Auto_Bed_Leveling: Enhanced G29 command

[AlexD]

Astek,

Thanks for the response, I have since realised this about the EPROM, is there any reason that the EPROM cannot be disabled? I would prefer the values to be loaded from the config as standard. When i get back from work i will update my firmware as the first post states.

PS I am also from the UK, i have got a few spare bits of glass that i have tried, so i am pretty confident there is no warping, i have also tested using a straight edge etc. I am using 4mm Toughened Glass from a local glass cutter (small business). It is only £7 a sheet for me and he cuts it to size (475 x 475mm) with a chamfer. Maybe worth looking into if theres anywhere local to you if you are struggling to source glass. I might also try a sheet of aluminium (fairly cheap online for 500 x 500mm sheet).

Cheers

Alex

[Astek777]

Hey Alex

That's a decent size print area you've got. Mine is only 250mm cubed. Your prints will be going on for days at max size! ;-p

To be honest, I don't know what else is loaded from the EPROM. In theory it's a good idea and running M502, M500 after each re-upload isn't a massive inconvenience. At some point it would be fantastic if the G29 map was saved to EPROM so you didn't have to scan it before every print, but like you, I'd happily settle for consistent level prints at the moment.

Found some glass online that was £13 for 300mm square (4mm thick), which I didn't think was too bad, but when I got to basket page they tried to charge me £63 for delivery...ouch! I've found a genuine replacement now which should arrive on Thursday.

Just out of interest which corners are high/low for you? Mine is generally low down the left hand side, worse at the front. Right side if anything is a little high, but not too bad.

Once you've updated the software you can also move your probe to different points and issue 'M48'. This performs a repeatable probe test which will tell you how accurate your probing is. Move it around to various points (corners, centre etc) and see what result it gives you.

Cheers

Sean.

[AlexD]

Haha at the moment only just over half is useable as i am using 2 210 x 240mm Silicone heated pads, i cant seem to find an appropriate heater to use, without importing a Monster Heat Bed (18" X 18") from the US for £100+. I have a feeling the heated pads may be affecting the level significantly. Yeah i can see the advantage of using the EPROM, i just always forget to update it as i'm used to it loading from the config.

Previously i had gone through 3 sheets of 4mm Standard 'Annealed' glass, but this cracked after about 10-20 heat cycles, then the glass supplier recommended toughened, as it is heat treated in manufacture apparently, Borosilicate glass is ridiculously expensive.

Hopefully it wont take long to get the firmware tweaked this evening, i might even just try to copy my configuration & configuration_adv into the firmware that has been posted by Roxy.

Alex

[Roxy]

Astek,

Thanks for the response, I have since realised this about the EPROM, is there any reason that the EPROM cannot be disabled? I would prefer the values to be loaded from the config as standard. When i get back from work i will update my firmware as the first post states.

The EEPROM is easy to disable. You just comment out these two lines in the Configuration.h file:

//#define EEPROM_SETTINGS
//#define EEPROM_CHITCHAT

And you don't need to add all the changes to get the G29 features. The right answer is to use this fork which has everything already added to it. It is configured for MakerFarm printers, but that isn't a big deal. You just make your changes to the configuration.h file.

https://github.com/beckdac/Marlin

PS I am also from the UK, i have got a few spare bits of glass that i have tried, so i am pretty confident there is no warping, i have also tested using a straight edge etc. I am using 4mm Toughened Glass from a local glass cutter (small business). It is only £7 a sheet for me and he cuts it to size (475 x 475mm) with a chamfer. Maybe worth looking into if theres anywhere local to you if you are struggling to source glass. I might also try a sheet of aluminium (fairly cheap online for 500 x 500mm sheet).

A 475mm x 475mm print bed is very big. That may be the source of the problem. But up above, you said you tried 3 point leveling with all 4 combinations of 3 corners. That is probably not going to give good results. The 'Right' answer is to use two corners along the same edge, and the mid-point of the opposite edge. Probably, that will give much better results.

[AlexD]

Hi Roxy, thanks for the reply. I have not tried the 3 Point Probing yet, i believe that was Astek, i did not try it because i thought the Grid was superior. I will flash that firmware this evening and disable the EPROM. Then begin the Probing... Thanks Again.

Alex

[eistee]

Hey Roxy,

Thank you for your ideas and enhancements to the Bed Leveling Process. I have added a inductive Sensor to my printer and included your improvements in my existing Marlin-FW. M48 works just fine and seeing the topology of my bed also helps a lot.
However, I have a few questions:
- I would like to make use of the L R F B parameters. It is obvious to me, that my slicer has to output the values. Where can i find the postprocessing scripts for Slic3r? Is there also a solution for Simplify3D (since that's my favourite software)?
- I've seen that newer versions of Marlin offer a mesh-probe to compensate uneven print beds. This is done by a bilinear interpolation approach and looks very interesting to me. Nevertheless, mesh probing is - to my knowledge - only available with manual leveling with M420 and M421. Is there a reason why there is no mesh auto leveling? In my eyes it seems like most work is already done, the combination of manual mesh probing and auto leveling shouldn't be too much work and i would like to give it a try. Am I missing something?

[Roxy]

Hey Roxy,

Thank you for your ideas and enhancements to the Bed Leveling Process. I have added a inductive Sensor to my printer and included your improvements in my existing Marlin-FW. M48 works just fine and seeing the topology of my bed also helps a lot.
However, I have a few questions:
- I would like to make use of the L R F B parameters. It is obvious to me, that my slicer has to output the values. Where can i find the postprocessing scripts for Slic3r? Is there also a solution for Simplify3D (since that's my favourite software)?

There are a couple of post processing scripts to do the LRFB stuff. If you grab the BeckDac fork up above, there is an extra folder in it with different flavors of the script. I never actually used those parameters for real prints. But I wrote the C version of the script.

- I've seen that newer versions of Marlin offer a mesh-probe to compensate uneven print beds. This is done by a bilinear interpolation approach and looks very interesting to me. Nevertheless, mesh probing is - to my knowledge - only available with manual leveling with M420 and M421. Is there a reason why there is no mesh auto leveling? In my eyes it seems like most work is already done, the combination of manual mesh probing and auto leveling shouldn't be too much work and i would like to give it a try. Am I missing something?

The Mesh leveling is interesting. But it isn't fully baked yet. It seems to me it should be able to probe the whole bed and then just use it. Kind of like the Auto Bed Leveling does. So far, it isn't that automated. The other thing I would like to see the Mesh stuff do is scale back the compensation at some selected rate. By the time it gets up 5 or 10 layers, I think it should not be doing any correction at all. Once again, it hasn't evolved to that point yet.

[AlexD]

Ok so this is the feedback from my first set of probing

RECEIVED: +0.39340 +0.05090 --0.27160 --0.61660 --1.33160
RECEIVED: +0.55590 +0.27840 --0.01410 --0.35160 --1.06660
RECEIVED: +0.65590 +0.36840 +0.14340 --0.17160 --0.81410
RECEIVED: +0.64590 +0.42840 +0.27090 +0.03590 --0.53160
RECEIVED: +0.61090 +0.46090 +0.36840 +0.19090 --0.28910
RECEIVED:
RECEIVED: planeNormal x: 0.01 y: 0.00 z: 1.00
RECEIVED:
RECEIVED:
RECEIVED: Bed Level Correction Matrix:
RECEIVED: 0.999987 0.000000 -0.005012
RECEIVED: -0.000009 0.999998 -0.001874
RECEIVED: 0.005012 0.001874 0.999986

From what i can gather its pretty out of level right?

[Roxy]

Yes... From the front left corner to the back right corner you have 2 mm of difference in height. The good news is it appears the bed is pretty flat. You should manually level the bed as best you can. Lower the front left corner .5 mm and raise the back right 1 mm. Then re-run the test and level it a little bit more. Even though it seems like you should be able to do the entire correction at once, it never works out that way. Just keep raising the low corners and lowering the high corners a couple of time.

When you get it much more level than it is now, you will be able to see if you have high or low spots on your bed. As you scan left to right or up to down, if you see the sign flip from + to - and then go back to + (or vice versa), that will indicate a low (or high) spot on the bed.

[printbus]

Just a reminder to those struggling to obtain nirvana with a level bed - what the process does is adjust for variation in the distance between the nozzle and the print surface. The source of the variation doesn't HAVE to be in the bed. One could have a perfectly flat and true print surface and still have it appear to be unflat if the xy movement alters the nozzle height due to bent or misaligned xy rails/rods, a hard to turn filament spool that tugs the hot end upwards, or...