Well, I put a couple of solar panels and a charge controller and a lithium Ion battery in a building I own. I put some lights off of it in the basement as kind of an always on low LED light so that I can walk over and turn on the main lighting, and I also want to put some lighting for some plants as I started seedlings this spring in this building and have the idea of keeping a few tomato plants and lettuce and such growing throughout the year in my temperate climate here. You know, the trimmings for a cheeseburger and such. It is nice to have fresh produce not too far away. I have done this before in a different building, but that was a much larger solar system and used a 400 Watt Metal Halide lamp. This is just a tiny system here.
While the battery BMS would kick in to protect the lithium cells, it has some drawbacks. First, if it kicks in my outback charge controller goes dead and loses its programming and I have to jump the solar to the batteries to turn the BMS back on and then set up the controller again, which takes a bit of time and is not something I want to do. Second, the BMS turns off at too low of a cell voltage for my liking. I really do not want to run these cells down below 3.3 volts, kind of between 3.3 and 4 volts in operation.
I could build an Arduino load controller as I have on larger systems, but that takes a bit of power to run and is overkill for what I need for this system for now. I could also rig a magnetic or solid state relay perhaps through a transistor and a pot to get it to turn on roughly where I want. An op-amp would be excellent to get the thing to snap on and off at a set point too, they do draw a bit of power, but for low power versions that draw is not significant. This is just lighting and I can ramp up or down the voltage as it turns on or off without harm, so I though I would see how cheap and quick I could get a transistor to turn them on and off using as few and inexpensive parts as I have in my junk bins. This was an exercise in cheapness and spare parts.
First I tried using a couple of 3904 NPN transistors in a darlington set up with a 5K pot to adjust the turn on voltage. Even though the gain was likely on the order of a thousand, the ramp up was still over several volts of range. I could make it work, but to get it essentially all on at 15 volts it still drew some current at 12V. I could probably reduce that a bit because I would basically need to use this to turn on a larger to-220 cased FET (Field Effect Transistor) anyway to switch the power that I would need. I was playing around on a breadboard with different iterations of this with my son.
I though well if it has this ramp up, I might as well just turn on a FET directly. I decided to try and sharpen the turn on by dropping the voltage from the high side with a zener. I used a 12V Zener there to the Gate and a 470K ohm resistor from the gate to ground to pull it down. This draws very little current when the FET is off. I found that with an IRF 610 FET the ramp up between off and on was still over a few volts. We substituted several other FET's and found that a 75339P gave the sharpest turn on. Thus with just three components we could control the lighting such that it was all off at roughly 13.5 Volts and all on at 15 Volts.