Hi Power LED Current Regulation

[bryan]

Hello:

Would like to drive a hi power LED without using a driver and requesting some help.

The LED is 100 watt and the specs are 32-34v forward voltage and a current draw of 3000ma. I have sourced a voltage regulated powers supply that is variable from 30-36 v.

If I set the power supply to say 34 volts can I safely use a 1 ohm 10 watt resistor. That is the suggested resistor and wattage combination recommended by this calculator.

 

[Mr RB]

The LED voltage will change as it heats up, so you can use a regulated voltage PSU as long as you set the final adjustment when hot, otherwise current will rise as the LED heats up and its Vf drops.

I think it's better to use a constant current supply, you could use a 3A LM338 (the big brother to the LM317 adjustable regulator) and a resistor to set the current at 1.25v. However that will cost you some 2.5-3v overhead and lots of watts.

The best performance solution would be to set the regulated PSU's internal current limit at 3A, and just connect it to the LED and run it as a constant current DC supply. Depending which IC is in the PSU, it is probably as simple as one resistor inside the PSU that sets its current limit (assuming this is a standard offline SMPS regulated DC supply). A photo of the insides of the PSU would help determine if it can be modified.

 

[bryan]

Thanks RB;

Was thinking of the LM338, but would it really be any more efficient heat dissipation wise than a resistor. Unfortunately the power supply is only voltage adjustable and I would need to add my own circuitry for constant current. There is a couple what seems to be simple circuits here that may be better than a resistor.

https://www.instructables.com/id/Circuits-for-using-High-Power-LED-s/#intro

 

[fernando_g]

What mr RB is saying is that a current limting resistor will allow more or less current as the LEDs voltage drop shifts with temperature. OTOH, a constant curent source will deliver always the same current, even if an acciident happens and you short out the output.
I like the suggestion of the LM338, as it is fully protected against abuse. And believe me, I have enough gray hair to tell you that erRors and accidents DO happen.

 

[bryan]

Thanks, probably should just go with the LM338, if a resistor was that easy everyone would be doing it. Dbut To confirm ,if the led is driven at ~32v I would need the power supply at ~36 v because of the overhead.

 

[fernando_g]

Yes, a 4 to 4.5 overhead is required.
A good way to ensure that the source is regulating is to monitor the voltage drop across the current setting resistor. If it is 1.25 V, it is regulatiing properly.

 

[bryan]

Actually don't think I can use the Lm338. If i am reading the datasheet correct the max input voltage is 35 v?. Or is it the differential, between input and output can only be max 35 volts?

 

[fernando_g]

It is differential voltage, you are ok.

 

[Mr RB]

Bryan do you have a phot of the PCB of the supply? To set the current limit inside the supply normally just means changing one resistor, it's pretty easy.

If you can make the supply itself be a "constant current supply" that solves all your problems and will save a lot of heat, and saves a LM338 and heatsink etc.

 

[bryan]

Hi RB:

I have not ordered the supply yet, but looks like I am back to square one, on careful examination of the supply description it is not variable and set at 36v, which is pushing the limit of the LED. It might be possible to limit the voltage as well as current as you suggest if i took a chance and ordered one. Wonder if something like this may be a better option, dc-dc step up voltage and current adjustable. Could probably use a used modified 12v atx computer power supply to power the driver and LED.

**broken link removed**

 

[Mr RB]

Most of those "fixed" SMPS supplies in the little metal cages have an internal trimpot for tweaking the voltage. They also use standard chips if you wanted to get a datasheet and change a resistor value to change the voltage output or current limit.

I think using an old PC PSU and separate 100W DC-DC converter sounds clunky and much less reliable. Also some PC PSUs need a decent current drawn from the 5v rail so they don't shutdown, which means you might need a big power resistor on the 5v rail wasting heat just to get the 12v rail to work.

 

[bryan]

Thanks RB,think you are right. Think I will order one of the 36v supplies,and tweak it if I can downwards. Could go for a 24volt, but then tweaking it upwards might be a issue with caps that are underrated .Failing that suppose a easy way to shave off a volt or two is a few high power diodes in series with the output. Will pop the lid and take a photo when it arrives.

This link has some schematics for some common power supplies.

http://hmin.tripod.com/als/andysm/pages/MiscAT-2.html#AT252

and this appers to be a schematic for a Meanwell ps, SVR1 and SVR2 for voltage and current

http://www.hordsoffun.com/ebike/pics/Meanwell_S-350-27.png

 

[Mike odom]

Actually don't think I can use the Lm338. If i am reading the datasheet correct the max input voltage is 35 v?. Or is it the differential, between input and output can only be max 35 volts?

the data sheet I'm reading says 40V and as fernando says, it's differential.