I have been thinking about backlash lately. Not having backlash is one of the selling points of flexures. The OFM otoh has like 200 steps of backlash - IIRC. This is not a big surprise given the way the drive train works. There is a 6 stages gear train - 5 inside the 28 BYJ stepper and one 3D printed after that.
Funny enough metal microscopes have similar gear trains in their focus axis without running into the same problem. They achieve this by keeping the gear train always loaded in one direction. So when lowering the stage the gears don’t push the stage lower but instead give way and the stage lowers by gravity - keeping the gears loaded to hold up the stage.
As the O-rings also pre-load the stage this would also work with the OFM. The thing that causes the backlash is the screw part of the actuator. The screw is not back-drivable. So relieving pressure from the stepper is not enough to make the stage move. We need to spin through the backlash and then actively push the screw to rotate in the other direction. The pre-load from the O-rings only prevent the backlash between the screw and the nut but not between the screw through the gears to the actual stepper motor.
The way to get rid of the backlash would be to replace the screw by some back-drivable gear train. Unfortunately the screw also offers a huge reduction. Turning the circumference of the gear to just 0.5mm of movement. With out looking up the exact size this should be a reduction of about 200:1. That’s kind of a lot.
Then there is the issue of how to turn the rotation of the gears to the motion of the level arm. Technically the arm already moves in an arc to we could just add and arc of gear teeth. But I doubt that we can 3D print such a gear with the necessary precision to make this work.
Another thing that I have actually used are winches. Just tie a piece of fishing line around a (motor) shaft. This works great with NEMA steppers and larger movements. Here we need a very thin shaft to not waste reduction ratio. A 2mm shaft has 6mm a circumference. A bit more if we add the diameter of the fishing line. So we are only off the screw drive by a factor of 14 or 16 (compared to the 0.5mm per rotation). As the new drive train is more efficient may be we can even lower the overall ration a bit. The rest can probably be compensated by a better stepper to shaft gear ratio and may be a 1:4 or 1:5 lever towards the stage.
Ok, this is far from being a thought out plan and I just made up all the numbers on the fly. But to me it looks like something that is not completely impossible.
Has anyone looked into that before? Or is a drive screw all that has ever been tried?
Edit: 2000 steps of backlash is obviously wrong. That’d be a whole rotation of the stepper. 200 also still sounds a lot. May be I still have the wrong number.