W10 H10    ; Small 2x2 mesh near origin
G80 R3 N7 X115 Y91 W32 H24      ; Small 3x3 mesh at the center of the bed
G80 R3 N7 X100 Y80 W70 H70      ; Small 4x4 mesh near the center of the bed
G80 R3 N7 X70 Y50 W109 H109     ; Small 5x5 mesh at the center of the bed

Useful tool for visualizing the bed data. Dump the mesh with G81. Beware that the Y axis is flipped:
http://lokspace.eu/3d-printer-auto-bed-leveling-mesh-visualizer/

leptun · 2023-05-05T23:49:26Z

Marked it as ready for review. Don't really have anything else I'd like to change besides maybe some optimizations here and there. Functionality wise, everything seems to be working fine. I still need to test how MBL performs with no Z calibration data. All other tests were performed on two printers: MK3S+_MMU3 and MK2.5_MMU2SR.

leptun · 2023-05-06T08:06:21Z

Up to date video of a centered small 3x3 mesh. Counter works fine, first point is no longer measured. Still a bit disappointed that the superpinda needs to slow down on certain points due to the higher hysteresis. Pinda V2 would not have those slowdowns as is the case on the MK2.5 printer.

IMG_1233.MOV

ojrik · 2023-05-07T22:38:08Z

Hello. Very nice work. It would be good to remove the return to Home, at the end of leveling.

leptun · 2023-05-07T22:56:43Z

@ojrik Thank you. I need to keep the return to origin for backwards compatibility with old gcodes. But I understand it could be a useful setting to disable somehow if the purge line is printed next to the object. I'm thinking I could add an "O<0,1>" parameter to specify if return to origin is needed after G80, with a default of 1. Still need to think about the implications of this.

ojrik · 2023-05-07T23:02:07Z

Nice thought.
I have this PR applied in Firmware 3.12.2, it works well on MK2.5S + MMU2S, MK3S + MMU2S.

leptun · 2023-05-08T07:22:21Z

@ojrik I added the O parameter like I described in my previous comment. You may now add O0 to the G80 command to disable the return to origin.

I have this PR applied in Firmware 3.12.2, it works well on MK2.5S + MMU2S, MK3S + MMU2S.

I'm surprised the PR can be rebased so easily onto 3.12. In any case this PR is for 3.14 at the earliest since the 3.13 release is already packed full of changes. Thank you also for testing the PR on your printers. Feedback is greatly appreciated. Do you mind if you also post the G81 output from both printers of the commands listed in this comment? #4183 (comment). I want to see how closely the mesh saved in eeprom matches the real hardware. Also, did you perform Z/XYZ calibration recently on any of those printers?

ojrik · 2023-05-08T08:21:37Z

Do you mean something like this?
The XYZ calibration was about a month ago when I mounted the Super Pinda on the MK3S MMU2S.
Test MBL.txt

leptun · 2023-05-08T08:28:56Z

Yes, something like that. I'm also interested in G80 O0 N7 W. This should load the Z calibration mesh from eeprom. This mesh data was stored there during Z calibration. The Z calibration is different than MBL and is performed at the end of XYZ calibration or manually via the menu ("Calibration->Calibrate Z"). That calibration probes 9 MBL points and stores them in eeprom. That data is later used by G80 while probing to validate measurements. With this PR, we rely much more on this eeprom data than in the past, so it's critical that the data in eeprom is correct.
Since you ran the XYZ calibration a month ago, the data in eeprom should be quite fresh still. Unless you did some work on the printer hardware in the meantime, it should still be fine.

ojrik · 2023-05-08T08:40:26Z

Zadal som príkaz G80 N7 O0 W, potom G81 my vyšlo toto.

