Yesterday, I did a quick test with my newly acquired Chinese ‘300W’ UV floodlight. It was immediately apparent that there is one thing very big about this floodlight: the exaggeration of its rated power. The whole thing made me curious, so I did two things: (1) opened up the unit and had a look at it, and (2) order 3 more of them. Yes, that’s right! Read on about what I found inside this unit and why I took up the ungodly plan to quadruple this mess.
This is the unit we’re talking about, and you can order it from AliExpress and probably eBay as well:
I immediately noticed that it does not consume anywhere near 300W and the closer actual figure going by my home energy monitor would be something around 100W. Of course, this is not a very exact way of measuring, and frankly, I can’t really be bothered to put an exact figure on it. Why not? Well, one stop difference would be a difference of a factor of 2, assuming equal efficiency etc. So whether this is really 85W or 125W, doesn’t make all that much difference to me. A little more distance between light source and printing frame and there goes the whole rated power difference out the window.
Anyway, I was just curious about the device’s contents. Fortunately, the contents aren’t particularly hard to get at; just unscrew the (many) screws of the lens array cover, which also acts as the weather seal. Cue the drum roll…here’s the device in all of its naked glory:
No surprises, at least at first glance. An array of LEDs, as anticipated, a rectifier and some net conditioning on the left-hand side, and what looks like an array of LED drivers on the right-hand side. Since this kind of aluminium PCB is single-sided, it’s childishly simple to retrace the schematic, which I actually did:
Center left is the net entry; it’s just phase and neutral coming in, each going through 2 resistors of 3R3 that together with the MOV/varistor between the poles would dissipate incoming voltage transients.
The working end consists of the LEDs, which turn out to be divided into two arrays which mostly consist of the individual LEDs put in series, but in 3 places on each string, two diodes are put in parallel. With a total of 150 LEDs, this makes for 75 LEDs per array, and each array has 72 LEDs in series and 3 additional ones in parallel. Given the forward voltage of a typical UV led, which will be around 4V give or take a bit, this works out as somewhere around 275V per led string to light up this Christmas tree. Since 230V rectified gives around 325V, this means there’s even room to spare.
Now, this puts us in an interesting situation, because up to a few years ago, you would encounter a buck topology led driver, which means an inductor and usually (always?) also a bulk capacitor would be present. This PCB houses neither. Apparently, the latest fad in driving LEDs for home/office lighting applications is to somehow (1) use on-chip inductors and capacitors (unlikely, as this would necessitate phenomenally high switching frequencies and therefore losses) or (2) basically just banging the LEDs on and off whenever the applied voltage is somewhat appropriate for the led string in question. Or maybe (3) banging the LEDs on and off rapidly whenever the voltage is high enough, in such a way that the RMS current through them does not exceed their design spec. Given this photo, my money is on option #3:
This picture suggests that the LEDs are on (or mostly on) during the positive half of the 230V input sine. I wouldn’t be surprised if they are actually PWM-ing at a few (tens of) kHz during this period. If you look at the current limiting/driving circuitry, we see 3 drivers for each LED array, apparently sharing the current between them. Each chip has a 9R10 resistor close to it which is nearly certainly a sense resistor through which the current passes.
I measured 325mV over that resistor, which is close enough to a beautifully round figure to assume that this is mentioned in the datasheet that I unfortunately could not find…the driver chip is marked LS1001X (or LS100 1X?), which I could not find anywhere.
Ohm’s Law gives a current of around 35mA through each chip, and the combined current through each string works out at around 107mA or so. My measurements with a very cheap and simple DMM are probably not very accurate, but this should work out as something like 35W RMS per array, for a total of 70W. This conforms fairly well with what I see happening on the home energy monitor if I switch the light source on. So my current, conservative estimate of its RMS power would be 70W. No, not 300W…oops!
Taking another angle on this, let’s have a look at the actual LEDs. They seem to be 2835 (or 3528?) SMD LEDs given the measurements and aspect ratio. Most of the power LEDs of this size are rated at max. 500mW, and I found none at all with a higher rating. At 150 LEDs, this would allow for 75W, which matches the measurements nearly perfectly.
All considered, you’d say all this is pretty disappointing. That’s at least what I would say, if you’re under the impression of buying a 300W unit. And indeed, while I anticipated a little optimism from Chinese sellers, the total lack of a relationship between the advertised power and actual performance is frankly shameful even to their standards. How come then I ordered 3 more of the same units?
Well, easy. I did a test print at 8×10″ yesterday, and as expected, it doesn’t cover at a useful distance (around 20cm in my test). I then did some searching and bean-counting, and figured out some ways to make a more powerful and satisfying LED unit. While not enormously costly, I figured it would take at least € 175 to build something in the 300W+ range and actually closer to € 250 to get something like 500W-600W RMS output. This would be a COB-led based approach, with borosilicate lenses and heatsinks + fans and ready-made LED drivers. There’s a couple of approaches to each of these components, depending on how much DIY-ing is acceptable.
While these little units here only put out let’s say 75W, they also cost only around € 30 each, and contrast and collimation seem to be good enough for my purposes. If I gang up 4 of them in a 2×2 grid, I get around 300 real Watts for around € 120, and very little to no fuss in building the thing. It’s just a matter of bolting them together and maybe fit a couple of fans above them to prevent them from setting the house on fire. So while these units pretty much suck in many was, they are also the most economical and straightforward solution at this point, it seems.
Let’s see if it really pans out that way; in the following weeks, I hope these units show up at my doorstep and then I’ll do some tests with all of them together. Maybe hook up a timer circuit to it for convenience; shouldn’t be too difficult either.