Hi!
So my esp32-based controller box for the Delta Stage is almost 100% complete.
I have added end stop switches into the feet and they work really well.
However, since the project is for 2P microscopy I can’t use any feedback from the images - the purpose is to make this a completely modular unit.
Hence, I need to calculate the backlash for each motor mechanically.
The process goes: (1) raise each actuator an arbitrary amount; (2) lower each until they hit the endstops; (3) count the number of steps until the switches turn back off.
My problem is the movement differential (distance that needs to be travelled back to un-trip the switch) is ~0.5mm which will account for most of the steps in (3), however I don’t know how this relates to the steps of the motors, since it’s vertical travel rather than the 0.5um of the stage itself.
If anyone who understands the levers and relevant principles would kindly share the information with me I would be SO grateful to put a bow on this and wrap it up!!
Just to confirm is your end stop above/below the stage or below the actuator levers? Could you do a really quick sketch, or a photo?
I gather that you are talking here about is 0.5mm hysteresis in the endstop switch? This will be a huge amount more than the stage backlash.
The motors themselves are 64 steps in the software per revolution, with a gearing of about 64 (I think it also varies between motors) which gives ~4000 steps per turn of the small gear. The small to large gear ratio is 1:2, so 8000 steps for each turn of the M3 hex head screw in the actuator. Your endstop in the foot will be directly in the direction of that screw, so the 0.5mm pitch of the M3 screw gives 0.5mm travel against the end stop for 8000 steps. The backlash is usually 70 or 100 steps, which is very small compared to 0.5mm. Variation in the 0.5mm from switch to switch, or even between actuations of the same switch is likely to be more than the backlash.
I think that I would measure the backlash using the camera-stage mapping in brightfield mode. Then just keep that number. Maybe even do a few tests in brightfield of different builds and then hard-code a backlash number big enough to cover everything - 200 steps?
Thanks for your reply! Perhaps these endstops are not the answer to my backlash challenge…
Because the module is not taking input from any computer, integrating automatically with camera input is hard (not to mention that I haven’t attached any lenses to this that would facilitate detection of 0.5um steps). My thinking is to set up a high-magnification camera and manually make the call on when the sample starts moving, using a single motor at a time. I am also thinking to take your advice and set a generic default backlash value ~100 and seeing how bad it could be. I suppose with biology, there isn’t so much of a need to return to exactly the same coordinates, when the samples themselves wobble and drift too…