Make light, not heat: LED lighting as 18650 battery test fixtures


Jan 12, 2017
Ooops. This thread really belongs in the "DIY Powerwall Builds" forum. (Or maybe, ideally, in a "Battery testing and sorting" forum) My bad.

Discharging batteries into a resistive load to test them here in Florida doesn't make a lot of sense to me. We are into this thing to reduce our carbon footprint, right? I would rather make light, not heat.

The innards of relatively inexpensive ($4 or so) PIR or radar LED lightbulbs from China generally consist of a pc board wafer with sensing circuitry on one side and an array of LEDs on the other. They run from a constant current switching power supply that is usually a separate unit about the size of a postage stamp in the base of the bulb.

Poweringup only the LED wafer from a bench supply, I have found different units run at voltages ranging from 9 to almost 100 VDC. The optimum voltage range is quite narrow (LEDs are diodes, after all!) The 23 volt units are especially interesting to me because that is 6 18650s in series at 3.83 volts per cell. Adding a 51ohm resistor between the battery pile and the light prevents burnout of the LEDs at the full battery voltage, and significantly widens the voltage range at which the luminary produces a useful volume of light.

Here is my main premise: If 6 cells discharge in series through the light, by the time the light dims, the lowest capacity cell will also have the lowest voltage, and can be diverted to a lower-rank usage, like a flashlight. I haven't done enough measurements yet to know if this premise might be wrong. The wildcard is the cell internal resistance; I don't know yet what effect that has on the voltage/capacity relationship.

Fail: Here are images of a very simple radar-activated luminary I have made with a light wafer, three cells, a 51ohm resistor, and a battery holder. As you can see from the first image, the light module incorporates a CdS sensor which keeps it from coming on during the daytime. The assembly fits into a little Talenti gelato container and I can hang it from a hook anywhere. The reason it failed: by the time I noticed the light is dead, the cells were down under 2V! Partly due to the resistor, which widens the voltage range at which the LEDs give good light, and partly due to the radar unit sucking milliamps even after the LEDs no longer shone. I no longer use radar-activated sensors to make battery-powered lamps.

I will be making some 6 cell units soon. I found a 6S LED wafer that lights at 36V with no resistors, comfortable with 9 cells (max 38V) but quite dim at 27V (9 x 3V). But since the little LiPo testers max out at 8 cells in series, that is my new target.




Update 170116: Waiting for PIR bulbs to arrive from China, might as well modify some of the HC-SR501 PIR sensors I have on hand. PIR sensors are an older technology than radar units, but they give you much better control of the activation space since they only activate line-of-sight. They also draw about 100 microamps, as opposed to about 6mA for the doppler radar sensors. The latter sense through walls, windows and doors and will activate a bathroom light when you toss in bed in the adjoining room!


These are popular units, featuring adjustable time delay and sensitivity, widely available for less than $1. Sadly, they only have a +3.3VDC activation output current limited with 1.5Kohm in series, so they won't drive area lighting directly, and although they have lands for a CdS daytime sensor, that unit is not provisioned on the board.

Adding a logic-level drive NMOS transistor fixes the drive problem. I thought one in a TO-92 package was unobtainium, but found some 40V 450mA TN0104N3-G-ND at $0.90 each from Digi-Key. Surface mount packages are a bit more elegant, however. I started using the 2A MTM232270LBFCT from Digi-Key, but its 20V rating is too low for 6 cells in series (25.2 max). I got some 30V 500mA NTR4003NT1GOSCT ($0.12 each for 100). Simply remove the 1.5K resistor (you won't need it because the NMOS has a high intrinsic input impedance), and solder in the NMOS in its place as shown in this image. Connect DRAIN to OUT, GATE to the output of the BISS0001 chip, and run a wire over to GND (0001 pin 7):


Update 170324: I now have several 6S and 8S lighting fixtures/test units operating, and should have data soon. Here is an under-counter unit:


Countersunk magnets on the housing allow the fixture to be easily relocated
to any cabinet that has a metal plate on the bottom surface.




General schematic of LED light



Oct 9, 2016
Good idea as LEDs are a consistent load, at least if you use the proper constant current LED driver. Looks like the circuitry on the back of the PCB does that.

I've got an odd 12V load I can discharge power into, still repairing it at the moment.


New member
Feb 22, 2021
It's true, I also use lamps for growing grass and not only that. I also like to grow tropical plants, which is my hobby. LED lamps are the best alternative to natural light. The main thing here is to choose the best LED light for a 4 × 4 tent, because everything depends on it. If you make the wrong choice, then nothing will work and you will be disappointed. On the site a great overview of the best LED lamps for growing plants in a tent. Who else grows weed in a tent? I am interested in your experience. I will be very happy if someone shares it with me.
Last edited: