Given that NiMH cells can be slow-charged at a constant current rate for a given amount of time, I built this circuit. Provided the timed slow-charge condition is verified, you can live without end-of-charge algorithms.
Some charging facts (if I remembered where I read this, I'd credit the author).
A NiMH cell, if charged at C/10, in 10 hours will reach approx. 66% of the fully charged state. You achieve the remaining 33% with 5 more hours. That's where the magic 15 hours come from.
My constant current circuit required a PNP transistor. I picked a BD646 from the junk box in a TO220 package. Since the transistor had to be floating from ground, I needed a way to attach it to the metallic enclosure for thermal dissipation without electrical contact.
Forgot about nylon screws, a search through the electronic junk showed this solution: wrap the transistor in what I believe to be a mica foil and then press it against the heatsink/box.
This is how it looks like:
I recharged a 9.6V pack while watching the temperature and the current. The transistor runs barely warm, while the power resistor heats up considerably.The pack under charge will increase its voltage, thus effectively reducing the charge current, in a auto-stop fashion. But since this occurred quite early in my charge cycle, I had to increase the input voltage to more than 14V.
One word of caution about the plugging sequence.
Charge start: connect the power supply and then the battery pack
Charge end: disconnect the battery pack and then the power supply
Otherwise the pack will discharge into the power supply. Probably a 1N400x diode on the +Vcc line blocks the process and allows to use a timed AC socket.
