Auto_Bed_Leveling: Enhanced G29 command

[beerdart]

Great, thank you Roxy. How do people usually find the offset between the probe and the tip of the nozzle? And have you seen this offset number change over time with all the mechanical stuff going on?

With pronterface G28 note z position then bump your z down to your nozzle clearance not the z position subtract from probe and that's the offset

[DarkAlchemist]

If the bed is not perfectly flat it never worked for me until I made it perfectly flat then it was able to compensate for it being unlevel. Thanks to the matrix I was able to see where hills and valleys were (something like -.01xxxx -.1xxxx -.05xxx, -.01xxx the ABL would show but never compensate for). That is what I was talking about that happened to me and ABL just hated that. I went to the forums (before this one) and was told it HAD to be perfectly flat and son of a gun they were right.

[Roxy]

Great, thank you Roxy. How do people usually find the offset between the probe and the tip of the nozzle? And have you seen this offset number change over time with all the mechanical stuff going on?

For any given setup... Once you are dialed in, it pretty much just works everytime. If I clog my nozzle, it really sucks because I have to take the hotend out of the carrage. And that pretty much means when it goes back together again, the spacing is going to be a little bit different. And I have to dial it in again.

With pronterface G28 note z position then bump your z down to your nozzle clearance not the z position subtract from probe and that's the offset

One more comment. Some people measure their spacing difference on the height of the probe and the height of the nozzle. That is a good place to start. But depending on what kind of switch you are using, it is hard to measure exactly where the setting needs to be. It seems better (for me anyway) to use that as the starting place and then use PronterFace to move the nozzle up and down a little bit until I get the number right. Some micro switches make an audible click when they trigger. If yours does that, you can get close using a ruler, and then use PronterFace to move it up and down until you hear the click. Then you will be super close to correct. (But you may still may need to fudge it up or down a few .1mm )

If the bed is not perfectly flat it never worked for me until I made it perfectly flat then it was able to compensate for it being unlevel. Thanks to the matrix I was able to see where hills and valleys were (something like -.01xxxx -.1xxxx -.05xxx, -.01xxx the ABL would show but never compensate for). That is what I was talking about that happened to me and ABL just hated that. I went to the forums (before this one) and was told it HAD to be perfectly flat and son of a gun they were right.

Well, I'm telling you my experience is it doesn't have to be perfectly flat. As you can see from my current numbers, I have some issues with my bed flatness. As long as the differences are smaller than the layer height, it should work.

Right now I'm printing with a layer height of .28mm. I do see a difference in the thickness of the bead going down on the first layer depending upon where the nozzle is on the bed. But I get good adhesion everywhere. And the 2nd layer is very uniform.

[DarkAlchemist]

That may be it as I use layer heights of .1 to .18 and for parts .2mm sometimes but what I was originally seeing were hills and valleys (in various spots) greater than my layer height.

[mwyrick]

For any given setup... Once you are dialed in, it pretty much just works everytime. If I clog my nozzle, it really sucks because I have to take the hotend out of the carrage. And that pretty much means when it goes back together again, the spacing is going to be a little bit different. And I have to dial it in again.



One more comment. Some people measure their spacing difference on the height of the probe and the height of the nozzle. That is a good place to start. But depending on what kind of switch you are using, it is hard to measure exactly where the setting needs to be. It seems better (for me anyway) to use that as the starting place and then use PronterFace to move the nozzle up and down a little bit until I get the number right. Some micro switches make an audible click when they trigger. If yours does that, you can get close using a ruler, and then use PronterFace to move it up and down until you hear the click. Then you will be super close to correct. (But you may still may need to fudge it up or down a few .1mm )


Well, I'm telling you my experience is it doesn't have to be perfectly flat. As you can see from my current numbers, I have some issues with my bed flatness. As long as the differences are smaller than the layer height, it should work.

Right now I'm printing with a layer height of .28mm. I do see a difference in the thickness of the bead going down on the first layer depending upon where the nozzle is on the bed. But I get good adhesion everywhere. And the 2nd layer is very uniform.

Roxy,

Can you clear something up for us. There seems to be a question of does the G29 command create a solution to correct for the hills and valleys of the print surface. The G29 command does create a topographic map but it seems to be used only in creating an over determined set of equations used to solve for a 3x3 co-variance matrix used to correct the plane of the print surface. I don't seen anywhere in the planner that it can correct for flatness of the bed, only the level of the bed. It would seem if this is correct the additional measurements in the topographic map only serve to average out the bed flatness to solve for the bed level.

If the bed is not perfectly flat it never worked for me until I made it perfectly flat then it was able to compensate for it being unlevel. Thanks to the matrix I was able to see where hills and valleys were (something like -.01xxxx -.1xxxx -.05xxx, -.01xxx the ABL would show but never compensate for). That is what I was talking about that happened to me and ABL just hated that. I went to the forums (before this one) and was told it HAD to be perfectly flat and son of a gun they were right.

This would back up DarkAlchemist's assertion that he needed the bed to be flat before ABL would work.

[DarkAlchemist]

Roxy, Can you clear something up for us. There seems to be a question of does the G29 command create a solution to correct for the hills and valleys of the print surface. The G29 command does create a topographic map but it seems to be used only in creating an over determined set of equations used to solve for a 3x3 co-variance matrix used to correct the plane of the print surface. I don't seen anywhere in the planner that it can correct for flatness of the bed, only the level of the bed. It would seem if this is correct the additional measurements in the topographic map only serve to average out the bed flatness to solve for the bed level. This would back up DarkAlchemist's assertion that he needed the bed to be flat before ABL would work.

