Problems Encountered in 3D Printing
We learnt how to use Autodesk Fusion 360 from scratch. Remote collaboration on Autodesk Fusion 360 was a chore as there were lags in the system. It was better to design and edit with group members face to face.
The printed model from our drawing on Autodesk Fusion 360 may not fit our purpose. This is also due to our inexperience in designing hardwares. For instance, in our design of the lens adaptor, we had to try multiple attempts of reprinting as the slot that is supposed to fit the lens was too tight based on our measurement. Our initial design also did not take some aspects of our 3D printer frame into account. The lens adaptor could not fit tightly on the 3D printer frame due to this lack of considerations. We need to drill holes to make the parts more compatible.
Time management is also important in preparing our prototype. As many of the parts we need to print take several hours to complete, we had to plan well so that the printing process does not interrupt our progress.
Evaluating Our Product
Our prototype is well-designed and is a decent modification from the original microscope.
In terms of the original microscope function:
The functions of the microscope are well-preserved after our modification. We are still able to obtain clear images from our microscope setup, and with the 3D printed microscope base, we can hold the specimen slide firmly in place and view it with the light source.
Our improvements on the microscope:
The microscope setup now has a camera fixed at the eyepiece, which became possible by our addition of a camera holder. The camera allows the field of view to be directly accessed from our computer or the Raspberry Pi screen. This was not possible with the previous biological microscope. Furthermore, to obtain cell images with the best quality, we took inspiration from the original microscope’s iris diaphragm and created our own light blocker. By adjusting the position of the light blocker we are able to obtain bright field and dark field images, and both types can be experimented with our machine learning model for cell counting.