Updated: Easier Leveling the Build Plate!

[cliffyk]

There is a potential conflict between the alignment and calibration procedure you describe, and the printer's fundamental operation.

The sequence of the first downward motion of the trolley at the start of a print job is as follows:

• the firmware commands the stepper motor to turn clockwise (facing the shaft) to lower the trolley;
  
• the trolley falls until it triggers the optical lower limit switch, causing firmware to stop the stepper motor;
  
• the firmware commands the stepper motor to turn anti-clockwise, raising the trolley ≈ 5-6 mm;
  
• as a validation the firmware again commands the motor to turn clockwise lowering the trolley 'til the limit switch once more triggers;
  
• the firmware resets it's internal "turn counter" variable; (a value of 0 = the physical zero trolley position;
  
• if a logical "zero" position has been set (via the "Z=0" GUI function), and the "print" file contains a "go to 0" G-Code command the motor is commanded to turn anti-clockwise the required number of steps¹ to relocate the platen to the set logical "zero" + one layer thickness (25 steps using your +0.2 mm logical "zero" setting + one 0.05 mm layer)--else it just moves up by one layer thickness (5 steps for a 50 micron layer)
  

It is vital to understand that in printing, in response to a G-Code "goto zero" command, the Z-axis ALWAYS first moves to the physical zero position (as determined by the limit switch) to establish the actual zero starting point--it has to to determine the trolley's actual zero reference. It and then rises to the logical "zero" position if set. Also, the "Home" function will always move the trolley to thee physical zero position regardless of how the logical "Z=0" position may have been set.

The only way to position the trolley to the logical "zero" position--other than manual re-positioning after the "Home" button has been pressed--is via G-Code commands ("G28 Z", "G 90 G1 Z0", etc.) in a "print" file.

I have validated this behaviour numerous time with a dial indicator.

After attaining the physical zero position the firmware maintains a running tally of the number of "up" and "down' motor steps it has commanded to maintain an accumulated value indicating the Z-axis current position I.e. where it is "supposed to/should be". If the trolley jams in the vertical rails, from being too tight in the rails or physically blocked, the actual position vs. the calculated "should be" position will be incorrect.

I have observed this in two MARS units I have repaired for others. In both the trolley was so tight in the rails that when the stepper motor attempted to raise it it could not, or "missed a step", causing the motor to "cog"² and not actually move one or more steps. When that happens the firmware thinks the Z-axis has moved but it has not; this fouls up the internal count and the firmware now has absolutely no idea of where the build-plate really is. Missing one or two steps may not be immediately audible as is the "grinding" of a complete jam. In both of the printers I repaired the Z-axis trolley was incorrectly adjusted so as to be absurdly tight within the rails, requiring 4 or 5 lbf to be moved. This is a pretty good YouTube video describing proper adjustment of the Z-axis trolley. The fellow calls it "calibration", which of course it is not as nothing is "calibrated" against any standard--it is simply an adjustment.

If I got a new MARS with the dual rail Z-axis this would be the first thing I'd check.

The purpose of using paper as a spacer (I use two sheets of common 20# copy paper, about 0.18 mm) is to establish the platen's position relative to the trolley's lowermost position--as determined by the limit switch. The paper is (as you have stated) intended as a stand-in for the FEP. When adjusted in this manner the build surface will be "one FEP thickness" above the LCD when the trolley is at the hardware determined absolute zero position. This is a physical calibration separate and distinct from setting a logical "zero" position using the "Z=0" function.

Not using the paper (or FEP) as a spacer will result in the build-plate being in contact with the LCD when the limit switch is triggered. With the resin vat in place the very first downward motion of the trolley--to the limit switch "physical zero" position-- will attempt to compress the FEP between the platen and LCD, and most likely flex the LCD as it is the more compliant of the two. The LCD is actually quite resilient for what it is however such repetitive flexing cannot do it any good. 


-----------------------------------------------
¹ -The stepper appears to be a pretty much standard 1.8° per step (200 steps/turn) unit. the lead screw ha a 2 mm pitch, so one step = 1/200th of 2 mm = .01 mm (10 microns); your .2 mm ("{zero" offset) + one .05 mm layer = 0.25 mm = 25 steps

² - "Cogging" is a condition common to stepper motors. Often described as "grinding", it occurs when a stepper is told to turn but does not have sufficient torque to overcome the load. In open-loop control systems that's when the motor cogs, the controller thinks the motor has turned, but in reality it has not.