Something is right for sure because if this really did account for hills and valleys it would lower in a valley and raise on a hill but it never did that. After I watched it try to handle my hills and valleys I had, at the time, I came to the conclusion it averages so if your point is too far away from the average it created for that point/spot you are in a world of hurt. After I figured out it was sort of averaging I decided to go to 9 points and with all of those extra points it worked better but not until I got rid of the valleys and hills did it work

Another thing about the ABL that forced me to ditch it finally was the fact that it was bulging the first few layers out meaning that regardless what I told it the distance was of the nozzle to the switch it was pressing into the glass too much. I found this when I noticed another person using the exact same version of the ABL (Roxy's) and had, basically, the same printer I had with the same bulge. So, he showed me some prints without the ABL engaged and the bulge was not there. I spotted that and told him and he never noticed it until then. Try as hard as I could I never managed to remove that bulge with ABL on so I turned ABL off and the bulge is now gone.

[Roxy]

Some Slicers (Like Sic3r which I use) have a setting for the amount of material to extrude on the first layer. You can change this number both positive and negative from 100%. Is it possible you had this number altered and it caused a bulge?

[DarkAlchemist]

Some Slicers (Like Sic3r which I use) have a setting for the amount of material to extrude on the first layer. You can change this number both positive and negative from 100%. Is it possible you had this number altered and it caused a bulge?

First place both of us looked and his, as well as mine, were manually set to 100% (default is like 200%). This bulge has been my biggest gripe of 3d printing and no matter what I tried it would never go away and people in the IRC channel couldn't understand it. Well, after many many months I finally found this other person who didn't even notice he had the issue until I showed it to him (thank goodness he was posting pics of his Wilson (very much like my Rework) and I spotted it because when he said the ABL was your version (no, not blaming you just explaining) I finally had something I could grasp onto.

I tried a tiny print where I had no dips, or valleys, or hills, right dead center and had a hollow cube printed with ABL and without and SOB the one without had no bulge though had other issues because of my non flat bed. Well, all of the above was exactly what I needed so I purchased an aluminum heat spreader and removed ABL once and for all and that bulge is long gone. Seriously, no matter what I tried, the other guy was trying stuff too, that bulge was always there so my only choice was to disable the matrix and not use G29.

I have a very flat bed now and I can print very fast and I can finally print interlocking pieces and gear sets like herringbone gears as that bulge went with the ABL. Oh, I even tried telling Slic3r 10% for first layer and that bulge stayed. As I said that bulge is because it was too close to the glass regardless what I tried to make it higher. No ABL (I still use G28 enhanced with a modifier of 1.55) G29 and I am printing finally.

[tstockgl]

No, it was a separate problem I noticed and fixed. The Y move on Z lift still occurs.

Hi! Not sure if you solved this or not, as I've only made it to page 12 of this thread, but I was running into the same issue. After much tinkering, I finally fixed the issue by simply defining the "Y_MIN_POS" under the "//Travel limits after homing" to something smaller than my extruders Y probe offset from the probe. My "Y_PROBE_OFFSET_FROM_EXTRUDER" was set to "-7" so I set my "Y_MIN_POS" to "-10" and that fixed the issue. I hope this helps if you haven't already figured it out.

-Tyson-

[tstockgl]

Just got it up and running last night! Everything seems to be running smooth. Here's my latest "G29 V4 n4" print out. Let me know if you see anything that looks out of wack. Thanks!

Roxy's Enhanced G29 Auto_Bed_Leveling Code V1.25:
Full support at http://3dprintboard.com
Bed x: 50.00 y: 10.00 z: -1.69
Bed x: 100.00 y: 10.00 z: -1.54
Bed x: 150.00 y: 10.00 z: -1.49
Bed x: 200.00 y: 10.00 z: -1.48
Bed x: 50.00 y: 48.00 z: -1.72
Bed x: 100.00 y: 48.00 z: -1.52
Bed x: 150.00 y: 48.00 z: -1.44
Bed x: 200.00 y: 48.00 z: -1.57
Bed x: 50.00 y: 86.00 z: -1.54
Bed x: 100.00 y: 86.00 z: -1.48
Bed x: 150.00 y: 86.00 z: -1.48
Bed x: 200.00 y: 86.00 z: -1.54
Bed x: 50.00 y: 124.00 z: -1.56
Bed x: 100.00 y: 124.00 z: -1.49
Bed x: 150.00 y: 124.00 z: -1.48
Bed x: 200.00 y: 124.00 z: -1.53
Eqn coefficients: a: 0.00 b: 0.00 d: -1.65
Mean of sampled points: -1.534406

Bed Height Topography:
+0.05366 +0.04641 --0.00784 --0.15734
--0.03834 +0.09716 +0.01116 --0.18734
--0.00184 +0.05691 +0.05541 --0.00984
+0.00866 +0.04991 +0.04566 --0.02234

planeNormal x: -0.00 y: -0.00 z: 1.00

Bed Level Correction Matrix:
1.000000 0.000000 0.000671
-0.000000 1.000000 0.000434
-0.000671 -0.000434 1.000000
echo:endstops hit: Z:-1.53