First Layer stil not leveled after G29 ABL.

[der buetts]

Hi printers,

finaly it got my proximity sensor with a fresh version of marlin working on my KC. 

G29 seems work just fine. It takes a lot of probes over the whole print bed. 

I have written the G29 Z-0.8 command in Cura "Start Code" AFTER G28 to make sure it keeps the data. 

''''''

G21 ;metric values

G90 ;absolute positioning

M82 ;set extruder to absolute mode

G28 ;Home all axes (max endstops)

G29 Z-0.8

G1 Z15.0 F4000 ;move the platform down 15mm

G92 E0 ;zero the extruded length

G1 F1000 X50 Y-90 Z30

G1 F200 E30 ;extrude 3mm of feed stock

G92 E0 ;zero the extruded length again

G1 F4000

M106

M117 Printing...

*********

The first layer ist still way too low in the corner of the X-tower. 

I have no idea where to start tweaking since every manual says that G29 is exactly there to avoid this problems. 

And since a managed to upgrade to a contactless sensor there can be no mechanical problem with the switch.

As far as i remember the KC tented to have this problem before the upgrade. 

I know from my attempts to print without G29 my bed isn´t in level. But in an other way. 

G29 compensates that very well. Exempt for the corner of the X-tower. 

Yes the aluminium plate ist flat. 

So i think the problem must be deeper. 

Maybe the steppers or the stepper settings? 

regards from germany and sorry for my poor english.

oliver 

[Haydn Huntley]

Hi Oliver,

You're quite ingenious figuring out how to attach the E3D hotend, your proximity sensor, and the blower fan for cooling!

The community doesn't have telepathic powers, so we can't tell which version of Marlin you're using.

If you using the one which BEL originally posted, then in order to get things level, you have to adjust the M3 screws which trigger the endstops.  I placed a stripe of white nail polish on mine to make it easier to keep track of how I was adjusting them.  One turn is 0.50mm.

I recommend using the version of Rich Cattell's Marlin.  You can use software commands to change any of the parameters, including the endstop offsets.  You can download it from this page:  http://www.maui-3d.com/kcZeroBacklashInstallation.php

On that page I posted instructions for how to do this.  Please note:  after making a change, you must use "G28" to home the printer, so that it will take effect.

Alternatively you can use Repetier Firmware, which you can also download from the above page.  It is much faster than Marlin, so it works better at higher speed.  When you use it with Repetier Host, it makes changing all of the software configurable parameters very easy.

Because you're using a non-contact proximity sensor, you might be able to get RC's Marlin's auto-calibration to work, but back when I was using the stock aluminum KC arms, my effector was so loose that the Bowden tube exerted enough pressure on it to twist it, so that's why I had to develop magnetic, zero-backlash arms.

Good luck!

:-)

[George P]

The biggest issue with delta kinematics is that it is hard to disclose which particular mechanical deviation is responsible for distortion of resulting working plane.

Or you have a good math model or you are experienced enough to find it.

So use of Rich Cattel Marlin fork is one of the easiest way to find a set of error compensating parameters to get the plane flat enough.

 

Dne středa 11. ledna 2017 9:15:00 UTC+1 der buetts napsal(a):

[der buetts]

Kossel Clear, 
Proximity Sensor, 
RAMPS 1.4
Marlin 1.1.0-RC8
Aluminium bed. 

[der buetts]

Ok thank you i will take a look at  Rich Cattell's Marlin. 

In the meantime i have really worked through Marlin FW. I really want to stick with it since i already know it pretty well. 

Anyway - a big Thank You to everyone who has tried to help me out. 

p.s. my workaround :-)

Tilting the bed with sticking 2 pieces of Paper between the aluminium plate and the heat element. 

G29 does his work.

Pulling the paper out right before the printing starts. 

Works like a charm. :-)

Am Freitag, 13. Januar 2017 08:28:10 UTC+1 schrieb Haydn Huntley:

[der buetts]

https://plus.google.com/u/0/101505725160036638895/posts/FAkZbzL2nz8?cfem=1

[Blue Eagle Labs]

HI Oliver,

double check that all the aluminum extrusions are flush against the acrylic. Also, if that doesnt solve the problem, try and get the heated bed on springs so that you can adjust them to be level. Then finally, when calibrating, use the concentric circles stl so you can manually tune the screws on the carriages. Attached is a calibration circle you might wanna try.

[Haydn Huntley]

I have also had this problem with my KC.

The way I thought about the problem is that the frame is three sided, and three points define a plane.  The print bed is four sided, and has to be secured at four points.  This is a little bit of a mismatch.

One of the first KC users, a fellow who's name might have been Hunter, posted a really simple different way of mounting the KC's bed.

This fellow figured out a way to lower the power supply, so that it was below the top of the horizontal extrusions.

Then he printed 8 little thumbwheels.  In the center of each wheel he placed an M3 nut, and then pressed/melted it into position.

With the old design, there is a long M3 (perhaps M3x20) bolt which goes through the MDF, through a spacer, and then into an M3 nut trapped in the horizontal extrusion.

With this fellow's design, the same M3 bolt goes through a thumbwheel, through the MDF, through a second thumbwheel, and then into an M3 nut trapped in the horizontal extrusion.

By rotating the two thumbwheels you can move the MDF board up or down.  One full rotation of the thumbwheels is 0.5mm.  This gives very fine control of each of the four corners of the bed, which makes it easy to solve the above problem.

I've used this design up until today, though I modified it slightly.  I've often needed to remove the bed, so that I can examine the wiring on the Arduino/RAMPS or to show people the inside of the machine, so I printed two little holders for 12mm" diameter disc shaped magnets.  Then the M3x20 bolts rest on the magnets.  It worked wonderfully reliably.

Here is a picture of the thumbwheel:

I've attached the STL file for the thumbwheel, as well as the OpenSCAD file to create it (in case you'd like to modify it), and my OpenSCAD file for the magneticThumbwheelScrewFoot.

These are some of the first designs I made, so please don't judge me too harshly for them!  :-)

I'd offer to send you the parts I just pulled off of my KC, but I believe you're overseas, so the postage would be something like $20.

[der buetts]

Ok guys, thank you for your effort.

If i am going to fine tune the level of the bed what is the point in order to use auto bed leveling?

I will go for a zero offset z-probe and ABL in the long run. 

But for now i will try to figure out if my KC is not proper mounted and i will shim the traxxas rod thingis. 

Feel free to follow the discusion over at G+ 

https://plus.google.com/u/0/101505725160036638895/posts/FAkZbzL2nz8?cfem=1

[Haydn Huntley]

You have to be careful with the words you're using.

Auto bed leveling is different from auto-calibration.

Right now your calibration is close, but still not quite correct (according to the photo you posted).

I don't think you need bed leveling.  Most Delta/Kossel printers don't.

When I tried using it, it just caused problems, so I turned it off.

You just need to correctly calibration your printer once, and then save and continue to use those settings.  You shouldn't need to recalibrate at the beginning of each print.  (Though if the temperature changes, your printer will thermally expand by about  0.017mm/°C.)

I don't recommend shimming the Traxxas rod joints, but you're welcome to do whatever you want.

At this point, I'd recommend that you either:

1)  Adjust the screws which trigger the endstops, so that you can print the first layer accurately.

2)  Adjust the software endstop offsets (with RC's Marlin or Repetier) so that you can print the first layer accurately.

3)  Shim the bed slightly to get it level with respect to your endstops so that you can print the first layer accurately.