Bed Leveling

[Woolf]

Hi

I have a problem with the Bed Leveling, i can not level the Bed.
I have bought a Dial Indicator and leveled the bed in 0.01 accurate. On all ( X,Y,Z and middle ) are the settings perfect but when I start a print then is the side across the Y tower much lower.

Things I've tried

- Bought an Aluminium Bed - totally flat.
- Leveled with a Dial Indicator
- Installed a new clear Repetier 0.91 firmware

but the same spot ( across from the Y tower ) is always to low.

Has someone an idea how can I fix this?

[Polygonhell]

If your measuring it flat then it appears not to be the odds are you are losing steps or something is slipping during the print. Most commonly this is either the drive pulley on one of the steppers being loose or if your unlucky a damaged belt, you want to check the belt 1/3 of the way up on the inside. My max ate it's first belt at the 300 or so hour mark, but it was almost certainly a side effect of ramming the effector into the build plate probably while I was debuging a different issue, because it's got 2500Hrs on it now and I've haven't had to touch the belt since.

The only other real option is the endstops aren't working consistently. But you would see that if you level, home then immediately check for level.

[Woolf]

I've checked the belts, they look good.
When I make an upgrade to hall-o endstops, would it be more precise or not?

[mhackney]

Do you know the difference between "bed leveling" and "delta calibration"? Not being facetious, it is confusing and many people do not really understand the difference. Let me continue...

First some assumptions - you are not skipping steps or have other electrical issues. Your firmware is "fresh" and has not been tweaked to account for asymmetries in the build. And if you are building on a glass plate you can pretty much be assured that it is flat (of course glass can warp, etc but for our purposes it is flat. Use a straightedge and check yours to verify).

Now...

Leveling is the simple process of making sure the build plate/bed is co-planar to the X-Y print nozzle movement. A plane is defined by 3 points so if you set the length of each tower the same and since there are 3 towers, you will be assured that the build plate and towers are "level" (or more precisely, co-planar) if the mechanical structure of the printer is "reasonable". We do this adjustment on our Rostocks and Orions using the end stop adjustment and measuring the Z travel at the base of each tower as per the calibration section in the assembly guide. That's "leveling".

Now, if you discover that your X-Y nozzle travel is NOT in the same plane as your build surface, perhaps it travels in an arc (either concave or convex). It is not the build surface that is not flat or level, it is either the mechanical geometry of the printer OR the software calibration for it. If you've "leveled" your bed as described earlier and you position the nozzle at X=Y=Z=0 (G0 X0 Y0 Z0) and discover that it is either too high OR crashes into your bed, it is not the bed that is the problem, it is the mechanical or software calibration. Since the mechanics are pretty much "fixed" during the build (although you can tweak these) let's focus on the software. Gene has done a great job capturing the calibration procedure for delta calibration in the Guide so I won't repeat it here - except to say - if you are seeing a low spot in one place (across from Y tower) then you have calibration issue and since it only appears in this one spot (assuming your calibration parameters in firmware to account for asymmetry are all "0") then you have a mechanical calibration issue related to tower position most likely. This can be "tweaked out" in software - i.e. there are configuration parameters in Configuration.h that allow you to compensate for mechanical issues in software. These are also described in the guide. Unfortunately, it is an iterative and somewhat trial and error process. Keep a written record of each change so you can go back to a previous state if you made things worse. Make ONE change at a time and verify it's effect until you understand what it does and how to use it. Once you've made a change, you may likely have to (and highly recommended that you at least check it) perform the "leveling" step again and then validate everything. The parameters are inter-related so a change in one place has an effect somewhere else. Go slow and take notes. Your dip in Y can most likely be tweaked out with these settings in Configuration.h:

Code: Select all

/** \brief column positions - change only to correct build imperfections! */
#define DELTA_ALPHA_A 210
#define DELTA_ALPHA_B 330
#define DELTA_ALPHA_C 90

// MLJH: start with "0"s here and calibrate once you get things moving
/** Correct radius by this value for each column. Perfect builds have 0 everywhere. */
#define DELTA_RADIUS_CORRECTION_A 0
#define DELTA_RADIUS_CORRECTION_B 0
#define DELTA_RADIUS_CORRECTION_C 0

If you try several times and things just spiral out of control, then get a large square, ruler and string and check your mechanical geometry. It should be obvious if it's that far off that you can't tweak calibration in firmware.

Hope this helps!

cheers,
Michael

[Woolf]

I have not seen any Mechanical problems and any problems with steps.
When I print a round thing its totally correct, and the square test print was square.

with a stroke angle I have measured from the tower to Print Bed, its perfekt.

when I change the values
#define DELTA_ALPHA_A 210
#define DELTA_ALPHA_B 330
#define DELTA_ALPHA_C 90

can you explain me please what happens when I change those values?

thanks to all

[mhackney]

Great that you've checked the mechanical setup. Those values in effect tell the firmware that the towers are not at the "perfect" positions of 210°, 330° and 90°. If the towers are not at the precise mechanical position the print head's travel path will include a Z component and will not travel in a plane and there can be a skew in the X-Y plane too depending on how the towers are misaligned.

[bvandiepenbos]

maybe one of your U-Joints are binding, check them.

[Woolf]

Today I've printed a few square prints more and they waren't god.
I have carbon arms, the Joints are looking good.
I've checked the mechanical and looks good, not any problem
Is it possible that one Nema 17 makes the problem?
Maybe when I change the belt ( Is the GT2 belt the right one? How Long should it be for one Tower? ).
When I upgrade to Hall-O Endstops, must I change anything in the Firmware?
It's maybe my Rambo Board broken?

I've checked my old prints, everyone of them has a tiny problem. I think that was also at the beginning but I've not noticed it.

[lordbinky]

If you have to change anything in firmware it would just be inverting the endstop. I forgot the G-Code command off the top of my head, but it's important to check they are functioning as intended before trying to home since they only work with a specific pole of the magnet. I swapped to just the straight sensor, since they're only ~$1, in my setup and do endstop adjustments in software.

[626Pilot]

You're not alone. See this thread: http://forum.seemecnc.com/viewtopic.php?f=37&t=4868