Auto_Bed_Leveling: Enhanced G29 command

[TopJimmyCooks]

Hi Roxy,
I am having a problem getting my G29 to probe four points only. This is the BeckDac fork from about 15 months ago. If it would help to update to newer marlin I could do so.

I have the following in configuration.h. Thanks for any help!

#define AUTO_BED_LEVELING_GRID  // with AUTO_BED_LEVELING_GRID, the bed is sampled in a
  // AUTO_BED_LEVELING_GRID_POINTSxAUTO_BED_LEVELING_GRID_POINTS grid
  // and least squares solution is calculated
  // Note: this feature occupies 10'206 byte
#ifdef AUTO_BED_LEVELING_GRID


    // set the rectangle in which to probe
    #define LEFT_PROBE_BED_POSITION 0
    #define RIGHT_PROBE_BED_POSITION 110
    #define BACK_PROBE_BED_POSITION 195
    #define FRONT_PROBE_BED_POSITION 0


    // set the max number of grid points per dimension
    // The G29 command defaults to 3 if nothing is specified.  But setting the number of probed 
    // points higher is very useful when getting a Bed Topology Report  (G29 n 5 T)
    #define AUTO_BED_LEVELING_GRID_POINTS 2


    // Uncomment one of the following four lines so the Bed Topology Report can produce a map
    // that relates accurately to your bed orientation.  


    //#define ORIGIN_FRONT_LEFT
    //#define ORIGIN_BACK_LEFT
    //#define ORIGIN_FRONT_RIGHT
    #define ORIGIN_BACK_RIGHT


  #else  // not AUTO_BED_LEVELING_GRID
    // with no grid, just probe 3 arbitrary points.  A simple cross-product
    // is used to esimate the plane of the print bed


      #define ABL_PROBE_PT_1_X 0
      #define ABL_PROBE_PT_1_Y 0
      #define ABL_PROBE_PT_2_X 115
      #define ABL_PROBE_PT_2_Y 0
      #define ABL_PROBE_PT_3_X 115
      #define ABL_PROBE_PT_3_Y 190


  #endif // AUTO_BED_LEVELING_GRID

[Roxy]

The default number of points (if you don't specify a number) is 3. You should be able to say: G29 n 2 to get 4 points (2x2) probed. Please be aware that it would be: G29 P 2 in the GitHub code. You should probably leave the

#define AUTO_BED_LEVELING_GRID_POINTS 5

Just because you have space declared for a bigger set of points doesn't mean you have to use them.

I do suggest you load up the RCBugFix when you have time. It has a lot of good stuff in it. And you will need to have your settings crossed over to it's Configuration.h format to run the Unified Bed Leveling that is coming out soon.

[TopJimmyCooks]

Ok, I did the change to #define AUTO_BED_LEVELING_GRID_POINTS 5

I went to pronterface, ran G28.
when I sent G29 n 2 i got this error message:

>>> g29 n 2SENDING:G29 N 2
Error:Line Number is not Last Line Number+1, Last Line: 0
[ERROR] Error:Line Number is not Last Line Number+1, Last Line: 0


Resend: 1

I'm searching for information on this error but not getting anything definite - Thanks for any ideas.

[TopJimmyCooks]

Ok, I transferred all the information from configuration.h to the new one for RC6 bugfix and flashed it. I can't get it to home to the back right as my endstops are set up for (and as a lot of makerfarm I3v's are set for) previously there was a place to define the home as back right which I don't find in the new configuration.h. I saw somewhere a positive/negative homing direction setting, I will go look for that.

[Roxy]

I went to pronterface, ran G28.
when I sent G29 n 2 i got this error message:

This is a problem with the GCode protocol. You can't send a capital N because that says what line of GCode was transferred. That is why a lower case 'n' is used. But because it was causing so many problems (like you just ran into), the parameter was changed to 'P' in the new code.

(If you look at what the error code says... You actually sent a Capital N. Probably a lower case 'n' would have behaved as expected.)

[TopJimmyCooks]

If I ignore pronterface, which forces all alpha characters to UC, and use a gcode file with the LC n will it work? i'm on 1.1 rc 6 bugfix now. is the P implemented in rc 6 or yet to be changed?

I would go ahead and just try it but I'm still wrestling with the allowable probed area of the bed vs probe offset vs homing at the right back thing.

[....later]
OK, reflashed back to old DACB fork and ran a test gcode file with "G29 n 2" and it worked, probed four points as expected. I'm going to stick with this until I get a better understand the new bed layout setup in RC bugfix. Thanks very much for the help and for your work as a Dev!

[Roxy]

The BeckDac fork is very old. But it is very stable. There are a couple of bugs, like you can't do 3-point leveling because of a small syntax problem. The BeckDac fork will work fine for you until you get the time to move to RC6 or RC7. Moving shouldn't be very hard, but you should use a Visual Diff program to cross your settings over to the new code base.

[inventabuild]

Roxy,

I have 2 Lulzbot TAZ 4's and I'm trying to implement auto bed leveling using Marlin the way Lulzbot does with their newly released TAZ 6's as shown here starting at 33 seconds:

https://www.youtube.com/watch?v=JKkerkp3l5w

My TAZ 4's are set up similar to the TAZ 6 in the video except I don't have the button switch which is the first thing you see being tapped in the video. I'm currently using Marlin's G29 to tap the four corner washers and create a print plane and I'm getting great prints, but first I have to wipe the nozzle w/ a Scotchbrite pad to make sure it's clean for electrical contact with the washers.

I want to implement tapping the button switch, but Lulzbot created a lot of new wiring so the switch and the corner washers are on separate circuits. The button is tapped by the hot end to approximate the bed height so if the nozzle is not clean it will not continue to drill down into the washers; instead it will back off and try wiping again.

I would like to keep it simple and keep the button switch and the corner washers on the same circuit. Is it possible to set up Marlin to start w/ a G28 to tap the button switch and then if a G29 on the corner washers causes the nozzle to go x mm below the G28 tap then the hot end will back off and try the wipe again? Thank you.

[Roxy]

I think what I would do is change the home() routine to not go so far below the bed while searching for the home position. But the switch should get pressed even if there is plastic on the nozzle. The plastic on the nozzle issue would only cause trouble during the auto bed level probing. In that case, you really do need a clean nozzle since you are probing the bed and looking for electrical conductivity.

Still, with a reliable home (because a switch is being pressed), it would seem you can wipe the nozzle sufficiently well to be sure you will get the electrical conductivity that you need.

[AbuMaia]

Roxy,

I'm using Marlin RC7 now, working fine so far. Just hooked up a new capacitive proximity sensor to replace my servo switch. How with RC7 do I get the topographic "map" of the printbed from G29? Is it the same as with the earlier code we had to add to Marlin ourselves?