I have been quiet the past few months about this project, because I was still struggling with power problems. Adding the boost converter was a great idea, and it works very well, but if you make the speaker play white noise, it will still overwhelm the batteries and make the voltage drop enough that the display resets, and you get a white screen of death.

I tried lowering the gain on the amplifier (I dedicated an extra GPIO pin for that, so that I can do it with software), but it didn't help. So I let the project sit for a while, while I waited for inspiration.

The inspiration struck a few days ago, and I was lucky that I could test it immediately, by cutting a trace on the board and soldering a resistor in series with the speaker. I know, it sounds obvious now, but you have to remember that I have very little experience with analog circuits. So yeah, the resistor limits the current drawn by the speaker, at the cost of lowering the volume somewhat. But now the screen doesn't reset, and the volume was too loud for my tastes anyways.

I think this was the last problem I was struggling with, and the next step is to bodge the remaining ten prototypes I have assembled and run a workshop with them, to see if the participants find any new and exciting problems. If they don't, and I will start looking into producing and selling them.