Author Options:

10 Watt LED Circuit Answered

I am a software developer by trade and have little or no experience regarding LED and/or electronic designs. I am hoping someone here can help with a project I am working on using high brightness LEDs. I have studied several 'instructables' and specifically dan's "Circuits for using High Power LED's." My project consists of a few high brightness LEDs that will be placed in a high ceiling room with the brightness controlled by a microcontroller based on the light in the room (time of day, sun in window, etc). The microcontroller will use photocells to determine the brightness in the room and then adjust the LED using PWM pins. The LED is 10 watt and approximately 450 lumens. Attached is a circuit I drew as a starting point and would like help in determining if it will work, I am close or does it need to be trashed. I am not sure what the value for the resistor should be. Below are some calculations but not sure if I am on the right track or not. No need to be kind, I am more interested in getting it right and not losing any 'magic puffs of smoke' from any of the components. Here are the specs: LED IF: 1.6 A Peak Forward Current: 1.7 A Forward Voltage: 8 V LM350 (heavy duty version of LM317 IO_MAX: 4.5 A 1.2 - 25 V adjustable regulator BC337-40 Collection Current - Continuous: 800 mA dc Total Device Dissipation: 625 mW Resistor: 5W or 10W Voltage Amp Ohms Watt 8 1.60 5.0 12.8 8 0.80 10.0 10.0 8 1.10 7.5 8.5 Note: LED and components will have adequate heat sinks.


Slowly, but by no means surely, my work with homemade LED drivers continues. I am attempting to put everything together into a weatherproof enclosure, and make it look pretty and stuff. Picture below is pretty much what it looks like: a yellow box filled with white light.


It's nice to see this thread is still going strong. I have been looking for a pre-made module that would be able to power the 10 watt LED as Jack recommended. I've got six potentials and once I narrow it down to 2-3 I will post links to the datasheets and/or post some circuits. Thanks again guys, desNotes

Cool. I am curious to see what you have found. If it's a complete module that does the job at a reasonable price, then that would excite me. I mean it would "make obsolete" much of the "fun" and hours of time spent via the harder DIY approach, but hey! Saving time and work can be fun too!

Hey everybody! I've got some more steamy, explicit, pics of homemade LED driver circuits.

The circuit shown in the pictures below is a Buck-type switching circuit featuring a NDP6020 P-channel MOSFET, a MBR2045 fast diode, a 150 uH inductor made from 44 turns of 24 AWG insulated wire wrapped on a ferrite donut I pulled out of a dead computer power supply, and a 0.1 ohm current sensing resistor. Those are the guts of this circuit.

The brain (or maybe it's the heart?) of this circuit is provided by TI's SG3524 PWM driver, plus two op-amps, TLC272, plus various resistors and capacitors.

Basically what's going on here is not too different from the negative feedback loop in the linear style LED driver circuit I posted previously.

The current through the LED is sensed with the 0.1 ohm current sensing resistor, then this signal is amplified with a non-invering amplifier, then compared to a desired reference level, and then this error is fed to an integrator. The output from the integrator is what controls the PWM-source.

The SG3524, the PWM-source, is basically an oscillator with constant frequency, but whose duty cycle is controlled by a voltage input. In this case, that voltage input is connected to the output of an error integrator.

The magic that makes this all work is that an integrator will not tolerate a DC steady state error. The only time its output ever stops changing, is when its input is zero. In this case zero is the difference between the reference (supplied by a pot in the diagram) and the amplified signal from the current sensing resistor.

Because this is a "Buck" style regulator Vload has to be less than or equal to Vsupply. Because I have chosen Vsupply=12VDC, that means the biggest series-stack of white LEDs I could drive is about 3. (3*3.6V = 10.8 V)

However in the picture I'm only driving one LED, at approx. 780 mA, or about 0.78A*3.6V = 2.8W


Of course the easiest way to solve the LED driver problem would be to find a pre-made module, professionally designed, mass produced, and hopefully cheap. Such modules do exist. I've seen them for 3W LEDs, e.g.

