9V NiMH charging

I think the circuit below would do it. It's based off of a simpler charger here:  Smart Doorbell System – Part 6 – Chime Module Complete Circuit Design  and based on information from a battery mnfr PDF doc which specifies that it should be OK to charge at 0.1C for at least a day. 

This circuit charges at 20 mA, and would take 15 hours to fully charge the battery. The battery would need to be manually disconnected within 24 hours. 

However, there is also a microcontroller interface, so that you could turn it on/off based on a timer if desired (or any other logic circuit that can perform the timing).

Without the microcontroller, you'd need to remember to disconnect after a day of charging. If you need a PCB for this, let me know (it's fairly easy for me to convert to a PCB, since I've already created the schematic in the same CAD software that would create the PCB).

There's no guarantee that it will work, but it would be easy to test and then that would provide reasonable confidence with it (I'd trust this more than a cheap charger).

It was bugging me that there wasn't a timeout. I've been in the position where I've forgotten to unplug a charger.

The timer needs to be somewhere in the region of 16-24 hours. The design below (untested) uses a load of divider chips. It looks a lot, but 74HC4060 are cheap (and probably there's no size restriction). All parts here shouldn't cost more than a few $ total.

The crystal output is just divided loads, and when 19 hours is reached, the TIMER_EXPIRY signal goes high, and that also prevents any more counting.

The timer isn't essential, and if there's a microcontroller attached, then it's redundant (unless such a hardware timer is preferred).

The main circuit is tweaked to add the TIMER_EXPIRY connection as an input:

Also, while I was at it, I replaced the constant current resistor with a rotary selector, since it appears there are PP3 batteries with a range of capacity. Probably overkill to have such granularity, but the rotary switch could be replaced with a wire link to permanently set a current if desired.

It was bugging me that there wasn't a timeout. I've been in the position where I've forgotten to unplug a charger.

The timer needs to be somewhere in the region of 16-24 hours. The design below (untested) uses a load of divider chips. It looks a lot, but 74HC4060 are cheap (and probably there's no size restriction). All parts here shouldn't cost more than a few $ total.

The crystal output is just divided loads, and when 19 hours is reached, the TIMER_EXPIRY signal goes high, and that also prevents any more counting.

The timer isn't essential, and if there's a microcontroller attached, then it's redundant (unless such a hardware timer is preferred).

The main circuit is tweaked to add the TIMER_EXPIRY connection as an input:

Also, while I was at it, I replaced the constant current resistor with a rotary selector, since it appears there are PP3 batteries with a range of capacity. Probably overkill to have such granularity, but the rotary switch could be replaced with a wire link to permanently set a current if desired.

Thank you, but to be honest, I think a single 8-bit 6-pin MCU would replace the timing circuit and maybe also the charging logic.

Yes, it would be greatly simplified with a microcontroller.

I'll document the rest somewhere and maybe finish it if it helps anyone.

I couldn't settle on a target microcontroller (and didn't fancy coding anything currently), so I figured I may as well make it microcontroller-agnostic, i.e., one could unpopulate those additions if the microcontroller interface is used.

It's just a mental exercise, I have no serious suggestion for anyone to construct this unless they enjoy soldering a lot of parts. Plus, my design decisions won't be the same as others; I just used what parts I was familiar with (I'd already tested that constant-current portion in that Bluetooth project, so I lifted that as-is).