Hello everyone, this is just a (longer than expected) introductory post to let you know about the additions to the microscope my lab partner (Simon Cox) and I are working on, with the goal to speed up imaging of hollow core optical fibre cross sections for anyone drawing fibre at the University of Bath’s facility.
We’ve only been working with the microscope for a few weeks, so still have a lot of exploring of the forum and GitLab pages to do, so please forgive any redundant ideas or questions.
The two main focusses of our project are:
How to repeatably position a fibre, with a diameter of ~100-200microns, so that it is in the frame of a 40 or 60x objective.
Processing the image from the microscope to quantify parameters of the fibre, such as core/capillary dimensions and symmetry.
The latter isn’t really something to ask about on this forum, but if anyone knows of any Python libraries/techniques for this, suggestions would be welcome.
As for the former, we have been prototyping a few different ideas of how to do this using only 3D printed or commonly available parts (M3 bolts, rubber bands etc.).
The first idea used a V-groove with square cut out sections and corresponding teeth, fixed to a springy flexure, as shown below. As the print direction was going to affect how smoothly the fibre and teeth could be slid into the groove, the two sides of the jaw we attached by simple dovetails so they could be printed separately to rest.
This design worked reasonably, but the end of the fibre would tend to deviate each time it was clamped, so we moved onto other ideas, but this could still work as a back up.
Next we wondered about using three hexagons, which if slid along their faces, would form a triangular aperture in between them. 3D printed hexagons tended to have slightly rounded corners, which wouldn’t close small enough to clamp the fibre. They could be filed down by hand, but we were concerned that this wouldn’t retain the exact angles between faces needed. Large machine nuts (e.g.M8) had similarly rounded corners, so we ended up working with M3s.
The image below shows our first idea using these. The smaller circle would be held still, whilst the larger is rotated, which in turn rotates each nut around its own centre. This causes the nuts to slide along their faces, whilst the plastic legs holding each nut flex, allowing for the centre separation to increase. The o-ring then pulls the aperture closed when the circle is twisted back.
This also worked reasonably, but had the practical problem of feeding a fibre down from the large circle (which would be at the top when on a microscope), to the nuts (just above the objective lens). A quick test of a 3D printed funnel didn’t work well, so a new prototype for positioning the 3 nuts is in the works (currently worked on by Simon).
The third design, which I’m currently working on, uses 4 cuboids, held together with a combination of single and double dovetails, which when slid against each other, open up a square aperture in between. The dovetails also prevent the fibre from slipping between gaps in the cuboids, meaning the faces don’t need to be perfectly pressed flat into each other. The difficulty with this idea is moving the cuboids in phase so that the aperture opens symmetrically, and closes to the same place each time.
This design is also the biggest victim of the ridges on vertical walls inevitable in FDM printing. The cuboids are printed so there is no ridge-over-ridge sliding, which minimises friction (not sure how to explain the relative orientations of 3D printed pieces sliding against each other, hopefully you get the idea).
This design can hold the fibre with a decent grip, but we are still working on smoothing the motion to make the aperture closing more accurate.
Anyway, if anyone has any advice, ideas, designs or questions on the fibre holding (or image analysis)
we’d be very happy to hear it.