Allow Zero Endstops (e.g., for CNC) #21120
Conversation
…cannot be homed. They must always be considered to be homed, and cannot be 'un-homed'.
|
Thanks for the submission. I've cleaned it up and extended it a bit, but haven't evaluated the question of encoders. However, since they are somewhat independent of each other, we should allow for something that has encoders but no endstops. |
|
I've just noticed that there are tests in the code for the toolhead being over the bed, with different behaviour (e.g. no skew correction) when the toolhead moves off the edge of the bed. Makes full sense in the context of a 3D printer, but I think in the context of a CNC that has no endstops (and hence the coordinates that correspond to "over the bed" are flexible) those tests should be bypassed. Going to look at it closer later. |
Fixed the conditional at the end of Conditionals_LCD.h - USE_ZMIN_PLUG was duplicated in the #if Having no endstops affects much more than just the planner, as checks are carried out in various places in the code to ensure the toolhead won't crash. But without endstops Marlin can never know if a crash is imminent, so the checks need to be disabled. Endstop-less machines (notably CNCs) can still have probes. Whether or how bed levelling works on such machines needs to be experimented on. Such machines also, for example, cannot reasonably have tool parking. Sanity checks have not been implemented to catch such probable anomalies.
|
Currently working on implementing G68 and G69 (rotated co-ordinate systems). Not sure how I should create another PR when its ready? My use case is PCB milling. With double-sided milling, it's important, and difficult, to get the second side of the PCB in the same orientation that it had when the first side was milled. An answer is to drill fiducials in the PCB, and use them to determine the actual orientation after flipping to the second side. |
For simplicity, you'll want to wait for this to be merged, and then build on top of |
thinkyhead
force-pushed
the
bugfix-2.0.x
branch
from
34cda52 to
ba301fd
Compare
|
Cool - thanks! Is there anything I should be doing to help matters along? (I'm new to this. I've done tests, and it all seems OK to my eye. I've tested leveling, and it also seems fine). |
Allow Zero Endstops (e.g., for CNC) (MarlinFirmware#21120)
Co-authored-by: Scott Lahteine <thinkyhead@users.noreply.github.com>
Co-authored-by: Scott Lahteine <thinkyhead@users.noreply.github.com>
Co-authored-by: Scott Lahteine <thinkyhead@users.noreply.github.com>
Co-authored-by: Scott Lahteine <thinkyhead@users.noreply.github.com>
Description
Some machines (e.g. cheap Chinese CNC machines) come without endstops. Such machines cannot be homed, and so need to be considered to always be in a known and trusted position. It is up to the user to ensure that there are no toolhead crashes.
My test for a machine without endstops is that none of
USE_[XYZ]MIN_PLUGorUSE_[XYZ]MAX_PLUGare set.The only thing I'm not sure of is in the
motion.cpp set_axis_never_homed(...)implementation. There is a test for I2C position encoders. I cannot see how a machine cheap enough to not have endstops would ever have position encoders. Should this combination throw an error in SanityCheck.h?Benefits
Marlin now becomes useful on the cheapest of machines that have no endstops!
Postscript
(p.s. my first ever PR - feedback welcome)
Concise Diff