It makes me wonder if there anyone out there selling the 10W version of this driver circuit?

BTW, desNotes, thanks for the link to sureelectronics. They've got some interesting stuff. Regarding sources, I got my 3W LEDs, and some 3W driver modules, from http://www.dealextreme.com/

Anyway, for or those of you adventurous enough to contemplate building your own Watt-sized LED driver, the other thing I was going to mention about the LED driver problem is the question of linear vs. switchmode. Linear regulators are simple, but waste power. Switchmode regulators are efficient in terms of power consumption, but they're more complicated to build.

The circuit diagram and pictures below show an example of a simple linear style constant-current regulator built from a big NPN transistor and some op-amps.

I'm still working on the switchmode style driver.


I've got a non-secret switch mode driver I did at work using Nat Semi simple swicher and Zetex parts if you're interested. IMHO Its obscenely efficient. Steve

Thanks for the reply, Steve. I am curious to see this circuit. Post a picture of this circuit diagram if you've got one handy. Or if it's less work, maybe just a summary would suffice. Some questions that come to mind: Which National switcher IC? External transistor? Inductor? Small current sensing resistor? What switching frequency? What topology: buck? boost? inverting? How much power ? Driving one LED, or a stack (i.e. in series) of them ?

A question about your schematic.....the second half of the op amp shown, it appears as though both inputs are shown as " - " is that right? I don't think I have ever seen that configuration before.

I think it might be the resolution at which Instructables is displaying this image. If the url has the string "MEDIUM.jpg"in it, try the similar url but with "LARGE.jpg" instead. Unless of course I've just uploaded a diagram nobody can read... Hey, that would suck. ;-) Uh, in that case, just wait till I get the thing written up all pretty in the form of a true instructable.

Hmmm, on closer examination, it does appear as though the output of the first half is fed in the + of the second half. If that is the case, you are right, it is very fuzzy, even when I look at it expanded, it still is easily mistaken for a - .

The easy question first: The 10W LED is available from www.sureelectronics.net. They are also available on Ebay. In addition to 10 Watt, there are now 20, 50 and 100 Watt. Needless to say the 100 Watt is good at burning through boxes and such. I took the circuit and modified it based on an old circuit here on Instructables. After putting it together (on paper) I looked at the Transistor and came to the conclusion that it could not handle the current. Since doing a little more research I found a couple of LED driver ICs that will drive an LED up to 2 Amps and provide a pin for dimming, which would be more than enough for the LED. Would it be better to go that way or is it possible to build it using a selection of components? Thanks to all for responding. desNotes

Careful! The BC337 cannot handle the current! Most any transistor will have an appreciable Vce at these currents which will make heat (power-handling) a problem! It may be possible to simply aim a photo-transistor at the lights and turn the (MOSFET?) driver off when they reach a pre-determined light level (by moving the sensor closer or further away, then they'll sense ambient lighting as well). I suggest you test them in place first. I find a pin-prick sized light sources EXTREMELY irritating! Not to mention limited lighting effect and serious shadows!

Let me guess, the voltage regulator with the resistor attached to its Vadj and Vout pins is intended as a constant-current source, and then the NPN transistor is for sort of switching this flow on and off?

It might work.

The main thing I don't like about it is the transistor and the regulator are kind of fighting against each other. That is to say the regulator is trying to keep the voltage at Vadj constant. When you switch the transistor OFF, the current will stop and Vadj = Vout, then the regulator will go full open trying to get current to flow to make Vadj lower. Then when you switch the transistor ON, the regulator will have go back to part-way open, back to being a regulator again.

I can probably come up with something prettier, and this is something I have been meaning to build anyway...

Where'd you find a 10W LED anyway?

I've got some 3W ones. They're what I've been meaning to build the driver for.

I am curiuos......to me this looks like the trigger to the transistor would turn it on full, and so the adjustment is either all the way on, or all the way off.