Luxeon area lighting

I am looking at using 1W stars to illuminate a corrider and under-counter task lighting. The problem I am having is finding a constant current source to drive them.

These are fixed devices so I can distribute whatever power I want (e.g., 12 or 24 VDC). I want a small and cheap constant current generator to drive 1 or more diodes in a series string. Everything I have found that does constant current assumes you are running from very low battery voltage and is a step-up regulator. They use inductors and are just more complicated that I want. Resistors use too much power.

Any suggestions for an IC or discrete approach?

If you want efficiency in driving a Luxeon and plan to power it from 12/24V then you need a switching regulator at each Luxeon or cluster of Luxeons. Any other scheme will be no better than your resistor. There are down converting switching regulators out there that will run from 12V or 24V. Also, if you want to implement dimming - you need to be able to control the current source too, either to lower the drive current or PWM the current.

LEDs just don't behave like incandescents - with incandescents you can put them in parallel with a voltage source. With LEDs you need to put them in series (or at a minimum have only 1 LED per driver) and drive them with a constant current.

You will also want to 'match' the colour of the LEDs you plan to use unless you can live with variability in your application.

george.

Have you looked into Xitanium Drivers ?

- Don

XItanium = $$$ and physically larger than I need.

Yes I know there are down-converting switching regulators. What I don't know is where to find constant current down-converting switching regulators. And preferably without inductors. Switching regulator chips tend to provide constant voltage.

Have you looked at the LM350?

3A capacity and a single resistor to set the current.

[ QUOTE ]
alternety said:
I am looking at using 1W stars to illuminate a corrider and under-counter task lighting. The problem I am having is finding a constant current source to drive them.

Any suggestions for an IC or discrete approach?

[/ QUOTE ]

Sure, the LM334 based Low Drop Out regulators 'Milkyspit' and I have been messing with. We've made up a few PCBs for fiddling with (.5 by 1.25 inches) that feature 3 switch selectable levels. You might want to contact him and get on his list for a couple to try out, they should be ready in a week or two. I've built up a few and tested them, they work fine for this use as long as the 'headroom' isn't too extreme. A 5 Volt distributed system would be 'just right' for singles (higher voltages to drive series strings of course).

Doug Owen

In the most general terms, I'd suggest that you use a power supply that gives you 5VDC, and use it to run a linear current regulation circuit. Chips such as the LM334 do a nice job of this. As far as the emitters themselves, you have many options, including parallel for all of them (make sure your power supply can give you enough total current) or series (increase the voltage of the power supply accordingly). You could, of course, use some sort of series-parallel combination, too.

Whatever you do, if you keep your power supply's voltage just a little above the Vf for your emitters (taking a series arrangement into account if it's present), the circuit will operate quite efficiently, even approaching 100% if designed carefully, and will generate little heat beyond whatever heat the emitters themselves create.

Hope this helps. /ubbthreads/images/graemlins/smile.gif

Well, as I took my time posting a reply, Doug Owen already said much the same thing. At least we're consistent! /ubbthreads/images/graemlins/grin.gif

You can certainly design a fine circuit for your needs yourself. Or if you perfer (and as Doug already said), please PM me about getting a prebuilt board or two once they're ready; then you'd only need to worry about the power supply itself. /ubbthreads/images/graemlins/buttrock.gif

[ QUOTE ]
milkyspit said:
Well, as I took my time posting a reply, Doug Owen already said much the same thing. At least we're consistent! /ubbthreads/images/graemlins/grin.gif


[/ QUOTE ]

Yup. You lie and I'll swear to it..... Works for me.

Doug Owen

Have you considered the Power Puck

It is designed to run a 5watt from a 12v supply but they supply a data sheet for running 4 1 watts from 12v too. It's on sale right now for $20 is looks like also.

I've not used this myself, but I have used their other product the micropuck and thats a great little device. I've read mostly positive reviews of the Power Puck. It's designed I think mostly for automtive applications.

Let me clarify a bit.

I am talking about a lot of lights (maybe 50 or so, maybe more). The current regulators need to be in the couple of dollar range (e.g., 1 IC and very few external components.

The reason I am not simply using resistors is to keep wasted power down. The same issue (power) is why it needs to be switching and not analog.

Power will be distributed with about 100' of wire (one way) so 5V distribution is a bit unhandy. If I have 50 lights this would be around 18A @ 5V with attendant voltage drops in the wire. Not impossible, but running to largish wire. Running at 5 V also implies a current regulator/LED. Too costly and real unhandy.

