Simple LED circuit (I thought)

[denno]

Hope everyone had a good festive season.

I'm looking for some information about a lamp I'm trying to make. It's incredibly simple, I've got an "LED Electronic Converter" (basically a power supply to convert the 240V to 12V), 3 High Power LED's (specifications at the end of the post), a 1W 1 ohm resistor, and some wires.

I've soldered everything together, and switched it on, and boy was it bright, blinded me for a while. But then the light started to flicker, so that was alright, figured it was just a bad solder joint (which you won't be surprised by when you see the photo below). So I went through and made sure that all of the solder joints were getting good contact with the LED star plate, and then switched it on again. This time it turned on, but the LEDs were incredibly..... dull. They were maybe around 10% of full brightness. So dull that I could actually look directly at the lights, and be able to see the wires and inspect everything.

So my question is, am I missing something? These are powerful LEDs, is my power supply insufficient? I really don't think it is, it's supposed to handle 2.5A, and the LEDs will be drawing around 700mA total, perhaps there isn't enough power draw?

Anyway, if someone could please point out what could possibly be going on here, that would be great.
Thanks.

LED Specs (copied from the eBay page):
Chips: Epistar
DC Forward Voltage: DC 3.2-3.8V
DC Forward Current: Typ 700mA

Power Supply Specs in attached image

 

[JimB]

I think that you have burned out the LEDS.

The Vf of the LEDs is 3.2 to 3.8v
Take the average of 3.5v, then three LEDs in series will drop 10.5v.

The PSU has an output of 12v, 12 - 10.5 = 2.5v. So 2.5 v has to be dropped across the 1 ohm resistor (and the wiring, say 0.5 Ohm)
Using a bit of Ohms law,
2.5v / 1.5 Ohm = 1.66 Amp flowing through the LEDs, maybe a bit much for LEDs rated at 0.7 Amp.

The situation will be worse if the Vf of the LEDs is on the low side of the specification, and the wiring resistance is less than the 0.5 Ohm I have guestimated..

JimB

Afterthoughts,
The LEDs are wired in series, aren't they? I it is a bit difficult to decide.
The heatsinking arrangements don't look that good. I am guessing that this lot got very hot.

 

[alec_t]

I agree with JimB. The LEDs have fried .
LEDs are ideally powered by a constant-current source rather than a fixed voltage source.
How did you decide on the 1 Ohm value?

 

[denno]

Oh man, I can't believe I made such a rookie mistake :/..

I used this website: **broken link removed**. I have used it many times, however I put in 3.8V as the voltage for the LEDs.. Hence the problem.

If I use that same wizard with 3.5V as the LED forward voltage, it says that the power dissipated in the resistors is a concern, however another website: http://ledcalc.com/, says that a 2.2Ohm 1W resistor should be okay.

I have another LED array that I've set up, but haven't wired in the power and ground connections yet, so I can still swap out the resistor for the correct one. Would a 1W 2.2Ohm be okay? Or should I go for a 5W, just to be safe? (I recall seeing 5W resistors as an option from Jaycar, which is why I'm suggesting it as an alternative).

Thanks for the replies guys, knew I could get some solid information here .

 

[alec_t]

You haven't answered Jim's question. Are the LEDs in series?

 

[JimB]

The problem is that the circuit is not "well behaved".
Small changes in the LED voltage lead to large changes in the required resistance.
The voltage of the LEDs is not well defined.
The power supply, does it give an EXACT 12volts?

The configuration does not lend itself to a simple one size fits all calculation.

The best suggestion I can make is to build the circut again using a 3.9 Ohm resistor.
Power up and measure the PSU output volts and the total LED volts.
Then calculate the LED current. Iled = (Vpsu -Vleds)/3.9
Or just measure the voltage across the resistor and divide by the resistance to get the LED current.

If the LED current is a bit high or low, you now have enough information to make a better calculation of the required resistor value.
Using the measured values of Vpsu and Vleds:
R = (Vpsu - Vleds)/0.7

JimB

 

[denno]

Oh sorry, didn't read the afterthoughts section, I thought that was a signature.

They are wired in series yes. I didn't feel any heat when I was handling the leds after having turned them on, so I don't know if there are going to be heat problems..
If I ran the LEDs are 350mA, they would still be rather bright, but then I wouldn't have too many heat concerns would I?

