Picture: GLM is... well... lets say it is attempt to create physical development environment for field use. It is self-education tool.

General Learning Machine: Build less, learn more.

There is nothing special in this project. Instead it is a general framework. That's why it is my favorite.

It is interesting and kind of easy to tinker with circuits and code. But if you want to make something not only working but really usable, thinks get hard and dirty. At least if your main interest is field communication like mine.

After some years of solder&tinker, some typical bottlenecks started to irritate me.
First, have you noticed that most "electrical problems" in projects are in fact mechanical failures: loose connections, bad connectors, broken wires...real failures in electronic part of device are real minority. Comparing to mechanics, electronic parts are very reliable. About only way get them damaged is over voltage.

Second, there are many interesting ideas that would be worth of trying. But they require too much mechanical work to make them properly. So your idea ends up to terrible spaghettidevice. It may work but leave you unsatisfied. In field you need seven hands to keep them working.

Third, hacks consist often same blocks: power (battery, regulators), some user interface (buttons, lcd, potentiometers), connectors, filters, and those magic parts you play with (my fafourites are arduino and raspberry pi).
That means that even your best homemade invention is going to lose some parts to your next hack. That is one reason why it was not really worth of proper work. After all, when it is is just not interesting any more.

Fourth, you think circuits as blocks. For example. you need amplifier for microphone. So you search suitable circuit from net and build it. You chance couple of parts in circuit, to the parts you happen to have in junkbox. If it does not work, just try next one. What you really have to pay attention is only characteristics of input and output of circuit. You really have not to worry how amplifier works. You really don't know. It is just a black box. So what have you really learn? Not much.

And how about documentation? You know the frustration when you do not understand how your own made gear is working? Or program you made? I know how it feels...

So I ended up to build this general machine for learning, or "General Learning Machine". Main parts of this framework are battery +regulators, Raspberry Pi, Arduino Mega2560, some general user interface parts, 2 adjustable general filters, audio unit with internal loudspeaker, si4735-radio chip, and adjustable whip antenna. Analog and digital sides have own shielded chambers, and although made of scrap aluminium, chassis is durable enough for field experiments. Size of equipment is quite little: it fits easily to backpack. instead of making it large enough to anything, it has some general fastening points to external parts if room inside is not enough.

My principle with this project is: Build less, learn more. I have started to really design simple circuits to understand their principles, instead of just copy-trying them from net. And I have started to really studying coding, instead of copy-paste. And I will try to make here some proper documentation also. Even design before building.