Any alternative needs to have at least 2-channel MIPI CSI, a compatible architecture like ARM Cortex-A53 or Cortex-A72, and a USB port. So, the Jetson Nano would work, the Rock Pi 4, ASUS Tinker Board, Banana Pi M3, NanoPi NEO4. Definitely others. I know the NanoPi can actually be ordered.
Here’s a different idea though - instead of using one of these boards, just use your desktop computer. You can get a CSI to USB adapter:
And use a USB-based Sangaboard for your motor controller.
Thanks Thomas. This sounds like a very interesting alternative. How do you install the openflexure software on a mac? do you create a virtual desktop? Can you control more than one microscope with 1 computer?
The drawback of this approach for me is portability. It looks like the libre ALL-H3-CC TRITIUM has a 4ARM Cortex A53 crypto architecture, 4 USB ports, and a camera port. I cannot find the camera port specs though, but in the picture it seems identical to the CSI port of the RPI.
The short answer is that it’s a generic parallel interface bus not compatible with CSI. It may be possible to use extremely low resolution parallel-interface cameras with that interface, but then again, the Pi’s GPIO can be used that way as well. There’s no evidence that anyone has used the Libre to interface to a camera.
Thanks @WilliamW and @tkircher!. I learned a lot today about this boards. The CSI-USB adapter or the USB camera alternative sounds like a very cool idea. I can avoid having a computer board all together and control the motor drivers and camera from the PC like you suggested. I am going to experiment with that!
How do you install the OFM software on an Mac or PC without using a raspberry pi?
The OFM server software is Raspberry Pi only. I think of it more as a firmware that enables all the functions of the hardware and allows them to be controlled from another computer using the web-app, Blockly or Python etc.
Some functions would be relatively easy to reproduce outside the OFM server, like sending the serial commands to the motor controllers. You can also easily take photographs with a USB camera. Combining into a series of picture and move needs an integration of those two functions. You can have a much lower cost manual microscope in this way.
What would be almost impossible to reproduce on a PC or Mac are the camera autocalibration and the fast autofocus algorithm. These both rely on access to the Raspberry Pi camera video stream and the camera image pre-processing. You can use the microscope without these functions, but the images really come alive with the flat field and colour correction, and fast autofocus is really necessary for slide scanning or long-term studies. There are other integrated single board computers that would be able to do these things, but none is directly software compatible.
I understand. Just trying to be creative and overcome the chip shortage and supply chain issues. I seems that RPi wont be available until the sometime in 2023 which may limit everyone’s ability to build a microscope.
The Pi supply is horrible. They do pop up quite regularly, but then are sold out very soon. However back-order does work with some/many approved resellers. Order and wait can be easier than wait, search and pounce on what becomes available.
Currently the Raspberry Pi is the only way to take advantage of the nice image pipeline in the Openflexure server that gives a colour-corrected flat illumination, and also to allow fast autofocus. The flat field and colour correction really make a huge difference to the quality of the images and so also to the usability of the microscope.
With the Delta stage currently the only optics modules that are available are designed for the Raspberry Pi camera.
It is very frustrating that the Raspberry Pi supply is still so difficult. Currently they project that availability for individual buyers will be better in the second half of this year, which is still a way off. I am having to modify some parts to fit Pi3 today as I have only used Pi3s anywhere in the lab.