As some of you know I am no longer working at the university on the microscope full time. I am now a freelancer, and am trying to find as much time as possible to work on open projects.
Recently experiment.com launched a low-cost tools for science challenge on their crowdfunding site. They want people to develop new tools. One thing we are regularly asked for in the OpenFlexure project is a microscope to look at bigger things like bugs and circuit boards. When we took the OpenFlexure to Panama, we met scientists studying orchid bees using dissection microscopes, they would love a low-cost portable microscope they could take into the field.
Working with Andy Quitmeyer who runs a field station in Panama, we have launched a fundraiser to develop an OpenFlexure-family Dissection Microscope:
The microscope will:
Be totally open so you can build one yourself
Have clear documentation to enable it to be built easily
Focus on good quality mechanics so the microscope is easy to use
If you are able to, please consider supporting us.
I guess a bigger working distance may be needed for this application. As part of the design process for a microfluidic workstation, we tested the RPi cam and other lenses to keep a compact design.
Here is a little contribution to community knowledge:
I’ve definitely spent quite a lot of time answering people who are curious about a lower-magnification system, and explaining that OpenFlexure could do it, but there are probably much better ways to do it. It would be awesome if Julian was able to come up with something that solved this problem with his customary quality of engineering
Playing with different lenes to increase the working distance is key. I am hoping to mechanises the lens to camera distance for some level of variability of the magnification. Some of this will involve some calculations, I think it might be possible to use pyoptic2 for this.
I think the first port of call is to create a sturdy frame with decent quality motion. And then to create a parametric design for the camera/lens mount that makes it easy for users to experiments with lens configurations.
Yes so these projects are what made me think we needed to start again from the mechanics side. Both of these show quite well that the performance of a simple camera and lens held above the sample is good and useful. However I don’t think @r.w.bowman or Andre would be insulted if I said that neither of these systems has been designed around robust ergonomic mechanics.
I want to create a system that has the main z motion control driven by a rack and pinion, via decent sized handles, in a way that is self locking so that the user can easily focus in the field. Variable magnification would be a bonus.
Definitely not offended The dissection microscope was knocked up in an afternoon in response to someone asking if they could look at microplastics. As @j.stirling says, the optics were more or less OK, but the mechanics were rubbish. I think a lot of the appeal of OpenFlexure is that it “gets the mechanics right” but currently that’s only really true for high-power microscopy (i.e. fields of view well below 1mm). For lower magnification work, something with a nice rack and pinion stage would be ideal - and I’m not aware of many examples of that.
FlyPi is probably closer - but again I don’t think it’s mechanically optimal. At least the version I built (which was a while ago) just slid the camera up and down on a rod, and locked in place with a screw - fine for doing a timelapse or taking videos (which, to be fair, is what it was designed for), but less ideal for interactive use where you’re adjusting focus frequently.
Talking of looking at microplastics (or other larger samples). If you want something less portable than what I am hopefully going to work on, but far more automated there is the 3D printer based microscope Niamh Burke is working on in Mark Pickering’s lab.
They are currently working through making the documentation in GitBuilding I don’t think there is a version online. But it is incredibly cool. She uses the fine motion control and electronics of an Ender 3D printer, with the optics clipped on where the print head goes. Even cooler, the clip is designed so you can quickly swap between printing and microscopy.
We have made some progress with the development of the field dissection microscope. Some more work is needed on the instructions before it is truly replicable, but we are excited to say that the prototype is starting to take some nicer images.
I have trouble imagining that one 5mm LED is sufficient illumination. Our cheap stereo microscope comes with one of these ring lights which has 144 of them: https://www.aliexpress.com/item/1862135697.html
I suppose it depends on the camera gains, but these were not too high, and it was certainly bright enough. Ring lights may give a more even illumination on objects casting shadows. It would be interesting to investigate ring lights, but many of them are much more complex to run than just powering up. Also as many of them are RGB, I am not sure we will get such a well rounded spectrum.
That is exactly a complicated as I thought it was.
This is the first prototype of a new microscope I am trying to get documented and out to people ASAP as I have limited time to spend on it at this time. The current design has:
no Arduinos
has no power running to Ardunios
has no mounting in the current frame to mount an Ardunio
has no documentation for how the user installs the Ardunio IDE
has no documentation for how the user compiles software and uploads it to an Arduino
uses the OpenFlexure software that currently has no plug-in for controlling the neopixel (if we wanted to have lighting control rather than it just being on).
Looking at the images above, the single LED clearly works pretty well. I don’t see why at this time I would divert effort from completing the documentation of the existing design and away from working on bigger issues with the design.
Once I have managed to get the first iteration well enough documented for replication then anyone else who already knows how to use an arduino is able to experiment with neopixels and contribute back the the project.
I’d be delighted if they do so, but my focus is getting complete documentation into the open as I promised I would to the people who backed the project.
Sure, every project has it’s own pace and time line. I did not want to suggest to drop everything and work on a new illumination module. All I wanted to say is that one 5mm LED is on the low side and it might be worth considering more light at some point.
The obvious thing to try first is if using a flash light makes a difference. This is the easiest way to replace the 60mW LED by a 1-5 W one. If this seems to help there are many ways to do a more powerful illumination. The simplest one is just replacing the 5mm LED by a 1W one - possibly running at only 0.5 W.
But as said above: If your camera is coping just fine with the current light conditions this is not a priority and I understand that there are many more important things to do.
To be fair I was shocked how well the single LED worked myself.
But I am taking a bit of an “ain’t broke don’t fix” mentality right now as I am behind schedule getting a documented first iteration to collaborators.
Absolutely not a suggestion that I’m expecting anyone to act on, but I recall ages ago at a Plymouth microscopy course, Brad Amos was very keen on some simpler ring lights. They’re made for car headlamps, and are “dumb” white LEDs - a ring of blue LEDs underneath a (continuous) phosphor layer. I think an example of what he meant would be these “angel eyes”. These may not be available from the usual “maker” suppliers, but seem easy to get on amazon. However, I doubt anyone will give you a spectrum. Also, at 3.6W they will require significantly more power electronics than you have at the moment!
This is more just a note in case anyone wants to do it in the future. I suspect if all you want is a non-programmable ring light, these will be more even, more powerful, and easier to use. Of course, a neopixel ring lets you do lots of other fun tricks with computational imaging, so I’m much more likely to play with one of them in the lab…
It works OK for the current prototype, but I can see it being an area of future investigation, as there is certainly some slip-stick. The slipstick is improved massively by lubrication of the dovetail and it is essential that the dovetail is printed parallel to the print bed.
Things that might want looking into eventually:
Experimenting with different tooth sizes. Too small and you run into issues with print resolution
Experimenting with smaller pinions (this would require smaller teeth too)
Adding a gibb system to adjust the dovetail, this would allow better control of the play/friction in the dovetail. Currently the friction to stop it dropping out is provided by the bearings.
