First build, High Resolution v7, with Arduino

[josefnovak]

Hi!
I am a scientist from the Czech Republic. The OpenFlexure project caught my attention. I do quite a lot of microscopy and image analysis, but frankly, I am not as familiar with the optics behind the microscopes as I would like to be. So, I decided to build a high-resolution version of the OpenFlexure to learn more, and I would like to thank the Openflexure community for this great project!

A photo of the microscope during its assembly is attached, the image is a dot from a permanent marker I used for initial calibrations:

The build is mostly according to the instructions using Bresser DIN plan-achromat 40x objective. Two notable changes are the bulky Arduino and the illumination. I am quite sure I have ordered the suggested Arduino, but this one has arrived. It works fine, but it probably won’t fit inside.
The second change is in the illumination. I decided to buy components locally as much as I could, and it turned out to be quite impossible to find condenser lens. So, I tried to use diffuse white LEDs from Raspberry 5 mm LED kit and it worked fine.
A small change is in the assembly of the actuators, as I could not find the Viton bands. Instead, I have regular rubber bands folded in half; so far they work fine. It was double the fun to assemble the actuators, though.

The software works really well, and I would like to applaud the programmers.
To make it work, I had to install specifically “Arduino UNO R4 Board”, which I somehow missed during the first attempt. Then, I had to run “sudo ofm upgradepipenv” as I found somewhere in the forum. At this point, the microscope runs only if I run it as a root, but runs nicely.

The printing was done using the default setting from the supplied files using Prusa MK 4. The print is great, and the microscope looks beautiful. The printing took two days and a night, but I forgot to print additional holders for Arduino electronics.

I planned to assemble the microscope over the long winter nights, but surprisingly, it took me just three evening to get to the state at the photo. Obviously, I need to do something about the wiring and probably buy a different Arduino, but the microscope works nicely. To my surprise, the assembly was not complicated at all; the instructions were really clear.

Concerning the sources of the material:
Prusa MK 4 and PLA filament was used to print everything. I used the files with default settings.
Objective is bresser-din-plan-achromat-40x from a local distributor of microscopes.
12.7 mm achromatic doublet lens (50 mm focal length) are from ThorLabs
Most of the electronics are from a local Raspberry Pi distributor.

In total, the material cost me approximately 400 EUR without shipments. The objective was roughly half of it.

After I built the microscope, I was quite busy and did not have time to tinker with it. Recently, I turned it on again just to test if it survived transfer between places. It did, but unfortunately, I again need to focus on something else now.

However, when I find time, I would like to use it for time-lapse microscopy. That would probably eventually mean to build a microfluidic module. Also, I would like to test fluorescence module. But that are plans for future

Josef

[j.stirling]

Awesome. Thanks for this update. Exciting to see once built in the Czech Republic. We will add you to our map!

The electronics are always an issue if you can’t get the Sangaboard. We don’t have a distributor in Czech Republic yet, but there are at least some in the UK. Hopefully if we can show that there are other use cases for the Sangaboard we can get larger distributors to pick it up!

On the topic of the arduino being too big, it looks like a uno rather than a nano, this may explain why it is so chunky?

Good to hear the elastic bands work. The issue we have found with elastic bands in the past is over time they become brittle and snap. If they do I think these are a local source for you:

OR 30-2 V | O-ring 80SH FKM (FPM) | HANSA-FLEX Shop

We look forward to hearing updates on your microscopic adventures!

[Drbrian]

This is great. I have pretty much printed everything, but the lack of a sangaboard was limiting me. Clever idea on just bypassing it, I pretty much have everything for it, including a spare uno lying around.

[josefnovak]

Thank you for your response!

I’ll check the supplier of the Viton rings. I actually saw some after I built the prototype, but not the correct size. I am quite sure the bands will snap sooner or later, so I’ll need a proper replacement.

Concerning an alternative to Sangaboard, I decided to try Arduino in part because I am also trying to use Arduino on some other project. Now, I have two of them, and both are indeed variants of Uno.

The other Arduino uses CNC shield to communicate with the drive boards for the motors. As I am looking at it, the pins look the same so they should be compatible. I have to disassemble two semi-working projects, switch the shield and switch to the different power source, though. I’ll let you know if it works, although I probably won’t find the time this week. It would solve the wiring issue, though.

I have a couple of additional images from the assembly of the microscope…

The Prusa MK4 printer with the body of the microscope, I liked the gray filament for this part. The base is black, small details are in fancy yellow.

All the parts at the start of the assembly. Note that I have ordered small parts in excess:

The fun part - assembly of the actuators with a rubber band:

The animals study the assembly instructions under the illumination module:

One of the first images, mice spleen stained with Hematoxylin&Eosin .

The purple part in the center is a white pulp region, not a problem with the color balance.

[WilliamW]

@josefnovak

The other Arduino uses CNC shield to communicate with the drive boards for the motors. As I am looking at it, the pins look the same so they should be compatible.

CNC shields are usually designed for bipolar stepper motors. The motors on the OpenFlexure microscope are wired as unipolar steppers. To make them work on a usual CNC shield you need to adapt the wiring where it joins to the motor.
You will also need to modify the Sangaboard firmware to get the microscooe to talk to the shield.

[WilliamW]

@josefnovak

The second change is in the illumination. I decided to buy components locally as much as I could, and it turned out to be quite impossible to find condenser lens. So, I tried to use diffuse white LEDs from Raspberry 5 mm LED kit and it worked fine.

With this setup you will get clear images, but you will not get the full performance out of your €200 lens. The resolution of a microscope depends on the maximum angle of light rays that are used in the image, known as the numerical aperture (NA). It is important to have a high enough NA in the illumination as well as in the objective lens. With your simple LED setup the maximum angle is a 5mm diameter LED at a distance of ~40mm. This is an NA less than 0.1. Your lens NA is, I think, 0.65 so you need that NA in the illumination to get the most out of the objective.

[josefnovak]

Dear @WilliamW
thank you very much, you are absolutely right. The CNC shield is indeed wired to bipolar stepper motors, so this is not a simple solution for my messy wiring.

I’ll try to get condenser lens and borrow some calibration slides to check the difference!

[josefnovak]

Well, one deadline down, one to go. I had a little time to work on the condenser, but not enough; I will continue next week.

The very first thing I actually bought for the microscope were glass beads with flat bottom I saw when I was looking for the potential lens. They have roughly the correct shape and size and considering that refractive index of glass is around 1.52 and PMMA 1.4893–1.4899, according to wikipedia, they may work. In retrospect, it would have been easier to order a kit with the microscope, but this is much more fun

As written somewhere in the instructions or on the forum, glass is not ideal. The lens are maybe slightly bigger than should be. Since they are from glass, they cannot be forced into the condenser.
So, I tried to glue the lens to the bottom of the condenser. I found that: 1) Superglue doesn’t glue glass to PLA at all 2) it does glue glass to the table within a second 3) it can be removed from glass nicely by isopropanol. So, I decided to tape it to the condenser instead.

With the lens, the image looks a bit better, although quantification have to wait. I took a couple of pictures from different positions of the condenser, but I took the pictures with lower resolution. So, only the best position with the maximum resolution follows.
I tested StarDist segmentation and it works fine. It struggles a bit in the area packed with cells and on the right side which is a bit out of focus, but that’s understandable and the results are still reasonable. So, we are looking at 4369 cells.