 

[denno]

Thanks Jim! That's great information.
I measured the output of the PSU with the LEDs connected, and it is a constant 12V. This was one of the first things I checked when the lights started to flicker.

I like the idea of going a little bit higher with the resistor value. As I said, it's for a lamp, so I don't need the LEDs to be pushed to their maximums

 

[JimB]

If you can measure the Vled and run at a lower current, like 350mA, that would make things less critical.

JimB

 

[denno]

Okay fantastic. I'll measure it tomorrow morning. 350mA should be more than sufficient for the purpose I need it

 

[4pyros]

You should be using a LED driver like this. No resister needed.

 

[denno]

You should be using a LED driver like this. No resister needed.

You're very right, however I had already purchased this PSU, so I'm hoping I can just get it to work. Future projects however, I will definitely be going for a current driver, instead of a voltage source

 

[4pyros]

Total LED forward voltage is 3.5volts times 3 = 10.5 volts.
That gives you 12 volts - 10.5 volts = 1.5 volts across the resister.
If running at 350ma or .350 amps x 1.5volts the resister will need to dissipate .525 watts.
So the 1 watt should be fine.

 

[MikeMl]

Even if the OP had used a constant current source, he still would have blown-up the LEDs. These are designed to be bolted to an aluminum heat-sink to get rid of the heat that they produce while running. If you run them with no heat sink, you will melt the LED.

 

[4pyros]

Even if the OP had used a constant current source, he still would have blown-up the LEDs. These are designed to be bolted to an aluminum heat-sink to get rid of the heat that they produce while running. If you run them with no heat sink, you will melt the LED.

Looks like he has them on some sort of a makeshift heat sink.

 

[MikeMl]

Looks like he has them on some sort of a makeshift heat sink.

A 2x4? Here is what Wiki says about heat sinking high power leds:

[URL='https://en.wikipedia.org/wiki/Light-emitting_diode#High-power']High power light-emitting diodes (LEDs) can use 350 milliwatts or more in a single LED. Most of the electricity in an LED becomes heat rather than light (about 70% heat and 30% light). If this heat is not removed, the LEDs run at high temperatures, which not only lowers their efficiency, but also makes the LED less reliable.[/URL]

 

[denno]

So this morning I went down to Jaycar and picked up some 4.7Ohm resistors. I first swapped out the resistor in the array that I had already lit up, as I already have the power wires for that, so it was going to be the easiest. I turned the switch on and they lit up, but weren't very bright at all (same as yesterday). Is this a sign that they have burnt out? I would have thought that if they had been burnt out, then they wouldn't light at all..

Anyway, while they were lit up, I measured the voltage across each LED. 2 of them were getting 2.5v, and the other was getting 7V.. This clearly doesn't seem right, but could be another sign that the LEDs are no good now.

As for the makeshift heatsink, that was kind of an accident. I just put it together that way because it fits into the lampshade that I have. I hadn't really taken any consideration of heat sinking, apart from using thermal tape to stick the LEDs to the metal brackets..

 

[audioguru]

I have never purchased cheap no-name-brand Chinese LEDs from e-bay and I have never fried an LED.
I think your LEDs are destroyed. I hope you replace them with name-brand ones, use a reasonable amount of current and use a suitable REAL heatsink.

 

[MikeMl]

...
As for the makeshift heatsink, that was kind of an accident. I just put it together that way because it fits into the lampshade that I have. I hadn't really taken any consideration of heat sinking, apart from using thermal tape to stick the LEDs to the metal brackets..

The metal brackets are way too small to be a proper heatsink. You will need an aluminum heatsink with a surface area about 100 times more than what your photograph showed.

Next time, start with an old CPU cooler heatsink. Mount the LEDs on the smooth side. Run it for a while, and then measure the temperature rise. Based on the temperature rise and the max temperture limits on the LED data sheet, if your rise is smaller than allowed, you can make the heatsink surface area smaller. I.E. start big, and then work down to the smallest heat sink that will keep the LEDs at <~60deg C.

 

[Diver300]

I used this website: **broken link removed**. I have used it many times, however I put in 3.8V as the voltage for the LEDs.. Hence the problem.

That website is flawed in a few ways.

1) It won't give a resistor less than 1 Ohm
2) It doesn't take any account of the variations in LED and supply voltage.
3) If there are too many LEDs for one string, there string lengths are made as uneven as possible.

I suggest you try ranges of LED and power supply voltages to get a feel for how much the current will change as the voltages change.