Recv: 0.12000 0.11556 0.11222 0.11000 0.10889 0.10889 0.11000
Recv: 0.10556 0.10296 0.10037 0.09778 0.09519 0.09259 0.09000
Recv: 0.09889 0.09556 0.09185 0.08824 0.08333 0.07852 0.07333
Recv: 0.10000 0.09333 0.08667 0.08000 0.07380 0.06667 0.06000
Recv: 0.10889 0.09630 0.08481 0.07528 0.06519 0.05704 0.05000
Recv: 0.12556 0.10444 0.08630 0.07111 0.05889 0.04963 0.04333
Recv: 0.15000 0.11778 0.09111 0.07000 0.05444 0.04444 0.04000

leptun · 2023-05-08T08:44:41Z

Wow, your printer is considerably straighter than mine :).
Were the measurements in Test.MBL.txt done while the printer is hot or cold?

ojrik · 2023-05-08T08:47:05Z

cold

ojrik · 2023-05-08T08:51:50Z

I use Prusa Leveling Guide in Octoprint to set BED.

ojrik · 2023-05-08T09:08:39Z

MK2.5S Test
Test MBL1.txt

leptun · 2023-05-08T09:12:08Z

Oy. That's one bent bed 😅. Does this PR help with rechecks during MBL? Both 7x7 and 3x3. In my case on the MK2.5, the points on the last row always had to be rechecked. With this PR, those issues were gone for me. Is this also the case for you?

ojrik · 2023-05-08T09:25:45Z

I rarely use MK2.5S, I just tested your PR with FIRMWARE 3.12.2, FW 3.13.0, I don't have FW for MMU2S

leptun · 2023-05-08T09:37:23Z

If all goes well, the new MMU firmware should be released at the end of this month. You can still use FW 3.13.0 on your printer, but you have to disable the MMU in settings

ojrik · 2023-05-08T10:02:01Z

I know that, I tried that too, I have it adjusted according to this PR.

ojrik · 2023-05-08T10:24:59Z

I use Visual Studio Code. With FW3.13.0 I have some problems compiling. MK2.5S date 1970-01-01, languages not applicable. MK3S also date, and some small things in FW. But that's a small point.

leptun · 2023-05-08T11:15:14Z

The compile date is intentional. It's done like this so that the FW build is consistent, so if you build now and again later, the same compiled hex will be output. The build server should use the correct time and date, so those builds are not affected.
MK2.5S language not applicable: which cmake target are you using for building?
What other fw issues are you encountering with the mk3s?

ojrik · 2023-05-08T12:51:19Z

Here is my Test MK2.5S FW3.12.2
Fantastic, because it's greasy and the bed is so bent.

Change in memory: Flash: -12 bytes SRAM: 0 bytes

Those constrains on t and s were useless. flash: -294

flash: -112

a

flash: -6 sram: 0? (somehow)

Flash: +80 SRAM: +1

github-actions · 2023-08-17T05:57:02Z

Target	ΔFlash (bytes)	ΔSRAM (bytes)
`MK3S_MULTILANG`	-264	-1
`MK3_MULTILANG`	-278	-1

3d-gussner

Tested on MK404 and real printer

leptun · 2023-09-13T08:36:46Z

Firmware/Marlin_main.cpp

@@ -4957,29 +4858,26 @@ void process_commands()
    Default 3x3 grid can be changed on MK2.5/s and MK3/s to 7x7 grid.
    #### Usage

-          G80 [ N | R | V | L | R | F | B ]
+          G80 [ N | R | L | R | F | B | X | Y | W | H ]


Just noticed I forgot to add O here

Clean up mesh definitions that where made obsolete by prusa3d#4183

sarusani · 2023-09-14T10:10:07Z

@leptun
I noticed that when I use "G80 C0 N3" the bed gets probed once at each point as expected. (1 probe, 0 repeats) But G81 wont return any data.

Log with "G80 C0 N3":

Send: G80 C0 N3
echo:Enqueing to the front: "G28 W"
Recv: tmc2130_home_enter(axes_mask=0x01)
Recv:   0 step=30 mscnt= 480
Recv: tmc2130_goto_step 0 61 2 1000
Recv: tmc2130_home_exit tmc2130_sg_homing_axes_mask=0x01
Recv: tmc2130_home_enter(axes_mask=0x02)
Recv: echo:busy: processing
Recv:   0 step=18 mscnt= 288
Recv: tmc2130_goto_step 1 33 2 1000
Recv: tmc2130_home_exit tmc2130_sg_homing_axes_mask=0x02

Send: G81
Recv: Num X,Y: 7,7
Recv: Z search height: 5.0f
Recv: Measured points:
Recv:   0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000
Recv:   0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000
Recv:   0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000
Recv:   0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000
Recv:   0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000
Recv:   0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000
Recv:   0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000