The solution needs to be switching vs analog, distribution voltage >= (4 X Vf + Vdrop) through current regulator (24V is a handy number), and cheap. There will be a bunch of clusters of LEDs containing 1 - 4 devices. Depending on light requirements at a particular spot.

I appreciate all your suggestions but I don't think any of them fix my problem. Perhaps it is not fixable with these parameters, although the power puck leads me to believe it can be done at some relatively small component cost.

Well, couple of dollars per regulator isn't going to happen. Not for 50 quantity. Not for a switcher. If you were in real production and talked 1000+, you can get it down to <$5 per driver. Of course that's the IC, the discretes, the PCB and assembly costs. I've built Luxeon drivers (switchers) and sell them - so I know the pricing structure for the components, especially in the 100-1000 piece pricing. It's not a couple of dollars...

george.

I agree totally with George.

Your goal and concept is right on spot. The low voltage lighting system I am installing is based on the same ideas you have. I will be distributing 12V throughout the house and I will be attaching efficient LED lighting for all locations.

For background and night vision lighting I will be using 3 3mm LEDs with a small resistor to set the LED current. This is very efficient since the resistor doesn't waste that much power in this configuration.

For Luxeon lighting I will be using my Downboy switching regulators. But, even my cost of the PCBs is way more than the target you have set.

The only way I can see you meeting your cost budget is to find the linear regulators at a surplus store. Generic Linears can be had for 0.30 or so each. Add a few resitors and capacitors and you should be able to keep the component costs down under a buck. But, you will be wasting a lot of power if you have a regulator for each LED.

My recommendation is to design around 3 Luxeons in series per node. In this way the Linear regulator or just a straight resistor drops only a small amount of voltage and power to give you constant current regulation to the LEDs.

I would also figure out the minimum voltage needed and what the voltage drop across the longest wire run will be and set the output of the power supply (wall wart) to be the minimum voltage needed to meet these requirements.

For example, If you can mix and match your LEDs so that they are all pretty close to an average of 3.6V/LED on each string your LED voltage requirements would be 3.6 * 3 or 10.8. You might be able to get away with a resistor at 12V regulated DC and the resistor value would be (12 - 10.8) / .35 or 3.4 ohms. The power in the resistor is then 1.2W which is a good compromise over cost.

Simply using a few choice resistor will allow you to tweek in each string to achieve the desired regulated current and if you use a switching wall wart with regulated output then the whole string will be regulated and constant.

In the cases where you only need one or two LEDs you could opt for a local switching regulator which might cost more than a few bucks. Using a couple of those along with resistors for the rest of the strings will average out to less than a buck or two per string.

Switching regulator ICs are usually anywhere from 0.80 to $3.00 depending on the manufacturer and complexity. Add an inductor for a buck, and two caps at a buck each and you can see that it is pretty hard in low volume to make a switching regulator for a couple of bucks. Of course you probably need some discrete components and a PCB to put the whole thing together.

Wayne

I think some false assumptions are being about the design criteria. The assumption being made is that the regulation required is of the same level as for battery operation which isn't the case if mains power is being used.
I would suggest a discrete approach using a suitable MOSFET such as a PHP3055E to drive the Luxeons. The input of the MOSFET would be controlled by a 7555 timer chip with its pulse width controlled by a preset resistor.
A little bit of playing around with the light dimmer circuit at
http://www.electronics-tutorials.com/devices/555-light-dimmer.htm
should provide what you need.
A 12 volt supply with 2 LEDs to a board gives .35 * 25 = 8.75 amps.
The actual printed circuit boards need only be about 2cm square. You could get 40 made for $21 plus shipping from
http://www.olimex.com/pcb/ so your total cost for 40 boards is maybe $120 - $150.
All you'd need extra would be your DC transformer and rectifier/smoothing + a current meter for setting up the boards.

[ QUOTE ]
Ralph_Hilton said:
I would suggest a discrete approach using a suitable MOSFET such as a PHP3055E to drive the Luxeons. The input of the MOSFET would be controlled by a 7555 timer chip with its pulse width controlled by a preset resistor.
A little bit of playing around with the light dimmer circuit at
http://www.electronics-tutorials.com/devices/555-light-dimmer.htm
should provide what you need.


[/ QUOTE ]

How does this limit the current to a pair of luxeons? The PWM only gives you the dimming capability. The time that the FET is turned on you will have the full 12V across a pair of Luxeons (minimal voltage drop across the FET) - with a power supply capable of providing at least 10Amps (since it has to run the rest of the 24 pairs of Luxeons). I think you'ed have 50 dead Luxeons within a short period of time.

