The stage I inherited was built for a different job
The mesoscope started life on the OpenFlexure platform, and OpenFlexure has a genuinely clever motion system: a flexure stage nudged around by tiny geared stepper motors. For the job it was designed for—small samples, gentle moves, a compact footprint—it’s elegant, and I had no reason to touch it.
Then the experiment changed what the stage had to do.
The experiment images cells on chips, and it needs to image along them—across the whole chip, repeatably, sample after sample. The obvious way to do that is to move the specimen table under a fixed microscope, which is exactly what the flexure stage does. But here the chips have to stay put. Disturbing them between captures isn’t acceptable, so the motion has to happen on the other side: instead of moving the sample under the instrument, the whole microscope travels along the chips.
That one inversion is what broke the original design. The moment the thing being moved is the entire optical assembly rather than a small specimen on a flexure, you’re no longer in the weight class the stock steppers were built for.
Why the small steppers had to go
The original geared steppers weren’t underpowered because someone chose badly—they were sized for the stage they came with. Once the structure around them had to grow to fit the experiment, they were being asked to move more mass, further, more repeatably than they were ever meant to. You can push a motor past its comfort zone, but in an imaging instrument that shows up as exactly the things you can’t tolerate: missed steps, drift between captures, and positions that don’t come back to where they were.
So the decision wasn’t really “better motors.” It was admitting that the whole motion structure—motors, drive mechanism, and the frame they mount to—was now the wrong shape for the problem, and that bolting a fix onto it would be slower than rebuilding it correctly.
Moving to NEMA 17
I moved the X and Y axes to NEMA 17 stepper motors, and redesigned the structure around them rather than the other way around.
Motion now runs on lead screws, with the microscope riding on a scissor-lift bench. The NEMA 17s turn the screws; the lead screws convert that into controlled, repeatable linear travel with enough mechanical advantage to carry the weight of the instrument rather than a bare specimen. The scissor bench gives the assembly a stable, adjustable platform to move on. It’s still in the mesoscope spirit—standard motors, lead screws anyone can source, and printed parts tying it together—just sized for a payload that’s an order of magnitude heavier than a cell on a slide.
The lead screws matter for more than load. They don’t back-drive, so once the microscope is positioned it stays there, and the pitch sets a predictable relationship between motor steps and millimetres travelled—which is exactly the repeatability an imaging run across many chips depends on.
The point of NEMA 17 wasn’t raw power for its own sake. It’s that they’re the boring, standard, well-supported choice: predictable torque, drivers everyone understands, mounts everyone has modelled. When the rest of the instrument is already doing something unusual, the motion system is exactly where I wanted no surprises.
Let the experiment set the shape
The lesson I keep relearning with hardware is the same one that applies to infrastructure: the requirements come first, and the design serves them—not the reverse. The OpenFlexure flexure stage was the right answer to its original question. My experiment asked a different question, and the honest response was to stop preserving an elegant design that no longer fit and rebuild the mechanics around what the work actually needed.
The microscope is more capable now not because the new parts are fancier, but because the structure finally matches the experiment it has to serve.