0
\$\begingroup\$

I purchased some 50W LED lamps intended for outdoor use. Each module has 70 LED chips, so I suppose they are about 700mW each.

To my surprise they turn out to be pulsed and hence no use for video lighting (which we intended them for.) I am wondering if I can power them off a DC source. I opened one and am reverese engineering it. It turns out that each chip is just about lit at 7.2V (less than 1mA) and gets pretty bright at somewhere around 8.3V (at which point they take about 30mA.)

Each module is 2 or 3 LEDs in series. Is this a common part? There are no chip markings that I can see. I would like to get an idea of the maximum rated current - I guess 80 to 100mA but it would be nice to find a (likely) datasheet.

\$\endgroup\$
5
  • 1
    \$\begingroup\$ Yes you can replace the power supply. You'll have to first figure out how the LEDs are driven and then you can go to an electronics supplier and buy a compatible replacement. \$\endgroup\$ Commented May 6, 2023 at 16:17
  • \$\begingroup\$ thanks, yes I'm aware. QUetsion of figuring out the matrix layout and buying or building a suitabe DC source. It would be nice to try to identify the LED chip - I guess it is a COB part with 2 series LEDs - is this a common configuration? \$\endgroup\$
    – danmcb
    Commented May 6, 2023 at 16:34
  • 1
    \$\begingroup\$ Without a datasheet, you must guess how much current will burn out the tiny wires inside each LED or how much heat will fry them. \$\endgroup\$
    – Audioguru
    Commented May 6, 2023 at 16:48
  • 1
    \$\begingroup\$ You won't be able to identify the specific LED chips so don't bother. Measure the voltage and current from the existing driver and buy a better one \$\endgroup\$ Commented May 6, 2023 at 17:14
  • \$\begingroup\$ Send them back. Much of that mail order crud is unsafe, non-listed and unfit for sale in western countries. The seller should be made to pay for that mistake, rather than you. \$\endgroup\$ Commented May 6, 2023 at 19:44

2 Answers 2

3
\$\begingroup\$

Your problem is thermal.

The LEDs will only work if you don't overheat them. The "heat sinking" in these cheapie lights is only so good - after all, there's a way they got so cheap.

So you need to figure out the power actually going into those LEDs and insert no more than that - derating as necessary to not exceed the original overall power. The good news is that not pulsing will only improve efficiency.

The problem is, it's more difficult to surveil the power usage of a pulsed DC source, at least from the DC side. Perhaps you can wire a 0.1 ohm resistor in series with the power supply and measure voltage across the resistor, giving current/10. Also measure voltage and take samples at high enough frequency to see the whole pulse. On the AC side you can install a Kill-a-Watt energy monitor but that may not be 100% accurate - not least the efficiency of the power supply is difficult to know.

One thing you will discover is whether the LED itself is doing the shimmering. There's a type of LED called Chip On Board (COB) where silicon right on the LED emitter manages the LED brightness. If that is the source of the shimmer, you are out of luck.

Bare naked LEDs are current devices

Think about any random appliance. Toaster, USB charging block, water heater, whole house air conditioner outdoor unit, TV remote. All of them are given a constant voltage - 120VAC, 240VAC, or 3V - and they "magically self-regulate how much current they take*. The incandescent bulb does the same. **And people get concept-locked into thinking "all loads are like this". Not true at all.

It's funny because Edison tried thousands of filament designs, trying to find (by brute force) one that would self-regulate on constant-voltage. Constant-current wasn't an option because in 1890 that was only practical with AC power, and that would mean conceding to Tesla's AC system. Had Tesla been the master marketer instead of Edison, your light sockets would all be constant-current sources :)

Lighting in particular is not that way. All technologies of lighting (except incandescent) must be current-limited - take fluorescent. It takes a high voltage shot to initially strike the arc, and then, the lamp acts like a "dead short". It needs a ballast to limit lamp current to a survivable amount. Same with all the HID types.

LED is only a little more friendly. There is a voltage/current curve, but it's very steep. Drive it constant-voltage and you increase voltage a little, current increases a lot. Worse, that curve moves around based on temperature, age and binning. You can get away with it if you choose very conservative voltage, but then you won't get very much light. Indicators generally put a resistor in series with the LED to make it more linear, but in LED lighting you can't afford to add even more heat when heat management is already a challenge.

So LEDs for lighting are generally driven in a constant-current mode similar to a fluorescent or HID. That allows the driver to keep the LED right on spec in a wide variety of conditions and manufacturing variances.

This "constant-current" mode is actually a great way to dim them. Vary the current to set the dimming level. In fact, in this mode they are the finest dimmable light available, short of running an incandescent on DC.

I don't know why your LED is pulsing. Either they are using a cheap-mode power supply and picking up 60 Hz power supply ripple (120Hz if full-wave rectifying)... or they are intentionally using PWM as a method of dimming (which is fine). I suspect the former, as PWM is usually at too high a frequency to bother TV equipment, and 60/120 Hz is a close multiple of popular TV frame rates of 24, 29.97, 30 or 60 Hz/FPS.

That said, using "cheap Cheese from overseas" LEDs is a bad idea with TV production. The reason is the CRI (Color Rendering Index). It will tend to be either "all over the map" or "just bad", and typically poorly documented in any case. Hold out for 90+ CRI from reputable suppliers - it will make you look better.

\$\endgroup\$
3
  • \$\begingroup\$ I notice that you have two very similar paragraphs describing fluorescent lamps and why they need a ballast. Not sure if that's intentional. \$\endgroup\$ Commented May 7, 2023 at 1:59
  • \$\begingroup\$ thank you. Yes, it seems that way. I'm familiar with what diode VI curves look like, and how dependent on temperature they are. I figured out the array and say that if I could find a supply giving constant current of about 30mA in the 200-300V range, that would work. Unfortunately no such supply is commonly available. I am going to try splitting the chain into 3 and running one from a supply of about 80-90V with some series resistance. I don't think we need them at anything like the rated 50W. If that works, they will be usable, if not ... they're trash. \$\endgroup\$
    – danmcb
    Commented May 8, 2023 at 5:54
  • \$\begingroup\$ @cassie thanks, I fixed that. \$\endgroup\$ Commented May 8, 2023 at 7:58
0
\$\begingroup\$

[each chip is] pretty bright at somewhere around 8.3V (at which point they take about 30mA.)

Ohms law says that if a 12VDC power supply is used, 12V-8.3V = 3.7V across a dropping resistor. If that resistor has 3.7V and 30mA across/through it, then it must be a value of 3.7V/0.03A = ~120 Ohms. The power through it would be 3.7V*0.03A = 0.111W, so a half-watt variety would be good (always go double the power rating or better for resistors.) And if there were say 20 of these resistors in total, then the total power loss (as resistive heating) would be 2.22W. The total power required would be 12V*30mA*20pcs = 7.2W. Since power is V*A and we're considering 12V, then the current needed is 7.2W/12A = 0.6A minimum.

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.