I'll tell you one thing - you apply 12V across a Luxeon (not sure about a pair) for even a few milliseconds and you'll haver a DEAD one. This is from experience!

Regardless of your dimming circuit (preferably PWM for colour maintenance - for White LEDs) you still need to regulate the current to the Luxeons. The regulation is either via voltage drop (dissipation) or via voltage conversion (switcher).

george.

Yes, as I said, the circuit needs playing with. All that needs adding are the LCR components which are quite cheap.
Perhaps I should have been more explicit.

George - If one took your circuit on
http://www.speleogroup.org/ATtiny-A3-Schematic.gif
and replaced the AVR chip with a 7555 and bits as per
http://www.electronics-tutorials.com/devices/555-light-dimmer.htm
then what would your component cost be in quantities of 50?
With the Luxeon stars at $200/50 elsewhere on the forum I daresay a few people, myself included, are looking at rewiring their home lighting.
It looks as if you have more test equipment around than I do. If you were to put together a working circuit then I'd be happy to do the pcb layout and get quotes.

[ QUOTE ]
Ralph_Hilton said:
George - If one took your circuit on
http://www.speleogroup.org/ATtiny-A3-Schematic.gif
and replaced the AVR chip with a 7555 and bits as per...

It looks as if you have more test equipment around than I do. If you were to put together a working circuit then I'd be happy to do the pcb layout and get quotes.

[/ QUOTE ]

That's not my circuit, but I'm using a similar scheme for my AVR based design. Of course mine can run from up to 20V and so it requires a level shifter & driver to efficiently control the FET.

The AVR is <$2 in 100 piece qty, so it's hardly that expensive. If you used a 555 you would still need a control system feedback to maintain the PWM in the right range. By the time you engineered that, you might as well put a switcher in there instead /ubbthreads/images/graemlins/grin.gif. In my new project I'm using an AVR to provide dimming features - being programmable I can easily add as many features as I want (maybe too many...).

I will have blank boards back from fab by the end of next week - 500 of them, for my AVR project /ubbthreads/images/graemlins/grin.gif

Yeah, just a bit of test equipment - and ebay makes it so tempting to add more /ubbthreads/images/graemlins/jpshakehead.gif

george.

[ QUOTE ]
alternety said:
Let me clarify a bit.

I am talking about a lot of lights (maybe 50 or so, maybe more). The current regulators need to be in the couple of dollar range (e.g., 1 IC and very few external components.

The reason I am not simply using resistors is to keep wasted power down. The same issue (power) is why it needs to be switching and not analog.

Power will be distributed with about 100' of wire (one way) so 5V distribution is a bit unhandy. If I have 50 lights this would be around 18A @ 5V with attendant voltage drops in the wire. Not impossible, but running to largish wire. Running at 5 V also implies a current regulator/LED. Too costly and real unhandy.

The solution needs to be switching vs analog, distribution voltage >= (4 X Vf + Vdrop) through current regulator (24V is a handy number), and cheap. There will be a bunch of clusters of LEDs containing 1 - 4 devices. Depending on light requirements at a particular spot.

I appreciate all your suggestions but I don't think any of them fix my problem. Perhaps it is not fixable with these parameters, although the power puck leads me to believe it can be done at some relatively small component cost.

[/ QUOTE ]

Thanks for the additional details.

Too bad you put the 'can't be analog' restriction on this. Let's consider:

12 Volt supply, 3 divices a set. 3.6 Vf per is 10.8 Volts, 1.2 across the pass element and sense resistor. 90% efficiency. Just what sort of efficiency did you expect to get from this 'two dollar' switcher? No way you're going to come close to 90%, IMO. The LM334, 3 resistors (two really, Rs is probably a foot of WW wire or even thin traces), capacitor and PNP are probably under a buck, easily under $1.50 for 50 sets. A simlilar amount for a PCB (is this included in your switcher cost?) and we're home safe and dry under $3 a copy.

You could, of course, go to sets of six at 24 volts with the same sort of efficiency. And half the regulators.

Seems to fit all the requirements, *except* it's not a switcher.....

Doug Owen

What I'm suggesting is that the feedback system isn't needed for use with a mains power supply. Its somewhat of a pain tweaking 25 preset resistors but it seemed as if the original poster really wanted to cut down on the costs!
Are you making the AVR firmware public? I'd be interested in playing with some AVR processors but starting from scratch is a bit daunting and time consuming.