Log with "G80 C1 N3":

Send: G80 C1 N3
Recv: echo:Enqueing to the front: "G28 W"
Recv: tmc2130_home_enter(axes_mask=0x01)
Recv:   0 step=29 mscnt= 479
Recv: tmc2130_goto_step 0 61 2 1000
Recv: tmc2130_home_exit tmc2130_sg_homing_axes_mask=0x01
Recv: tmc2130_home_enter(axes_mask=0x02)
Recv:   0 step=18 mscnt= 290
Recv: tmc2130_goto_step 1 33 2 1000
Recv: tmc2130_home_exit tmc2130_sg_homing_axes_mask=0x02

Send: G81
Recv: Num X,Y: 7,7
Recv: Z search height: 5.0f
Recv: Measured points:
Recv:   0.14750  0.20333  0.24167  0.26250  0.26583  0.25167  0.22000
Recv:   0.14167  0.19525  0.22775  0.23917  0.22951  0.19877  0.14694
Recv:   0.13917  0.18654  0.21127  0.21333  0.19275  0.14951  0.08361
Recv:   0.14000  0.17722  0.19222  0.18500  0.15556  0.10389  0.03000
Recv:   0.14417  0.16728  0.17062  0.15417  0.11793  0.06191  -0.01389
Recv:   0.15167  0.15673  0.14645  0.12083  0.07988  0.02358  -0.04806
Recv:   0.16250  0.14556  0.11972  0.08500  0.04139  -0.01111  -0.07250

leptun · 2023-09-14T10:25:00Z

C0 is not a valid parameter. Only C >= 1 is allowed. The first probing try is always discarded. That's why C1 probes twice. That initial probe is used for detecting deviations only afaik. In any case, this PR didn't touch that part of the code.

sarusani · 2023-09-14T10:27:59Z

👍 Thanks for the info!

🚧 Add new MBL message for #4183 and translation

leptun · 2023-11-05T09:37:13Z

Fyi, PrusaSlicer 2.7.0-alpha1 now has the updated startup gcode for the MK3/S/+ and MK2.5/S variants to make use of the MBL print area feature.

leptun force-pushed the mbl_print_area branch 3 times, most recently from ff558d2 to 75d5d56 Compare May 5, 2023 10:40

gudnimg reviewed May 5, 2023

View reviewed changes

Firmware/Marlin_main.cpp Outdated Show resolved Hide resolved

leptun changed the title ~~🚧 MBL print area~~ MBL print area May 5, 2023

leptun marked this pull request as ready for review May 5, 2023 23:47

leptun force-pushed the mbl_print_area branch from b9f4310 to 9bf0c0a Compare May 7, 2023 16:54

leptun force-pushed the mbl_print_area branch from db68c04 to 2e6c0d9 Compare May 8, 2023 06:31

gudnimg and others added 8 commits August 17, 2023 07:52

optimisation: Move divison into constexpr

a9d0cc5

Change in memory: Flash: -12 bytes SRAM: 0 bytes

Remove unused code

42c27ca

Fix Z calibration points

dd1bde7

mbl.get_z() optimizations

aba0450

Those constrains on t and s were useless. flash: -294

Combine BED_X/Y() and mbl::get_x/y()

a15f536

flash: -112

G80 O parameter

a984b2e

a

Do not store a global mbl_z_probe_nr

7740a81

flash: -6 sram: 0? (somehow)

Add a limit to how many times G80 can fail

f5f09f1

Flash: +80 SRAM: +1

leptun force-pushed the mbl_print_area branch from a9cb87a to 0cd376f Compare August 17, 2023 05:55

leptun added 2 commits August 17, 2023 08:02

Make the G80 Z threshold tighter

45823e7

Update error messages

ede4cc5

leptun force-pushed the mbl_print_area branch from 0cd376f to ede4cc5 Compare August 17, 2023 06:03

3d-gussner added the Mesh bed leveling label Sep 1, 2023

3d-gussner approved these changes Sep 13, 2023

View reviewed changes

3d-gussner merged commit 2e70697 into prusa3d:MK3 Sep 13, 2023

leptun commented Sep 13, 2023

View reviewed changes

leptun deleted the mbl_print_area branch September 13, 2023 08:37

sarusani added a commit to sarusani/Prusa-Firmware that referenced this pull request Sep 13, 2023

Remove obsolete mesh definitions

dc129e0

Clean up mesh definitions that where made obsolete by prusa3d#4183

sarusani mentioned this pull request Sep 13, 2023

Remove obsolete mesh definitions #4385

Merged

sarusani added a commit to sarusani/Prusa-Firmware that referenced this pull request Sep 13, 2023

Remove obsolete mesh definitions

cdadde1

Clean up mesh definitions that where made obsolete by prusa3d#4183

3d-gussner added a commit to 3d-gussner/Prusa-Firmware that referenced this pull request Sep 13, 2023

Add new MBL message for prusa3d#4183 and translate these messages

ca27841

sarusani added a commit to sarusani/Prusa-Firmware that referenced this pull request Sep 13, 2023

Remove obsolete mesh definitions

29194d1

Clean up mesh definitions that where made obsolete by prusa3d#4183

AttilaSVK mentioned this pull request Sep 13, 2023

Add new MBL message for #4183 and translation #4386

Merged

13 tasks

3d-gussner added the Post release tasks open Tasks after release label Sep 13, 2023

3d-gussner added a commit that referenced this pull request Sep 15, 2023

Merge pull request #4386 from 3d-gussner/MK3_mbl_print_area_translations

9e794a4

🚧 Add new MBL message for #4183 and translation

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

MBL print area #4183

MBL print area #4183

leptun commented May 1, 2023 •

edited

Loading

leptun commented May 3, 2023 •

edited

Loading

leptun commented May 5, 2023 •

edited

Loading

leptun commented May 5, 2023

leptun commented May 6, 2023

ojrik commented May 7, 2023

leptun commented May 7, 2023

ojrik commented May 7, 2023 •

edited

Loading

leptun commented May 8, 2023

ojrik commented May 8, 2023

leptun commented May 8, 2023

ojrik commented May 8, 2023

leptun commented May 8, 2023

ojrik commented May 8, 2023

ojrik commented May 8, 2023

ojrik commented May 8, 2023

leptun commented May 8, 2023

ojrik commented May 8, 2023

leptun commented May 8, 2023

ojrik commented May 8, 2023

ojrik commented May 8, 2023

leptun commented May 8, 2023

ojrik commented May 8, 2023

github-actions bot commented Aug 17, 2023 •

edited

Loading

3d-gussner left a comment

leptun Sep 13, 2023

sarusani commented Sep 14, 2023 •

edited

Loading

leptun commented Sep 14, 2023

sarusani commented Sep 14, 2023

leptun commented Nov 5, 2023

MBL print area #4183

MBL print area #4183

Conversation

leptun commented May 1, 2023 • edited Loading

leptun commented May 3, 2023 • edited Loading

leptun commented May 5, 2023 • edited Loading

leptun commented May 5, 2023

leptun commented May 6, 2023

ojrik commented May 7, 2023

leptun commented May 7, 2023

ojrik commented May 7, 2023 • edited Loading

leptun commented May 8, 2023

ojrik commented May 8, 2023

leptun commented May 8, 2023

ojrik commented May 8, 2023

leptun commented May 8, 2023

ojrik commented May 8, 2023

ojrik commented May 8, 2023

ojrik commented May 8, 2023

leptun commented May 8, 2023

ojrik commented May 8, 2023

leptun commented May 8, 2023

ojrik commented May 8, 2023

ojrik commented May 8, 2023

leptun commented May 8, 2023

ojrik commented May 8, 2023

github-actions bot commented Aug 17, 2023 • edited Loading

3d-gussner left a comment

Choose a reason for hiding this comment

leptun Sep 13, 2023

Choose a reason for hiding this comment

sarusani commented Sep 14, 2023 • edited Loading

leptun commented Sep 14, 2023

sarusani commented Sep 14, 2023

leptun commented Nov 5, 2023

leptun commented May 1, 2023 •

edited

Loading

leptun commented May 3, 2023 •

edited

Loading

leptun commented May 5, 2023 •

edited

Loading

ojrik commented May 7, 2023 •

edited

Loading

github-actions bot commented Aug 17, 2023 •

edited

Loading

sarusani commented Sep 14, 2023 •

edited

Loading