OK, let me try this question here.
I am designing a minimal component circuit to light two or three white LEDs from a 3V battery power supply using some small hearing aid-type batteries (needs to be very small and porable).
Since the components will be inaccessible (actually sealed - except in emergencies to change-out the batteries). I need a remote switch.
I've seen something similar that uses a magnetic reed switch and a magnetic "wand" to turn the LEDs on and off by waving the wand over the magnetic reed switch which is enclosed in a sealed case (needs to be waterproof).
Problem is, this circuit has an IC potted in black epoxy, so I can't figure out what component they are using for the LED driver/switch. It is obviously some sort of a push-on, push-off circuit, but using only 1 small IC, not the usual handful of components or bistable logic circuits I have seen bandied around in circuit design websites (one using one four pin IC is what I am looking for).
This whole deal has to be very small and efficient, since the batteries are not very accessible to change out (waterproof case).
Also, this is not for NASA, so overkill on voltage or current regulation are not needed, however, getting some life out of the small batteries is (while not frying the LEDs).
I just need something that will latch on when pulsed from the reed switch and unlatch when the reed switch is triggered again (I think the circuit I saw takes advantage of the switch ringing to go from one state to another).
I tried an SCR for this (which engineers usually DON'T want to operate in a latched state), and ran into two problems: 1) with 3V input the LED barely remained lit (voltage drop through the SCR probably excessive) and; 2) as soon as the magnet was moved away, the trigger current dropped too much to hold the SCR open and the light went out (e.g., the gate didn't latch).
Oh by the way, I measured the current through these LEDs when running free off 2-AAA batteries (no resistors) and it came out to 7.5mA per LED (they were still VERY bright though).
Any one-chip solutions?