Now that remote control of the camera, backlight, and movement axes can be scripted, the next challenge was to combine these in a way that allows us to actually take photographs of sections of a 24-well microplate without having to manually move the microscope or the plate.
The overall goal is to mount the camera, optics, and backlight in a "C" shape, with the backlight at the bottom and the camera optics above. The resulting rig should be able to move along the X axis between different plates in a given row. It should also be able to move precisely along the Z axis in order to focus the camera. At the same time, the microplate needs to be suspended between the camera and backlight, and movable along the Y axis, so that any row of wells can be positioned below the camera. Taken together, this should allow for us to take photographs of any well on our plate, or any other plate that fits within our apparatus.
Because the "hot end" and extruder of the 3d printer aren't necessary, they were removed from the print head, leaving us with just a sturdy piece of metal with a few screw holes. A prototype of K'Nex pieces was constructed to hold the camera with optics and the backlight below (need a photo here). Another prototype was constructed to hold the well plate at approximately the correct level, without interfering with the first part's X or Z movement.
The first and most obvious challenge encountered was that most K'Nex pieces are made of flexible ABS plastic and so the overall construction was very wobbly and the well plate could not be held out level. While it allows for proof-of-concept with regards to camera positioning, this will probably result in unacceptable image quality, and so different construction materials and methods will be needed to make a reliable and useful device.
The second challenge encountered is that the auto-homing function of the 3d printer needs to work since all subsequent movement is relative to that point. The positions of the switches that allow the printer firmware to find this home point didn't account for the K'Nex scaffolding attached to the usual print head location, and so the printer would smash our prototype to bits if allowed to auto-home normally. To address this, additional bars need to be added or the switches repositioned to account for the new size of the "print" head, but this has also proven difficult with the flexibility of the K'Nex pieces and can probably also be addressed better with a switch to different materials and construction methods.