Initial build of Delta Stage. This was assembled quickly with a breadboard, Arduino Mega instead of Nano, bad pin header soldering and some missing screws. Heat sink was added to Pi 4. Future updates incoming…
Configuration: Grid Illumination and reflection optics module (Plan 40x, only beamsplitter).
Some Images
Full Build Image:
In Action:
Calibration Slide (Smallest Div 0.01mm)(Brightfield)
Normal Hair (Brightfield):
Thinner hair (suspected eyebrow/eyelash) (Brightfield):
Grid Illumination (Saliva left on a coverslip for 90 min):
Brightfield with 4 central LED (Color 5,5,5)
Darkfield with outer LED (Color 10,10,10)
Oblique Illumination with 2 central LED (Color 5,5,5)
Sketchy Reflection Transmission:
- Still have some missing parts for the illumination (variable resistor to control brightness), so tried shinning a torch into the hole.
- Weird (but cool) flourecense effect when not recalibrated (it was previously calibrated for grid illumination.
Issues (In case anyone encounters them):
- Screw thread for the sample clips got stripped. Solution in another post (search for stripped thread in the forum). ALSO, do not panic, it still works fine without the clips.
- Camera to PCB cable/socket loosened pulling on CSI cable. Solution: Unscrew with camera facing down and push the connector back. The loosening might not be visible by eye.
- Initally seemed like there’s serious dust issues but calibrations seems to have removed those artifacts.
- 40x mounting height is really hard to get; especially when I switched from an upsidedown slide to a coverslip to slide converter. A coarse focus with larger range of motion will be good.
- Tried viewing hairs but seems like 40x is too high and the entire hair can’t get in focus. Depth of field of 40x is around 1μm while thickness of hair is around 17-181μm. Stacking might fix this issue.
- Because I followed the reflection BOM and also used a dovetail, I missed some screws
1 Like
Grid LED Color & Image Stitching (Onion skin cells)
Color set to (10,5,0) for 1 LED in the center. Stitching done using instructions from OpenFlexure / openflexure-stitching · GitLab
Before stitching:
After stitching (5 step in xy, fast focus):
Some stitching artefacts can be seen but it is insignificant and most of the cell features remains visible at a high resolution. View area is approximately 4 times (doubled diagonal field of view, matching that of a 20x objective).
1 Like
Nice.
Just to clarify, there should be no artefacts introduced in stitching. There is no processing of the overlapping regions, the output selects one image or the other. This is deliberate. It gives some discontinuitoes if the stitching has placed images in the wrong place, or if colour or focus change. Leaving the sharp change between images means that you can easily see where that happens.
In this case it looks to be focus being different in different images.
Could it be because of fast autofocus being used instead of find?
It is most likely to be because the sample is not really thin, so there is not a single plane that can be considered to be in focus.
There are also some potential problems with the way the microscope returns to the focal position after scanning z. @JohemianKnapsody has worked on that, but I think that the latest server v2 does not have any of those improvements.
It may also be different for the Delta Stage. The three motors move together for focus, but the same motors also do x and y, and may have to change direction between x and z or y and z moves. There could be artefacts from backlash that mean that when all three motors move together, they don’t actually all start the stage moving at the same time. This could either be because only one motor is changing direction, or all motors change direction, but the backlash is slightly different. I don’t know whether any of this is handled in autofocus.
I agree with @WilliamW , that looks to me like two images that are being stitched together, but from different focus positions. So the membrane(?) is thick in one, and thin in the other
