My Tech Wishes for 2025

I know it’s a bit late in the year for this kind of post, but I have been busy with other things. In any case, I decided to make it about general technology related wishes, and not just CircuitPython, but large part of this is going to be about CircuitPython anyways. Or at least related.

Media Transfer Protocol

Exposing your microcontroller to the computer it is connected to as a disk drive is getting really popular now. And it makes a lot of sense too — it doesn’t require you to install any drivers or programs, it easy to understand, and very convenient. CircuitPython does it too and I love it. But it does introduce a number of problems:

  • You have to use a filesystem that is supported on most operating system, and with the small sizes you have, it’s usually FAT12, which is not great for flash wearing off or fragmentation.

  • Since diskettes are not as common anymore, proprietary operating systems don’t have their implementations of FAT12 tested as well as other things, which means there are lots of bugs in MacOS and Windows.

  • Tools like virus scanners or disk monitors see those drives as regular disks and often interact with them in ways that breaks them.

  • You can write to the disk only from one side, either through USB, or from your code on the device, but not both at the same time.

All of those problems could be trivially solved if we could only use the MTP protocol instead of exposing the disk as a physical device with USB MSC. MTP uses commands to manipulate the files, kinda like FTP, instead of writing to the disk directly, and writes themselves are done by the microcontroller itself, so it’s easy to use any custom filesystem you want, and to ensure proper interactions when multiple operations are happening from different sources.

There is only one problem, and that is iTunes. You see, back when Apple’s flag product was a music player, they were very jealous about it, and they tried very hard to “protect” their users from using any other music player. They made sure their player uses a custom, proprietary protocol for transferring the music collection, and they also made sure that Apple computers didn’t support transferring of arbitrary files using MTP. They had a problem in that most cameras were already using MTP to transfer photos, and their users needed that, so they made it an exception: you can use MTP on a Mac, but only for transferring image files, and not any other file. Neat, eh?

Today you can only find iPads in museums, and iTunes is just a memory in some boomer nightmares, but the decision they made back then stays. No arbitrary files with MTP, if you use a Mac, you have to install a special program to use MTP for anything else than photos. So any microcontroller device that uses MTP would not work out of the box on Apple products, which means that nobody does it.

So my wish is for someone at Apple to revise that decision, recognize how stupid and harmful it was, and remove that stupid limitation from all of their operating systems, thank you very much. I would also like to see some ready to use libraries and examples for using MTP, but that will surely come once it’s usable.

CircuitPython on CH32V208

This is something I would love to do myself, but I’m scared to even start, as it is a huge amount of work, and just outside of my comfort zone. Basically, the bluetooth versions of the CH32 chips tend to have a lot of memory for the bluetooth stack, and that makes it theoretically possible to run CircuitPython or LUA on them (as long as we don’t need the bluetooth). It’s a nice cheap chip, and it would be perfect as a replacement for the SAMD21 chips I currently use for the PewPew devices.

Of course porting whole CircuitPython to a new chip is a lot of work, and it’s something I have never done before, so naturally it’s scary. But people do it, so it’s definitely possible.

Voice Synthesis

After my failed attempts to port the SAM voice synthesiser from MicroBit to CircuitPython, I’m happy to see that someone ported it to MicroPython, but not as a C library, as I wanted to do, but literally rewrote it in Python. Now I need to take it and see if it works in CircuitPython as well. Then I can make my robots speak.

Servo Controller on CH32V003

This one I’m actually perfectly capable of finishing on my own, I just need to find the energy to work on it. I basically want to control 16 servos with good resolution using PWM signals specified through I2C. I already have the PWM part working, and there are plenty of examples for I2C slave. It’s just a question of sitting down and actually writing the code.

Once I have it, I can use it in the future versions of FluffBug and WeeBug, and also a robot for the new small OpenMV board. It would free up 8 pins on the Xiao, which could then be used for a display or I2S amplifier. It would also allow for connecting additional servos, for things like arms or moving features.

LED Matrices on CH32V003

I also have two projects with LED matrices for the CH32V003 that I still need to program. One is a spectrum analyzer thing, and the hard part is the FFT code for it (though there are plenty of examples), and the other is a business card fake PewPew—not really running CircuitPython, but with some of the PewPew games implemented directly in C and hardcoded in the device.

GIF Pendant

Finally, I’m also still working on my GIF pendant project. Now with a round screen and CH32V208 as the brains. I have a TinyUSB example for USB MSC in RAM working on it. What I need to figure out how to put the storage on the internal flash instead, and how to access it from the code to render those GIF files onto the screen. Again, it’s something I have never done, and there isn’t much documentation about it around, so it’s a lot of trial and error. I wish someone who is more knowledgable about TinyUSB and USB MSC could help me with that.

And that’s pretty much it for me for 2025. It’s mostly projects I would like to finish (or would like someone to finish for me), except for that Apple MTP thing. Will I get any of my wishes come true? We will see.