thinkyhead
commented
9 years ago Great description, and thanks for posting as a diff. Let me see if we're on the same page…
First, you should really be using the latest Development code. A lot has changed since 1.0.2, including adding support for a Z endstop that is also a probe, which makes sense for your machine.
Currently, M206 is used to set the home_offset values. Meanwhile M851 is used to set the Z offset of the probe relative to the nozzle, zprobe_zoffset. So, you might want to try using M851 Z-1.19 in addition to setting the home_offset with M206, since your "probe" is that far from Z=0. This gets a little confusing with your type of setup, so correct me if I'm wrong about setting both options or just one of them. You will also need to enable the new Z_PROBE_ENDSTOP option.
brainscan
WheresWaldo
AnHardt
Wackerbarth
Just below the idler on both sides there is a switch that is physically attached to a nut that rides up and down the 8 mm threaded rods. When the head hits the build plate, the rods continue moving until the small micro switches are disengaged.
The X and Y have similar micro switch endstops, the X_MIN is mounted on the front right side of the X Carriage and hits a plastic stop located on the left X Idler. The Y is located under the bed and is triggered by the frame that holds the glass build plate. The only Thingiverse model available is this one here: 
paulusjacobus
Background: Robo3D R1 (R1) with auto-level and 8 mm threaded rods using stock firmware based on 1.0.0 works after a fashion. This is a report about auto-leveling in 1.0.2 release with the proper modifications for use on the R1.
Issue stems from the use of Z_PROBE_OFFSET_FROM_EXTRUDER. I will explain as best as I can. The R1 uses two switches under the x-carriage in parallel to determine when it has hit the endstop for the Z axis. These switches are normally closed and actuating either results in the endstop being triggered. They are affixed to nuts that can travel independently from the Z axis mounts. When the lead screws are turned beyond the point that the extruder head hits the bed, one or both switches activate (open). This is how the R1 knows it has hit the endstop. The distance the lead screws move between the time that the hotend hits the bed and the switch actually triggers the endstop was measured on my machine, but should apply to all R1's with auto-leveling, the Z_PROBE_OFFSET_FROM_EXTRUDER will be a negative value. In my specific case one or both switches trigger -1.19 mm from the top of the bed.
On line 402 this offset was inserted in the proper line. Among other changes in Configuration.h to make it work for the R1.
Here is the issue. After uploading and resetting the printer I perform a M503 to make sure that I get the following output M206 X0.00 Y0.00 Z-1.19. Once it is determined the correct offset is both in Memory and in EEPROM. I proceed with a G28 followed by a G29.
The output of the G28 is noted in a serial monitor as well as on an LCD Smart Controller installed on the printer. After issuing the G28 but before issuing the G29 the printer moves appropriately to home X then home Y then center the hotend to home Z. The result is that my location is listed as X:120 Y:127 Z:-001.1 (output is truncated by the software). This is behavior that is expected. If I then issue a G1 Z0 I end up with the hotend exactly where I expect it hovering at the top of the bed and Z0 being the actual point it should be. Here is where the problem begins. Now executing a G29 Auto-leveling command, the printer moves to the appropriate nine locations on the print bed and performs a Z axis endstop probe. When it has reached its 9th and final location the firmware makes the assumption that it is resting at the Z0 location when in fact it is actually at the Z axis trigger point. The more egregious part is that the firmware actually believes this point is Z0 + -Z_OFFSET_FROM_EXTRUDER. So not only is it 1.19 mm too low, it is telling me that it is 1.19 mm in the air. Any command to move the Z axis is now too low by double the amount of the offset. This is repeatable with any amount of negative offset you want to put in the Configuration.h file.
To verify that it isn't simply assuming that the Z_PROBE_OFFSET_FROM_EXTRUDER should have been entered as a positive number I edited the source with a +1.19, recompiled and performed the same experiment. This time the result was that the final resting place of the G29 command was still the trigger point but the firmware reported that I was actually at Z:+002.3. Any Z axis moves are now too low by over 3.5 mm. No matter what is placed in Z_PROBE_OFFSET_FROM_EXTRUDER, G29 appears to always assume that the last Z axis endstop trigger point is Z0 and then it adds the positive value of the offset instead of what it really is.
I am willing to provide any additional information and test various solutions to solve this issue. I would not have reported this until I understood the problem and saw no solution in any of the configuration file options.