In order to lengthen battery life of a portable thingy, I am building a couple of Powerbank Switches, and learning electronics at the same time. The idea is that the switch, controlled by a PIC, switches between up to seven USB Powerbanks to a common output. Maybe one switch operation a day, round robin, so that they all can charge the internal Li-Ion batteries of the thingy.
I made the first switch with five 2-pole bistable relays coupled in a tree pattern so that only one path led from powerbank to battery regardless of the state of the relays and the PIC (I felt clever there!). And it works. It even "tickles" the switched-to powerbank with a 50 ohm load, senses if the powerbank wakens, and switches to the next one if it is "dead".
In the process I found out that a powerbank drains completely in 14 days if kept awake, since charge indicator LEDs and the internal voltage converter designed for 2A are power hungry. And as others have found, it needs to be "tickled" about every 30s with a higher load if only a little microcontroller is connected. So I use instead three Size D drycells as "logic battery" to save a powerbank. The first such battery pack I rolled in a brown cardboard tube with wires and tape around. A so-called "IED" - Improvised Electric Device.
But there were so many components, two PIC pins, two drive transistors for every relay. And fancy LEDs and lots of resistors. So I decided to use MOSFET only. And stepping up by stepping down to feed the logic through a 3.3V converter to save battery.
I drew a schema, and soldered wildly a dense construction on a pertinax board and it didn't work. So I tested the design on a breadboard. 3 Powerbanks, 3 IRLZ44 MOSFET, 3 used dry cells (>=4V) for circuit power, some cut USB cables, and a little lightbulb (12V 10W) as artificial load. And no 3.3V. The MOSFETS are put low-side (another new word for me, yet understandable), so I have connected all powerbank "+" to output "+" rail, individual powerbank "-" to MOSFET Source(s), all Drains together to output "-" rail, and a 1K resistor net to tie all MOSFET Source to the "-" side of the battery operated logic. If I didn't do this the Gates would float, leaving the MOSFETS only half closed!
But the breadboard design only works intermittently. The MOSFETs can only switch on if one powerbank is "awake", i.e. already outputs 5V. Or I believe they all switch, VSG voltage is 4V which is enough, but I think that the MOSFET does not "open" if there is no VSD voltage, and the powerbank to provide this has of course turned its internals off to save power. I turn the MOSFET on, by touching the Gate pin with a 4V wire, nothing happens, I then turn on the powerbank with its On button, and the light bulb goes on. So the FET has indeed switched (or rather the FET capacitor is charged). I touch the gate with the battery "-" wire and it closes so the light goes out. And then I can open it again with the "+" wire. But only while the powerbank is awake.
It wakens sporadically if I connect/disconnect wires, which I interpret as the powerbanks maybe sense changes in capacitance and wake up and work.
This was no problem when I used relays, the powerbank would always wake up and drive the light bulb.
Can I solve this? Maybe switch on a FET, then (dis)charge a capacitor between battery "+" and output "+" so that the powerbanks get "hiccups", wake up, feed 5V so FET opens, and everything is fine. But what happens when the thingy turns off charging for a while or doesn't want to charge right now, will the connection then "collapse"? I.e. powerbank shuts off, MOSFET closes too and won't open. Is it a bad idea to let the PIC try to drive the Drain high? Pullup resistors on all Drain? Ugly ideas, since I am no expert.
Or should I go back to my relay tree? After all, it's potential free. Fail safe in that it works in the last state even without battery. And it goes "click"!!